| Doc. no. | D???? |
| Date: | 2017-06-19 |
| Project: | Programming Language C++ |
| Reply to: | Marshall Clow <lwgchair@gmail.com> |
Revised 2017-06-19 at 02:20:00 UTC
Reference ISO/IEC IS 14882:2014(E)
Also see:
The purpose of this document is to record the status of issues which have come before the Library Working Group (LWG) of the INCITS PL22.16 and ISO WG21 C++ Standards Committee. Issues represent potential defects in the ISO/IEC IS 14882:2014(E) document.
This document contains only library issues which are actively being considered by the Library Working Group, i.e., issues which have a status of New, Open, Ready, or Review. See Library Defect Reports List for issues considered defects and Library Closed Issues List for issues considered closed.
The issues in these lists are not necessarily formal ISO Defect Reports (DR's). While some issues will eventually be elevated to official Defect Report status, other issues will be disposed of in other ways. See Issue Status.
Prior to Revision 14, library issues lists existed in two slightly different versions; a Committee Version and a Public Version. Beginning with Revision 14 the two versions were combined into a single version.
This document includes [bracketed italicized notes] as a reminder to the LWG of current progress on issues. Such notes are strictly unofficial and should be read with caution as they may be incomplete or incorrect. Be aware that LWG support for a particular resolution can quickly change if new viewpoints or killer examples are presented in subsequent discussions.
For the most current official version of this document see http://www.open-std.org/jtc1/sc22/wg21/. Requests for further information about this document should include the document number above, reference ISO/IEC 14882:2014(E), and be submitted to Information Technology Industry Council (ITI), 1250 Eye Street NW, Washington, DC 20005.
Public information as to how to obtain a copy of the C++ Standard, join the standards committee, submit an issue, or comment on an issue can be found in the comp.std.c++ FAQ.
Issues reported to the LWG transition through a variety of statuses, indicating their progress towards a resolution. Typically, most issues will flow through the following stages.
New - The issue has not yet been reviewed by the LWG. Any Proposed Resolution is purely a suggestion from the issue submitter, and should not be construed as the view of LWG.
Open - The LWG has discussed the issue but is not yet ready to move the issue forward. There are several possible reasons for open status:
A Proposed Resolution for an open issue is still not be construed as the view of LWG. Comments on the current state of discussions are often given at the end of open issues in an italic font. Such comments are for information only and should not be given undue importance.
Review - Exact wording of a Proposed Resolution is now available for review on an issue for which the LWG previously reached informal consensus.
Ready - The LWG has reached consensus that the issue is a defect in the Standard, the Proposed Resolution is correct, and the issue is ready to forward to the full committee for further action as a Defect Report (DR).
Typically, an issue must have a proposed resolution in the currently published issues list, whose wording does not change during LWG review, to move to the Ready status.
Voting - This status should not be seen in a published issues list, but is a marker for use during meetings to indicate an issues was Ready in the pre-meeting mailing, the Proposed Resolution is correct, and the issue will be offered to the working group at the end of the current meeting to apply to the current working paper (WP) or to close in some other appropriate manner. This easily distinguishes such issues from those moving to Ready status during the meeting itself, that should not be forwarded until the next meeting. If the issue does not move forward, it should fall back to one of the other open states before the next list is published.
Immediate - This status should not be seen in a published issues list, but is a marker for use during meetings to indicate an issues was not Ready in the pre-meeting mailing, but the Proposed Resolution is correct, and the issue will be offered to the working group at the end of the current meeting to apply to the current working paper (WP) or to close in some other appropriate manner. This status is used only rarely, typically for fixes that are both small and obvious, and usually within a meeting of the expected publication of a revised standard. If the issue does not move forward, it should fall back to one of the other open states before the next list is published.
In addition, there are a few ways to categorise and issue that remains open to a resolution within the library, but is not actively being worked on.
Deferred - The LWG has discussed the issue, is not yet ready to move the issue forward, but neither does it deem the issue significant enough to delay publishing a standard or Technical Report. A typical deferred issue would be seeking to clarify wording that might be technically correct, but easily mis-read.
A Proposed Resolution for a deferred issue is still not be construed as the view of LWG. Comments on the current state of discussions are often given at the end of open issues in an italic font. Such comments are for information only and should not be given undue importance.
Core - The LWG has discussed the issue, and feels that some key part of resolving the issue is better handled by a cleanup of the language in the Core part of the standard. The issue is passed to the Core Working Group, which should ideally open a corresponding issue that can be linked from the library issue. Such issues will be revisitted after Core have made (or declined to make) any changes.
EWG - The LWG has discussed the issue, and wonder that some key part of resolving the issue is better handled by some (hopefully small) extension to the language. The issue is passed to the Evolution Working Group, which should ideally open a corresponding issue that can be linked from the library issue. Such issues will be revisitted after Evoltion have made (or declined to make) any recommendations. Positive recommendations from EWG will often mean the issue transition to Core status while we wait for some proposed new feature to land in the working paper.
LEWG - The LWG has discussed the issue, and deemd the issue is either an extension, however small, or changes the library design in some fundamental way, and so has delegated the initial work to the Library Evolution Working Group.
Ultimately, all issues should reach closure with one of the following statuses.
DR - (Defect Report) - The full WG21/PL22.16 committee has voted to forward the issue to the Project Editor to be processed as a Potential Defect Report. The Project Editor reviews the issue, and then forwards it to the WG21 Convenor, who returns it to the full committee for final disposition. This issues list accords the status of DR to all these Defect Reports regardless of where they are in that process.
WP - (Working Paper) - The proposed resolution has not been accepted as a Technical Corrigendum, but the full WG21/PL22.16 committee has voted to apply the Defect Report's Proposed Resolution to the working paper.
C++14 - (C++ Standard, as revised for 2014) - The full WG21/PL22.16 committee has voted to accept the Defect Report's Proposed Resolution into the published 2014 revision to the C++ standard, ISO/IEC IS 14882:2014(E).
C++11 - (C++ Standard, as revised for 2011) - The full WG21/PL22.16 committee has voted to accept the Defect Report's Proposed Resolution into the published 2011 revision to the C++ standard, ISO/IEC IS 14882:2011(E).
CD1 - (Committee Draft 2008) - The full WG21/PL22.16 committee has voted to accept the Defect Report's Proposed Resolution into the Fall 2008 Committee Draft.
TC1 - (Technical Corrigenda 1) - The full WG21/PL22.16 committee has voted to accept the Defect Report's Proposed Resolution as a Technical Corrigenda. Action on this issue is thus complete and no further action is possible under ISO rules.
TRDec - (Decimal TR defect) - The LWG has voted to accept the Defect Report's Proposed Resolution into the Decimal TR. Action on this issue is thus complete and no further action is expected.
Resolved - The LWG has reached consensus that the issue is a defect in the Standard, but the resolution adopted to resolve the issue came via some other mechanism than this issue in the list - typically by applying a formal paper, occasionally as a side effect of consolidating several interacting issue resolutions into a single issue.
Dup - The LWG has reached consensus that the issue is a duplicate of another issue, and will not be further dealt with. A Rationale identifies the duplicated issue's issue number.
NAD - The LWG has reached consensus that the issue is not a defect in the Standard.
NAD Editorial - The LWG has reached consensus that the issue can either be handled editorially, or is handled by a paper (usually linked to in the rationale).
Tentatively - This is a status qualifier. The issue has been reviewed online, or at an unofficial meeting, but not in an official meeting, and some support has been formed for the qualified status. Tentatively qualified issues may be moved to the unqualified status and forwarded to full committee (if Ready) within the same meeting. Unlike Ready issues, Tentatively Ready issues will be reviewed in subcommittee prior to forwarding to full committee. When a status is qualified with Tentatively, the issue is still considered active.
Pending - This is a status qualifier. When prepended to a status this indicates the issue has been processed by the committee, and a decision has been made to move the issue to the associated unqualified status. However for logistical reasons the indicated outcome of the issue has not yet appeared in the latest working paper.
The following statuses have been retired, but may show up on older issues lists.
NAD Future - In addition to the regular status, the LWG believes that this issue should be revisited at the next revision of the standard. That is now an ongoing task managed by the Library Evolution Working Group, and most issues in this status were reopended with the status LEWG.
NAD Concepts - This status reflects an evolution of the language during the development of C++11, where a new feature entered the language, called concepts, that fundamentally changed the way templates would be specified and written. While this language feature was removed towards the end of the C++11 project, there is a clear intent to revisit this part of the language design. During that development, a number of issues were opened against the updated library related to use of that feature, or requesting fixes that would require explicit use of the concepts feature. All such issues have been closed with this status, and may be revisitted should this or a similar language feature return for a future standard.
NAD Arrays - This status reflects an evolution of the language during the development of C++14/17, where work on a Technical Specification, called the Arrays TS was begun. In early 2016, this work was abandoned, and the work item was officially withdrawn. During development of the TS, a number of issues were opened the features in the TS. All such issues have been closed with this status, and may be revisitted should this or a similar language feature return for a future standard.
Issues are always given the status of New when they first appear on the issues list. They may progress to Open or Review while the LWG is actively working on them. When the LWG has reached consensus on the disposition of an issue, the status will then change to Dup, NAD, or Ready as appropriate. Once the full PL22.16 committee votes to forward Ready issues to the Project Editor, they are given the status of Defect Report (DR). These in turn may become the basis for Technical Corrigenda (TC1), an updated standard (C++11, C++14), or are closed without action other than a Record of Response (Resolved) where the desired effect has already been achieved by some other process. The intent of this LWG process is that only issues which are truly defects in the Standard move to the formal ISO DR status.
Section: 27 [input.output] Status: LEWG Submitter: Martin Sebor Opened: 2003-09-18 Last modified: 2017-02-03
Priority: Not Prioritized
View all other issues in [input.output].
View all issues with LEWG status.
Discussion:
A third party test suite tries to exercise istream::ignore(N) with a negative value of N and expects that the implementation will treat N as if it were 0. Our implementation asserts that (N >= 0) holds and aborts the test.
I can't find anything in section 27 that prohibits such values but I don't see what the effects of such calls should be, either (this applies to a number of unformatted input functions as well as some member functions of the basic_streambuf template).
[ 2009-07 Frankfurt ]
This is related to LWG 255.
Move to NAD Future.
Proposed resolution:
I propose that we add to each function in clause 27 that takes an argument, say N, of type streamsize a Requires clause saying that "N >= 0." The intent is to allow negative streamsize values in calls to precision() and width() but disallow it in calls to streambuf::sgetn(), istream::ignore(), or ostream::write().
[Kona: The LWG agreed that this is probably what we want. However, we need a review to find all places where functions in clause 27 take arguments of type streamsize that shouldn't be allowed to go negative. Martin will do that review.]
Section: 24.2.3 [input.iterators] Status: Open Submitter: Chris Jefferson Opened: 2004-09-16 Last modified: 2017-06-15
Priority: Not Prioritized
View other active issues in [input.iterators].
View all other issues in [input.iterators].
View all issues with Open status.
Discussion:
From comp.std.c++:
I note that given an input iterator a for type T, then *a only has to be "convertable to T", not actually of type T.
Firstly, I can't seem to find an exact definition of "convertable to T". While I assume it is the obvious definition (an implicit conversion), I can't find an exact definition. Is there one?
Slightly more worryingly, there doesn't seem to be any restriction on the this type, other than it is "convertable to T". Consider two input iterators a and b. I would personally assume that most people would expect *a==*b would perform T(*a)==T(*b), however it doesn't seem that the standard requires that, and that whatever type *a is (call it U) could have == defined on it with totally different symantics and still be a valid inputer iterator.
Is this a correct reading? When using input iterators should I write T(*a) all over the place to be sure that the object I'm using is the class I expect?
This is especially a nuisance for operations that are defined to be "convertible to bool". (This is probably allowed so that implementations could return say an int and avoid an unnessary conversion. However all implementations I have seen simply return a bool anyway. Typical implemtations of STL algorithms just write things like while(a!=b && *a!=0). But strictly speaking, there are lots of types that are convertible to T but that also overload the appropriate operators so this doesn't behave as expected.
If we want to make code like this legal (which most people seem to expect), then we'll need to tighten up what we mean by "convertible to T".
[Lillehammer: The first part is NAD, since "convertible" is well-defined in core. The second part is basically about pathological overloads. It's a minor problem but a real one. So leave open for now, hope we solve it as part of iterator redesign.]
[ 2009-07-28 Reopened by Alisdair. No longer solved by concepts. ]
[ 2009-10 Santa Cruz: ]
Mark as NAD Future. We agree there's an issue, but there is no proposed solution at this time and this will be solved by concepts in the future.
[2017-02 in Kona, LEWG recommends NAD]
Has been clarified by 14. By design. Ranges might make it go away. Current wording for input iterators is more constrained.
[2017-06-02 Issues Telecon]
Move to Open. This is very similar to 2962, possibly a duplicate.
Marshall to research
Proposed resolution:
Rationale:
[ San Francisco: ]
Solved by N2758.
Section: 28 [re] Status: LEWG Submitter: Eric Niebler Opened: 2005-07-01 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [re].
View all other issues in [re].
View all issues with LEWG status.
Discussion:
A problem with TR1 regex is currently being discussed on the Boost developers list. It involves the handling of case-insensitive matching of character ranges such as [Z-a]. The proper behavior (according to the ECMAScript standard) is unimplementable given the current specification of the TR1 regex_traits<> class template. John Maddock, the author of the TR1 regex proposal, agrees there is a problem. The full discussion can be found at http://lists.boost.org/boost/2005/06/28850.php (first message copied below). We don't have any recommendations as yet.
-- Begin original message --
The situation of interest is described in the ECMAScript specification (ECMA-262), section 15.10.2.15:
"Even if the pattern ignores case, the case of the two ends of a range is significant in determining which characters belong to the range. Thus, for example, the pattern /[E-F]/i matches only the letters E, F, e, and f, while the pattern /[E-f]/i matches all upper and lower-case ASCII letters as well as the symbols [, \, ], ^, _, and `."
A more interesting case is what should happen when doing a case-insentitive match on a range such as [Z-a]. It should match z, Z, a, A and the symbols [, \, ], ^, _, and `. This is not what happens with Boost.Regex (it throws an exception from the regex constructor).
The tough pill to swallow is that, given the specification in TR1, I don't think there is any effective way to handle this situation. According to the spec, case-insensitivity is handled with regex_traits<>::translate_nocase(CharT) -- two characters are equivalent if they compare equal after both are sent through the translate_nocase function. But I don't see any way of using this translation function to make character ranges case-insensitive. Consider the difficulty of detecting whether "z" is in the range [Z-a]. Applying the transformation to "z" has no effect (it is essentially std::tolower). And we're not allowed to apply the transformation to the ends of the range, because as ECMA-262 says, "the case of the two ends of a range is significant."
So AFAICT, TR1 regex is just broken, as is Boost.Regex. One possible fix is to redefine translate_nocase to return a string_type containing all the characters that should compare equal to the specified character. But this function is hard to implement for Unicode, and it doesn't play nice with the existing ctype facet. What a mess!
-- End original message --
[ John Maddock adds: ]
One small correction, I have since found that ICU's regex package does implement this correctly, using a similar mechanism to the current TR1.Regex.
Given an expression [c1-c2] that is compiled as case insensitive it:
Enumerates every character in the range c1 to c2 and converts it to it's case folded equivalent. That case folded character is then used a key to a table of equivalence classes, and each member of the class is added to the list of possible matches supported by the character-class. This second step isn't possible with our current traits class design, but isn't necessary if the input text is also converted to a case-folded equivalent on the fly.
ICU applies similar brute force mechanisms to character classes such as [[:lower:]] and [[:word:]], however these are at least cached, so the impact is less noticeable in this case.
Quick and dirty performance comparisons show that expressions such as "[X-\\x{fff0}]+" are indeed very slow to compile with ICU (about 200 times slower than a "normal" expression). For an application that uses a lot of regexes this could have a noticeable performance impact. ICU also has an advantage in that it knows the range of valid characters codes: code points outside that range are assumed not to require enumeration, as they can not be part of any equivalence class. I presume that if we want the TR1.Regex to work with arbitrarily large character sets enumeration really does become impractical.
Finally note that Unicode has:
Three cases (upper, lower and title). One to many, and many to one case transformations. Character that have context sensitive case translations - for example an uppercase sigma has two different lowercase forms - the form chosen depends on context(is it end of a word or not), a caseless match for an upper case sigma should match either of the lower case forms, which is why case folding is often approximated by tolower(toupper(c)).
Probably we need some way to enumerate character equivalence classes, including digraphs (either as a result or an input), and some way to tell whether the next character pair is a valid digraph in the current locale.
Hoping this doesn't make this even more complex that it was already,
[ Portland: Alisdair: Detect as invalid, throw an exception. Pete: Possible general problem with case insensitive ranges. ]
[ 2009-07 Frankfurt ]
We agree that this is a problem, but we do not know the answer.
We are going to declare this NAD until existing practice leads us in some direction.
No objection to NAD Future.
Move to NAD Future.
Proposed resolution:
Section: 20.5.3.8 [tuple.rel], 99 [tr.tuple.rel] Status: LEWG Submitter: David Abrahams Opened: 2005-11-29 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [tuple.rel].
View all other issues in [tuple.rel].
View all issues with LEWG status.
Duplicate of: 348
Discussion:
Where possible, tuple comparison operators <,<=,=>, and > ought to be defined in terms of std::less rather than operator<, in order to support comparison of tuples of pointers.
[ 2009-07-28 Reopened by Alisdair. No longer solved by concepts. ]
[ 2009-10 Santa Cruz: ]
If we solve this for tuple we would have to solve it for pair algorithms, etc. It is too late to do that at this time. Move to NAD Future.
Proposed resolution:
change 6.1.3.5/5 from:
Returns: The result of a lexicographical comparison between t and u. The result is defined as: (bool)(get<0>(t) < get<0>(u)) || (!(bool)(get<0>(u) < get<0>(t)) && ttail < utail), where rtail for some tuple r is a tuple containing all but the first element of r. For any two zero-length tuples e and f, e < f returns false.
to:
Returns: The result of a lexicographical comparison between t and u. For any two zero-length tuples e and f, e < f returns false. Otherwise, the result is defined as: cmp( get<0>(t), get<0>(u)) || (!cmp(get<0>(u), get<0>(t)) && ttail < utail), where rtail for some tuple r is a tuple containing all but the first element of r, and cmp(x,y) is an unspecified function template defined as follows.
Where T is the type of x and U is the type of y:
if T and U are pointer types and T is convertible to U, returns less<U>()(x,y)
otherwise, if T and U are pointer types, returns less<T>()(x,y)
otherwise, returns (bool)(x < y)
[ Berlin: This issue is much bigger than just tuple (pair, containers, algorithms). Dietmar will survey and work up proposed wording. ]
Rationale:
Recommend NAD. This will be fixed with the next revision of concepts.
[ San Francisco: ]
Solved by N2770.
Section: 22 [localization] Status: LEWG Submitter: Peter Dimov Opened: 2007-07-28 Last modified: 2017-02-03
Priority: Not Prioritized
View all other issues in [localization].
View all issues with LEWG status.
Discussion:
The POSIX "Extended API Set Part 4,"
introduces extensions to the C locale mechanism that allow multiple concurrent locales to be used in the same application by introducing a type locale_t that is very similar to std::locale, and a number of _l functions that make use of it.
The global locale (set by setlocale) is now specified to be per- process. If a thread does not call uselocale, the global locale is in effect for that thread. It can install a per-thread locale by using uselocale.
There is also a nice querylocale mechanism by which one can obtain the name (such as "de_DE") for a specific facet, even for combined locales, with no std::locale equivalent.
std::locale should be harmonized with the new POSIX locale_t mechanism and provide equivalents for uselocale and querylocale.
[ Kona (2007): Bill and Nick to provide wording. ]
[ San Francisco: Bill and Nick still intend to provide wording, but this is a part of the task to be addressed by the group that will look into issue 860. ]
[ 2009-07 Frankfurt: ]
It's our intention to stay in sync with WG14. If WG14 makes a decision that requires a change in WG21 the issue will be reopened.
Move to NAD Future.
Proposed resolution:
Section: 23.2 [container.requirements] Status: Tentatively NAD Submitter: Paolo Carlini Opened: 2007-11-11 Last modified: 2017-02-03
Priority: 2
View all other issues in [container.requirements].
View all issues with Tentatively NAD status.
Discussion:
In an emplace member function the function parameter pack may be bound to a priori unlimited number of objects: some or all of them can be elements of the container itself. Apparently, in order to conform to the blanket statement 23.2 [container.requirements]/11, the implementation must check all of them for that possibility. A possible solution can involve extending the exception in 23.2 [container.requirements]/12 also to the emplace member. As a side note, the push_back and push_front member functions are luckily not affected by this problem, can be efficiently implemented anyway.
[ Related to 767 and to 2164 ]
[ Bellevue: ]
The proposed addition (13) is partially redundant with the existing paragraph 12. Why was the qualifier "rvalues" added to paragraph 12? Why does it not cover subelements and pointers?
Resolution: Alan Talbot to rework language, then set state to Review.
[ 2009-07 Frankfurt ]
The problem is broader than emplace. The LWG doesn't feel that it knows how to write wording that prohibits all of the problematic use cases at this time.
NAD Future.
[2015-02 Cologne]
LWG believes that 2164 addresses this issue and therefore considers 760 as NAD.
Proposed resolution:
Add after 23.2 [container.requirements]/12:
-12- Objects passed to member functions of a container as rvalue references shall not be elements of that container. No diagnostic required.
-13- Objects bound to the function parameter pack of the emplace member function shall not be elements or sub-objects of elements of the container. No diagnostic required.
Section: 20.17.7 [time.clock] Status: LEWG Submitter: Beman Dawes Opened: 2008-11-24 Last modified: 2017-02-03
Priority: Not Prioritized
View all issues with LEWG status.
Discussion:
Each of the three clocks specified in Clocks 20.17.7 [time.clock] provides the member function:
static time_point now();
The semantics specified by Clock requirements 20.17.3 [time.clock.req] make no mention of error handling. Thus the function may throw bad_alloc or an implementation-defined exception (17.5.5.12 [res.on.exception.handling] paragraph 4).
Some implementations of these functions on POSIX, Windows, and presumably on other operating systems, may fail in ways only detectable at runtime. Some failures on Windows are due to supporting chipset errata and can even occur after successful calls to a clock's now() function.
These functions are used in cases where exceptions are not appropriate or where the specifics of the exception or cause of error need to be available to the user. See N2828, Library Support for hybrid error handling (Rev 1), for more specific discussion of use cases. Thus some change in the interface of now is required.
The proposed resolution has been implemented in the Boost version of the chrono library. No problems were encountered.
[ Batavia (2009-05): ]
We recommend this issue be deferred until the next Committee Draft has been issued and the prerequisite paper has been accepted.
Move to Open.
[ 2009-10 Santa Cruz: ]
Mark as NAD future. Too late to make this change without having already accepted the hybrid error handling proposal.
Proposed resolution:
Accept the proposed wording of N2828, Library Support for hybrid error handling (Rev 1).
Change Clock requirements 20.17.3 [time.clock.req] as indicated:
-2- In Table 55 C1 and C2 denote clock types. t1 and t2 are values returned by C1::now() where the call returning t1 happens before (1.10) the call returning t2 and both of these calls happen before C1::time_point::max(). ec denotes an object of type error_code (19.5.3.1 [syserr.errcode.overview]).
Table 55 — Clock requirements Expression Return type Operational semantics ... ... ... C1::now() C1::time_point Returns a time_point object representing the current point in time. C1::now(ec) C1::time_point Returns a time_point object representing the current point in time.
Change class system_clock 20.17.7.1 [time.clock.system] as indicated:
static time_point now(error_code& ec=throws());
Change class monotonic_clock 99 [time.clock.monotonic] as indicated:
static time_point now(error_code& ec=throws());
Change class high_resolution_clock 20.17.7.3 [time.clock.hires] as indicated:
static time_point now(error_code& ec=throws());
Section: 30.4.3 [thread.mutex.requirements] Status: LEWG Submitter: Pete Becker Opened: 2008-12-05 Last modified: 2017-03-05
Priority: Not Prioritized
View other active issues in [thread.mutex.requirements].
View all other issues in [thread.mutex.requirements].
View all issues with LEWG status.
Duplicate of: 961
Discussion:
30.4.3 [thread.mutex.requirements] describes the requirements for a type to be a "Mutex type". A Mutex type can be used as the template argument for the Lock type that's passed to condition_variable_any::wait (although Lock seems like the wrong name here, since Lock is given a different formal meaning in 30.4.4 [thread.lock]) and, although the WD doesn't quite say so, as the template argument for lock_guard and unique_lock.
The requirements for a Mutex type include:
Also, a Mutex type "shall not be copyable nor movable".
The latter requirement seems completely irrelevant, and the three requirements on return types are tighter than they need to be. For example, there's no reason that lock_guard can't be instantiated with a type that's copyable. The rule is, in fact, that lock_guard, etc. won't try to copy objects of that type. That's a constraint on locks, not on mutexes. Similarly, the requirements for void return types are unnecessary; the rule is, in fact, that lock_guard, etc. won't use any returned value. And with the return type of bool, the requirement should be that the return type is convertible to bool.
[ Summit: ]
Move to open. Related to conceptualization and should probably be tackled as part of that.
- The intention is not only to place a constraint on what types such as lock_guard may do with mutex types, but on what any code, including user code, may do with mutex types. Thus the constraints as they are apply to the mutex types themselves, not the current users of mutex types in the standard.
- This is a low priority issue; the wording as it is may be overly restrictive but this may not be a real issue.
[ Post Summit Anthony adds: ]
Section 30.4.3 [thread.mutex.requirements] conflates the requirements on a generic Mutex type (including user-supplied mutexes) with the requirements placed on the standard-supplied mutex types in an attempt to group everything together and save space.
When applying concepts to chapter 30, I suggest that the concepts Lockable and TimedLockable embody the requirements for *use* of a mutex type as required by unique_lock/lock_guard/condition_variable_any. These should be relaxed as Pete describes in the issue. The existing words in 30.4.3 [thread.mutex.requirements] are requirements on all of std::mutex, std::timed_mutex, std::recursive_mutex and std::recursive_timed_mutex, and should be rephrased as such.
[2017-03-01, Kona]
SG1: Agreement that we need a paper.
Proposed resolution:
Section: 30.4.3 [thread.mutex.requirements] Status: LEWG Submitter: Pete Becker Opened: 2009-01-07 Last modified: 2017-03-05
Priority: Not Prioritized
View other active issues in [thread.mutex.requirements].
View all other issues in [thread.mutex.requirements].
View all issues with LEWG status.
Duplicate of: 936
Discussion:
30.4.3 [thread.mutex.requirements] describes required member functions of mutex types, and requires that they throw exceptions under certain circumstances. This is overspecified. User-defined types can abort on such errors without affecting the operation of templates supplied by standard-library.
[ Summit: ]
Move to open. Related to conceptualization and should probably be tackled as part of that.
[ 2009-10 Santa Cruz: ]
Would be OK to leave it as is for time constraints, could loosen later.
Mark as NAD Future.
[2017-03-01, Kona]
SG1: Agreement that we need a paper.
Proposed resolution:
Section: 20.14.15 [unord.hash] Status: LEWG Submitter: Alisdair Meredith Opened: 2009-03-11 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [unord.hash].
View all other issues in [unord.hash].
View all issues with LEWG status.
Discussion:
Addresses UK 208 [CD1]
std::hash should be implemented for much more of the standard library. In particular for pair, tuple and all the standard containers.
Proposed resolution:
Section: 24.5.1.3.5 [reverse.iter.opref] Status: LEWG Submitter: Alisdair Meredith Opened: 2009-03-12 Last modified: 2017-02-03
Priority: Not Prioritized
View all other issues in [reverse.iter.opref].
View all issues with LEWG status.
Duplicate of: 2775
Discussion:
Addresses UK 281 [CD1]
The current specification for return value for reverse_iterator::operator-> will always be a true pointer type, but reverse_iterator supports proxy iterators where the pointer type may be some kind of 'smart pointer'.
[ Summit: ]
move_iterator avoids this problem by returning a value of the wrapped Iterator type. study group formed to come up with a suggested resolution.
move_iterator solution shown in proposed wording.
[ 2009-07 post-Frankfurt: ]
Howard to deconceptize. Move to Review after that happens.
[ 2009-08-01 Howard deconceptized: ]
[ 2009-10 Santa Cruz: ]
We can't think of any reason we can't just define reverse iterator's pointer types to be the same as the underlying iterator's pointer type, and get it by calling the right arrow directly.
Here is the proposed wording that was replaced:
template <class Iterator> class reverse_iterator { ... typedeftypename iterator_traits<Iterator>::pointerpointer;Change 24.5.1.3.5 [reverse.iter.opref]:
pointer operator->() const;Returns:
&(operator*());this->tmp = current; --this->tmp; return this->tmp;
[ 2010-03-03 Daniel opens: ]
- There is a minor problem with the exposition-only declaration of the private member deref_tmp which is modified in a const member function (and the same problem occurs in the specification of operator*). The fix is to make it a mutable member.
The more severe problem is that the resolution for some reasons does not explain in the rationale why it was decided to differ from the suggested fix (using deref_tmp instead of tmp) in the [ 2009-10 Santa Cruz] comment:
this->deref_tmp = current; --this->deref_tmp; return this->deref_tmp;combined with the change of
typedef typename iterator_traits<Iterator>::pointer pointer;to
typedef Iterator pointer;The problem of the agreed on wording is that the following rather typical example, that compiled with the wording before 1052 had been applied, won't compile anymore:
#include <iterator> #include <utility> int main() { typedef std::pair<int, double> P; P op; std::reverse_iterator<P*> ri(&op + 1); ri->first; // Error }Comeau online returns (if a correspondingly changed reverse_iterator is used):
"error: expression must have class type return deref_tmp.operator->(); ^ detected during instantiation of "Iterator reverse_iterator<Iterator>::operator->() const [with Iterator=std::pair<int, double> *]""Thus the change will break valid, existing code based on std::reverse_iterator.
IMO the suggestion proposed in the comment is a necessary fix, which harmonizes with the similar specification of std::move_iterator and properly reflects the recursive nature of the evaluation of operator-> overloads.
Suggested resolution:
In the class template reverse_iterator synopsis of 24.5.1.1 [reverse.iterator] change as indicated:
namespace std { template <class Iterator> class reverse_iterator : public iterator<typename iterator_traits<Iterator>::iterator_category, typename iterator_traits<Iterator>::value_type, typename iterator_traits<Iterator>::difference_type,typename iterator_traits<Iterator>::pointer, typename iterator_traits<Iterator>::reference> { public: [..] typedeftypename iterator_traits<Iterator>::pointerpointer; [..] protected: Iterator current; private: mutable Iterator deref_tmp; // exposition only };- Change 24.5.1.3.5 [reverse.iter.opref]/1 as indicated:
pointer operator->() const;1
ReturnsEffects:&(operator*()).deref_tmp = current; --deref_tmp; return deref_tmp;
[ 2010 Pittsburgh: ]
We prefer to make to use a local variable instead of deref_tmp within operator->(). And although this means that the mutable change is no longer needed, we prefer to keep it because it is needed for operator*() anyway.
Here is the proposed wording that was replaced:
Change 24.5.1.3.5 [reverse.iter.opref]:
pointer operator->() const;Returns:
&(operator*());deref_tmp = current; --deref_tmp; return deref_tmp::operator->();
[ 2010-03-10 Howard adds: ]
Here are three tests that the current proposed wording passes, and no other solution I've seen passes all three:
Proxy pointer support:
#include <iterator> #include <cassert> struct X { int m; }; X x; struct IterX { typedef std::bidirectional_iterator_tag iterator_category; typedef X& reference; struct pointer { pointer(X& v) : value(v) {} X& value; X* operator->() const {return &value;} }; typedef std::ptrdiff_t difference_type; typedef X value_type; // additional iterator requirements not important for this issue reference operator*() const { return x; } pointer operator->() const { return pointer(x); } IterX& operator--() {return *this;} }; int main() { std::reverse_iterator<IterX> ix; assert(&ix->m == &(*ix).m); }Raw pointer support:
#include <iterator> #include <utility> int main() { typedef std::pair<int, double> P; P op; std::reverse_iterator<P*> ri(&op + 1); ri->first; // Error }Caching iterator support:
#include <iterator> #include <cassert> struct X { int m; }; struct IterX { typedef std::bidirectional_iterator_tag iterator_category; typedef X& reference; typedef X* pointer; typedef std::ptrdiff_t difference_type; typedef X value_type; // additional iterator requirements not important for this issue reference operator*() const { return value; } pointer operator->() const { return &value; } IterX& operator--() {return *this;} private: mutable X value; }; int main() { std::reverse_iterator<IterX> ix; assert(&ix->m == &(*ix).m); }
[ 2010 Pittsburgh: ]
Moved to NAD Future, rationale added.
Rationale:
The LWG did not reach a consensus for a change to the WP.
Proposed resolution:
In the class template reverse_iterator synopsis of 24.5.1.1 [reverse.iterator] change as indicated:
namespace std {
template <class Iterator>
class reverse_iterator : public
iterator<typename iterator_traits<Iterator>::iterator_category,
typename iterator_traits<Iterator>::value_type,
typename iterator_traits<Iterator>::difference_type,
typename iterator_traits<Iterator&>::pointer,
typename iterator_traits<Iterator>::reference> {
public:
[..]
typedef typename iterator_traits<Iterator&>::pointer pointer;
[..]
protected:
Iterator current;
private:
mutable Iterator deref_tmp; // exposition only
};
pointer operator->() const;1
ReturnsEffects:&(operator*()).deref_tmp = current; --deref_tmp; return deref_tmp;
[Alternate Proposed Resolution from 2775, which was closed as a dup of this issue]
This wording is relative to N4606.
Modify 24.5.1.3.5 [reverse.iter.opref] as indicated:
constexpr pointer operator->() const;-1-
Returns: addressof(operator*()).Effects: If Iterator is a pointer type, as if by:Iterator tmp = current; return --tmp;Otherwise, as if by:
Iterator tmp = current; --tmp; return tmp.operator->();
Section: 20.16.4 [ratio.arithmetic] Status: LEWG Submitter: Alisdair Meredith Opened: 2009-05-25 Last modified: 2017-02-03
Priority: Not Prioritized
View all other issues in [ratio.arithmetic].
View all issues with LEWG status.
Discussion:
Both add and multiply could sensibly be called with more than two arguments. The variadic template facility makes such declarations simple, and is likely to be frequently wrapped by end users if we do not supply the variant ourselves.
We deliberately ignore divide at this point as it is not transitive. Likewise, subtract places special meaning on the first argument so I do not suggest extending that immediately. Both could be supported with analogous wording to that for add/multiply below.
Note that the proposed resolution is potentially incompatible with that proposed for 921, although the addition of the typedef to ratio would be equally useful.
[ 2009-10-30 Alisdair adds: ]
The consensus of the group when we reviewed this in Santa Cruz was that 921 would proceed to Ready as planned, and the multi-paramater add/multiply templates should be renamed as ratio_sum and ratio_product to avoid the problem mixing template aliases with partial specializations.
It was also suggested to close this issue as NAD Future as it does not correspond directly to any NB comment. NBs are free to submit a specific comment (and re-open) in CD2 though.
Walter Brown also had concerns on better directing the order of evaluation to avoid overflows if we do proceed for 0x rather than TR1, so wording may not be complete yet.
[ Alisdair updates wording. ]
[ 2009-10-30 Howard: ]
Moved to Tentatively NAD Future after 5 positive votes on c++std-lib.
Rationale:
Does not have sufficient support at this time. May wish to reconsider for a future standard.
Proposed resolution:
Add the following type traits to p3 20.16 [ratio]
// ratio arithmetic template <class R1, class R2> struct ratio_add; template <class R1, class R2> struct ratio_subtract; template <class R1, class R2> struct ratio_multiply; template <class R1, class R2> struct ratio_divide; template <class R1, class ... RList> struct ratio_sum; template <class R1, class ... RList> struct ratio_product;
after 20.16.4 [ratio.arithmetic] p1: add
template <class R1, class ... RList> struct ratio_sum; // declared, never defined template <class R1> struct ratio_sum<R1> : R1 {};Requires: R1 is a specialization of class template ratio
template <class R1, class R2, class ... RList> struct ratio_sum<R1, R2, RList...> : ratio_add< R1, ratio_sum<R2, RList...>> { };Requires: R1 and each element in parmater pack RList is a specialization of class template ratio
after 20.16.4 [ratio.arithmetic] p3: add
template <class R1, class ... RList> struct ratio_product; // declared, never defined template <class R1> struct ratio_product<R1> : R1 {};Requires: R1 is a specialization of class template ratio
template <class R1, class R2, class ... RList> struct ratio_sum<R1, R2, RList...> : ratio_add< R1, ratio_product<R2, RList...>> { };Requires: R1 and each element in parmater pack RList is a specialization of class template ratio
Section: 27.9.5 [fstream] Status: LEWG Submitter: LWG Opened: 2009-06-28 Last modified: 2017-02-03
Priority: Not Prioritized
View all issues with LEWG status.
Discussion:
Addresses JP 73
Description
It is a problem from C++98, fstream cannot appoint a filename of wide character string(const wchar_t and const wstring&).
Suggestion
Add interface corresponding to wchar_t, char16_t and char32_t.
[ 2009-07-01 Alisdair notes that this is a duplicate of 454 which has more in-depth rationale. ]
[ 2009-09-21 Daniel adds: ]
I suggest to mark this issue as NAD Future with the intend to solve the issue with a single file path c'tor template assuming a provision of a TR2 filesystem library.
[ 2009 Santa Cruz: ]
NAD Future. This is a duplicate of 454.
Proposed resolution:
Section: 26.5 [complex.numbers] Status: LEWG Submitter: LWG Opened: 2009-06-28 Last modified: 2017-03-05
Priority: Not Prioritized
View other active issues in [complex.numbers].
View all other issues in [complex.numbers].
View all issues with LEWG status.
Discussion:
Addresses FR 35
Description
Instantiations of the class template complex<> have to be allowed for integral types, to reflect existing practice and ISO standards (LIA-III).
Suggestion
[ 2009-10-26 Proposed wording in N3002. ]
[ 2010 Pittsburgh: ]
Moved to NAD Future. Rationale added.
[2017-03-03, Kona]
SG6 suggests this issue is a new feature, not a problem with the existing standard, and should therefore be closed NAD. However, SG6 invites papers that bring the proposal up to date with the current standard.
Rationale:
There is no consensus for making this change at this time.
Proposed resolution:
Adopt N3002.
Section: 23.2.7 [unord.req] Status: Open Submitter: Pablo Halpern Opened: 2009-07-17 Last modified: 2017-02-03
Priority: 3
View other active issues in [unord.req].
View all other issues in [unord.req].
View all issues with Open status.
Discussion:
When I look at the unordered_* constructors, I think the complexity is poorly described and does not follow the style of the rest of the standard.
The complexity for the default constructor is specified as constant. Actually, it is proportional to n, but there are no invocations of value_type constructors or other value_type operations.
For the iterator-based constructor the complexity should be:
Complexity: exactly n calls to construct value_type from InputIterator::value_type (where n = distance(f,l)). The number of calls to key_equal::operator() is proportional to n in the average case and n*n in the worst case.
[ 2010 Rapperswil: ]
Concern that the current wording may require O(1) where that cannot be delivered. We need to look at both the clause 23 requirements tables and the constructor description of each unordered container to be sure.
Howard suggests NAD Editorial as we updated the container requirement tables since this issue was written.
Daniel offers to look deeper, and hopefully produce wording addressing any outstanding concerns at the next meeting.
Move to Open.
[2011-02-26: Daniel provides wording]
I strongly suggest to clean-up the differences between requirement tables and individual specifications. In the usual way, the most specific specifications wins, which is in this case the wrong one. In regard to the concern expressed about missing DefaultConstructible requirements of the value type I disagree: The function argument n is no size-control parameter, but only some effective capacity parameter: No elements will be value-initialized by these constructors. The necessary requirement for the value type, EmplaceConstructible into *this, is already listed in Table 103 — Unordered associative container requirements. Another part of the proposed resolution is the fact that there is an inconsistency of the complexity counting when both a range and a bucket count is involved compared to constructions where only bucket counts are provided: E.g. the construction X a(n); has a complexity of n bucket allocations, but this part of the work is omitted for X a(i, j, n);, even though it is considerable larger (in the average case) for n ≫ distance(i, j).
[2011-03-24 Madrid meeting]
Move to deferred
[ 2011 Bloomington ]
The proposed wording looks good. Move to Review.
[2012, Kona]
Fix up some presentation issues with the wording, combining the big-O expressions into single expressions rather than the sum of two separate big-Os.
Strike "constant or linear", prefer "linear in the number of buckets". This allows for number of buckets being larger than requested n as well.
Default n to "unspecified" rather than "implementation-defined". It seems an un-necessary burden asking vendors to document a quantity that is easily determined through the public API of these classes.
Replace distance(f,l) with "number of elements in the range [f,l)"
Retain in Review with the updated wording
[2012, Portland: Move to Open]
The wording still does not call out Pablo's original concern, that the element constructor is called no more than N times, and that the N squared term applies to moves during rehash.
Inconsistent use of O(n)+O(N) vs. O(n+N), with a preference for the former.
AJM to update wording with a reference to "no more than N element constructor calls".
Matt concerned that calling out the O(n) requirements is noise, and dangerous noise in suggesting a precision we do not mean. The cost of constructing a bucket is very different to constructing an element of user-supplied type.
AJM notes that if there are multiple rehashes, the 'n' complexity is probably not linear.
Matt suggests back to Open, Pablo suggests potentially NAD if we keep revisitting without achieving a resolution.
Matt suggests complexity we are concerned with is the number of operations, such as constructing elements, moving nodes, and comparing/hashing keys. We are less concerned with constructing buckets, which are generally noise in this bigger picture.
[2015-01-29 Telecon]
AM: essentially correct, but do we want to complicate the spec?
HH: Pablo has given us permission to NAD it JM: when I look at the first change in the P/R I find it mildly disturbing that the existing wording says you have a constant time constructor with a single element even if your n is 10^6, so I think adding this change makes people aware there might be a large cost in initializing the hash table, even though it doesn't show up in user-visible constructions. HH: one way to avoid that problem is make the default ctor noexcept. Then the container isn't allowed to create an arbitrarily large hash table AM: but this is the constructor where the user provides n MC: happy with the changes, except I agree with the editorial recommendation to keep the two 𝒪s separate. JW: yes, the constant 'k' is different in 𝒪(n) and 𝒪(N) GR: do we want to talk about buckets at all JM: yes, good to highlight that bucket construction might be a significant cost HH: suggest we take the suggestion to split 𝒪(n+N) to 𝒪(n)+𝒪(N) and move to Tentatively Ready GR: 23.2.1p2 says all complexity requirements are stated solely in terms of the number of operations on the contained object, so we shouldn't be stating complexity in terms of the hash table initialization HH: channeling Pete, there's an implicit "unless otherwise specified" everywhere. VV: seem to be requesting modifications that render this not Tentatively Ready GR: I think it can't be T/R AM: make the editorial recommendation, consider fixing 23.2.1/3 to give us permission to state complexity in terms of bucket initialization HH: only set it to Review after we get new wording to review[2015-02 Cologne]
Update wording, revisit later.
Proposed resolution:
Modify the following rows in Table 103 — Unordered associative container requirements to add the explicit bucket allocation overhead of some constructions. As editorial recommendation it is suggested not to shorten the sum 𝒪(n) + 𝒪(N) to 𝒪(n + N), because two different work units are involved.
Table 103 — Unordered associative container requirements (in addition to container) Expression Return type Assertion/note pre-/post-condition Complexity … X(i, j, n, hf, eq)
X a(i, j, n, hf, eq)X …
Effects: Constructs an empty container with at least n
buckets, using hf as the hash function and eq as the key
equality predicate, and inserts elements from [i, j) into it.Average case 𝒪(n + N) (N is distance(i, j)),
worst case 𝒪(n) + 𝒪(N2)X(i, j, n, hf)
X a(i, j, n, hf)X …
Effects: Constructs an empty container with at least n
buckets, using hf as the hash function and key_equal() as the key
equality predicate, and inserts elements from [i, j) into it.Average case 𝒪(n + N) (N is distance(i, j)),
worst case 𝒪(n + N2)X(i, j, n)
X a(i, j, n)X …
Effects: Constructs an empty container with at least n
buckets, using hasher() as the hash function and key_equal() as the key
equality predicate, and inserts elements from [i, j) into it.Average case 𝒪(n + N) (N is distance(i, j)),
worst case 𝒪(n + N2)…
Modify 23.5.4.2 [unord.map.cnstr] p. 1-4 as indicated (The edits of p. 1 and p. 3 attempt to fix some editorial oversight.):
explicit unordered_map(size_type n = see below, const hasher& hf = hasher(), const key_equal& eql = key_equal(), const allocator_type& a = allocator_type());1 Effects: Constructs an empty unordered_map using the specified hash function, key equality function, and allocator, and using at least n buckets. If n is not provided, the number of buckets is unspecified
impldefdefault number of buckets in unordered_map. max_load_factor() returns 1.0.2 Complexity:
ConstantLinear in the number of buckets.
template <class InputIterator> unordered_map(InputIterator f, InputIterator l, size_type n = see below, const hasher& hf = hasher(), const key_equal& eql = key_equal(), const allocator_type& a = allocator_type());3 Effects: Constructs an empty unordered_map using the specified hash function, key equality function, and allocator, and using at least n buckets. If n is not provided, the number of buckets is unspecified
impldefdefault number of buckets in unordered_map. Then inserts elements from the range [f, l). max_load_factor() returns 1.0.4 Complexity:
Average case linear, worst case quadraticLinear in the number of buckets. In the average case linear in N and in the worst case quadratic in N to insert the elements, where N is equal to number of elements in the range [f,l).
Modify 23.5.5.2 [unord.multimap.cnstr] p. 1-4 as indicated (The edits of p. 1 and p. 3 attempt to fix some editorial oversight.):
explicit unordered_multimap(size_type n = see below, const hasher& hf = hasher(), const key_equal& eql = key_equal(), const allocator_type& a = allocator_type());1 Effects: Constructs an empty unordered_multimap using the specified hash function, key equality function, and allocator, and using at least n buckets. If n is not provided, the number of buckets is unspecified
impldefdefault number of buckets in unordered_multimap. max_load_factor() returns 1.0.2 Complexity:
ConstantLinear in the number of buckets.
template <class InputIterator> unordered_multimap(InputIterator f, InputIterator l, size_type n = see below, const hasher& hf = hasher(), const key_equal& eql = key_equal(), const allocator_type& a = allocator_type());3 Effects: Constructs an empty unordered_multimap using the specified hash function, key equality function, and allocator, and using at least n buckets. If n is not provided, the number of buckets is unspecified
impldefdefault number of buckets in unordered_multimap. Then inserts elements from the range [f, l). max_load_factor() returns 1.0.4 Complexity:
Average case linear, worst case quadraticLinear in the number of buckets. In the average case linear in N and in the worst case quadratic in N to insert the elements, where N is equal to number of elements in the range [f,l).
Modify 23.5.6.2 [unord.set.cnstr] p. 1-4 as indicated (The edits of p. 1 and p. 3 attempt to fix some editorial oversight.):
explicit unordered_set(size_type n = see below, const hasher& hf = hasher(), const key_equal& eql = key_equal(), const allocator_type& a = allocator_type());1 Effects: Constructs an empty unordered_set using the specified hash function, key equality function, and allocator, and using at least n buckets. If n is not provided, the number of buckets is unspecified
impldefdefault number of buckets in unordered_set. max_load_factor() returns 1.0.2 Complexity:
ConstantLinear in the number of buckets.
template <class InputIterator> unordered_set(InputIterator f, InputIterator l, size_type n = see below, const hasher& hf = hasher(), const key_equal& eql = key_equal(), const allocator_type& a = allocator_type());3 Effects: Constructs an empty unordered_set using the specified hash function, key equality function, and allocator, and using at least n buckets. If n is not provided, the number of buckets is unspecified
impldefdefault number of buckets in unordered_set. Then inserts elements from the range [f, l). max_load_factor() returns 1.0.4 Complexity:
Average case linear, worst case quadraticLinear in the number of buckets. In the average case linear in N and in the worst case quadratic in N to insert the elements, where N is equal to number of elements in the range [f,l).
Modify 23.5.7.2 [unord.multiset.cnstr] p. 1-4 as indicated (The edits of p. 1 and p. 3 attempt to fix some editorial oversight.):
explicit unordered_multiset(size_type n = see below, const hasher& hf = hasher(), const key_equal& eql = key_equal(), const allocator_type& a = allocator_type());1 Effects: Constructs an empty unordered_multiset using the specified hash function, key equality function, and allocator, and using at least n buckets. If n is not provided, the number of buckets is unspecified
impldefdefault number of buckets in unordered_multiset. max_load_factor() returns 1.0.2 Complexity:
ConstantLinear in the number of buckets.
template <class InputIterator> unordered_multiset(InputIterator f, InputIterator l, size_type n = see below, const hasher& hf = hasher(), const key_equal& eql = key_equal(), const allocator_type& a = allocator_type());3 Effects: Constructs an empty unordered_multiset using the specified hash function, key equality function, and allocator, and using at least n buckets. If n is not provided, the number of buckets is unspecified
impldefdefault number of buckets in unordered_multiset. Then inserts elements from the range [f, l). max_load_factor() returns 1.0.4 Complexity:
Average case linear, worst case quadraticLinear in the number of buckets. In the average case linear in N and in the worst case quadratic in N to insert the elements, where N is equal to number of elements in the range [f,l).
Section: 23.2.7 [unord.req], 23.5 [unord] Status: LEWG Submitter: Matt Austern Opened: 2009-08-10 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [unord.req].
View all other issues in [unord.req].
View all issues with LEWG status.
Discussion:
Unordered associative containers have a notion of a maximum load factor: when the number of elements grows large enough, the containers automatically perform a rehash so that the number of elements per bucket stays below a user-specified bound. This ensures that the hash table's performance characteristics don't change dramatically as the size increases.
For similar reasons, Google has found it useful to specify a minimum load factor: when the number of elements shrinks by a large enough, the containers automatically perform a rehash so that the number of elements per bucket stays above a user-specified bound. This is useful for two reasons. First, it prevents wasting a lot of memory when an unordered associative container grows temporarily. Second, it prevents amortized iteration time from being arbitrarily large; consider the case of a hash table with a billion buckets and only one element. (This was discussed even before TR1 was published; it was TR issue 6.13, which the LWG closed as NAD on the grounds that it was a known design feature. However, the LWG did not consider the approach of a minimum load factor.)
The only interesting question is when shrinking is allowed. In principle the cleanest solution would be shrinking on erase, just as we grow on insert. However, that would be a usability problem; it would break a number of common idioms involving erase. Instead, Google's hash tables only shrink on insert and rehash.
The proposed resolution allows, but does not require, shrinking in rehash, mostly because a postcondition for rehash that involves the minimum load factor would be fairly complicated. (It would probably have to involve a number of special cases and it would probably have to mention yet another parameter, a minimum bucket count.)
The current behavior is equivalent to a minimum load factor of 0. If we specify that 0 is the default, this change will have no impact on backward compatibility.
[ 2010 Rapperswil: ]
This seems to a useful extension, but is too late for 0x. Move to Tentatively NAD Future.
[ Moved to NAD Future at 2010-11 Batavia ]
Proposed resolution:
Add two new rows, and change rehash's postcondition in the unordered associative container requirements table in 23.2.7 [unord.req]:
Table 87 — Unordered associative container requirements (in addition to container) Expression Return type Assertion/note pre-/post-condition Complexity a.min_load_factor() float Returns a non-negative number that the container attempts to keep the load factor greater than or equal to. The container automatically decreases the number of buckets as necessary to keep the load factor above this number. constant a.min_load_factor(z) void Pre: z shall be non-negative. Changes the container's minimum load factor, using z as a hint. [Footnote: the minimum load factor should be significantly smaller than the maximum. If z is too large, the implementation may reduce it to a more sensible value.] constant a.rehash(n) void Post: a.bucket_count() >= n, and a.size() <= a.bucket_count() * a.max_load_factor(). [Footnote: It is intentional that the postcondition does not mention the minimum load factor. This member function is primarily intended for cases where the user knows that the container's size will increase soon, in which case the container's load factor will temporarily fall below a.min_load_factor().] a.bucket_cout > a.size() / a.max_load_factor() and a.bucket_count() >= n.Average case linear in a.size(), worst case quadratic.
Add a footnote to 23.2.7 [unord.req] p12:
The insert members shall not affect the validity of references to container elements, but may invalidate all iterators to the container. The erase members shall invalidate only iterators and references to the erased elements.
[A consequence of these requirements is that while insert may change the number of buckets, erase may not. The number of buckets may be reduced on calls to insert or rehash.]
Change paragraph 13:
The insert members shall not affect the validity of iterators if
(N+n) < z * Bzmin * B <= (N+n) <= zmax * B, where N is the number of elements in the container prior to the insert operation, n is the number of elements inserted, B is the container's bucket count, zmin is the container's minimum load factor, and zmax is the container's maximum load factor.
Add to the unordered_map class synopsis in section 23.5.4 [unord.map], the unordered_multimap class synopsis in 23.5.5 [unord.multimap], the unordered_set class synopsis in 23.5.6 [unord.set], and the unordered_multiset class synopsis in 23.5.7 [unord.multiset]:
float min_load_factor() const; void min_load_factor(float z);
In 23.5.4.2 [unord.map.cnstr], 23.5.5.2 [unord.multimap.cnstr], 23.5.6.2 [unord.set.cnstr], and 23.5.7.2 [unord.multiset.cnstr], change:
... max_load_factor() returns 1.0 and min_load_factor() returns 0.
Section: 20.5.3.4 [tuple.creation], 20.4 [pairs] Status: LEWG Submitter: Alisdair Meredith Opened: 2009-09-05 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [tuple.creation].
View all other issues in [tuple.creation].
View all issues with LEWG status.
Discussion:
Spotting a recent thread on the boost lists regarding collapsing optional representations in optional<optional<T>> instances, I wonder if we have some of the same issues with make_tuple, and now make_pair?
Essentially, if my generic code in my own library is handed a reference_wrapper by a user, and my library in turn delegates some logic to make_pair or make_tuple, then I am going to end up with a pair/tuple holding a real reference rather than the intended reference wrapper.
There are two things as a library author I can do at this point:
(There may be some metaprogramming approaches my library can use to wrap the make_tuple call, but all will be significantly more complex than simply implementing a simplified make_tuple.)
Now I don't propose we lose this library facility, I think unwrapping references will be the common behaviour. However, we might want to consider adding another overload that does nothing special with ref-wrappers. Note that we already have a second overload of make_tuple in the library, called tie.
[ 2009-09-30 Daniel adds: ]
I suggest to change the currently proposed paragraph for make_simple_pair
template<typename... Types> pair<typename decay<Types>::type...> make_simple_pair(Types&&... t);
Type requirements: sizeof...(Types) == 2.Remarks: The program shall be ill-formed, if sizeof...(Types) != 2....
or alternatively (but with a slightly different semantic):
Remarks: If sizeof...(Types) != 2, this function shall not participate in overload resolution.
to follow a currently introduced style and because the library does not have yet a specific "Type requirements" element. If such thing would be considered as useful this should be done as a separate issue. Given the increasing complexity of either of these wordings it might be preferable to use the normal two-argument-declaration style again in either of the following ways:
Let V1 be typename decay<T1>::type and V2 be typename decay<T2>::type.
[ 2009-10 post-Santa Cruz: ]
Mark as Tentatively NAD Future.
Rationale:
Does not have sufficient support at this time. May wish to reconsider for a future standard.
Proposed resolution:
Add the following function to 20.4 [pairs] and signature in appropriate synopses:
template<typename... Types> pair<typename decay<Types>::type...> make_simple_pair(Types&&... t);Type requirements: sizeof...(Types) == 2.
Returns: pair<typename decay<Types>::type...>(std::forward<Types>(t)...).
[ Draughting note: I chose a variadic representation similar to make_tuple rather than naming both types as it is easier to read through the clutter of metaprogramming this way. Given there are exactly two elements, the committee may prefer to draught with two explicit template type parameters instead ]
Add the following function to 20.5.3.4 [tuple.creation] and signature in appropriate synopses:
template<typename... Types> tuple<typename decay<Types>::type...> make_simple_tuple(Types&&... t);Returns: tuple<typename decay<Types>::type...>(std::forward<Types>(t)...).
Section: 27.7.5.5 [ostream.rvalue], 27.7.4.5 [istream.rvalue] Status: LEWG Submitter: Howard Hinnant Opened: 2009-09-06 Last modified: 2017-02-03
Priority: Not Prioritized
View all other issues in [ostream.rvalue].
View all issues with LEWG status.
Discussion:
27.7.5.5 [ostream.rvalue] was created to preserve the ability to insert into (and extract from 27.7.4.5 [istream.rvalue]) rvalue streams:
template <class charT, class traits, class T> basic_ostream<charT, traits>& operator<<(basic_ostream<charT, traits>&& os, const T& x);1 Effects: os << x
2 Returns: os
This is good as it allows code that wants to (for example) open, write to, and close an ofstream all in one statement:
std::ofstream("log file") << "Some message\n";
However, I think we can easily make this "rvalue stream helper" even easier to use. Consider trying to quickly create a formatted string. With the current spec you have to write:
std::string s = static_cast<std::ostringstream&>(std::ostringstream() << "i = " << i).str();
This will store "i = 10" (for example) in the string s. Note the need to cast the stream back to ostringstream& prior to using the member .str(). This is necessary because the inserter has cast the ostringstream down to a more generic ostream during the insertion process.
I believe we can re-specify the rvalue-inserter so that this cast is unnecessary. Thus our customer now has to only type:
std::string s = (std::ostringstream() << "i = " << i).str();
This is accomplished by having the rvalue stream inserter return an rvalue of the same type, instead of casting it down to the base class. This is done by making the stream generic, and constraining it to be an rvalue of a type derived from ios_base.
The same argument and solution also applies to the inserter. This code has been implemented and tested.
[ 2009 Santa Cruz: ]
NAD Future. No concensus for change.
Proposed resolution:
Change 27.7.4.5 [istream.rvalue]:
template <classcharT, class traitsIstream, class T>basic_istream<charT, traits>&Istream&& operator>>(basic_istream<charT, traits>Istream&& is, T& x);1 Effects: is >> x
2 Returns: std::move(is)
3 Remarks: This signature shall participate in overload resolution if and only if Istream is not an lvalue reference type and is derived from ios_base.
Change 27.7.5.5 [ostream.rvalue]:
template <classcharT, class traitsOstream, class T>basic_ostream<charT, traits>&Ostream&& operator<<(basic_ostream<charT, traits>Ostream&& os, const T& x);1 Effects: os << x
2 Returns: std::move(os)
3 Remarks: This signature shall participate in overload resolution if and only if Ostream is not an lvalue reference type and is derived from ios_base.
Section: 24.2 [iterator.requirements] Status: Open Submitter: Daniel Krügler Opened: 2009-09-19 Last modified: 2017-02-03
Priority: 4
View all other issues in [iterator.requirements].
View all issues with Open status.
Discussion:
The terms valid iterator and singular aren't properly defined. The fuzziness of those terms became even worse after the resolution of 208 (including further updates by 278). In 24.2 [iterator.requirements] as of N2723 the standard says now:
5 - These values are called past-the-end values. Values of an iterator i for which the expression *i is defined are called dereferenceable. The library never assumes that past-the-end values are dereferenceable. Iterators can also have singular values that are not associated with any container. [...] Results of most expressions are undefined for singular values; the only exceptions are destroying an iterator that holds a singular value and the assignment of a non-singular value to an iterator that holds a singular value. [...] Dereferenceable values are always non-singular.
10 - An invalid iterator is an iterator that may be singular.
First, issue 208 intentionally removed the earlier constraint that past-the-end values are always non-singular. The reason for this was to support null pointers as past-the-end iterators of e.g. empty sequences. But there seem to exist different views on what a singular (iterator) value is. E.g. according to the SGI definition a null pointer is not a singular value:
Dereferenceable iterators are always nonsingular, but the converse is not true. For example, a null pointer is nonsingular (there are well defined operations involving null pointers) even thought it is not dereferenceable.
and proceeds:
An iterator is valid if it is dereferenceable or past-the-end.
Even if the standard prefers a different meaning of singular here, the change was incomplete, because by restricting feasible expressions of singular iterators to destruction and assignment isn't sufficient for a past-the-end iterator: Of-course it must still be equality-comparable and in general be a readable value.
Second, the standard doesn't clearly say whether a past-the-end value is a valid iterator or not. E.g. 20.10.10 [specialized.algorithms]/1 says:
In all of the following algorithms, the formal template parameter ForwardIterator is required to satisfy the requirements of a forward iterator (24.1.3) [..], and is required to have the property that no exceptions are thrown from [..], or dereference of valid iterators.
The standard should make better clear what "singular pointer" and "valid iterator" means. The fact that the meaning of a valid value has a core language meaning doesn't imply that for an iterator concept the term "valid iterator" has the same meaning.
Let me add a final example: In 99 [allocator.concepts.members] of N2914 we find:
pointer X::allocate(size_type n);11 Returns: a pointer to the allocated memory. [Note: if n == 0, the return value is unspecified. —end note]
[..]
void X::deallocate(pointer p, size_type n);Preconditions: p shall be a non-singular pointer value obtained from a call to allocate() on this allocator or one that compares equal to it.
If singular pointer value would include null pointers this make the preconditions unclear if the pointer value is a result of allocate(0): Since the return value is unspecified, it could be a null pointer. Does that mean that programmers need to check the pointer value for a null value before calling deallocate?
[ 2010-11-09 Daniel comments: ]
A later paper is in preparation.
[ 2010 Batavia: ]
Doesn't need to be resolved for Ox
[2014-02-20 Re-open Deferred issues as Priority 4]
Consider to await the paper.
Proposed resolution:
Section: 26.5 [complex.numbers] Status: LEWG Submitter: Ted Shaneyfelt Opened: 2009-09-26 Last modified: 2017-03-05
Priority: Not Prioritized
View other active issues in [complex.numbers].
View all other issues in [complex.numbers].
View all issues with LEWG status.
Discussion:
Concerning mathematically proper operation of the type:
complex<complex<T> >
Generally accepted mathematical semantics of such a construct correspond to quaternions through Cayly-Dickson construct
(w+xi) + (y+zi) j
The proper implementation seems straightforward by adding a few declarations like those below. I have included operator definition for combining real scalars and complex types, as well, which seems appropriate, as algebra of complex numbers allows mixing complex and real numbers with operators. It also allows for constructs such as complex<double> i=(0,1), x = 12.34 + 5*i;
Quaternions are often used in areas such as computer graphics, where, for example, they avoid the problem of Gimbal lock when rotating objects in 3D space, and can be more efficient than matrix multiplications, although I am applying them to a different field.
/////////////////////////ALLOW OPERATORS TO COMBINE REAL SCALARS AND COMPLEX VALUES /////////////////////////
template<typename T,typename S> complex<T> operator+(const complex<T> x,const S a) {
complex<T> result(x.real()+a, x.imag());
return result;
}
template<typename T,typename S> complex<T> operator+(const S a,const complex<T> x) {
complex<T> result(a+x.real(), x.imag());
return result;
}
template<typename T,typename S> complex<T> operator-(const complex<T> x,const S a) {
complex<T> result(x.real()-a, x.imag());
return result;
}
template<typename T,typename S> complex<T> operator-(const S a,const complex<T> x) {
complex<T> result(a-x.real(), x.imag());
return result;
}
template<typename T,typename S> complex<T> operator*(const complex<T> x,const S a) {
complex<T> result(x.real()*a, x.imag()*a);
return result;
}
template<typename T,typename S> complex<T> operator*(const S a,const complex<T> x) {
complex<T> result(a*x.real(), a*x.imag());
return result;
}
/////////////////////////PROPERLY IMPLEMENT QUATERNION SEMANTICS/////////////////////////
template<typename T> double normSq(const complex<complex<T> >q) {
return q.real().real()*q.real().real()
+ q.real().imag()*q.real().imag()
+ q.imag().real()*q.imag().real()
+ q.imag().imag()*q.imag().imag();
}
template<typename T> double norm(const complex<complex<T> >q) {
return sqrt(normSq(q));
}
/////// Cayley-Dickson Construction
template<typename T> complex<complex<T> > conj(const complex<complex<T> > x) {
complex<complex<T> > result(conj(x.real()),-x.imag());
return result;
}
template<typename T> complex<complex<T> > operator*(const complex<complex<T> > ab,const complex<complex<T> > cd) {
complex<T> re(ab.real()*cd.real()-conj(cd.imag())*ab.imag());
complex<T> im(cd.imag()*ab.real()+ab.imag()*conj(cd.real()));
complex<complex<double> > q(re,im);
return q;
}
//// Quaternion division
template<typename S,typename T> complex<complex<T> > operator/(const complex<complex<T> > q,const S a) {
return q * (1/a);
}
template<typename S,typename T> complex<complex<T> > operator/(const S a,const complex<complex<T> > q) {
return a*conj(q)/normSq(q);
}
template<typename T> complex<complex<T> > operator/(const complex<complex<T> > n, const complex<complex<T> > d) {
return n * (conj(d)/normSq(d));
}
[ 2009-10 Santa Cruz: ]
NAD Future. There is no consensus or time to move this into C++0X.
[2017-03-03, Kona]
SG6 suggests this issue is a new feature, not a problem with the existing standard, and should therefore be closed NAD. However, SG6 invites papers that bring the proposal up to date with the current standard.
Proposed resolution:
Section: 26.6.1.6 [rand.req.dist] Status: LEWG Submitter: Matthias Troyer Opened: 2009-10-12 Last modified: 2017-05-31
Priority: Not Prioritized
View all other issues in [rand.req.dist].
View all issues with LEWG status.
Discussion:
There exist optimized, vectorized vendor libraries for the creation of random number generators, such as Intel's MKL [1] and AMD's ACML [2]. In timing tests we have seen a performance gain of a factor of up to 80 (eighty) compared to a pure C++ implementation (in Boost.Random) when using these generator to generate a sequence of normally distributed random numbers. In codes dominated by the generation of random numbers (we have application codes where random number generation is more than 50% of the CPU time) this factor 80 is very significant.
To make use of these vectorized generators, we use a C++ class modeling the RandomNumberEngine concept and forwarding the generation of random numbers to those optimized generators. For example:
namespace mkl {
class mt19937 {.... };
}
For the generation of random variates we also want to dispatch to optimized vectorized functions in the MKL or ACML libraries. See this example:
mkl::mt19937 eng; std::normal_distribution<double> dist; double n = dist(eng);
Since the variate generation is done through the operator() of the distribution there is no customization point to dispatch to Intel's or AMD's optimized functions to generate normally distributed numbers based on the mt19937 generator. Hence, the performance gain of 80 cannot be achieved.
Contrast this with TR1:
mkl::mt19937 eng; std::tr1::normal_distribution<double> dist; std::tr1::variate_generator<mkl::mt19937,std::tr1::normal_distribution<double> > rng(eng,dist); double n = rng();
This - admittedly much uglier from an aestethic point of view - design allowed optimization by specializing the variate_generator template for mkl::mt19937:
namespace std { namespace tr1 {
template<>
class variate_generator<mkl::mt19937,std::tr1::normal_distribution<double> > { .... };
} }
A similar customization point is missing in the C++0x design and prevents the optimized vectorized version to be used.
Suggested resolution:
Add a customization point to the distribution concept. Instead of the variate_generator template this can be done through a call to a free function generate_variate found by ADL instead of operator() of the distribution:
template <RandomNumberDistribution, class RandomNumberEngine> typename RandomNumberDistribution ::result_type generate_variate(RandomNumberDistribution const& dist, RandomNumberEngine& eng);
This function can be overloaded for optimized enginges like mkl::mt19937.
[ 2009-10 Santa Cruz: ]
NAD Future. No time to add this feature for C++0X.
Proposed resolution:
Section: 25 [algorithms] Status: LEWG Submitter: Alisdair Meredith Opened: 2009-10-15 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [algorithms].
View all other issues in [algorithms].
View all issues with LEWG status.
Discussion:
The library has many algorithms that take a source range represented by a pair of iterators, and the start of some second sequence given by a single iterator. Internally, these algorithms will produce undefined behaviour if the second 'range' is not as large as the input range, but none of the algorithms spell this out in Requires clauses, and there is no catch-all wording to cover this in clause 17 or the front matter of 25.
There was an attempt to provide such wording in paper n2944 but this seems incidental to the focus of the paper, and getting the wording of this issue right seems substantially more difficult than the simple approach taken in that paper. Such wording will be removed from an updated paper, and hopefully tracked via the LWG issues list instead.
It seems there are several classes of problems here and finding wording to solve all in one paragraph could be too much. I suspect we need several overlapping requirements that should cover the desired range of behaviours.
Motivating examples:
A good initial example is the swap_ranges algorithm. Here there is a clear requirement that first2 refers to the start of a valid range at least as long as the range [first1, last1). n2944 tries to solve this by positing a hypothetical last2 iterator that is implied by the signature, and requires distance(first2,last2) < distance(first1,last1). This mostly works, although I am uncomfortable assuming that last2 is clearly defined and well known without any description of how to obtain it (and I have no idea how to write that).
A second motivating example might be the copy algorithm. Specifically, let us image a call like:
copy(istream_iterator<int>(is),istream_iterator(),ostream_iterator<int>(os));
In this case, our input iterators are literally simple InputIterators, and the destination is a simple OutputIterator. In neither case am I happy referring to std::distance, in fact it is not possible for the ostream_iterator at all as it does not meet the requirements. However, any wording we provide must cover both cases. Perhaps we might deduce last2 == ostream_iterator<int>{}, but that might not always be valid for user-defined iterator types. I can well imagine an 'infinite range' that writes to /dev/null and has no meaningful last2.
The motivating example in n2944 is std::equal, and that seems to fall somewhere between the two.
Outlying examples might be partition_copy that takes two output iterators, and the _n algorithms where a range is specified by a specific number of iterations, rather than traditional iterator pair. We should also not accidentally apply inappropriate constraints to std::rotate which takes a third iterator that is not intended to be a separate range at all.
I suspect we want some wording similar to:
For algorithms that operate on ranges where the end iterator of the second range is not specified, the second range shall contain at least as many elements as the first.
I don't think this quite captures the intent yet though. I am not sure if 'range' is the right term here rather than sequence. More awkwardly, I am not convinced we can describe an Output sequence such as produce by an ostream_iterator as "containing elements", at least not as a precondition to the call before they have been written.
Another idea was to describe require that the trailing iterator support at least distance(input range) applications of operator++ and may be written through the same number of times if a mutable/output iterator.
We might also consider handling the case of an output range vs. an input range in separate paragraphs, if that simplifies how we describe some of these constraints.
[ 2009-11-03 Howard adds: ]
Moved to Tentatively NAD Future after 5 positive votes on c++std-lib.
Rationale:
Does not have sufficient support at this time. May wish to reconsider for a future standard.
Proposed resolution:
Section: 23 [containers] Status: LEWG Submitter: Herb Sutter Opened: 2009-10-21 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [containers].
View all other issues in [containers].
View all issues with LEWG status.
Discussion:
See N2980.
[ 2009-10 Santa Cruz ]
The paper was lengthy discussed but considerable concern remained to add this feature to C++0x. Strong consensus was found to consider it for C++1x, though.
Proposed resolution:
The LWG does not wish to make a change at this time.
Section: 25 [algorithms] Status: LEWG Submitter: Igor Semenov Opened: 2009-12-07 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [algorithms].
View all other issues in [algorithms].
View all issues with LEWG status.
Discussion:
Motivation and Scope
Splitting strings into parts by some set of delimiters is an often task, but there is no simple and generalized solution in C++ Standard. Usually C++ developers use std::basic_stringstream<> to split string into parts, but there are several inconvenient restrictions:
Impact on the Standard
This algorithm doesn't interfere with any of current standard algorithms.
Design Decisions
This algorithm is implemented in terms of input/output iterators. Also, there is one additional wrapper for const CharType * specified delimiters.
Example implementation
template< class It, class DelimIt, class OutIt >
void split( It begin, It end, DelimIt d_begin, DelimIt d_end, OutIt out )
{
while ( begin != end )
{
It it = std::find_first_of( begin, end, d_begin, d_end );
*out++ = std::make_pair( begin, it );
begin = std::find_first_of( it, end, d_begin, d_end,
std::not2( std::equal_to< typename It::value_type >() ) );
}
}
template< class It, class CharType, class OutIt >
void split( It begin, It end, const CharType * delim, OutIt out )
{
split( begin, end, delim, delim + std::strlen( delim ), out );
}
Usage
std::string ss( "word1 word2 word3" );
std::vector< std::pair< std::string::const_iterator, std::string::const_iterator > > v;
split( ss.begin(), ss.end(), " ", std::back_inserter( v ) );
for ( int i = 0; i < v.size(); ++i )
{
std::cout << std::string( v[ i ].first, v[ i ].second ) << std::endl;
}
// word1
// word2
// word3
[ 2010-01-22 Moved to Tentatively NAD Future after 5 positive votes on c++std-lib. Rationale added below. ]
Rationale:
The LWG is not considering completely new features for standardization at this time. We would like to revisit this good suggestion for a future TR and/or standard.
Proposed resolution:
Add to the synopsis in 25.1 [algorithms.general]:
template< class ForwardIterator1, class ForwardIterator2, class OutputIterator >
void split( ForwardIterator1 first, ForwardIterator1 last,
ForwardIterator2 delimiter_first, ForwardIterator2 delimiter_last,
OutputIterator result );
template< class ForwardIterator1, class CharType, class OutputIterator >
void split( ForwardIterator1 first, ForwardIterator1 last,
const CharType * delimiters, OutputIterator result );
Add a new section [alg.split]:
template< class ForwardIterator1, class ForwardIterator2, class OutputIterator > void split( ForwardIterator1 first, ForwardIterator1 last, ForwardIterator2 delimiter_first, ForwardIterator2 delimiter_last, OutputIterator result );1. Effects: splits the range [first, last) into parts, using any element of [delimiter_first, delimiter_last) as a delimiter. Results are pushed to output iterator in the form of std::pair<ForwardIterator1, ForwardIterator1>. Each of these pairs specifies a maximal subrange of [first, last) which does not contain a delimiter.
2. Returns: nothing.
3. Complexity: Exactly last - first assignments.
template< class ForwardIterator1, class CharType, class OutputIterator > void split( ForwardIterator1 first, ForwardIterator1 last, const CharType * delimiters, OutputIterator result );1. Effects: split the range [first, last) into parts, using any element of delimiters (interpreted as zero-terminated string) as a delimiter. Results are pushed to output iterator in the form of std::pair<ForwardIterator1, ForwardIterator1>. Each of these pairs specifies a maximal subrange of [first, last) which does not contain a delimiter.
2. Returns: nothing.
3. Complexity: Exactly last - first assignments.
Section: 20.2 [utility] Status: LEWG Submitter: Ion Gaztañaga Opened: 2009-12-14 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [utility].
View all other issues in [utility].
View all issues with LEWG status.
Discussion:
In section 17.5.3.5 [allocator.requirements], Table 40 — Allocator requirements, the following expression is required for allocator pointers:
Table 40 — Allocator requirements Expression Return type Assertion/note
pre-/post-conditionDefault static_cast<X::pointer>(w) X::pointer static_cast<X::pointer>(w) == p
To achieve this expression, a smart pointer writer must introduce an explicit conversion operator from smart_ptr<void> to smart_ptr<T> so that static_cast<pointer>(void_ptr) is a valid expression. Unfortunately this explicit conversion weakens the safety of a smart pointer since the following expression (invalid for raw pointers) would become valid:
smart_ptr<void> smart_v = ...; smart_ptr<T> smart_t(smart_v);
On the other hand, shared_ptr also defines its own casting functions in 20.11.2.2.9 [util.smartptr.shared.cast], and although it's unlikely that a programmer will use shared_ptr as allocator::pointer, having two different ways to do the same cast operation does not seem reasonable. A possible solution would be to replace static_cast<X::pointer>(w) expression with a user customizable (via ADL) static_pointer_cast<value_type>(w), and establish the xxx_pointer_cast functions introduced by shared_ptr as the recommended generic casting utilities of the standard.
Unfortunately, we've experienced problems in Boost when trying to establish xxx_pointer_cast as customization points for generic libraries (http://objectmix.com/c/40424-adl-lookup-explicit-template-parameters.html) because these casting functions are called with explicit template parameters and the standard says in 14.8.1 [temp.arg.explicit] p.8 "Explicit template argument specification":
8 ...But when a function template with explicit template arguments is used, the call does not have the correct syntactic form unless there is a function template with that name visible at the point of the call. If no such name is visible, the call is not syntactically well-formed and argument-dependent lookup does not apply.
So we can do this:
template<class BasePtr>
void generic_ptr_swap(BasePtr p)
{
//ADL customization point
swap(p, p);
//...
}
but not the following:
template<class BasePtr>
void generic_ptr_algo(BasePtr p)
{
typedef std::pointer_traits<BasePtr>::template
rebind<Derived> DerivedPtr;
DerivedPtr dp = static_pointer_cast<Derived>(p);
}
The solution to make static_pointer_cast a customization point is to add a generic declaration (no definition) of static_pointer_cast in a namespace (like std) and apply "using std::static_pointer_cast" declaration to activate ADL:
namespace std{
template<typename U, typename T>
unspecified
static_pointer_cast(T&&) = delete;
}
template<class BasePtr>
void generic_ptr_algo(BasePtr p)
{
typedef std::pointer_traits<BasePtr>::template
rebind<Derived> DerivedPtr;
//ADL applies because static_pointer_cast is made
// visible according to [temp.arg.explicit]/8
using std::static_pointer_cast;
DerivedPtr dp = static_pointer_cast<Derived>(p);
//...
}
A complete solution will need also the definition of static_pointer_cast for raw pointers, and this definition has been present in Boost (http://www.boost.org/boost/ pointer_cast.hpp) for years.
[ 2010-03-26 Daniel made editorial adjustments to the proposed wording. ]
[ Moved to NAD Future at 2010-11 Batavia ]
This is a new feature rather than a defect. It can be added later: "this is such a hairy area that people will put up with changes"
Proposed resolution:
Add to section 20.2 [utility] Utility components, Header <utility> synopsis:
// 20.3.X, generic pointer cast functions template<typename U, typename T> unspecified static_pointer_cast(T&&) = delete; template<typename U, typename T> unspecified dynamic_pointer_cast(T&&) = delete; template<typename U, typename T> unspecified const_pointer_cast(T&&) = delete; //Overloads for raw pointers template<typename U, typename T> auto static_pointer_cast(T* t) -> decltype(static_cast<U*>(t)); template<typename U, typename T> auto dynamic_pointer_cast(T* t) -> decltype(dynamic_cast<U*>(t)); template<typename U, typename T> auto const_pointer_cast(T* t) -> decltype(const_cast<U*>(t));
Add to section 20.2 [utility] Utility components, a new subclause 20.3.X Pointer cast utilities [pointer.cast]:
20.3.X Pointer cast utilities [pointer.cast]
1 The library defines generic pointer casting function templates so that template code can explicitly make these names visible and activate argument-dependent lookup for pointer cast calls.
//Generic declarations template<typename U, typename T> unspecified static_pointer_cast(T&&) = delete; template<typename U, typename T> unspecified dynamic_pointer_cast(T&&) = delete; template<typename U, typename T> unspecified const_pointer_cast(T&&) = delete;2 The library also defines overloads of these functions for raw pointers.
//Overloads for raw pointers template<typename U, typename T> auto static_pointer_cast(T* t) -> decltype(static_cast<U*>(t));Returns: static_cast<U*>(t)
template<typename U, typename T> auto dynamic_pointer_cast(T* t) -> decltype(dynamic_cast<U*>(t));Returns: dynamic_cast<U*>(t)
template<typename U, typename T> auto const_pointer_cast(T* t) -> decltype(const_cast<U*>(t));Returns: const_cast<U*>(t)
[Example:
#include <utility> //static_pointer_cast #include <memory> //pointer_traits class Base{}; class Derived : public Base{}; template<class BasePtr> void generic_pointer_code(BasePtr b) { typedef std::pointer_traits<BasePtr>::template rebind<Derived> DerivedPtr; using std::static_pointer_cast; //ADL applies now that static_pointer_cast is visible DerivedPtr d = static_pointer_cast<Derived>(b); }— end example]
Replace in section 17.5.3.5 [allocator.requirements] Table 40 — Allocator requirements, the following table entries for allocator pointers:
Table 40 — Allocator requirements Expression Return type Assertion/note
pre-/post-conditionDefault static_pointer_cast< X::pointerT>(w)X::pointer static_pointer_cast< X::pointerT>(w) == pstatic_pointer_cast< X::const_pointerconst T>(w)X::const_pointer static_pointer_cast< X::const_pointerconst T>(z) == q
Section: 20.14.15 [unord.hash] Status: LEWG Submitter: Nicolai M. Josuttis Opened: 2010-02-10 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [unord.hash].
View all other issues in [unord.hash].
View all issues with LEWG status.
Discussion:
Currently, the library lacks a convenient way to provide a hash function that can be used with the provided unordered containers to allow the usage of non trivial element types.
While we can easily declare an
std::unordered_set<int>
or
std::unordered_set<std::string>
we have no easy way to declare an unordered_set for a user defined type. IMO, this is a big obstacle to use unordered containers in practice. Note that in Java, the wide usage of HashMap is based on the fact that there is always a default hash function provided.
Of course, a default hash function implies the risk to provide poor hash functions. But often even poor hash functions are good enough.
While I really would like to see a default hash function, I don't propose it here because this would probably introduce a discussion that's too big for this state of C++0x.
However, I strongly suggest at least to provide a convenience variadic template function make_hash<>() to allow an easy definition of a (possibly poor) hash function.
As a consequence for a user-defined type such as
class Customer {
friend class CustomerHash;
private:
string firstname;
string lastname;
long no;
...
};
would allow to specify:
class CustomerHash : public std::unary_function<Customer, std::size_t>
{
public:
std::size_t operator() (const Customer& c) const {
return make_hash(c.firstname,c.lastname,c.no);
}
};
instead of:
class CustomerHash : public std::unary_function<Customer, std::size_t>
{
public:
std::size_t operator() (const Customer& c) const {
return std::hash<std::string>()(c.firstname) +
std::hash<std::string>()(c.lastname) +
std::hash<long>()(c.no);
}
};
Note that, in principle, we can either specify that
make_hash returns the sum of a call of std::hash<T>()(x) for each argument x of type T
or we can specify that
make_hash provides a hash value for each argument, for which a std::hash() function is provided
with the possible note that the hash value may be poor or only a good hash value if the ranges of all passed arguments is equally distributed.
For my convenience, I propose wording that describes the concrete implementation.
[ 2010 Pittsburgh: Moved to NAD Editorial, rationale added below. ]
Rationale:
There is no consensus to make this change at this time.
Proposed resolution:
In Function objects 20.14 [function.objects] in paragraph 2 at the end of the Header <functional> synopsis insert:
// convenience functions template <class T> size_t make_hash (const T&); template <class T, class... Types> size_t make_hash (const T&, const Types&...);
In Class template hash 20.14.15 [unord.hash] add:
20.7.16.1 Hash creation functions [hash.creation]
template <class T> size_t make_hash (const T& val);Returns: hash<T>()(val);
template <class T, class... Types> size_t make_hash (const T& val, const Types&... args);Returns: hash<T>()(val) + std::make_hash(args...)
Section: 28.8 [re.regex] Status: LEWG Submitter: INCITS Opened: 2010-08-25 Last modified: 2017-02-03
Priority: Not Prioritized
View all other issues in [re.regex].
View all issues with LEWG status.
Duplicate of: 1451
Discussion:
Addresses US-104, US-141
std::basic_regex should have an allocator for all the reasons that a std::string does. For example, I can use boost::interprocess to put a string or vector in shared memory, but not a regex.
[ Resolution proposed by ballot comment ]
Add allocators to regexes
[ 2010-10-24 Daniel adds: ]
Accepting n3171 would solve this issue.
[2011-03-22 Madrid]
Close 1396 as NAD Future.
Rationale:
No consensus for a change at this time
Proposed resolution:
Section: 20.11.2.2 [util.smartptr.shared] Status: LEWG Submitter: Japan Opened: 2010-08-25 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [util.smartptr.shared].
View all other issues in [util.smartptr.shared].
View all issues with LEWG status.
Discussion:
Addresses JP-5
Hash support based on ownership sharing should be supplied for shared_ptr and weak_ptr. For two shared_ptr objects p and q, two distinct equivalence relations can be defined. One is based on equivalence of pointer values, which is derived from the expression p.get() == q.get() (hereafter called address based equivalence relation), the other is based on equivalence of ownership sharing, which is derived from the expression !p.owner_before(q) && !q.owner_before(p) (hereafter called ownership-based equivalence relation). These two equivalence relations are independent in general. For example, a shared_ptr object created by the constructor of the signature shared_ptr(shared_ptr<U> const &, T *) could reveal a difference between these two relations. Therefore, hash support based on each equivalence relation should be supplied for shared_ptr. However, while the standard library provides the hash support for address-based one (20.9.11.6 paragraph 2), it lacks the hash support for ownership-based one. In addition, associative containers work well in combination with the shared_ptr's ownership-based comparison but unordered associative containers don't. This is inconsistent.
For the case of weak_ptr, hash support for the ownership based equivalence relation can be safely defined on weak_ptrs, and even on expired ones. The absence of hash support for the ownership-based equivalence relation is fatal, especially for expired weak_ptrs. And the absence of such hash support precludes some quite effective use-cases, e.g. erasing the unordered_map entry of an expired weak_ptr key from a customized deleter supplied to shared_ptrs.
Hash support for the ownership-based equivalence relation cannot be provided by any user-defined manner because information about ownership sharing is not available to users at all. Therefore, the only way to provide ownership-based hash support is to offer it intrusively by the standard library.
As far as we know, such hash support is implementable. Typical implementation of such hash function could return the hash value of the pointer of the counter object that is internally managed by shared_ptr and weak_ptr.
[2010 Rapperswil:]
No consensus to make this change at this time.
Proposed resolution:
Add the following non-static member functions to shared_ptr and weak_ptr class template;
Update [util.smartptr.shared], 20.9.11.2 paragraph 1
namespace std{
template<class T> class shared_ptr {
public:
...
size_t owner_hash() const;
...
};
}
Update [util.smartptr.weak], 20.9.11.3 paragraph 1
namespace std{
template<class T> class weak_ptr {
public:
...
size_t owner_hash() const;
...
};
}
These functions satisfy the following requirements. Let p and q be objects of either shared_ptr or weak_ptr, H be a hypothetical function object type that satisfies the hash requirements ([hash.requirements], 20.2.4) and h be an object of the type H. The expression p.owner_hash() behaves as if it were equivalent to the expression h(p). In addition, h(p) == h(q) must become true if p and q share ownership.
Section: 23.3.12 [vector.bool] Status: Open Submitter: BSI Opened: 2010-08-25 Last modified: 2017-06-15
Priority: 3
View all other issues in [vector.bool].
View all issues with Open status.
Discussion:
Addresses GB-118
vector<bool> iterators are not random access iterators because their reference type is a special class, and not bool &. All standard libary operations taking iterators should treat this iterator as if it was a random access iterator, rather than a simple input iterator.
[ Resolution proposed in ballot comment ]
Either revise the iterator requirements to support proxy iterators (restoring functionality that was lost when the Concept facility was removed) or add an extra paragraph to the vector<bool> specification requiring the library to treat vector<bool> iterators as-if they were random access iterators, despite having the wrong reference type.
[ Rapperswil Review ]
The consensus at Rapperswil is that it is too late for full support for proxy iterators, but requiring the library to respect vector<bool> iterators as-if they were random access would be preferable to flagging this container as deliberately incompatible with standard library algorithms.
Alisdair to write the note, which may become normative Remark depending on the preferences of the project editor.
[ Post-Rapperswil Alisdair provides wording ]
Initial wording is supplied, deliberately using Note in preference to Remark although the author notes his preference for Remark. The issue of whether iterator_traits<vector<bool>>::iterator_category is permitted to report random_access_iterator_tag or must report input_iterator_tag is not addressed.
[ Old Proposed Resolution: ]
Insert a new paragraph into 23.3.12 [vector.bool] between p4 and p5:
[Note All functions in the library that take a pair of iterators to denote a range shall treat vector<bool> iterators as-if they were random access iterators, even though the reference type is not a true reference.-- end note]
[ 2010-11 Batavia: ]
Closed as NAD Future, because the current iterator categories cannot correctly describe vector<bool>::iterator. But saying that they are Random Access Iterators is also incorrect, because it is not too hard to create a corresponding test that fails. We should deal with the more general proxy iterator problem in the future, and see no benefit to take a partial workaround specific to vector<bool> now.
[2017-02 in Kona, LEWG recommends NAD]
D0022 Proxy Iterators for the Ranges Extensions - as much a fix as we’re going to get for vector<bool>.
[2017-06-02 Issues Telecon]
P0022 is exploring a resolution. We consider this to be fairly important issue
Move to Open, set priority to 3
Proposed resolution:
Rationale:
No consensus to make this change at this time.
Section: 29.3 [atomics.order] Status: LEWG Submitter: Canada Opened: 2010-08-25 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [atomics.order].
View all other issues in [atomics.order].
View all issues with LEWG status.
Duplicate of: 1458
Discussion:
Addresses CA-21, GB-131
29.4 [atomics.lockfree] p.8 states:
An atomic store shall only store a value that has been computed from constants and program input values by a finite sequence of program evaluations, such that each evaluation observes the values of variables as computed by the last prior assignment in the sequence.
... but 1.9 [intro.execution] p.13 states:
If A is not sequenced before B and B is not sequenced before A, then A and B are unsequenced. [ Note: The execution of unsequenced evaluations can overlap. — end note ]
Overlapping executions can make it impossible to construct the sequence described in 29.4 [atomics.lockfree] p.8. We are not sure of the intention here and do not offer a suggestion for change, but note that 29.4 [atomics.lockfree] p.8 is the condition that prevents out-of-thin-air reads.
For an example, suppose we have a function invocation f(e1,e2). The evaluations of e1 and e2 can overlap. Suppose that the evaluation of e1 writes y and reads x whereas the evaluation of e2 reads y and writes x, with reads-from edges as below (all this is within a single thread).
e1 e2
Wrlx y-- --Wrlx x
rf\ /rf
X
/ \
Rrlx x<- ->Rrlx y
This seems like it should be allowed, but there seems to be no way to produce a sequence of evaluations with the property above.
In more detail, here the two evaluations, e1 and e2, are being executed as the arguments of a function and are consequently not sequenced-before each other. In practice we'd expect that they could overlap (as allowed by 1.9 [intro.execution] p.13), with the two writes taking effect before the two reads. However, if we have to construct a linear order of evaluations, as in 29.4 [atomics.lockfree] p.8, then the execution above is not permited. Is that really intended?
[ Resolution proposed by ballot comment ]
Please clarify.
[2011-03-09 Hans comments:]
I'm not proud of 29.3 [atomics.order] p9 (formerly p8), and I agree with the comments that this isn't entirely satisfactory. 29.3 [atomics.order] p9 was designed to preclude out-of-thin-air results for races among memory_order_relaxed atomics, in spite of the fact that Java experience has shown we don't really know how to do that adequately. In the long run, we probably want to revisit this.
However, in the short term, I'm still inclined to declare this NAD, for two separate reasons:1.9 [intro.execution] p15 states: "If a side effect on a scalar object is unsequenced relative to either another side effect on the same scalar object or a value computation using the value of the same scalar object, the behavior is undefined." I think the examples presented here have undefined behavior as a result. It's not completely clear to me whether examples can be constructed that exhibit this problem, and don't have undefined behavior.
This comment seems to be using a different meaning of "evaluation" from what is used elsewhere in the standard. The sequence of evaluations here doesn't have to consist of full expression evaluations. They can be evaluations of operations like lvalue to rvalue conversion, or individual assignments. In particular, the reads and writes executed by e1 and e2 in the example could be treated as separate evaluations for purposes of producing the sequence. The definition of "sequenced before" in 1.9 [intro.execution] makes little sense if the term "evaluation" is restricted to any notion of complete expression. Perhaps we should add yet another note to clarify this? 29.3 [atomics.order] p10 probably leads to the wrong impression here.
An alternative resolution would be to simply delete our flakey attempt at preventing out-of-thin-air reads, by removing 29.3 [atomics.order] p9-11, possibly adding a note that explains that we technically allow, but strongly discourage them. If we were starting this from scratch now, that would probably be my preference. But it seems like too drastic a resolution at this stage.[2011-03-24 Madrid]
Moved to NAD Future
Proposed resolution:
Section: 30.3.2 [thread.thread.class] Status: LEWG Submitter: INCITS Opened: 2010-08-25 Last modified: 2017-03-05
Priority: Not Prioritized
View all issues with LEWG status.
Discussion:
Addresses US-183
There is no way to join a thread with a timeout.
[ Resolution proposed by ballot comment: ]
Add join_for and join_until. Or decide one should never join a thread with a timeout since pthread_join doesn't have a timeout version.
[ 2010 Batavia ]
The concurrency working group deemed this an extension beyond the scope of C++0x.
Rationale:
The LWG does not wish to make a change at this time.
[2017-03-01, Kona]
SG1 recommends: Close as NAD
There has not been much demand for it, and it would usually be difficult to deal with thread_local destructor races. It can be approximated with a condition variable wait followed by an unconditional join. Adding it would create implementation issues on Posix. As always, this may be revisited if we have a paper exploring the issues in detail.Proposed resolution:
Section: 30.4 [thread.mutex] Status: LEWG Submitter: INCITS Opened: 2010-08-25 Last modified: 2017-03-05
Priority: Not Prioritized
View all other issues in [thread.mutex].
View all issues with LEWG status.
Discussion:
Addresses US-185
Cooperate with WG14 to improve interoperability between the C++0x and C1x threads APIs. In particular, C1x mutexes should be conveniently usable with a C++0x lock_guard. Performance overheads for this combination should be considered.
[ Resolution proposed by ballot comment: ]
Remove C++0x timed_mutex and timed_recursive_mutex if that facilitates development of more compatible APIs.
[ 2010 Batavia ]
The concurrency sub-group reviewed the options, and decided that closer harmony should wait until both standards are published.
Rationale:
The LWG does not wish to make any change at this time.
[2017-03-01, Kona]
SG1 recommends: Close as NAD
Papers about C compatibility are welcome, but there may be more pressing issues. C threads are not consistently available at this point, so there seems to be little demand to fix this particular problem.Proposed resolution:
Section: 30.4.3 [thread.mutex.requirements] Status: LEWG Submitter: INCITS Opened: 2010-08-25 Last modified: 2017-03-05
Priority: Not Prioritized
View other active issues in [thread.mutex.requirements].
View all other issues in [thread.mutex.requirements].
View all issues with LEWG status.
Discussion:
Addresses US-189
mutex and recursive_mutex should have an is_locked() member function. is_locked allows a user to test a lock without acquiring it and can be used to implement a lightweight try_try_lock.
[ Resolution proposed by ballot comment: ]
Add a member function:
bool is_locked() const;to std::mutex and std::recursive_mutex. These functions return true if the current thread would not be able to obtain a mutex. These functions do not synchronize with anything (and, thus, can avoid a memory fence).
[ 2010 Batavia ]
The Concurrency subgroup reviewed this issue and deemed it to be an extension to be handled after publishing C++0x.
Rationale:
The LWG does not wish to make a change at this time.
[2017-03-01, Kona]
SG1 recommends: Close as NAD
Several participants voiced strong objections, based on either memory model issues or lock elision. No support. It is already possible to write a wrapper that explicitly tracks ownership for testing in the owning thread, which may have been part of the intent here.Proposed resolution:
Section: 30.5 [thread.condition] Status: LEWG Submitter: INCITS Opened: 2010-08-25 Last modified: 2017-03-05
Priority: Not Prioritized
View all other issues in [thread.condition].
View all issues with LEWG status.
Discussion:
Addresses US-193
Condition variables preclude a wakeup optimization.
[ Resolution proposed by ballot comment: ]
Change condition_variable to allow such optimization. See Appendix 1 - Additional Details
[ 2010 Batavia ]
The Concurrency subgroup reviewed the issue, and deemed it an extension to be handled after C++0x.
Rationale:
The LWG does not wish to make the change at this time.
[2017-03-01, Kona]
SG1 recommends: Close as NAD
The approach suggested there raises all sorts of issues about thread_locals, etc. It's probably way too late to change this anyway, but this would have required a careful paper.Proposed resolution:
Section: 23.2.1 [container.requirements.general] Status: LEWG Submitter: Mike Spertus Opened: 2010-10-16 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [container.requirements.general].
View all other issues in [container.requirements.general].
View all issues with LEWG status.
Discussion:
Addresses US-104, US-141
The standard doesn't say that containers should use abstract pointer types internally. Both Howard and Pablo agree that this is the intent. Further, it is necessary for containers to be stored, for example, in shared memory with an interprocess allocator (the type of scenario that allocators are intended to support).
In spite of the (possible) agreement on intent, it is necessary to make this explicit:
An implementations may like to store the result of dereferencing the pointer (which is a raw reference) as an optimization, but that prevents the data structure from being put in shared memory, etc. In fact, a container could store raw references to the allocator, which would be a little weird but conforming as long as it has one by-value copy. Furthermore, pointers to locales, ctypes, etc. may be there, which also prevents the data structure from being put in shared memory, so we should make explicit that a container does not store raw pointers or references at all.
[ Pre-batavia ]
This issue is being opened as part of the response to NB comments US-104/141. See paper N3171 in the pre-Batavia mailing.
[2011-03-23 Madrid meeting]
Deferred
[ 2011 Batavia ]
This may be an issue, but it is not clear. We want to gain a few years experience with the C++11 allocator model to see if this is already implied by the existing specification.
Proposed resolution:
Add to the end of 23.2.1 [container.requirements.general] p. 8:
[..] In all container types defined in this Clause, the member get_allocator() returns a copy of the allocator used to construct the container or, if that allocator has been replaced, a copy of the most recent replacement. The container may not store internal objects whose types are of the form T * or T & except insofar as they are part of the item type or members.
Section: 24.2.4 [output.iterators] Status: Open Submitter: Daniel Krügler Opened: 2011-02-27 Last modified: 2017-02-03
Priority: 3
View other active issues in [output.iterators].
View all other issues in [output.iterators].
View all issues with Open status.
Discussion:
During the Pittsburgh meeting the proposal N3066 became accepted because it fixed several severe issues related to the iterator specification. But the current working draft (N3225) does not reflect all these changes. Since I'm unaware whether every correction can be done editorial, this issue is submitted to take care of that. To give one example: All expressions of Table 108 — "Output iterator requirements" have a post-condition that the iterator is incrementable. This is impossible, because it would exclude any finite sequence that is accessed by an output iterator, such as a pointer to a C array. The N3066 wording changes did not have these effects.
[2011-03-01: Daniel comments:]
This issue has some overlap with the issue 2038 and I would prefer if we could solve both at one location. I suggest the following approach:
The terms dereferencable and incrementable could be defined in a more general way not restricted to iterators (similar to the concepts HasDereference and HasPreincrement from working draft N2914). But on the other hand, all current usages of dereferencable and incrementable are involved with types that satisfy iterator requirements. Thus, I believe that it is sufficient for C++0x to add corresponding definitions to 24.2.1 [iterator.requirements.general] and to let all previous usages of these terms refer to this sub-clause. Since the same problem occurs with the past-the-end iterator, this proposal suggest providing similar references to usages that precede its definition as well.
We also need to ensure that all iterator expressions get either an operational semantics in terms of others or we need to add missing pre- and post-conditions. E.g. we have the following ones without semantics:
*r++ = o // output iterator *r-- // bidirectional iterator
According to the SGI specification these correspond to
{ *r = o; ++r; } // output iterator
{ reference tmp = *r; --r; return tmp; } // bidirectional iterator
respectively. Please note especially the latter expression for bidirectional iterator. It fixes a problem that we have for forward iterator as well: Both these iterator categories provide stronger guarantees than input iterator, because the result of the dereference operation is reference, and not only convertible to the value type (The exact form from the SGI documentation does not correctly refer to reference).
[2011-03-14: Daniel comments and updates the suggested wording]
In addition to the before mentioned necessary changes there is another one need, which became obvious due to issue 2042: forward_list<>::before_begin() returns an iterator value which is not dereferencable, but obviously the intention is that it should be incrementable. This leads to the conclusion that imposing dereferencable as a requirement for the expressions ++r is wrong: We only need the iterator to be incrementable. A similar conclusion applies to the expression --r of bidirectional iterators.
[ 2011 Bloomington ]
Consensus this is the correct direction, but there are (potentially) missing incrementable preconditions on some table rows, and the Remarks on when an output iterator becomes dereferencable are probably better handled outside the table, in a manner similar to the way we word for input iterators.
There was some concern about redundant pre-conditions when the operational semantic is defined in terms of operations that have preconditions, and a similar level of concern over dropping such redundancies vs. applying a consistent level of redundant specification in all the iterator tables. Wording clean-up in either direction would be welcome.
[2011-08-18: Daniel adapts the proposed resolution to honor the Bloomington request]
There is only a small number of further changes suggested to get rid of superfluous requirements and essentially non-normative assertions. Operations should not have extra pre-conditions, if defined by "in-terms-of" semantics, see e.g. a != b or a->m for Table 107. Further, some remarks, that do not impose anything or say nothing new have been removed, because I could not find anything helpful they provide. E.g. consider the remarks for Table 108 for the operations dereference-assignment and preincrement: They don't provide additional information say nothing surprising. With the new pre-conditions and post-conditions it is implied what the remarks intend to say.
[ 2011-11-03: Some observations from Alexander Stepanov via c++std-lib-31405 ]
The following sentence is dropped from the standard section on OutputIterators:
"In particular, the following two conditions should hold: first, any iterator value should be assigned through before it is incremented (this is, for an output iterator i, i++; i++; is not a valid code sequence); second, any value of an output iterator may have at most one active copy at any given time (for example, i = j; *++i = a; *j = b; is not a valid code sequence)."[ 2011-11-04: Daniel comments and improves the wording ]
In regard to the first part of the comment, the intention of the newly proposed wording was to make clear that for the expression
*r = o
we have the precondition dereferenceable and the post-condition incrementable. And for the expression
++r
we have the precondition incrementable and the post-condition dereferenceable or past-the-end. This should not allow for a sequence like i++; i++; but I agree that it doesn't exactly say that.
In regard to the second point: To make this point clearer, I suggest to add a similar additional wording as we already have for input iterator to the "Assertion/note" column of the expression ++r: "Post: any copies of the previous value of r are no longer required to be dereferenceable or incrementable." The proposed has been updated to honor the observations of Alexander Stepanov.[2015-02 Cologne]
The matter is complicated, Daniel volunteers to write a paper.
Proposed resolution:
Add a reference to 24.2.1 [iterator.requirements.general] to the following parts of the library preceding Clause 24 Iterators library: (I stopped from 23.2.7 [unord.req] on, because the remaining references are the concrete containers)
17.5.3.2 [swappable.requirements] p5:
-5- A type X satisfying any of the iterator requirements (24.2) is ValueSwappable if, for any dereferenceable (24.2.1 [iterator.requirements.general]) object x of type X, *x is swappable.
17.5.3.5 [allocator.requirements], Table 27 — "Descriptive variable definitions", row with the expression c:
a dereferenceable (24.2.1 [iterator.requirements.general]) pointer of type C*
20.10.3.2 [pointer.traits.functions]:
Returns: The first template function returns a dereferenceable (24.2.1 [iterator.requirements.general]) pointer to r obtained by calling Ptr::pointer_to(r); […]
21.3.2.3 [string.iterators] p. 2:
Returns: An iterator which is the past-the-end value (24.2.1 [iterator.requirements.general]).
22.4.5.1.2 [locale.time.get.virtuals] p. 11:
iter_type do_get(iter_type s, iter_type end, ios_base& f, ios_base::iostate& err, tm *t, char format, char modifier) const;Requires: t shall be dereferenceable (24.2.1 [iterator.requirements.general]).
23.2.1 [container.requirements.general] p. 6:
[…] end() returns an iterator which is the past-the-end (24.2.1 [iterator.requirements.general]) value for the container. […]
23.2.3 [sequence.reqmts] p. 3:
[…] q denotes a valid dereferenceable (24.2.1 [iterator.requirements.general]) const iterator to a, […]
23.2.6 [associative.reqmts] p. 8 (I omit intentionally one further reference in the same sub-clause):
[…] q denotes a valid dereferenceable (24.2.1 [iterator.requirements.general]) const iterator to a, […]
23.2.7 [unord.req] p. 10 (I omit intentionally one further reference in the same sub-clause):
[…] q and q1 are valid dereferenceable (24.2.1 [iterator.requirements.general]) const iterators to a, […]
Edit 24.2.1 [iterator.requirements.general] p. 5 as indicated (The intent is to properly define incrementable and to ensure some further library guarantee related to past-the-end iterator values):
-5- Just as a regular pointer to an array guarantees that there is a pointer value pointing past the last element of the array, so for any iterator type there is an iterator value that points past the last element of a corresponding sequence. These values are called past-the-end values. Values of an iterator i for which the expression *i is defined are called dereferenceable. Values of an iterator i for which the expression ++i is defined are called incrementable. The library never assumes that past-the-end values are dereferenceable or incrementable. Iterators can also have singular values that are not associated with any sequence. […]
Modify the column contents of Table 106 — "Iterator requirements", 24.2.2 [iterator.iterators], as indicated:
Table 106 — Iterator requirements Expression Return type Operational semantics Assertion/note
pre-/post-condition*r reference pre: r is dereferenceable. ++r X& pre: r is incrementable.
Modify the column contents of Table 107 — "Input iterator requirements", 24.2.3 [input.iterators], as indicated [Rationale: The wording changes attempt to define a minimal "independent" set of operations, namely *a and ++r, and to specify the semantics of the remaining ones. This approach seems to be in agreement with the original SGI specification — end rationale]:
Table 107 — Input iterator requirements (in addition to Iterator) Expression Return type Operational semantics Assertion/note
pre-/post-conditiona != b contextually
convertible to bool!(a == b) pre: (a, b) is in the domain
of ==.*a convertible to T pre: a is dereferenceable.
The expression
(void)*a, *a is equivalent
to *a.
If a == b and (a,b) is in
the domain of == then *a is
equivalent to *b.a->m (*a).m pre: a is dereferenceable.++r X& pre: r is dereferenceableincrementable.
post: r is dereferenceable or
r is past-the-end.
post: any copies of the
previous value of r are no
longer required either to be
dereferenceable, incrementable,
or to be in the domain of ==.(void)r++ (void)++r equivalent to (void)++r*r++ convertible to T { T tmp = *r;
++r;
return tmp; }
Modify the column contents of Table 108 — "Output iterator requirements", 24.2.4 [output.iterators], as indicated [Rationale: The wording changes attempt to define a minimal "independent" set of operations, namely *r = o and ++r, and to specify the semantics of the remaining ones. This approach seems to be in agreement with the original SGI specification — end rationale]:
Table 108 — Output iterator requirements (in addition to Iterator) Expression Return type Operational semantics Assertion/note
pre-/post-condition*r = o result is not used pre: r is dereferenceable.
Remark: After this operation
r is not required to be
dereferenceable and any copies of
the previous value of r are no
longer required to be dereferenceable
or incrementable.
post: r is incrementable.++r X& pre: r is incrementable.
&r == &++r.
Remark: After this operationRemark: After this operation
r is not required to be
dereferenceable.
r is not required to be
incrementable and any copies of
the previous value of r are no
longer required to be dereferenceable
or incrementable.
post: r is dereferenceable
or r is past-the-endincrementable.
r++ convertible to const X& { X tmp = r;
++r;
return tmp; }Remark: After this operation
r is not required to be
dereferenceable.
post: r is incrementable.*r++ = o result is not used { *r = o; ++r; } Remark: After this operation
r is not required to be
dereferenceable.
post: r is incrementable.
Modify the column contents of Table 109 — "Forward iterator requirements", 24.2.5 [forward.iterators], as indicated [Rationale: Since the return type of the expression *r++ is now guaranteed to be type reference, the implied operational semantics from input iterator based on value copies is wrong — end rationale]
Table 109 — Forward iterator requirements (in addition to input iterator) Expression Return type Operational semantics Assertion/note
pre-/post-conditionr++ convertible to const X& { X tmp = r;
++r;
return tmp; }*r++ reference { reference tmp = *r;
++r;
return tmp; }
Modify the column contents of Table 110 — "Bidirectional iterator requirements", 24.2.6 [bidirectional.iterators], as indicated:
Table 110 — Bidirectional iterator requirements (in addition to forward iterator) Expression Return type Operational semantics Assertion/note
pre-/post-condition--r X& pre: there exists s such that
r == ++s.
post: r isdereferenceableincrementable.
--(++r) == r.
--r == --s implies r == s.
&r == &--r.r-- convertible to const X& { X tmp = r;
--r;
return tmp; }*r-- reference { reference tmp = *r;
--r;
return tmp; }
Section: 24.2.4 [output.iterators] Status: Open Submitter: Pete Becker Opened: 2011-02-27 Last modified: 2017-02-03
Priority: 3
View other active issues in [output.iterators].
View all other issues in [output.iterators].
View all issues with Open status.
Discussion:
In comp.lang.c++, Vicente Botet raises the following questions:
"In "24.2.4 Output iterators" there are 3 uses of incrementable. I've not found the definition. Could some one point me where it is defined?
Something similar occurs with dereferenceable. While the definition is given in "24.2.1 In general" it is used several times before. Shouldn't these definitions be moved to some previous section?"
He's right: both terms are used without being properly defined.
There is no definition of "incrementable". While there is a definition of "dereferenceable", it is, in fact, a definition of "dereferenceable iterator". "dereferenceable" is used throughout Clause 23 (Containers) before its definition in Clause 24. In almost all cases it's referring to iterators, but in 17.5.3.2 [swappable.requirements] there is a mention of "dereferenceable object"; in 17.5.3.5 [allocator.requirements] the table of Descriptive variable definitions refers to a "dereferenceable pointer"; 20.10.3.2 [pointer.traits.functions] refers to a "dereferenceable pointer"; in 22.4.5.1.2 [locale.time.get.virtuals]/11 (do_get) there is a requirement that a pointer "shall be dereferenceable". In those specific cases it is not defined.[2011-03-02: Daniel comments:]
I believe that the currently proposed resolution of issue 2035 solves this issue as well.
[ 2011 Bloomington ]
Agree with Daniel, this will be handled by the resolution of 2035.
Proposed resolution:
Section: 20.15 [meta] Status: LEWG Submitter: Daniel Krügler Opened: 2011-03-03 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [meta].
View all other issues in [meta].
View all issues with LEWG status.
Discussion:
When n3142 was suggested, it concentrated on constructions, assignments, and destructions, but overlooked to complement the single remaining compiler-support trait
template <class From, class To> struct is_convertible;
with the no-throw and triviality related aspects as it had been done with the other expression-based traits. Specifically, the current specification misses to add the following traits:
template <class From, class To> struct is_nothrow_convertible; template <class From, class To> struct is_trivially_convertible;
In particular the lack of is_nothrow_convertible is severly restricting. This was recently recognized when the proposal for decay_copy was prepared by n3255. There does not exist a portable means to define the correct conditional noexcept specification for the decay_copy function template, which is declared as:
template <class T> typename decay<T>::type decay_copy(T&& v) noexcept(???);
The semantics of decay_copy bases on an implicit conversion which again influences the overload set of functions that are viable here. In most circumstances this will have the same effect as comparing against the trait std::is_nothrow_move_constructible, but there is no guarantee for that being the right answer. It is possible to construct examples, where this would lead to the false result, e.g.
struct S {
S(const S&) noexcept(false);
template<class T>
explicit S(T&&) noexcept(true);
};
std::is_nothrow_move_constructible will properly honor the explicit template constructor because of the direct-initialization context which is part of the std::is_constructible definition and will in this case select it, such that std::is_nothrow_move_constructible<S>::value == true, but if we had the traits is_nothrow_convertible, is_nothrow_convertible<S, S>::value would evaluate to false, because it would use the copy-initialization context that is part of the is_convertible definition, excluding any explicit constructors and giving the opposite result.
The decay_copy example is surely not one of the most convincing examples, but is_nothrow_convertible has several use-cases, and can e.g. be used to express whether calling the following implicit conversion function could throw an exception or not:
template<class T, class U>
T implicit_cast(U&& u) noexcept(is_nothrow_convertible<U, T>::value)
{
return std::forward<U>(u);
}
Therefore I suggest to add the missing trait is_nothrow_convertible and for completeness also the missing trait is_trivially_convertible to 20.15 [meta].
[2011-03-24 Madrid meeting]
Daniel K: This is a new feature so out of scope.
Pablo: Any objections to moving 2040 to Open? No objections.[Bloomington, 2011]
Move to NAD Future, this would be an extension to existing functionality.
Proposed resolution:
Ammend the following declarations to the header <type_traits> synopsis in 20.15.2 [meta.type.synop]:
namespace std {
…
// 20.9.6, type relations:
template <class T, class U> struct is_same;
template <class Base, class Derived> struct is_base_of;
template <class From, class To> struct is_convertible;
template <class From, class To> struct is_trivially_convertible;
template <class From, class To> struct is_nothrow_convertible;
…
}
Modify Table 51 — "Type relationship predicates" as indicated. The removal of the remaining traces of the trait is_explicitly_convertible is an editorial step, it was removed by n3047:
Table 51 — Type relationship predicates Template Condition Comments … template <class From, class To>
struct is_convertible;see below From and To shall be complete
types, arrays of unknown bound, or
(possibly cv-qualified) void
types.template <class From, class To>
struct is_explicitly_convertible;is_constructible<To, From>::valuea synonym for a two-argument
version of is_constructible.
An implementation may define it
as an alias template.template <class From, class To>
struct is_trivially_convertible;is_convertible<From,
To>::value is true and the
conversion, as defined by
is_convertible, is known
to call no operation that is
not trivial ([basic.types], [special]).From and To shall be complete
types, arrays of unknown bound,
or (possibly cv-qualified) void
types.template <class From, class To>
struct is_nothrow_convertible;is_convertible<From,
To>::value is true and the
conversion, as defined by
is_convertible, is known
not to throw any
exceptions ([expr.unary.noexcept]).From and To shall be complete
types, arrays of unknown bound,
or (possibly cv-qualified) void
types.…
Section: 26.8 [numeric.ops] Status: Open Submitter: Chris Jefferson Opened: 2011-01-01 Last modified: 2017-06-15
Priority: 3
View all other issues in [numeric.ops].
View all issues with Open status.
Discussion:
The C++0x draft says std::accumulate uses: acc = binary_op(acc, *i).
Eelis van der Weegen has pointed out, on the libstdc++ mailing list, that using acc = binary_op(std::move(acc), *i) can lead to massive improvements (particularly, it means accumulating strings is linear rather than quadratic). Consider the simple case, accumulating a bunch of strings of length 1 (the same argument holds for other length buffers). For strings s and t, s+t takes time length(s)+length(t), as you have to copy both s and t into a new buffer. So in accumulating n strings, step i adds a string of length i-1 to a string of length 1, so takes time i. Therefore the total time taken is: 1+2+3+...+n = O(n2) std::move(s)+t, for a "good" implementation, is amortized time length(t), like vector, just copy t onto the end of the buffer. So the total time taken is: 1+1+1+...+1 (n times) = O(n). This is the same as push_back on a vector. I'm trying to decide if this implementation might already be allowed. I suspect it might not be (although I can't imagine any sensible code it would break). There are other algorithms which could benefit similarly (inner_product, partial_sum and adjacent_difference are the most obvious). Is there any general wording for "you can use rvalues of temporaries"? The reflector discussion starting with message c++std-lib-29763 came to the conclusion that above example is not covered by the "as-if" rules and that enabling this behaviour would seem quite useful.[ 2011 Bloomington ]
Moved to NAD Future. This would be a larger change than we would consider for a simple TC.
[2017-02 in Kona, LEWG responds]
Like the goal.
What is broken by adding std::move() on the non-binary-op version?
A different overload might be selected, and that would be a breakage. Is it breakage that we should care about?
We need to encourage value semantics.
Need a paper. What guidance do we give?
Use std::reduce() (uses generalized sum) instead of accumulate which doesn’t suffer it.
Inner_product and adjacent_difference also. adjacent_difference solves it half-way for “val” object, but misses the opportunity for passing acc as std::move(acc).
[2017-06-02 Issues Telecon]
Ville to encourage Eelis to write a paper on the algorithms in <numeric>, not just for accumulate.
Howard pointed out that this has already been done for the algorithms in <algorithm>
Status to Open; Priority 3
Proposed resolution:
Section: 20.11.2.2.6 [util.smartptr.shared.create] Status: Open Submitter: Jonathan Wakely Opened: 2011-07-11 Last modified: 2017-02-03
Priority: 2
View all other issues in [util.smartptr.shared.create].
View all issues with Open status.
Discussion:
20.11.2.2.6 [util.smartptr.shared.create] says:
-2- Effects: Allocates memory suitable for an object of type T and constructs an object in that memory via the placement new expression ::new (pv) T(std::forward<Args>(args)...). The template allocate_shared uses a copy of a to allocate memory. If an exception is thrown, the functions have no effect.
This explicitly requires placement new rather than using allocator_traits<A>::construct(a, (T*)pv, std::forward<Args>(args)...) In most cases that would result in the same placement new expression, but would allow more control over how the object is constructed e.g. using scoped_allocator_adaptor to do uses-allocator construction, or using an allocator declared as a friend to construct objects with no public constructors.
[2011-08-16 Bloomington:]
Agreed to fix in principle, but believe that make_shared and allocate_shared have now diverged enough that their descriptions should be separated. Pablo and Stefanus to provide revised wording.
Daniel's (old) proposed resolution:
This wording is relative to the FDIS.
Change the following paragraphs of 20.11.2.2.6 [util.smartptr.shared.create] as indicated (The suggested removal of the last sentence of p1 is not strictly required to resolve this issue, but is still recommended, because it does not say anything new but may give the impression that it says something new):
template<class T, class... Args> shared_ptr<T> make_shared(Args&&... args); template<class T, class A, class... Args> shared_ptr<T> allocate_shared(const A& a, Args&&... args);-1- Requires: For the template make_shared, t
-2- Effects: Allocates memory suitable for an object of type T and constructs an object in that memory. The template make_shared constructs the object via the placement new expression ::new (pv) T(std::forward<Args>(args)...). The template allocate_shared uses a copy of a to allocate memory and constructs the object by calling allocator_traits<A>::construct(a, pt, std::forward<Args>(args)...). If an exception is thrown, the functions have no effect. -3- Returns: A shared_ptr instance that stores and owns the address of the newly constructed object of type T. -4- Postconditions: get() != 0 && use_count() == 1 -5- Throws: bad_alloc, or, for the template make_shared, an exception thrown from the constructor of T, or, for the template allocate_shared, an exception thrown from A::allocate or from allocator_traits<A>::constructThe expression ::new (pv) T(std::forward<Args>(args)...), where pv has type void* and points to storage suitable to hold an object of type T, shall be well formed. For the template allocate_shared, the expression allocator_traits<A>::construct(a, pt, std::forward<Args>(args)...), where pt has type T* and points to storage suitable to hold an object of type T, shall be well formed. A shall be an allocator ([allocator.requirements]).The copy constructor and destructor of A shall not throw exceptions.from the constructor of T. -6- Remarks: Implementations are encouraged, but not required, to perform no more than one memory allocation. [ Note: This provides efficiency equivalent to an intrusive smart pointer. — end note ] -7- [ Note: These functions will typically allocate more memory than sizeof(T) to allow for internal bookkeeping structures such as the reference counts. — end note ]
[2011-12-04: Jonathan and Daniel improve wording]
See also c++std-lib-31796
[2013-10-13, Ville]
This issue is related to 2089.
[2014-02-15 post-Issaquah session : move to Tentatively NAD]
STL: This takes an allocator, but then ignores its construct. That's squirrely.
Alisdair: The convention is when you take an allocator, you use its construct.
STL: 23.2.1 [container.requirements.general]/3, argh! This fills me with despair, but I understand it now.
STL: Ok, this is some cleanup.
STL: You're requiring b to be of type A and not being rebound, is that an overspecification?
Pablo: Good point. Hmm, that's only a requirement on what must be well-formed.
STL: If it's just a well-formed requirement, then why not just use a directly?
Pablo: Yeah, the well-formed requirement is overly complex. It's not a real call, we could just use a directly. It makes it harder to read.
Alisdair: b should be an allocator in the same family as a.
Pablo: This is a well-formed requirement, I wonder if it's the capital A that's the problem here. It doesn't matter here, this is way too much wording.
Alisdair: It's trying to tie the constructor arguments into the allocator requirements.
Pablo: b could be struck, that's a runtime quality. The construct will work with anything that's in the family of A.
Alisdair: The important part is the forward of Args.
Pablo: A must be an allocator, and forward Args must work with that.
Alisdair: First let's nail down A.
Pablo: Then replace b with a, and strike the rest.
STL: You need pt's type, at least.
Pablo: There's nothing to be said about runtime constraints here, this function doesn't even take a pt.
STL: Looking at the Effects, I believe b is similarly messed up, we can use a2 to construct an object.
Alisdair: Or any allocator in the family of a.
STL: We say this stuff for the deallocate too, it should be lifted up.
STL: "owns the address" is weird.
Alisdair: shared_ptr owns pointers, although it does sound funky.
Walter: "to destruct" is ungrammatical.
STL: "When ownership is given up" is not what we usually say.
Alisdair: I think the Returns clause is the right place to say this.
STL: The right place to say this is shared_ptr's dtor, we don't want to use Core's "come from" convention.
Alisdair: I'm on the hook to draft cleaner wording.
[2015-10, Kona Saturday afternoon]
AM: I was going to clean up the wording, but haven't done it yet.
Defer until we have new wording.
[2016-03, Jacksonville]
Alisdair: we need to figure out whether we should call construct or not; major implementation divergence
STL: this does not grant friendship, does it?
Jonathan: some people want it.
Thomas: scoped allocator adapter should be supported, so placement new doesn't work
Alisdair: this makes the make_ functions impossible
Thomas: you don't want to use those though.
Alisdair: but people use that today, at Bloomberg
Alisdair: and what do we do about fancy pointers?
Jonathan: we constrain it to only non-fancy pointers.
STL: shared_ptr has never attempted to support fancy pointers; seems like a paper is needed.
Poll: call construct:6 operator new: 0 don't care: 4
Poll: should we support fancy pointers? Yes: 1 No: 4 don't care: 4
STL: 20.8.2.2.6p2: 'and pv->~T()' is bogus for void
STL: 20.8.2.2.6p4: is this true even if we're going to allocate a bit more?
Alisdair: yes
Alisdair: coming up with new wording
[2016-08, Chicago Monday PM]
Alisdair to provide new wording this week
Proposed resolution:
This wording is relative to the FDIS.
Change the following paragraphs of 20.11.2.2.6 [util.smartptr.shared.create] as indicated:
template<class T, class... Args> shared_ptr<T> make_shared(Args&&... args);template<class T, class A, class... Args> shared_ptr<T> allocate_shared(const A& a, Args&&... args);
-1- Requires: The expression ::new (pv) T(std::forward<Args>(args)...), where pv
has type void* and points to storage suitable to hold an object of type T, shall be well
formed. A shall be an allocator (17.5.3.5 [allocator.requirements]). The copy constructor
and destructor of A shall not throw exceptions.
return allocate_shared<T>(allocator<T>(), std::forward<Args>(args)...);
Allocates memory suitable for an object of type T
and constructs an object in that memory via the placement new expression
::new (pv) T(std::forward<Args>(args)...). The template allocate_shared uses a copy
of a to allocate memory. If an exception is thrown, the functions have no effect.
Add the following set of new paragraphs immediately following the previous paragraph 7 of 20.11.2.2.6 [util.smartptr.shared.create]:
template<class T, class A, class... Args> shared_ptr<T> allocate_shared(const A& a, Args&&... args);
-?- Requires: The expressions allocator_traits<A>::construct(b, pt, std::forward<Args>(args)...) and allocator_traits<A>::destroy(b, pt) shall be well-formed and well-defined, where b has type A and is a copy of a and where pt has type T* and points to storage suitable to hold an object of type T. A shall meet the allocator requirements (17.5.3.5 [allocator.requirements]).
-?- Effects: Uses an object a2 of type allocator_traits<A>::rebind_alloc<unspecified> that compares equal to a to allocate memory suitable for an object of type T. Uses a copy b of type A from a to construct an object of type T in that memory by calling allocator_traits<A>::construct(b, pt, std::forward<Args>(args)...). If an exception is thrown, the function has no effect. -?- Returns: A shared_ptr instance that stores and owns the address of the newly constructed object of type T. When ownership is given up, the effects are as follows: Uses a copy b2 of type A from a to destruct an object of type T by calling allocator_traits<A>::destroy(b2, pt2) where pt2 has type T* and refers to the newly constructed object. Then uses an object of type allocator_traits<A>::rebind_alloc<unspecified> that compares equal to a to deallocate the allocated memory. -?- Postconditions: get() != 0 && use_count() == 1 -?- Throws: Nothing unless memory allocation or allocator_traits<A>::construct throws an exception. -?- Remarks: Implementations are encouraged, but not required, to perform no more than one memory allocation. [Note: Such an implementation provides efficiency equivalent to an intrusive smart pointer. — end note] -?- [Note: This function will typically allocate more memory than sizeof(T) to allow for internal bookkeeping structures such as the reference counts. — end note]Section: 20.15.4.3 [meta.unary.prop] Status: Open Submitter: Daniel Krügler Opened: 2011-08-20 Last modified: 2017-02-03
Priority: 3
View other active issues in [meta.unary.prop].
View all other issues in [meta.unary.prop].
View all issues with Open status.
Discussion:
The currently agreed on proposed wording for 2015 using remove_all_extents<T>::type instead of the "an array of unknown bound" terminology in the precondition should be extended to some further entries especially in Table 49, notably the is_*constructible, is_*assignable, and is_*destructible entries. To prevent ODR violations, incomplete element types of arrays must be excluded for value-initialization and destruction for example. Construction and assignment has to be honored, when we have array-to-pointer conversions or pointer conversions of incomplete pointees in effect.
[2012, Kona]
The issue is that in three type traits, we are accidentally saying that in certain circumstances the type must give a specified answer when given an incomplete type. (Specifically: an array of unknown bound of incomplete type.) The issue asserts that there's an ODR violation, since the trait returns false in that case but might return a different version when the trait is completed.
Howard argues: no, there is no risk of an ODR violation. is_constructible<A[]> must return false regardless of whether A is complete, so there's no reason to forbid an array of unknown bound of incomplete types. Same argument applies to is_assignable. General agreement with Howard's reasoning.
There may be a real issue for is_destructible. None of us are sure what is_destructible is supposed to mean for an array of unknown bound (regardless of whether its type is complete), and the standard doesn't make it clear. The middle column doesn't say what it's supposed to do for incomplete types.
In at least one implementation, is_destructible<A[]> does return true if A is complete, which would result in ODR violation unless we forbid it for incomplete types.
Move to open. We believe there is no issue for is_constructible or is_assignable, but that there is a real issue for is_destructible.
Proposed resolution:
Section: 18.8.4 [exception.terminate] Status: Open Submitter: Daniel Krügler Opened: 2011-09-25 Last modified: 2017-02-03
Priority: 3
View all issues with Open status.
Discussion:
Andrzej Krzemienski reported the following on comp.std.c++:
In N3290, which is to become the official standard, in 18.8.4.4 [terminate], paragraph 1 reads
Remarks: Called by the implementation when exception handling must be abandoned for any of several reasons (15.5.1), in effect immediately after evaluating the throw-expression (18.8.3.1). May also be called directly by the program.
It is not clear what is "in effect". It was clear in previous drafts where paragraphs 1 and 2 read:
Called by the implementation when exception handling must be abandoned for any of several reasons (15.5.1). May also be called directly by the program.
Effects: Calls the terminate_handler function in effect immediately after evaluating the throw-expression (18.8.3.1), if called by the implementation, or calls the current terminate_handler function, if called by the program.It was changed by N3189. The same applies to function unexpected (D. 11.4, paragraph 1).
Assuming the previous wording is still intended, the wording can be read "unless std::terminate is called by the program, we will use the handler that was in effect immediately after evaluating the throw-expression". This assumes that there is some throw-expression connected to every situation that triggers the call to std::terminate. But this is not the case:
- In case std::thread is assigned to or destroyed while being joinable there is no throw-expression involved.
- In case std::unexpected is called by the program, std::terminate is triggered by the implementation - no throw-expression involved.
- In case a destructor throws during stack unwinding we have two throw-expressions involved.
Which one is referred to?
In case std::nested_exception::rethrow_nested is called for an object that has captured no exception, there is no throw-expression involved directly (and may no throw be involved even indirectly). Next, 18.8.4.1 [terminate.handler], paragraph 2 saysRequired behavior: A terminate_handler shall terminate execution of the program without returning to the caller.
This seems to allow that the function may exit by throwing an exception (because word "return" implies a normal return).
One could argue that words "terminate execution of the program" are sufficient, but then why "without returning to the caller" would be mentioned. In case such handler throws, noexcept specification in function std::terminate is violated, and std::terminate would be called recursively - should std::abort not be called in case of recursive std::terminate call? On the other hand some controlled recursion could be useful, like in the following technique.
The here mentioned wording changes by N3189 in regard to 18.8.4.4 [terminate] p1 were done for a better separation of effects (Effects element) and additional normative wording explanations (Remarks element), there was no meaning change intended. Further, there was already a defect existing in the previous wording, which was not updated when further situations where defined, when std::terminate where supposed to be called by the implementation.
The part "in effect immediately after evaluating the throw-expression" should be removed and the quoted reference to 18.8.4.1 [terminate.handler] need to be part of the effects element where it refers to the current terminate_handler function, so should be moved just after "Effects: Calls the current terminate_handler function." It seems ok to allow a termination handler to exit via an exception, but the suggested idiom should better be replaced by a more simpler one based on evaluating the current exception pointer in the terminate handler, e.g.
void our_terminate (void) {
std::exception_ptr p = std::current_exception();
if (p) {
... // OK to rethrow and to determine it's nature
} else {
... // Do something else
}
}
[2011-12-09: Daniel comments]
A related issue is 2111.
[2012, Kona]
Move to Open.
There is an interaction with Core issues in this area that Jens is already supplying wording for. Review this issue again once Jens wording is available.
Alisdair to review clause 15.5 (per Jens suggestion) and recommend any changes, then integrate Jens wording into this issue.
Proposed resolution:
Section: 20.10.9.1 [allocator.members] Status: EWG Submitter: David Krauss Opened: 2011-10-07 Last modified: 2017-02-03
Priority: 2
View all other issues in [allocator.members].
View all issues with EWG status.
Discussion:
When the EmplaceConstructible (23.2.1 [container.requirements.general]/13) requirement is used to initialize an object, direct-initialization occurs. Initializing an aggregate or using a std::initializer_list constructor with emplace requires naming the initialized type and moving a temporary. This is a result of std::allocator::construct using direct-initialization, not list-initialization (sometimes called "uniform initialization") syntax.
Altering std::allocator<T>::construct to use list-initialization would, among other things, give preference to std::initializer_list constructor overloads, breaking valid code in an unintuitive and unfixable way — there would be no way for emplace_back to access a constructor preempted by std::initializer_list without essentially reimplementing push_back.
std::vector<std::vector<int>> v;
v.emplace_back(3, 4); // v[0] == {4, 4, 4}, not {3, 4} as in list-initialization
The proposed compromise is to use SFINAE with std::is_constructible, which tests whether direct-initialization is well formed. If is_constructible is false, then an alternative std::allocator::construct overload is chosen which uses list-initialization. Since list-initialization always falls back on direct-initialization, the user will see diagnostic messages as if list-initialization (uniform-initialization) were always being used, because the direct-initialization overload cannot fail.
I can see two corner cases that expose gaps in this scheme. One occurs when arguments intended for std::initializer_list satisfy a constructor, such as trying to emplace-insert a value of {3, 4} in the above example. The workaround is to explicitly specify the std::initializer_list type, as in v.emplace_back(std::initializer_list<int>(3, 4)). Since this matches the semantics as if std::initializer_list were deduced, there seems to be no real problem here. The other case is when arguments intended for aggregate initialization satisfy a constructor. Since aggregates cannot have user-defined constructors, this requires that the first nonstatic data member of the aggregate be implicitly convertible from the aggregate type, and that the initializer list have one element. The workaround is to supply an initializer for the second member. It remains impossible to in-place construct an aggregate with only one nonstatic data member by conversion from a type convertible to the aggregate's own type. This seems like an acceptably small hole. The change is quite small because EmplaceConstructible is defined in terms of whatever allocator is specified, and there is no need to explicitly mention SFINAE in the normative text.[2012, Kona]
Move to Open.
There appears to be a real concern with initializing aggregates, that can be performed only using brace-initialization. There is little interest in the rest of the issue, given the existence of 'emplace' methods in C++11.
Move to Open, to find an acceptable solution for intializing aggregates. There is the potential that EWG may have an interest in this area of language consistency as well.
[2013-10-13, Ville]
This issue is related to 2070.
[2015-02 Cologne]
Move to EWG, Ville to write a paper.
[2015-09, Telecon]
Ville: N4462 reviewed in Lenexa. EWG discussion to continue in Kona.
[2016-08 Chicago]
See N4462
The notes in Lenexa say that Marshall & Jonathan volunteered to write a paper on this
Proposed resolution:
This wording is relative to the FDIS.
Change 20.10.9.1 [allocator.members] p12 as indicated:
template <class U, class... Args> void construct(U* p, Args&&... args);12 Effects: ::new((void *)p) U(std::forward<Args>(args)...) if is_constructible<U, Args...>::value is true, else ::new((void *)p) U{std::forward<Args>(args)...}
Section: 30.6.6 [futures.promise], 30.6.10 [futures.task] Status: LEWG Submitter: Jonathan Wakely Opened: 2011-11-01 Last modified: 2017-02-03
Priority: 4
View other active issues in [futures.promise].
View all other issues in [futures.promise].
View all issues with LEWG status.
Discussion:
This example is ill-formed according to C++11 because uses_allocator<promise<R>, A>::value is true, but is_constructible<promise<R>, A, promise<R>&&>::value is false. Similarly for packaged_task.
#include <future>
#include <memory>
#include <tuple>
using namespace std;
typedef packaged_task<void()> task;
typedef promise<void> prom;
allocator<task> a;
tuple<task, prom> t1{ allocator_arg, a };
tuple<task, prom> t2{ allocator_arg, a, task{}, prom{} };
[2012, Portland]
This is an allocator issue, and should be dealt with directly by LWG.
[2013-03-06]
Jonathan suggests to make the new constructors non-explicit and makes some representational improvements.
[2013-09 Chicago]
Move to deferred.
This issue has much in common with similar problems with std::function that are being addressed by the polymorphic allocators proposal currently under evaluation in LEWG. Defer further discussion on this topic until the final outcome of that paper and its proposed resolution is known.
[2014-02-20 Re-open Deferred issues as Priority 4]
[2016-08 Chicago]
Fri PM: Send to LEWG - and this also applies to function in LFTS.
Proposed resolution:
[This wording is relative to the FDIS.]
Add to 30.6.6 [futures.promise], class template promise synopsis, as indicated:
namespace std {
template <class R>
class promise {
public:
promise();
template <class Allocator>
promise(allocator_arg_t, const Allocator& a);
template <class Allocator>
promise(allocator_arg_t, const Allocator& a, promise&& rhs) noexcept;
promise(promise&& rhs) noexcept;
promise(const promise& rhs) = delete;
~promise();
[…]
};
[…]
}
Change 30.6.6 [futures.promise] as indicated:
promise(promise&& rhs) noexcept; template <class Allocator> promise(allocator_arg_t, const Allocator& a, promise&& rhs) noexcept;-5- Effects: constructs a new promise object and transfers ownership of the shared state of rhs (if any) to the newly-constructed object.
-6- Postcondition: rhs has no shared state. -?- [Note: a is not used — end note]
Add to 30.6.10 [futures.task], class template packaged_task synopsis, as indicated:
namespace std {
template<class> class packaged_task; // undefined
template<class R, class... ArgTypes>
class packaged_task<R(ArgTypes...)> {
public:
// construction and destruction
packaged_task() noexcept;
template <class Allocator>
packaged_task(allocator_arg_t, const Allocator& a) noexcept;
template <class F>
explicit packaged_task(F&& f);
template <class F, class Allocator>
explicit packaged_task(allocator_arg_t, const Allocator& a, F&& f);
~packaged_task();
// no copy
packaged_task(const packaged_task&) = delete;
template<class Allocator>
packaged_task(allocator_arg_t, const Allocator& a, const packaged_task&) = delete;
packaged_task& operator=(const packaged_task&) = delete;
// move support
packaged_task(packaged_task&& rhs) noexcept;
template <class Allocator>
packaged_task(allocator_arg_t, const Allocator& a, packaged_task&& rhs) noexcept;
packaged_task& operator=(packaged_task&& rhs) noexcept;
void swap(packaged_task& other) noexcept;
[…]
};
[…]
}
Change 30.6.10.1 [futures.task.members] as indicated:
packaged_task() noexcept; template <class Allocator> packaged_task(allocator_arg_t, const Allocator& a) noexcept;-1- Effects: constructs a packaged_task object with no shared state and no stored task.
-?- [Note: a is not used — end note]
[…]
packaged_task(packaged_task&& rhs) noexcept; template <class Allocator> packaged_task(allocator_arg_t, const Allocator& a, packaged_task&& rhs) noexcept;-5- Effects: constructs a new packaged_task object and transfers ownership of rhs's shared state to *this, leaving rhs with no shared state. Moves the stored task from rhs to *this.
-6- Postcondition: rhs has no shared state. -?- [Note: a is not used — end note]
Section: 17.5.3.3 [nullablepointer.requirements], 24.2.3 [input.iterators], 24.2.7 [random.access.iterators], 25.1 [algorithms.general], 25.7 [alg.sorting], 30.2.1 [thread.req.paramname] Status: Open Submitter: Daniel Krügler Opened: 2011-12-09 Last modified: 2017-02-03
Priority: 3
View all issues with Open status.
Discussion:
As of 17.5.3.1 [utility.arg.requirements] Table 17/18, the return types of the expressions
a == b
or
a < b
for types satisfying the EqualityComparable or LessThanComparable types, respectively, are required to be "convertible to bool" which corresponds to a copy-initialization context. But several newer parts of the library that refer to such contexts have lowered the requirements taking advantage of the new terminology of "contextually convertible to bool" instead, which corresponds to a direct-initialization context (In addition to "normal" direct-initialization constructions, operands of logical operations as well as if or switch conditions also belong to this special context).
One example for these new requirements are input iterators which satisfy EqualityComparable but also specify that the expressiona != b
shall be just "contextually convertible to bool". The same discrepancy exists for requirement set NullablePointer in regard to several equality-related expressions.
For random access iterators we havea < b contextually convertible to bool
as well as for all derived comparison functions, so strictly speaking we could have a random access iterator that does not satisfy the LessThanComparable requirements, which looks like an artifact to me.
Even if we keep with the existing requirements based on LessThanComparable or EqualityComparable we still would have the problem that some current specifications are actually based on the assumption of implicit convertibility instead of "explicit convertibility", e.g. 20.11.1.5 [unique.ptr.special] p3:template <class T1, class D1, class T2, class D2> bool operator!=(const unique_ptr<T1, D1>& x, const unique_ptr<T2, D2>& y);-3- Returns: x.get() != y.get().
Similar examples exist in 20.11.1.2.2 [unique.ptr.single.dtor] p2, 20.11.1.2.3 [unique.ptr.single.asgn] p9, 20.11.1.2.4 [unique.ptr.single.observers] p1+3+8, etc.
In all these places the expressions involving comparison functions (but not those of the conversion of a NullablePointer to bool!) assume to be "convertible to bool". I think this is a very natural assumption and all delegations of the comparison functions of some type X to some other API type Y in third-party code does so assuming that copy-initialization semantics will just work. The actual reason for using the newer terminology can be rooted back to LWG 556. My hypotheses is that the resolution of that issue also needs a slight correction. Why so? The reason for opening that issue were worries based on the previous "convertible to bool" wording. An expressions like "!pred(a, b)" might not be well-formed in those situations, because operator! might not be accessible or might have an unusual semantics (and similarly for other logical operations). This can indeed happen with unusual proxy return types, so the idea was that the evaluation of Predicate, BinaryPredicate (25.1 [algorithms.general] p8+9), and Compare (25.7 [alg.sorting] p2) should be defined based on contextual conversion to bool. Unfortunately this alone is not sufficient: In addition, I think, we also want the predicates to be (implicitly) convertible to bool! Without this wording, several conditions are plain wrong, e.g. 25.5.5 [alg.find] p2, which talks about "pred(*i) != false" (find_if) and "pred(*i) == false" (find_if_not). These expressions are not within a boolean context! While we could simply fix all these places by proper wording to be considered in a "contextual conversion to bool", I think that this is not the correct solution: Many third-party libraries already refer to the previous C++03 Predicate definition — it actually predates C++98 and is as old as the SGI specification. It seems to be a high price to pay to switch to direct initialization here instead of fixing a completely different specification problem. A final observation is that we have another definition for a Predicate in 30.2.1 [thread.req.paramname] p2:If a parameter is Predicate, operator() applied to the actual template argument shall return a value that is convertible to bool.
The problem here is not that we have two different definitions of Predicate in the standard — this is confusing, but this fact alone is not a defect. The first (minor) problem is that this definition does not properly apply to function objects that are function pointers, because operator() is not defined in a strict sense. But the actually worse second problem is that this wording has the very same problem that has originally lead to LWG 556! We only need to look at 30.5.3 [thread.condition.condvar] p15 to recognice this:
while (!pred()) wait(lock);
The negation expression here looks very familiar to the example provided in LWG 556 and is sensitive to the same "unusual proxy" problem. Changing the 30.2.1 [thread.req.paramname] wording to a corresponding "contextual conversion to bool" wouldn't work either, because existing specifications rely on "convertible to bool", e.g. 30.5.3 [thread.condition.condvar] p32+33+42 or 30.5.4 [thread.condition.condvarany] p25+26+32+33.
To summarize: I believe that LWG 556 was not completely resolved. A pessimistic interpretation is, that even with the current wording based on "contextually convertible to bool" the actual problem of that issue has not been fixed. What actually needs to be required here is some normative wording that basically expresses something along the lines of:The semantics of any contextual conversion to bool shall be equivalent to the semantics of any implicit conversion to bool.
This is still not complete without having concepts, but it seems to be a better approximation. Another way of solving this issue would be to define a minimum requirements table with equivalent semantics. The proposed wording is a bit simpler but attempts to express the same thing.
[2012, Kona]
Agree with Daniel that we potentially broke some C++03 user code, accept the changes striking "contextually" from tables. Stefan to provide revised wording for section 25, and figure out changes to section 30.
Move to open, and then to Review when updated wording from Stefan is available.
[2012-10-12, STL comments]
The current proposed resolution still isn't completely satisfying. It would certainly be possible for the Standard to require these various expressions to be implicitly and contextually convertible to bool, but that would have a subtle consequence (which, I will argue, is undesirable - regardless of the fact that it dates all the way back to C++98/03). It would allow users to provide really wacky types to the Standard Library, with one of two effects:
Standard Library implementations would have to go to great lengths to respect such wacky types, essentially using static_cast<bool> when invoking any predicates or comparators.
Otherwise, such wacky types would be de facto nonportable, because they would make Standard Library implementations explode.
Effect B is the status quo we're living with today. What Standard Library implementations want to do with pred(args) goes beyond "if (pred(args))" (C++03), contextually converting pred(args) to bool (C++11), or implicitly and contextually converting pred(args) to bool (the current proposed resolution). Implementations want to say things like:
if (pred(args)) if (!pred(args)) if (cond && pred(args)) if (cond && !pred(args))
These are real examples taken from Dinkumware's implementation. There are others that would be realistic ("pred(args) && cond", "cond || pred(args)", etc.)
Although negation was mentioned in this issue's Discussion section, and in LWG 556's, the current proposed resolution doesn't fix this problem. Requiring pred(args) to be implicitly and contextually convertible to bool doesn't prevent operator!() from being overloaded and returning std::string (as a wacky example). More ominously, it doesn't prevent operator&&() and operator||() from being overloaded and destroying short-circuiting.I would like LWG input before working on Standardese for a new proposed resolution. Here's an outline of what I'd like to do:
Introduce a new "concept" in 17.5.3 [utility.requirements], which I would call BooleanTestable in the absence of better ideas.
Centralize things and reduce verbosity by having everything simply refer to BooleanTestable when necessary. I believe that the tables could say "Return type: BooleanTestable", while Predicate/BinaryPredicate/Compare would need the incantation "shall satisfy the requirements of BooleanTestable".
Resolve the tug-of-war between users (who occasionally want to do weird things) and implementers (who don't want to have to contort their code) by requiring that:
Given a BooleanTestable x, x is both implicitly and contextually convertible to bool.
Given a BooleanTestable x, !x is BooleanTestable. (This is intentionally "recursive".)
Given a BooleanTestable x, bool t = x, t2(x), f = !x; has the postcondition t == t2 && t != f.
Given a BooleanTestable x and a BooleanTestable y of possibly different types, "x && y" and "x || y" invoke the built-in operator&&() and operator||(), triggering short-circuiting.
bool is BooleanTestable.
I believe that this simultaneously gives users great latitude to use types other than bool, while allowing implementers to write reasonable code in order to get their jobs done. (If I'm forgetting anything that implementers would want to say, please let me know.)
About requirement (I): As Daniel patiently explained to me, we need to talk about both implicit conversions and contextual conversions, because it's possible for a devious type to have both "explicit operator bool()" and "operator int()", which might behave differently (or be deleted, etc.).
About requirement (IV): This is kind of tricky. What we'd like to say is, "BooleanTestable can't ever trigger an overloaded logical operator". However, given a perfectly reasonable type Nice - perhaps even bool itself! - other code (perhaps a third-party library) could overload operator&&(Nice, Evil). Therefore, I believe that the requirement should be "no first use" - the Standard Library will ask for various BooleanTestable types from users (for example, the result of "first != last" and the result of "pred(args)"), and as long as they don't trigger overloaded logical operators with each other, everything is awesome.
About requirement (V): This is possibly redundant, but it's trivial to specify, makes it easier for users to understand what they need to do ("oh, I can always achieve this with bool"), and provides a "base case" for requirement (IV) that may or may not be necessary. Since bool is BooleanTestable, overloading operator&&(bool, Other) (etc.) clearly makes the Other type non-BooleanTestable.
This wording is relative to the FDIS.
Change Table 25 — "NullablePointer requirements" in 17.5.3.3 [nullablepointer.requirements] as indicated:
Table 25 — NullablePointer requirements Expression Return type Operational semantics […] a != b contextuallyconvertible to bool!(a == b) a == np
np == acontextuallyconvertible to boola == P() a != np
np != acontextuallyconvertible to bool!(a == np) Change Table 107 — "Input iterator requirements" in 24.2.3 [input.iterators] as indicated:
Table 107 — Input iterator requirements (in addition to Iterator) Expression Return type Operational semantics Assertion/note
pre-/post-conditiona != b contextuallyconvertible to bool!(a == b) pre: (a, b) is in the domain of ==. […] Change Table 111 — "Random access iterator requirements" in 24.2.7 [random.access.iterators] as indicated:
Table 111 — Random access iterator requirements (in addition to bidirectional iterator) Expression Return type Operational semantics Assertion/note
pre-/post-condition[…] a < b contextuallyconvertible to boolb - a > 0 < is a total ordering relation a > b contextuallyconvertible to boolb < a > is a total ordering relation opposite to <. a >= b contextuallyconvertible to bool!(a < b) a <= b contextuallyconvertible to bool!(a > b) Change 25.1 [algorithms.general] p8+9 as indicated:
-8- The Predicate parameter is used whenever an algorithm expects a function object (20.14 [function.objects]) that, when applied to the result of dereferencing the corresponding iterator, returns a value testable as true. In other words, if an algorithm takes Predicate pred as its argument and first as its iterator argument, it should work correctly in the construct pred(*first) implicitly or contextually converted to bool (Clause 4 [conv]). The function object pred shall not apply any non-constant function through the dereferenced iterator.
-9- The BinaryPredicate parameter is used whenever an algorithm expects a function object that when applied to the result of dereferencing two corresponding iterators or to dereferencing an iterator and type T when T is part of the signature returns a value testable as true. In other words, if an algorithm takes BinaryPredicate binary_pred as its argument and first1 and first2 as its iterator arguments, it should work correctly in the construct binary_pred(*first1, *first2) implicitly or contextually converted to bool (Clause 4 [conv]). BinaryPredicate always takes the first iterator's value_type as its first argument, that is, in those cases when T value is part of the signature, it should work correctly in the construct binary_pred(*first1, value) implicitly or contextually converted to bool (Clause 4 [conv]). binary_pred shall not apply any non-constant function through the dereferenced iterators.Change 25.7 [alg.sorting] p2 as indicated:
-2- Compare is a function object type (20.14 [function.objects]). The return value of the function call operation applied to an object of type Compare, when implicitly or contextually converted to bool (4 [conv]), yields true if the first argument of the call is less than the second, and false otherwise. Compare comp is used throughout for algorithms assuming an ordering relation. It is assumed that comp will not apply any non-constant function through the dereferenced iterator.
Change 30.2.1 [thread.req.paramname] p2 as indicated:
-2-
If a parameter is Predicate, operator() applied to the actual template argument shall return a value that is convertible to boolPredicate is a function object type (20.14 [function.objects]). The return value of the function call operation applied to an object of type Predicate, when implicitly or contextually converted to bool (4 [conv]), yields true if the corresponding test condition is satisfied, and false otherwise.
[2014-05-20, Daniel suggests concrete wording based on STL's proposal]
The presented wording follows relatively closely STL's outline with the following notable exceptions:
A reference to BooleanTestable in table "Return Type" specifications seemed very unusual to me and I found no "prior art" for this in the Standard. Instead I decided to follow the usual style to add a symbol with a specific meaning to a specific paragraph that specifies symbols and their meanings.
STL's requirement IV suggested to directly refer to built-in operators && and ||. In my opinion this concrete requirement isn't needed if we simply require that two BooleanTestable operands behave equivalently to two those operands after conversion to bool (each of them).
I couldn't find a good reason to require normatively that type bool meets the requirements of BooleanTestable: My assertion is that after having defined them, the result simply falls out of this. But to make this a bit clearer, I added also a non-normative note to these effects.
[2014-06-10, STL comments]
In the current wording I would like to see changed the suggested changes described by bullet #6:
In 23.2.1 [container.requirements.general] p4 undo the suggested change
Then change the 7 occurrences of "convertible to bool" in the denoted tables to "bool".
[2015-05-05 Lenexa]
STL: Alisdair wanted to do something here, but Daniel gave us updated wording.
[2015-07 Telecon]
Alisdair: Should specify we don't break short circuiting.
Ville: Looks already specified because that's the way it works for bool.
Geoffrey: Maybe add a note about the short circuiting.
B2/P2 is somewhat ambiguous. It implies that B has to be both implicitly convertible to bool and contextually convertible to bool.
We like this, just have nits.
Status stays Open.
Marshall to ping Daniel with feedback.
[2016-02-27, Daniel updates wording]
The revised wording has been updated from N3936 to N4567.
To satisfy the Kona 2015 committee comments, the wording in [booleantestable.requirements] has been improved to better separate the two different requirements of "can be contextually converted to bool" and "can be implicitly converted to bool. Both are necessary because it is possible to define a type that has the latter property but not the former, such as the following one:
2016-08-07, Daniel: The below example has been corrected to reduce confusion about the performed conversions as indicated by the delta markers:
using Bool = int;
struct OddBoolean
{
explicit operator bool() const = delete;
operator Bool() const;
OddBoolean(bool) = delete;
OddBoolean(Bool){}
} ob;
bool b2 = ob; // OK
bool b1(ob); // Error
OddBoolean b2 = true; // OK
OddBoolean b1(true); // Error
In [booleantestable.requirements] a note has been added to ensure that an implementation is not allowed to break any short-circuiting semantics.
I decided to separate LWG 2587/2588 from this issue. Both these issues aren't exactly the same but depending on the committee's position, their resolution might benefit from the new vocabulary introduced here.
Proposed resolution:
This wording is relative to N4567.
Change 17.5.3.1 [utility.arg.requirements] p1, Table 17 — "EqualityComparable requirements", and Table 18 — "LessThanComparable requirements" as indicated:
-1- […] In these tables, T is an object or reference type to be supplied by a C++ program instantiating a template; a, b, and c are values of type (possibly const) T; s and t are modifiable lvalues of type T; u denotes an identifier; rv is an rvalue of type T;
[…]andv is an lvalue of type (possibly const) T or an rvalue of type const T; and BT denotes a type that meets the BooleanTestable requirements ([booleantestable.requirements]).
Table 17 — EqualityComparable requirements [equalitycomparable] Expression Return type Requirement a == b convertible toBT
bool== is an equivalence relation, that is, it has the following properties: […] […]
Table 18 — LessThanComparable requirements [lessthancomparable] Expression Return type Requirement a < b convertible toBT
bool< is a strict weak ordering relation (25.7 [alg.sorting])
Between 17.5.3.2 [swappable.requirements] and 17.5.3.3 [nullablepointer.requirements] insert a new sub-clause as indicated:
?.?.?.? BooleanTestable requirements [booleantestable.requirements]-?- A BooleanTestable type is a boolean-like type that also supports conversions to bool. A type B meets the BooleanTestable requirements if the expressions described in Table ?? are valid and have the indicated semantics, and if B also satisfies all the other requirements of this sub-clause [booleantestable.requirements].
An object b of type B can be implicitly converted to bool and in addition can be contextually converted to bool (Clause 4). The result values of both kinds of conversions shall be equivalent. [Example: The types bool, std::true_type, and std::bitset<>::reference are BooleanTestable types. — end example] For the purpose of Table ??, let B2 and Bn denote types (possibly both equal to B or to each other) that meet the BooleanTestable requirements, let b1 denote a (possibly const) value of B, let b2 denote a (possibly const) value of B2, and let t1 denote a value of type bool. [Note: These rules ensure what an implementation can rely on but doesn't grant it license to break short-circuiting behavior of a BooleanTestable type. — end note]
Somewhere within the new sub-clause [booleantestable.requirements] insert the following new Table (?? denotes the assigned table number):
Table ?? — BooleanTestable requirements [booleantestable] Expression Return type Operational semantics bool(b1) bool Remarks: bool(b1) == t1 for every value
b1 implicitly converted to t1.!b1 Bn Remarks: bool(b1) == !bool(!b1) for
every value b1.b1 && b2 bool bool(b1) && bool(b2) b1 || b2 bool bool(b1) || bool(b2)
Change 17.5.3.3 [nullablepointer.requirements] p5 and Table 25 — "NullablePointer requirements" as indicated:
[…]
-5- In Table 25, u denotes an identifier, t denotes a non-const lvalue of type P, a and b denote values of type (possibly const) P,andnp denotes a value of type (possibly const) std::nullptr_t, and BT denotes a type that meets the BooleanTestable requirements ([booleantestable.requirements]). […]
Table 25 — NullablePointer requirements [nullablepointer] Expression Return type Operational semantics … a != b contextually convertible to boolBT[…] a == np
np == acontextually convertible to boolBT[…] a != np
np != acontextually convertible to boolBT[…]
Change 20.5.3.8 [tuple.rel] as indicated;
template<class... TTypes, class... UTypes> constexpr bool operator==(const tuple<TTypes...>& t, const tuple<UTypes...>& u);-1- Requires: For all i, where 0 <= i and i < sizeof...(TTypes), get<i>(t) == get<i>(u) is a valid expression returning a type that
[…]is convertible to boolmeets the BooleanTestable requirements ([booleantestable.requirements]). sizeof...(TTypes) == sizeof...(UTypes).template<class... TTypes, class... UTypes> constexpr bool operator<(const tuple<TTypes...>& t, const tuple<UTypes...>& u);-4- Requires: For all i, where 0 <= i and i < sizeof...(TTypes), get<i>(t) < get<i>(u) and get<i>(u) < get<i>(t) are valid expressions returning types that
[…]are convertible to boolmeet the BooleanTestable requirements ([booleantestable.requirements]). sizeof...(TTypes) == sizeof...(UTypes).
Change 23.2.1 [container.requirements.general], Table 95 — "Container requirements", and Table 97 — "Optional container operations" as indicated:
-4- In Tables 95, 96, and 97 X denotes a container class containing objects of type T, a and b denote values of type X, u denotes an identifier, r denotes a non-const value of type X,
andrv denotes a non-const rvalue of type X, and BT denotes a type that meets the BooleanTestable requirements ([booleantestable.requirements]).
Table 95 — Container requirements Expression Return type […] … a == b convertible toBT
bool[…] a != b convertible toBT
bool[…] … a.empty() convertible toBT
bool[…] […]
Table 97 — Optional container requirements Expression Return type […] … a < b convertible toBT
bool[…] a > b convertible toBT
bool[…] a <= b convertible toBT
bool[…] a >= b convertible toBT
bool[…]
Change 24.2.1 [iterator.requirements.general], Table 106 — "Input iterator requirements", and Table 110 — "Random access iterator requirements" as indicated:
-12- In the following sections, a and b denote values of type X or const X, difference_type and reference refer to the types iterator_traits<X>::difference_type and iterator_traits<X>::reference, respectively, n denotes a value of difference_type, u, tmp, and m denote identifiers, r denotes a value of X&, t denotes a value of value type T, o denotes a value of some type that is writable to the output iterator, and BT denotes a type that meets the BooleanTestable requirements ([booleantestable.requirements]).
Table 106 — Input iterator requirements Expression Return type […] a != b contextually convertible toBT
bool[…] […]
Table 110 — Random access iterator requirements Expression Return type […] … a < b contextually convertible toBT
bool[…] a > b contextually convertible toBT
bool[…] a >= b contextually convertible toBT
bool[…] a <= b contextually convertible toBT
bool[…]
Change 25.1 [algorithms.general] p8+p9 as indicated:
[Drafting note: The wording changes below also fix (a) unusual wording forms used ("should work") which are unclear in which sense they are imposing normative requirements and (b) the problem, that the current wording seems to allow that the predicate may mutate a call argument, if that is not a dereferenced iterator. Upon applying the new wording it became obvious that the both the previous and the new wording has the effect that currently algorithms such as adjacent_find, search_n, unique, and unique_copy are not correctly described (because they have no iterator argument named first1), which could give raise to a new library issue. — end drafting note]
-8- The Predicate parameter is used whenever an algorithm expects a function object (20.9) that, when applied to the result of dereferencing the corresponding iterator, returns a value testable as true.
-9- The BinaryPredicate parameter is used whenever an algorithm expects a function object that when applied to the result of dereferencing two corresponding iterators or to dereferencing an iterator and type T when T is part of the signature returns a value testable as true.In other words, iIf an algorithm takes Predicate pred as its argument and first as its iterator argument,it should work correctly in the construct pred(*first) contextually converted to bool (Clause 4)the expression pred(*first) shall have a type that meets the BooleanTestable requirements ( [booleantestable.requirements]). The function object pred shall not apply any non-constant function throughthe dereferenced iteratorits argument.In other words, iIf an algorithm takes BinaryPredicate binary_pred as its argument and first1 and first2 as its iterator arguments,it should work correctly in the construct binary_pred(*first1, *first2) contextually converted to bool (Clause 4)the expression binary_pred(*first1, *first2) shall have a type that meets the BooleanTestable requirements ( [booleantestable.requirements]). BinaryPredicate always takes the first iterator's value_type as its first argument, that is, in those cases when T value is part of the signature,it should work correctly in the construct binary_pred(*first1, value) contextually converted to bool (Clause 4)the expression binary_pred(*first1, value) shall have a type that meets the BooleanTestable requirements ( [booleantestable.requirements]). binary_pred shall not apply any non-constant function throughthe dereferenced iteratorsany of its arguments.
Change 25.7 [alg.sorting] p2 as indicated:
[…]
-2- Compare is a function object type (20.9).The return value of the function call operation applied to an object of type Compare, when contextually converted to bool(Clause 4), yields true if the first argument of the call is less than the second, and false otherwise.Compare comp is used throughout for algorithms assuming an ordering relation. Let a and b denote two argument values whose types depend on the corresponding algorithm. Then the expression comp(a, b) shall have a type that meets the BooleanTestable requirements ( [booleantestable.requirements]). The return value of comp(a, b), converted to bool, yields true if the first argument a is less than the second argument b, and false otherwise. It is assumed that comp will not apply any non-constant function throughthe dereferenced iteratorany of its arguments. […]
Change 27.5.4.2 [fpos.operations] and Table 126 — "Position type requirements" as indicated:
-1- Operations specified in Table 126 are permitted. In that table,
P refers to an instance of fpos,
[…]
o refers to a value of type streamoff,
BT refers to a type that meets the BooleanTestable requirements ([booleantestable.requirements]),
[…]
Table 126 — Position type requirements Expression Return type […] … p == q convertible to boolBT[…] p != q convertible to boolBT[…]
Change 30.2.1 [thread.req.paramname] p1 as indicated:
-1- Throughout this Clause, the names of template parameters are used to express type requirements.
If a template parameter is named Predicate, operator() applied to the template argument shall return a value that is convertible to boolPredicate is a function object type (20.14 [function.objects]). Let pred denote an lvalue of type Predicate. Then the expression pred() shall have a type that meets the BooleanTestable requirements ( [booleantestable.requirements]). The return value of pred(), converted to bool, yields true if the corresponding test condition is satisfied, and false otherwise.
Section: 26.7.8 [template.mask.array] Status: Open Submitter: Thomas Plum Opened: 2011-12-10 Last modified: 2017-02-03
Priority: 4
View all issues with Open status.
Discussion:
Recently I received a Service Request (SR) alleging that one of our testcases causes an undefined behavior. The complaint is that 26.7.8 [template.mask.array] in C++11 (and the corresponding subclause in C++03) are interpreted by some people to require that in an assignment "a[mask] = b", the subscript mask and the rhs b must have the same number of elements.
IMHO, if that is the intended requirement, it should be stated explicitly. In any event, there is a tiny editorial cleanup that could be made: In C++11, 26.7.8.1 [template.mask.array.overview] para 2 mentions"the expression a[mask] = b;"
but the semicolon cannot be part of an expression. The correction could omit the semicolon, or change the word "expression" to "assignment" or "statement".
Here is the text of the SR, slightly modified for publication:Subject: SR01174 LVS _26322Y31 has undefined behavior [open]
[Client:]
The test case t263.dir/_26322Y31.cpp seems to be illegal as it has an undefined behaviour. I searched into the SRs but found SRs were not related to the topic explained in this mail (SR00324, SR00595, SR00838).const char vl[] = {"abcdefghijklmnopqrstuvwxyz"}; const char vu[] = {"ABCDEFGHIJKLMNOPQRSTUVWXYZ"}; const std::valarray<char> v0(vl, 27), vm5(vu, 5), vm6(vu, 6); std::valarray<char> x = v0; […] const bool vb[] = {false, false, true, true, false, true}; const std::valarray<bool> vmask(vb, 6); x = v0; x[vmask] = vm5; // ***** HERE.... steq(&x[0], "abABeCghijklmnopqrstuvwxyz"); x2 = x[vmask]; // ***** ....AND HERE […]This problem has already been discussed between [experts]: See thread http://gcc.gnu.org/ml/libstdc++/2009-11/threads.html#00051 Conclusion http://gcc.gnu.org/ml/libstdc++/2009-11/msg00099.html
[Plum Hall:]
Before I log this as an SR, I need to check one detail with you. I did read the email thread you mentioned, and I did find a citation (see INCITS ISO/IEC 14882-2003 Section 26.3.2.6 on valarray computed assignments): Quote: "If the array and the argument array do not have the same length, the behavior is undefined", But this applies to computed assignment (*=, +=, etc), not to simple assignment. Here is the C++03 citation re simple assignment: 26.3.2.2 valarray assignment [lib.valarray.assign]valarray<T>& operator=(const valarray<T>&);1 Each element of the *this array is assigned the value of the corresponding element of the argument array. The resulting behavior is undefined if the length of the argument array is not equal to the length of the *this array.
In the new C++11 (N3291), we find ...
26.6.2.3 valarray assignment [valarray.assign]valarray<T>& operator=(const valarray<T>& v);1 Each element of the *this array is assigned the value of the corresponding element of the argument array. If the length of v is not equal to the length of *this, resizes *this to make the two arrays the same length, as if by calling resize(v.size()), before performing the assignment.
So it looks like the testcase might be valid for C++11 but not for C++03; what do you think?
[Client:]
I quite agree with you but the two problems I mentioned:x[vmask] = vm5; // ***** HERE.... […] x2 = x[vmask]; // ***** ....AND HERErefer to mask_array assignment hence target the C++03 26.3.8 paragraph. Correct?
[Plum Hall:]
I mentioned the contrast between C++03 26.3.2.2 para 1 versus C++11 26.6.2.3 para 1. But in C++03 26.3.8, I don't find any corresponding restriction. Could you quote the specific requirement you're writing about? [Client:]
I do notice the difference between c++03 26.3.2.2 and c++11 26.6.2.3 about assignments between different sized valarray and I perfectly agree with you. But, as already stated, this is not a simple valarray assignment but a mask_array assignment (c++03 26.3.8 / c++11 26.6.8). See c++11 quote below: 26.6.8 Class template mask_array
26.6.8.1 Class template mask_array overview
[....]
This template is a helper template used by the mask subscript operator: mask_array<T> valarray<T>::operator[](const valarray<bool>&).
It has reference semantics to a subset of an array specified by a boolean mask. Thus, the expression a[mask] = b; has the effect of assigning the elements of b to the masked elements in a (those for which the corresponding element in mask is true.)
26.6.8.2 mask_array assignment
void operator=(const valarray<T>&) const; const mask_array& operator=(const mask_array&) const;1 These assignment operators have reference semantics, assigning the values of the argument array elements to selected elements of the valarray<T> object to which it refers.
In particular, [one of the WG21 experts] insisted on the piece "the elements of b".
That is why I reported the test t263.dir/_26322Y31.cpp having an undefined behaviour. [Plum Hall:]
OK, I can see that I will have to ask WG21; I will file an appropriate issue with the Library subgroup. In the meantime, I will mark this testcase as "DISPUTED" so that it is not required for conformance testing, until we get a definitive opinion.
[2012, Kona]
Moved to Open.
There appears to be a real need for clarification in the standard, and implementations differ in their current interpretation. This will need some research by implementers and a proposed resolution before further discussion is likely to be fruitful.
Proposed resolution:
Section: 20.15.4.3 [meta.unary.prop] Status: Open Submitter: Dave Abrahams Opened: 2011-12-09 Last modified: 2017-02-03
Priority: 3
View other active issues in [meta.unary.prop].
View all other issues in [meta.unary.prop].
View all issues with Open status.
Discussion:
IMO if we specified is_[nothrow_]constructible in terms of a variable declaration whose validity requires destructibility, it is clearly a bug in our specification and a failure to realize the actual original intent. The specification should have been in terms of placement-new.
Daniel:A defaulted copy/move constructor for a class X is defined as deleted (8.4.3 [dcl.fct.def.delete]) if X has:
[…]
— any direct or virtual base class or non-static data member of a type with a destructor that is deleted or inaccessible from the defaulted constructor,
[…]
Dave:
This is about is_nothrow_constructible in particular. The fact that it is
foiled by not having a noexcept dtor is a defect.
[2012, Kona]
Move to Open.
is_nothrow_constructible is defined in terms of is_constructible, which is defined by looking at a hypothetical variable and asking whether the variable definition is known not to throw exceptions. The issue claims that this also examines the type's destructor, given the context, and thus will return false if the destructor can potentially throw. At least one implementation (Howard's) does return false if the constructor is noexcept(true) and the destructor is noexcept(false). So that's not a strained interpretation. The issue is asking for this to be defined in terms of placement new, instead of in terms of a temporary object, to make it clearer that is_nothrow_constructible looks at the noexcept status of only the constructor, and not the destructor.
Sketch of what the wording would look like:
require is_constructible, and then also require that a placement new operation does not throw. (Remembering the title of this issue... What does this imply for swap?
If we accept this resolution, do we need any changes to swap?
STL argues: no, because you are already forbidden from passing anything with a throwing desturctor to swap.
Dietmar argues: no, not true. Maybe statically the destructor can conceivably throw for some values, but maybe there are some values known not to throw. In that case, it's correct to pass those values to swap.
[2017-01-27 Telecon]
Gave the issue a better title
This issue interacts with 2827
Ville would like "an evolution group" to take a look at this issue.
Proposed resolution:
Section: 22.4.2.2.2 [facet.num.put.virtuals], 27.5.3.1.2 [ios::fmtflags], 27.5.6.1 [fmtflags.manip] Status: Open Submitter: Benjamin Kosnik Opened: 2011-12-15 Last modified: 2017-02-03
Priority: 3
View other active issues in [facet.num.put.virtuals].
View all other issues in [facet.num.put.virtuals].
View all issues with Open status.
Discussion:
Iostreams should include a manipulator to toggle grouping on/off for locales that support grouped digits. This has come up repeatedly and been deferred. See LWG 826 for the previous attempt.
If one is using a locale that supports grouped digits, then output will always include the generated grouping characters. However, very plausible scenarios exist where one might want to output the number, un-grouped. This is similar to existing manipulators that toggle on/off the decimal point, numeric base, or positive sign. See some user commentary here.[21012, Kona]
Move to Open.
This is a feature request.
Walter is slightly uncomfortable with processing feature requests through the issues lists.
Alisdair says this is far from the first feature request that has come in from the issues list.
STL: The fact that you can turn off grouping on hex output is compelling.
Marshall: if we add this flag, we'll need to update tables 87-91 as well.
STL: If it has been implemented somewhere, and it works, we'd be glad to add it.
Howard: We need to say what the default is.
Alisdair sumarizes:
(1) We want clear wording that says what the effect is of turning the flag off;
(2) what the default values are, and
(3) how this fits into tables 87-90. (and 128)
[Issaquah 2014-02-10-12: Move to LEWG]
Since this issue was filed, we have grown a new working group that is better placed to handle feature requests.
We will track such issues with an LEWG status until we get feedback from the Library Evolution Working Group.
[Issaquah 2014-02-12: LEWG discussion]
| SF | F | N | A | SA |
| 2 | 4 | 1 | 0 | 0 |
Think about the ABI break for adding a flag. But this could be mitigated by putting the data into an iword instead of a flag.
This needs to change Stage 2 in [facet.num.put.virtuals].
Previous resolution, which needs the above corrections:
This wording is relative to the FDIS.
Insert in 22.4.2.2.2 [facet.num.put.virtuals] paragraph 5:
Stage 1: The first action of stage 1 is to determine a conversion specifier. The tables that describe this determination use the following local variables
fmtflags flags = str.flags() ; fmtflags basefield = (flags & (ios_base::basefield)); fmtflags uppercase = (flags & (ios_base::uppercase)); fmtflags floatfield = (flags & (ios_base::floatfield)); fmtflags showpos = (flags & (ios_base::showpos)); fmtflags showbase = (flags & (ios_base::showbase)); fmtflags showgrouping = (flags & (ios_base::showgrouping));Change header <ios> synopsis, [iostreams.base.overview] as indicated:
#include <iosfwd> namespace std { […] // 27.5.6, manipulators: […] ios_base& showpoint (ios_base& str); ios_base& noshowpoint (ios_base& str); ios_base& showgrouping (ios_base& str); ios_base& noshowgrouping(ios_base& str); ios_base& showpos (ios_base& str); ios_base& noshowpos (ios_base& str); […] }Change class ios_base synopsis, 27.5.3 [ios.base] as indicated:
namespace std { class ios_base { public: class failure; // 27.5.3.1.2 fmtflags typedef T1 fmtflags; […] static constexpr fmtflags showpoint = unspecified ; static constexpr fmtflags showgrouping = unspecified ; static constexpr fmtflags showpos = unspecified ; […] }; }Add a new entry to Table 122 — "fmtflags effects" as indicated:
Table 122 — fmtflags effects Element Effect(s) if set […] showpoint generates a decimal-point character unconditionally in generated floatingpoint output showgrouping generates grouping characters unconditionally in generated output […] After 27.5.3.1.2 [ios::fmtflags] p12 insert the following:
ios_base& showgrouping(ios_base& str);-?- Effects: Calls str.setf(ios_base::showgrouping).
-?- Returns: str.ios_base& noshowgrouping(ios_base& str);-?- Effects: Calls str.unsetf(ios_base::showgrouping).
-?- Returns: str.
Proposed resolution:
Section: 27.8.5.1 [stringstream.cons] Status: New Submitter: Nicolai Josuttis Opened: 2012-01-15 Last modified: 2017-02-03
Priority: 3
View all issues with New status.
Discussion:
This issue was raised while discussing issue 1448.
Note the following program:
string s("s1: 123456789");
ostringstream s1(s, ios_base::out|ios_base::app);
s1 << "hello";
cout << s1.str() << endl;
With g++4.x it prints:
s1: 123456789hello
With VisualC++10 it prints:
hello23456789
From my intuitive understanding the flag "app" should result in the output of g++4.x. I also would read that from 27.5.3.1.4 [ios::openmode] claiming:
app seek to end before each write
However in issue 1448 P.J.Plauger comments:
I think we should say nothing special about app at construction time (thus leaving the write pointer at the beginning of the buffer). Leave implementers wiggle room to ensure subsequent append writes as they see fit, but don't change existing rules for initial seek position.
Note that the flag ate on both platforms appends "hello" to s.
Proposed resolution:
Section: 17.4.1 [structure] Status: Open Submitter: Jens Maurer Opened: 2012-03-08 Last modified: 2017-02-03
Priority: 3
View all issues with Open status.
Discussion:
The front matter in clause 17 should clarify that postconditions will not hold if a standard library function exits via an exception. Postconditions or guarantees that apply when an exception is thrown (beyond the basic guarantee) are described in an "Exception safety" section.
[ 2012-10 Portland: Move to Open ]
Consensus that we do not clearly say this, and that we probably should. A likely location to describe the guarantees of postconditions could well be a new sub-clause following 17.5.4.11 [res.on.required] which serves the same purpose for requires clauses. However, we need such wording before we can make progress.
Also, see 2137 for a suggestion that we want to see a paper resolving both issues together.
[2015-05-06 Lenexa: EirkWF to write paper addressing 2136 and 2137]
MC: Idea is to replace all such "If no exception" postconditions with "Exception safety" sections.
Proposed resolution:
Section: 28.8.3 [re.regex.assign] Status: Open Submitter: Jonathan Wakely Opened: 2012-03-08 Last modified: 2017-02-03
Priority: 3
View all other issues in [re.regex.assign].
View all issues with Open status.
Discussion:
The post-conditions of basic_regex<>::assign 28.8.3 [re.regex.assign] p16 say:
If no exception is thrown, flags() returns f and mark_count() returns the number of marked sub-expressions within the expression.
The default expectation in the library is that post-conditions only hold, if there is no failure (see also 2136), therefore the initial condition should be removed to prevent any misunderstanding.
[ 2012-10 Portland: Move to Open ]
A favorable resolution clearly depends on a favorable resolution to 2136. There is also a concern that this is just one example of where we would want to apply such a wording clean-up, and which is really needed to resolve both this issue and 2136 is a paper providing the clause 17 wording that gives the guarantee for postcondition paragaraphs, and then reviews clauses 18-30 to apply that guarantee consistently. We do not want to pick up these issues piecemeal, as we risk openning many issues in an ongoing process.
[2015-05-06 Lenexa: EirkWF to write paper addressing 2136 and 2137]
Proposed resolution:
This wording is relative to N3376.
template <class string_traits, class A> basic_regex& assign(const basic_string<charT, string_traits, A>& s, flag_type f = regex_constants::ECMAScript);[…]
-15- Effects: Assigns the regular expression contained in the string s, interpreted according the flags specified in f. If an exception is thrown, *this is unchanged. -16- Postconditions:If no exception is thrown,flags() returns f and mark_count() returns the number of marked sub-expressions within the expression.
Section: 17.5.4.2.1 [namespace.std], 19.5 [syserr], 20.10.7.1 [allocator.uses.trait], 20.14.11.1 [func.bind.isbind], 20.14.11.2 [func.bind.isplace], 20.14.15 [unord.hash], 20.15.7.6 [meta.trans.other], 22.3.1 [locale], 22.4.1.4 [locale.codecvt], 28.12.1.4 [re.regiter.incr] Status: Open Submitter: Loïc Joly Opened: 2012-03-08 Last modified: 2017-02-03
Priority: 4
View all other issues in [namespace.std].
View all issues with Open status.
Discussion:
The expression "user-defined type" is used in several places in the standard, but I'm not sure what it means. More specifically, is a type defined in the standard library a user-defined type?
From my understanding of English, it is not. From most of the uses of this term in the standard, it seem to be considered as user defined. In some places, I'm hesitant, e.g. 17.5.4.2.1 [namespace.std] p1:A program may add a template specialization for any standard library template to namespace std only if the declaration depends on a user-defined type and the specialization meets the standard library requirements for the original template and is not explicitly prohibited.
Does it mean we are allowed to add in the namespace std a specialization for std::vector<std::pair<T, U>>, for instance?
Additional remarks from the reflector discussion: The traditional meaning of user-defined types refers to class types and enum types, but the library actually means here user-defined types that are not (purely) library-provided. Presumably a new term - like user-provided type - should be introduced and properly defined.[ 2012-10 Portland: Move to Deferred ]
The issue is real, in that we never define this term and rely on a "know it when I see it" intuition. However, there is a fear that any attempt to pin down a definition is more likely to introduce bugs than solve them - getting the wording for this precisely correct is likely far more work than we are able to give it.
There is unease at simple closing as NAD, but not real enthusiasm to provide wording either. Move to Deferred as we are not opposed to some motivated individual coming back with full wording to review, but do not want to go out of our way to encourage someone to work on this in preference to other issues.
[2014-02-20 Re-open Deferred issues as Priority 4]
[2015-03-05 Jonathan suggests wording]
I dislike the suggestion to change to "user-provided" type because I already find the difference between user-declared / user-provided confusing for special member functions, so I think it would be better to use a completely different term. The core language uses "user-defined conversion sequence" and "user-defined literal" and similar terms for things which the library provides, so I think we should not refer to "user" at all to distinguish entities defined outside the implementation from things provided by the implementation.
I propose "program-defined type" (and "program-defined specialization"), defined below. The P/R below demonstrates the scope of the changes required, even if this name isn't adopted. I haven't proposed a change for "User-defined facets" in [locale].[Lenexa 2015-05-06]
RS, HT: The core language uses "user-defined" in a specific way, including library things but excluding core language things, let's use a different term.
MC: Agree.
RS: "which" should be "that", x2
RS: Is std::vector<MyType> a "program-defined type"?
MC: I think it should be.
TK: std::vector<int> seems to take the same path.
JW: std::vector<MyType> isn't program-defined, we don't need it to be, anything that depends on that also depends on =MyType.
TK: The type defined by an "explicit template specialization" should be a program-defined type.
RS: An implicit instantiation of a "program-defined partial specialization" should also be a program-defined type.
JY: This definition formatting is horrible and ugly, can we do better?
RS: Checking ISO directives.
RS: Define "program-defined type" and "program-defined specialization" instead, to get rid of the angle brackets.
JW redrafting.
Proposed resolution:
This wording is relative to N4296.
Add a new sub-clause to 17.3 [definitions]:
17.3.? [defns.program.defined]
program-defined
<type> a class type or enumeration type which is not part of the C++ standard library and not defined by the implementation. [Note: Types defined by the implementation include extensions (1.4 [intro.compliance]) and internal types used by the library. — end note]program-defined
<specialization> an explicit template specialization or partial specialization which is not part of the C++ standard library and not defined by the implementation.Change 17.5.4.2.1 [namespace.std] paragraph 1+2:
-1- The behavior of a C++ program is undefined if it adds declarations or definitions to namespace std or to a
namespace within namespace std unless otherwise specified. A program may add a template specialization
for any standard library template to namespace std only if the declaration depends on a
userprogram-defined type and the specialization meets the standard library requirements for the
original template and is not explicitly prohibited.
Change 19.5 [syserr] paragraph 4:
-4- The is_error_code_enum and is_error_condition_enum may be specialized for
userprogram-defined types to indicate that such types are eligible for class error_code
and class error_condition automatic conversions, respectively.
Change 20.10.7.1 [allocator.uses.trait] paragraph 1:
-1- Remarks: automatically detects […]. A program may specialize this template to derive from
true_type for a userprogram-defined type T that does not have a nested
allocator_type but nonetheless can be constructed with an allocator where either: […]
Change 20.14.11.1 [func.bind.isbind] paragraph 2:
-2- Instantiations of the is_bind_expression template […]. A program may specialize
this template for a userprogram-defined type T to have a BaseCharacteristic
of true_type to indicate that T should be treated as a subexpression in a bind call.
Change 20.14.11.2 [func.bind.isplace] paragraph 2:
-2- Instantiations of the is_placeholder template […]. A program may specialize this template for a
userprogram-defined type T to have a BaseCharacteristic of
integral_constant<int, N> with N > 0 to indicate that T should be
treated as a placeholder type.
Change 20.14.15 [unord.hash] paragraph 1:
The unordered associative containers defined in 23.5 use specializations of the class template hash […], the instantiation hash<Key> shall:
[…]
[…]
[…]
[…]
satisfy the requirement that the expression h(k), where h is an object of type
hash<Key> and k is an object of type Key, shall not throw an exception unless
hash<Key> is a userprogram-defined specialization that depends on at least one
userprogram-defined type.
Change 20.15.7.5 [meta.trans.ptr] Table 57 (common_type row):
Table 57 — Other transformations Template Condition Comments … template <class... T>
struct common_type;The member typedef type shall be
defined or omitted as specified below.
[…]. A program may
specialize this trait if at least one
template parameter in the
specialization is auserprogram-defined type.
[…]…
Change 22.4.1.4 [locale.codecvt] paragraph 3:
-3- The specializations required in Table 81 (22.3.1.1.1) […]. Other encodings can be converted
by specializing on a userprogram-defined stateT type.[…]
Change 28.12.1.4 [re.regiter.incr] paragraph 8:
-8- [Note: This means that a compiler may call an implementation-specific search function, in which case
a userprogram-defined specialization of regex_search will not be called. —
end note]
Section: 17.5.3.1 [utility.arg.requirements] Status: Open Submitter: Nikolay Ivchenkov Opened: 2012-03-23 Last modified: 2017-02-03
Priority: 2
View all other issues in [utility.arg.requirements].
View all issues with Open status.
Discussion:
According to 17.5.3.1 [utility.arg.requirements] p1
The template definitions in the C++ standard library refer to various named requirements whose details are set out in tables 17-24. In these tables, T is an object or reference type to be supplied by a C++ program instantiating a template; a, b, and c are values of type (possibly const) T; s and t are modifiable lvalues of type T; u denotes an identifier; rv is an rvalue of type T; and v is an lvalue of type (possibly const) T or an rvalue of type const T.
Is it really intended that T may be a reference type? If so, what should a, b, c, s, t, u, rv, and v mean? For example, are "int &" and "int &&" MoveConstructible?
As far as I understand, we can explicitly specify template arguments for std::swap and std::for_each. Can we use reference types there?
#include <iostream>
#include <utility>
int main()
{
int x = 1;
int y = 2;
std::swap<int &&>(x, y); // undefined?
std::cout << x << " " << y << std::endl;
}
#include <algorithm>
#include <iostream>
#include <iterator>
#include <utility>
struct F
{
void operator()(int n)
{
std::cout << n << std::endl;
++count;
}
int count;
} f;
int main()
{
int arr[] = { 1, 2, 3 };
auto&& result = std::for_each<int *, F &&>( // undefined?
std::begin(arr),
std::end(arr),
std::move(f));
std::cout << "count: " << result.count << std::endl;
}
Are these forms of usage well-defined?
Let's also consider the following constructor of std::thread:template <class F, class ...Args> explicit thread(F&& f, Args&&... args);Requires: F and each Ti in Args shall satisfy the MoveConstructible requirements.
When the first argument of this constructor is an lvalue (e.g. a name of a global function), template argument for F is deduced to be lvalue reference type. What should "MoveConstructible" mean with regard to an lvalue reference type? Maybe the wording should say that std::decay<F>::type and each std::decay<Ti>::type (where Ti is an arbitrary item in Args) shall satisfy the MoveConstructible requirements?
[2013-03-15 Issues Teleconference]
Moved to Open.
The questions raised by the issue are real, and should have a clear answer.
[2015-10, Kona Saturday afternoon]
STL: std::thread needs to be fixed, and anything behaving like it needs to be fixed, rather than reference types. std::bind gets this right. We need to survey this. GR: That doesn't sound small to me. STL: Seach for CopyConstructible etc. It may be a long change, but not a hard one.
MC: It seems that we don't have a PR. Does anyone have one? Is anyone interested in doing a survey?
[2016-03, Jacksonville]
Casey volunteers to make a survey
[2016-06, Oulu]
During an independent survey performed by Daniel as part of the analysis of LWG 2716, some overlap was found between these two issues. Daniel suggested to take responsibility for surveying LWG 2146 and opined that the P/R of LWG 2716 should restrict to forwarding references, where the deduction to lvalue references can happen without providing an explicit template argument just by providing an lvalue function argument.
Proposed resolution:
Section: 21.3.2.1 [string.require] Status: Open Submitter: Robert Shearer Opened: 2012-04-13 Last modified: 2017-02-03
Priority: 3
View all other issues in [string.require].
View all issues with Open status.
Discussion:
In C++11, basic_string is not described as a "container", and is not governed by the allocator-aware container semantics described in sub-clause 23.2 [container.requirements]; as a result, and requirements or contracts for the basic_string interface must be documented in Clause 21 [strings].
Sub-clause 21.3.2.6.8 [string.swap] defines the swap member function with no requirements, and with guarantees to execute in constant time without throwing. Fulfilling such a contract is not reasonable in the presence of unequal non-propagating allocators. In contrast, 23.2.1 [container.requirements.general] p7 declares the behavior of member swap for containers with unequal non-propagating allocators to be undefined. Resolution proposal: Additional language from Clause 23 [containers] should probably be copied to Clause 21 [strings]. I will refrain from an exactly recommendation, however, as I am raising further issues related to the language in Clause 23 [containers].[2013-03-15 Issues Teleconference]
Moved to Open.
Alisdair has offered to provide wording.
Telecon notes that 23.2.1 [container.requirements.general]p13 says that string is an allocator-aware container.
Proposed resolution:
Section: 17.5.3.2 [swappable.requirements], 23.2.1 [container.requirements.general] Status: LEWG Submitter: Robert Shearer Opened: 2012-04-13 Last modified: 2017-06-15
Priority: 3
View all other issues in [swappable.requirements].
View all issues with LEWG status.
Discussion:
Sub-clause 17.5.3.2 [swappable.requirements] defines two notions of swappability: a binary version defining when two objects are swappable with one another, and a unary notion defining whether an object is swappable (without qualification), with the latter definition requiring that the object satisfy the former with respect to all values of the same type.
Let T be a container type based on a non-propagating allocator whose instances do not necessarily compare equal. Then sub-clause 23.2.1 [container.requirements.general] p7 implies that no object t of type T is swappable (by the unary definition). Throughout the standard it is the unary definition of "swappable" that is listed as a requirement (with the exceptions of 20.2.3 [utility.swap] p4, 20.4.2 [pairs.pair] p31, 20.5.3.3 [tuple.swap] p2, 25.6.3 [alg.swap] p2, and 25.6.3 [alg.swap] p6, which use the binary definition). This renders many of the mutating sequence algorithms of sub-clause 25.6 [alg.modifying.operations], for example, inapplicable to sequences of standard container types, even where every element of the sequence is swappable with every other. Note that this concern extends beyond standard containers to all future allocator-based types. Resolution proposal: I see two distinct straightforward solutions:I favor the latter solution, for reasons detailed in the following issue.
[ 2012-10 Portland: Move to Open ]
The issue is broader than containers with stateful allocotors, although they are the most obvious example contained within the standard itself. The basic problem is that once you have a stateful allocator, that does not propagate_on_swap, then whether two objects of this type can be swapped with well defined behavior is a run-time property (the allocators compare equal) rather than a simple compile-time property that can be deduced from the type. Strictly speaking, any type where the nature of swap is a runtime property does not meet the swappable requirements of C++11, although typical sequences of such types are going to have elements that are all swappable with any other element in the sequence (using our other term of art for specifying requirements) as the common case is a container of elements who all share the same allocator.
The heart of the problem is that the swappable requirments demand that any two objects of the same type be swappable with each other, so if any two such objects would not be swappable with each other, then the whole type is never swappable. Many algorithms in clause 25 are specified in terms of swappable which is essentially an overspecification as all they actually need is that any element in the sequence is swappable with any other element in the sequence.
At this point Howard joins the discussion and points out that the intent of introducing the two swap-related terms was to support vector<bool>::reference types, and we are reading something into the wording that was never intended. Consuses is that regardless of the intent, that is what the words today say.
There is some support to see a paper reviewing the whole of clause 25 for this issue, and other select clauses as may be necessary.
There was some consideration to introducing a note into the front of clause 25 to indicate swappable requirements in the clause should be interpreted to allow such awkward types, but ultimately no real enthusiasm for introducing a swappable for clause 25 requirement term, especially if it confusingly had the same name as a term used with a subtly different meaning through the rest of the standard.
There was no enthusiasm for the alternate resolution of requiring containers with unequal allocators that do not propagate provide a well-defined swap behavior, as it is not believed to be possible without giving swap linear complexity for such values, and even then would require adding the constraint that the container element types are CopyConstructible.
Final conclusion: move to open pending a paper from a party with a strong interest in stateful allocators.
[2016-03 Jacksonville]
Alisdair says that his paper P0178 addresses this.
[2016-06 Oulu]
P0178 reviewed, and sent back to LEWG for confirmation.
[2017-02 in Kona, LEWG responds]
Note in the issue that this is tracked here
[2017-06-02 Issues Telecon]
Leave as LEWG; priority 3
Proposed resolution:
Apply P0178.
Section: 20.2.3 [utility.swap], 17.5.3.2 [swappable.requirements], 23.2.1 [container.requirements.general] Status: LEWG Submitter: Robert Shearer Opened: 2012-04-13 Last modified: 2017-02-03
Priority: 2
View other active issues in [utility.swap].
View all other issues in [utility.swap].
View all issues with LEWG status.
Discussion:
Sub-clause 20.2.3 [utility.swap] defines a non-member 'swap' function with defined behavior for all MoveConstructible and MoveAssignable types. It does not guarantee constant-time complexity or noexcept in general, however this definition does render all objects of MoveConstructible and MoveAssignable type swappable (by the unary definition of sub-clause 17.5.3.2 [swappable.requirements]) in the absence of specializations or overloads.
The overload of the non-member swap function defined in Table 96, however, defines semantics incompatible with the generic non-member swap function, since it is defined to call a member swap function whose semantics are undefined for some values of MoveConstructible and MoveAssignable types. The obvious (perhaps naive) interpretation of sub-clause 17.5.3.2 [swappable.requirements] is as a guide to the "right" semantics to provide for a non-member swap function (called in the context defined by 17.5.3.2 [swappable.requirements] p3) in order to provide interoperable user-defined types for generic programming. The standard container types don't follow these guidelines. More generally, the design in the standard represents a classic example of "contract narrowing". It is entirely reasonable for the contract of a particular swap overload to provide more guarantees, such as constant-time execution and noexcept, than are provided by the swap that is provided for any MoveConstructible and MoveAssignable types, but it is not reasonable for such an overload to fail to live up to the guarantees it provides for general types when it is applied to more specific types. Such an overload or specialization in generic programming is akin to an override of an inherited virtual function in OO programming: violating a superclass contract in a subclass may be legal from the point of view of the language, but it is poor design and can easily lead to errors. While we cannot prevent user code from providing overloads that violate the more general swap contract, we can avoid doing so within the library itself. My proposed resolution is to draw a sharp distinction between member swap functions, which provide optimal performance but idiosyncratic contracts, and non-member swap functions, which should always fulfill at least the contract of 20.2.3 [utility.swap] and thus render objects swappable. The member swap for containers with non-propagating allocators, for example, would offer constant-time guarantees and noexcept but would only offer defined behavior for values with allocators that compare equal; non-member swap would test allocator equality and then dispatch to either member swap or std::swap depending on the result, providing defined behavior for all values (and rendering the type "swappable"), but offering neither the constant-time nor the noexcept guarantees.[2013-03-15 Issues Teleconference]
Moved to Open.
This topic deserves more attention than can be given in the telocon, and there is no proposed resolution.
[2013-03-15 Issues Teleconference]
Moved to Open.
This topic deserves more attention than can be given in the telocon, and there is no proposed resolution.
[2016-03 Jacksonville]
Alisdair says that his paper P0178 addresses this.
[2016-08 Chicago]
Send to LEWG
[2016-06 Oulu]
P0178 reviewed, and sent back to LEWG for confirmation.
Proposed resolution:
Apply P0178.Section: 26.6.1.3 [rand.req.urng] Status: New Submitter: John Salmon Opened: 2012-04-26 Last modified: 2017-02-03
Priority: 4
View all other issues in [rand.req.urng].
View all issues with New status.
Discussion:
The expressions G::min() and G::max() in Table 116 in 26.6.1.3 [rand.req.urng] are specified as having "compile-time" complexity.
It is not clear what, exactly, this requirement implies. If a URNG has a method:static int min();
then is the method required to have a constexpr qualifier? I believe the standard would benefit from clarification of this point.
Proposed resolution:
Section: 18.10 [support.runtime] Status: Open Submitter: Thomas Plum Opened: 2012-04-30 Last modified: 2017-03-13
Priority: 4
View all other issues in [support.runtime].
View all issues with Open status.
Discussion:
Since C99, the C standard describes a macro named __bool_true_false_are_defined.
In the process of harmonizing C++11 with C99, this name became part of the C++ standard. I propose that all mention of this name should be removed from the C and C++ standards. Here's the problem: The name was originally proposed as a transition tool, so that the headers for a project could contain lines like the following.#if !defined(__bool_true_false_are_defined) #define bool int /* or whatever */ #define true 1 #define false 0 #endif
Then when the project was compiled by a "new" compiler that implemented bool as defined by the evolving C++98 or C99 standards, those lines would be skipped; but when compiled by an "old" compiler that didn't yet provide bool, true, and false, then the #define's would provide a simulation that worked for most purposes.
It turns out that there is an unfortunate ambiguity in the name. One interpretation is as shown above, but a different reading says "bool, true, and false are #define'd", i.e. that the meaning of the macro is to assert that these names are macros (not built-in) ... which is true in C, but not in C++. In C++11, the name appears in parentheses followed by a stray period, so some editorial change is needed in any event: 18.10 [support.runtime] para 1:Headers <csetjmp> (nonlocal jumps), <csignal> (signal handling), <cstdalign> (alignment), <cstdarg> (variable arguments), <cstdbool> (__bool_true_false_are_defined). <cstdlib> (runtime environment getenv(), system()), and <ctime> (system clock clock(), time()) provide further compatibility with C code.
However, para 2 says
"The contents of these headers are the same as the Standard C library headers <setjmp.h>, <signal.h>, <stdalign.h>, <stdarg.h>, <stdbool.h>, <stdlib.h>, and <time.h>, respectively, with the following changes:",
and para 8 says
"The header <cstdbool> and the header <stdbool.h> shall not define macros named bool, true, or false."
Thus para 8 doesn't exempt the C++ implementation from the arguably clear requirement of the C standard, to provide a macro named __bool_true_false_are_defined defined to be 1.
Real implementations of the C++ library differ, so the user cannot count upon any consistency; furthermore, the usefulness of the transition tool has faded long ago. That's why my suggestion is that both C and C++ standards should eliminate any mention of __bool_true_false_are_defined. In that case, the name belongs to implementers to provide, or not, as they choose.[2013-03-15 Issues Teleconference]
Moved to Open.
While not strictly necessary, the clean-up look good.
We would like to hear from our C liaison before moving on this issue though.
[2015-05 Lenexa]
LWG agrees. Jonathan provides wording.
[2017-03-04, Kona]
The reference to <cstdbool> in p2 needs to be struck as well. Continue the discussion on the reflector once the DIS is available.
Proposed resolution:
This wording is relative to N4296.
Edit the footnote on 17.5.1.2 [headers] p7:
176) In particular, including any of the standard headers <stdbool.h>, <cstdbool>, <iso646.h> or <ciso646> has no effect.
Edit 18.10 [support.runtime] p1 as indicated (and remove the index entry for __bool_true_false_are_defined):
-1- Headers <csetjmp> (nonlocal jumps), <csignal> (signal handling), <cstdalign> (alignment), <cstdarg> (variable arguments), <cstdbool>,
(__bool_true_false_are_defined).<cstdlib> (runtime environment getenv(), system()), and <ctime> (system clock clock(), time()) provide further compatibility with C code.
Remove Table 38 — Header <cstdbool> synopsis [tab:support.hdr.cstdbool] from 18.10 [support.runtime]
Table 38 — Header <cstdbool> synopsis Type Name(s) Macro: __bool_true_false_are_defined
Section: 23.3.7.8 [array.zero] Status: Open Submitter: Daryle Walker Opened: 2012-05-08 Last modified: 2017-02-03
Priority: 3
View all other issues in [array.zero].
View all issues with Open status.
Discussion:
Objects of std::array<T,N> are supposed to be initialized with aggregate initialization (when not the destination of a copy or move). This clearly works when N is positive. What happens when N is zero? To continue using an (inner) set of braces for initialization, a std::array<T,0> implementation must have an array member of at least one element, and let default initialization take care of those secret elements. This cannot work when T has a set of constructors and the default constructor is deleted from that set. Solution: Add a new paragraph in 23.3.7.8 [array.zero]:
The unspecified internal structure of array for this case shall allow initializations like:
array<T, 0> a = { };and said initializations must be valid even when T is not default-constructible.
[2012, Portland: Move to Open]
Some discussion to understand the issue, which is that implementations currently have freedom to implement an empty array by holding a dummy element, and so might not support value initialization, which is surprising when trying to construct an empty container. However, this is not mandated, it is an unspecified implementation detail.
Jeffrey points out that the implication of 23.3.7.1 [array.overview] is that this initialization syntax must be supported by empty array objects already. This is a surprising inference that was not obvious to the room, but consensus is that the reading is accurate, so the proposed resolution is not necessary, although the increased clarity may be useful.
Further observation is that the same clause effectively implies that T must always be DefaultConstructible, regardless of N for the same reasons - as an initializer-list may not supply enough values, and the remaining elements must all be value initialized.
Concern that we are dancing angels on the head of pin, and that relying on such subtle implications in wording is not helpful. We need a clarification of the text in this area, and await wording.
[2015-02 Cologne]
DK: What was the outcome of Portland? AM: Initially we thought we already had the intended behaviour. We concluded that T must always be DefaultConstructible, but I'm not sure why. GR: It's p2 in std::array, "up to N". AM: That wording already implies that "{}" has to work when N is zero. But the wording of p2 needs to be fixed to make clear that it does not imply that T must be DefaultConstructible.
Conclusion: Update wording, revisit later.[2015-10, Kona Saturday afternoon]
MC: How important is this? Can you not just use default construction for empty arrays?
TK: It needs to degenerate properly from a pack. STL agrees.
JW: Yes, this is important, and we have to make it work.
MC: I hate the words "initialization like".
JW: I'll reword this.
WEB: Can I ask that once JW has reworded this we move it to Review rather than Open?
MC: We'll try to review it in a telecon and hopefully get it to tentatively ready.
STL: Double braces must also work: array<T, 0> a = {{}};.
Jonathan to reword.
Proposed resolution:
This wording is relative to N3376.
Add the following new paragraph between the current 23.3.7.8 [array.zero] p1 and p2:
-1- array shall provide support for the special case N == 0.
-?- The unspecified internal structure of array for this case shall allow initializations like:array<T, 0> a = { };and said initializations must be valid even when T is not default-constructible.
-2- In the case that N == 0, begin() == end() == unique value. The return value of data() is unspecified. -3- The effect of calling front() or back() for a zero-sized array is undefined. -4- Member function swap() shall have a noexcept-specification which is equivalent to noexcept(true).
Section: 23.3.11.3 [vector.capacity] Status: Open Submitter: Nikolay Ivchenkov Opened: 2012-05-08 Last modified: 2017-02-03
Priority: 2
View all other issues in [vector.capacity].
View all issues with Open status.
Discussion:
There are various operations on std::vector that can cause elements of the vector to be moved from one location to another. A move operation can use either rvalue or const lvalue as argument; the choice depends on the value of !is_nothrow_move_constructible<T>::value && is_copy_constructible<T>::value, where T is the element type. Thus, some operations on std::vector (e.g. 'resize' with single parameter, 'reserve', 'emplace_back') should have conditional requirements. For example, let's consider the requirement for 'reserve' in N3376 – 23.3.11.3 [vector.capacity]/2:
Requires: T shall be MoveInsertable into *this.
This requirement is not sufficient if an implementation is free to select copy constructor when !is_nothrow_move_constructible<T>::value && is_copy_constructible<T>::value evaluates to true. Unfortunately, is_copy_constructible cannot reliably determine whether T is really copy-constructible. A class may contain public non-deleted copy constructor whose definition does not exist or cannot be instantiated successfully (e.g., std::vector<std::unique_ptr<int>> has copy constructor, but this type is not copy-constructible). Thus, the actual requirements should be:
if !is_nothrow_move_constructible<T>::value && is_copy_constructible<T>::value then T shall be CopyInsertable into *this;
otherwise T shall be MoveInsertable into *this.
Maybe it would be useful to introduce a new name for such conditional requirement (in addition to "CopyInsertable" and "MoveInsertable").
[2016-08 Chicago]
The problem does not appear to be as severe as described. The MoveInsertable requirements are consistently correct, but an issue may arise on the exception-safety guarantees when we check for is_copy_constructible_v<T>. The problem, as described, is typically for templates that appear to have a copy constructor, but one that fails to compile once instantiated, and so gives a misleading result for the trait.
In general, users should not provide such types, and the standard would not serve users well by trying to address support for such types. However, the standard should not be providing such types either, such as vector<unique_ptr<T>>. A possible resolution would be to tighten the constraints in Table 80 — Container Requirements, so that if the Requirements for the copy constructor/assingment operator of a container are not satisfied, that operation shall be deleted.
A futher problem highlighted by this approach is that there are no constraints on the copy-assignment operator, so that vector<unique_ptr<T>> should be CopyAssignable! However, we can lift the equivalent constraints from the Allocator-aware container requirements.
[08-2016, Chicago]
Fri PM: Move to OPen
Proposed resolution:
This wording is relative to N4606.
| Expression | Return type | Operational semantics | Assertion/note/pre-/post-condition | Complexity |
| X(a) |
Requires: T is CopyInsertable into
X (see below) post: a == X(a). |
linear | ||
| X u(a) X u = a; |
Requires: T is CopyInsertable into
X (see below) post: u == a. |
linear | ||
| ... | ... | ... | ... | ... |
| r = a | X& |
Requires: T is CopyInsertable into X
and CopyAssignable, otherwise this expression shall be ill-formed. post: r == a. |
linear |
| Expression | Return type | Operational semantics | Assertion/note/pre-/post-condition | Complexity |
| a = t | X& |
Requires: T is CopyInsertable into X and
CopyAssignable post: r == a. |
linear |
Section: 23.3.11.5 [vector.modifiers], 23.2 [container.requirements] Status: Open Submitter: Howard Hinnant Opened: 2012-07-07 Last modified: 2017-02-03
Priority: 2
View all other issues in [vector.modifiers].
View all issues with Open status.
Discussion:
Nikolay Ivchenkov recently brought the following example on the std-discussion newsgroup, asking whether the following program well-defined:
#include <iostream>
#include <vector>
int main()
{
std::vector<int> v;
v.reserve(4);
v = { 1, 2, 3 };
v.emplace(v.begin(), v.back());
for (int x : v)
std::cout << x << std::endl;
}
Nikolay Ivchenkov:
I think that an implementation of vector's 'emplace' should initialize an intermediate object with v.back() before any shifts take place, then perform all necessary shifts and finally replace the value pointed to by v.begin() with the value of the intermediate object. So, I would expect the following output:3 1 2 3
GNU C++ 4.7.1 and GNU C++ 4.8.0 produce other results:
2 1 2 3
Howard Hinnant:
I believe Nikolay is correct that vector should initialize an intermediate object with v.back() before any shifts take place. As Nikolay pointed out in another email, this appears to be the only way to satisfy the strong exception guarantee when an exception is not thrown by T's copy constructor, move constructor, copy assignment operator, or move assignment operator as specified by 23.3.11.5 [vector.modifiers]/p1. I.e. if the emplace construction throws, the vector must remain unaltered. That leads to an implementation that tolerates objects bound to the function parameter pack of the emplace member function may be elements or sub-objects of elements of the container. My position is that the standard is correct as written, but needs a clarification in this area. Self-referencing emplace should be legal and give the result Nikolay expects. The proposed resolution of LWG 760 is not correct.[2015-02 Cologne]
LWG agrees with the analysis including the assessment of LWG 760 and would appreciate a concrete wording proposal.
[2015-04-07 dyp comments]
The Standard currently does not require that creation of such intermediate objects is legal. 23.2.3 [sequence.reqmts] Table 100 — "Sequence container requirements" currently specifies:
Table 100 — Sequence container requirements Expression Return type Assertion/note
pre-/post-condition… a.emplace(p, args); iterator Requires: T is EmplaceConstructible into X from args. For vector and deque, T is also MoveInsertable into X and MoveAssignable. […] …
The EmplaceConstructible concept is defined via allocator_traits<A>::construct in 23.2.1 [container.requirements.general] p15.5 That's surprising to me since the related concepts use the suffix Insertable if they refer to the allocator. An additional requirement such as std::is_constructible<T, Args...> is necessary to allow creation of intermediate objects.
The creation of intermediate objects also affects other functions, such as vector.insert. Since aliasing the vector is only allowed for the single-element forms of insert and emplace (see 526), the range-forms are not affected. Similarly, aliasing is not allowed for the rvalue-reference overload. See also LWG 2266.
There might be a problem with a requirement of std::is_constructible<T, Args...> related to the issues described in LWG 2461. For example, a scoped allocator adapter passes additional arguments to the constructor of the value type. This is currently not done in recent implementations of libstdc++ and libc++ when creating the intermediate objects, they simply create the intermediate object by perfectly forwarding the arguments. If such an intermediate object is then moved to its final destination in the vector, a change of the allocator instance might be required — potentially leading to an expensive copy. One can also imagine worse problems, such as run-time errors (allocators not comparing equal at run-time) or compile-time errors (if the value type cannot be created without the additional arguments). I have not looked in detail into this issue, but I'd be reluctant adding a requirement such as std::is_constructible<T, Args...> without further investigation.
It should be noted that the creation of intermediate objects currently is inconsistent in libstdc++ vs libc++. For example, libstdc++ creates an intermediate object for vector.insert, but not vector.emplace, whereas libc++ does the exact opposite in this respect.
A live demo of the inconsistent creation of intermediate objects can be found here.
[2015-10, Kona Saturday afternoon]
HH: If it were easy, it'd have wording. Over the decades I have flipped 180 degrees on this. My current position is that it should work even if the element is in the same container.
TK: What's the implentation status? JW: Broken in GCC. STL: Broken in MSVS. Users complain about this every year.
MC: 526 says push_back has to work.
HH: I think you have to make a copy of the element anyway for reasons of exception safety. [Discussion of exception guarantees]
STL: vector has strong exception guarantees. Could we not just provide the Basic guarantee here.
HH: It would terrify me to relax that guarantee. It'd be an ugly, imperceptible runtime error.
HH: I agree if we had a clean slate that strong exception safety is costing us here, and we shouldn't provide it if it costs us.
STL: I have a mail here, "how can vector provide the strong guarantee when inserting in the middle".
HH: The crucial point is that you only get the strong guarantee if the exception is thrown by something other than the copy and move operations that are used to make the hole.
STL: I think we need to clean up the wording. But it does mandate currently that the self-emplacement must work, because nothings says that you can't do it. TK clarifies that a) self-emplacement must work, and b) you get the strong guarantee only if the operations for making the hole don't throw, otherwise basic. HH agrees. STL wants this to be clear in the Standard.
STL: Should it work for deque, too? HH: Yes.
HH: I will attempt wording for this.
TK: Maybe mail this to the reflector, and maybe someone has a good idea?
JW: I will definitely not come up with anything better, but I can critique wording.
Moved to Open; Howard to provide wording, with feedback from Jonathan.
Proposed resolution:
operator + in the description of the algorithmsSection: 25 [algorithms] Status: New Submitter: Nikolay Ivchenkov Opened: 2012-08-01 Last modified: 2017-02-03
Priority: 4
View other active issues in [algorithms].
View all other issues in [algorithms].
View all issues with New status.
Discussion:
According to 25.1 [algorithms.general]/12,
In the description of the algorithms operators + and - are used for some of the iterator categories for which they do not have to be defined. In these cases the semantics of a+n is the same as that of
X tmp = a; advance(tmp, n); return tmp;
There are several places where such operator + is applied to an output iterator — for example, see the description of std::copy:
template<class InputIterator, class OutputIterator> OutputIterator copy(InputIterator first, InputIterator last, OutputIterator result);-1- Effects: Copies elements in the range [first,last) into the range [result,result + (last - first)) starting from first and proceeding to last. For each non-negative integer n < (last - first), performs *(result + n) = *(first + n).
std::advance is not supposed to be applicable to output iterators, so we need a different method of description.
See also message c++std-lib-32908.[2014-06-07 Daniel comments and provides wording]
The specification for output iterators is somewhat tricky, because here a sequence of increments is required to be combined with intervening assignments to the dereferenced iterator. I tried to respect this fact by using a conceptual assignment operation as part of the specification.
Another problem in the provided as-if-code is the question which requirements are imposed on n. Unfortunately, the corresponding function advance is completely underspecified in this regard, so I couldn't borrow wording from it. We cannot even assume here that n is the difference type of the iterator, because for output iterators there is no requirements for this associated type to be defined. The presented wording attempts to minimize assumptions, but still can be considered as controversial.Proposed resolution:
This wording is relative to N4606.
Change 25.1 [algorithms.general] around p12 as indicated:
-12- In the description of the algorithms operators + and - are used for some of the iterator categories for which they do not have to be defined. In these cases the semantics of a+n is the same as that of
X tmp = a; advance(tmp, n); return tmp;when X meets the input iterator requirements (24.2.3 [input.iterators]), otherwise it is the same as that of
X tmp = a; for (auto i = n; i; ++tmp, (void) --i) *tmp = Expr(i); return tmp;where Expr(i) denotes the (n-i)th expression that is assigned to for the corresponding algorithm; and that of b-a is the same as of
return distance(a, b);
Section: 28.10.1 [re.results.const], 28.10.6 [re.results.all] Status: New Submitter: Pete Becker Opened: 2012-08-29 Last modified: 2017-02-03
Priority: 3
View other active issues in [re.results.const].
View all other issues in [re.results.const].
View all issues with New status.
Discussion:
28.10.1 [re.results.const] p1 says:
In all match_results constructors, a copy of the Allocator argument shall be used for any memory allocation performed by the constructor or member functions during the lifetime of the object.
There are three constructors:
match_results(const Allocator& = Allocator()); match_results(const match_results& m); match_results(match_results&& m) noexcept;
The second and third constructors do no have an Allocator argument, so despite the "all match_results constructors", it is not possible to use "the Allocator argument" for the second and third constructors.
The requirements for those two constructors also does not give any guidance. The second constructor has no language about allocators, and the third states that the stored Allocator value is move constructed from m.get_allocator(), but doesn't require using that allocator to allocate memory. The same basic problem recurs in 28.10.6 [re.results.all], which gives the required return value for get_allocator():Returns: A copy of the Allocator that was passed to the object's constructor or, if that allocator has been replaced, a copy of the most recent replacement.
Again, the second and third constructors do not take an Allocator, so there is nothing that meets this requirement when those constructors are used.
Proposed resolution:
Section: 28.10.1 [re.results.const], 28.10.6 [re.results.all] Status: New Submitter: Pete Becker Opened: 2012-08-29 Last modified: 2017-02-03
Priority: 3
View other active issues in [re.results.const].
View all other issues in [re.results.const].
View all issues with New status.
Discussion:
The effects of the two assignment operators are specified in Table 141. Table 141 makes no mention of allocators, so, presumably, they don't touch the target object's allocator. That's okay, but it leaves the question: match_results::get_allocator() is supposed to return "A copy of the Allocator that was passed to the object's constructor or, if that allocator has been replaced, a copy of the most recent replacement"; if assignment doesn't replace the allocator, how can the allocator be replaced?
Proposed resolution:
Section: 23.2.7.1 [unord.req.except] Status: Open Submitter: Alisdair Meredith Opened: 2012-09-23 Last modified: 2017-02-03
Priority: 3
View all issues with Open status.
Discussion:
The hash functor and key-comparison functor of unordered containers are allowed to throw on swap.
23.2.7.1 [unord.req.except]p3 "For unordered associative containers, no swap function throws an exception unless that exception is thrown by the swap of the container's Hash or Pred object (if any)."
In such a case we must offer the basic exception safety guarantee, where both objects are left in valid but unspecified states, and no resources are leaked. This yields a corrupt, un-usable container if the first swap succeeds, but the second fails by throwing, as the functors form a matched pair.
So our basic scenario is first, swap the allocators if the allocators propagate on swap, according to allocator_traits. Next we swap the pointers to our internal hash table data structures, so that they match the allocators that allocated them. (Typically, this operation cannot throw). Now our containers are back in a safely destructible state if an exception follows.
Next, let's say we swap the hash functor, and that throws. We have a corrupt data structure, in that the buckets are not correctly indexed by the correct functors, lookups will give unpredicatable results etc. We can safely restore a usable state by forcibly clearing each container - which does not leak resources and leaves us with two (empty but) usable containers.
Now let us assume that the hasher swap succeeds. Next we swap the equality comparator functor, and this too could throw. The important point to bear in mind is that these two functors form an important pairing - two objects that compare equal by the equality functor must also hash to the same value. If we swap one without the other, we most likely leave the container in an unusable state, even if we clear out all elements.
1. A colleague pointed out that the solution for this is to dynamically allocate the two functors, and then we need only swap pointers, which is not a throwing operation. And if we don't want to allocate on default construction (a common QoI request), we might consider moving to a dynamically allocated functors whenever swap is called, or on first insertion. Of course, allocating memory in swap is a whole new can of worms, but this does not really sound like the design we had intended.
2. The simplest option is to say that we do not support hasher or equality functors that throw on ADL swap. Note that the requirement is simply to not throw, rather than to be explicitly marked as noexcept. Throwing functors are allowed, so long as we never use values that would actually manifest a throw when used in an unordered container.
Pablo went on to give me several more options, to be sure we have a full set to consider:
3. Disallow one or the other functor from throwing. In that case, the possibly-throwing functor must be swapped first, then the other functor, the allocator, and the data pointer(s) afterwards (in any order -- there was a TC that allocator assignment and swap may not throw if the corresponding propagation trait is true.). Of course, the question becomes: which functor is allowed to throw and which one is not?
4. Require that any successful functor swap be reliably reversible. This is very inventive. I know of no other place in the standard where such a requirement is stated, though I have occasionally wanted such a guarantee.
5. Allow a failed swap to leave the containers in a state where future insertions may fail for reasons other than is currently allowed. Specifically, if the hash and equality functors are out of sync, all insertions will fail. Presumably some "incompletely swapped" exception would be thrown. This is "slightly" inventive, although people have been discussing "radioactive" states for a while.
[2013-03-15 Issues Teleconference]
Moved to Open.
Proposed resolution:
Section: 28.10.1 [re.results.const] Status: New Submitter: Pete Becker Opened: 2012-10-02 Last modified: 2017-02-03
Priority: 4
View other active issues in [re.results.const].
View all other issues in [re.results.const].
View all issues with New status.
Discussion:
28.10.1 [re.results.const]/3: "Move-constructs an object of class match_results satisfying the same postconditions as Table 141."
Table 141 lists various member functions and says that their results should be the results of the corresponding member function calls on m. But m has been moved from, so the actual requirement ought to be based on the value that m had before the move construction, not on m itself.
In addition to that, the requirements for the copy constructor should refer to Table 141.
Ganesh: Also, the requirements for move-assignment should refer to Table 141. Further it seems as if in Table 141 all phrases of "for all integers n < m.size()" should be replaced by "for all unsigned integers n < m.size()".Proposed resolution:
Section: 28.10 [re.results] Status: Open Submitter: Daniel Krügler Opened: 2012-10-06 Last modified: 2017-02-03
Priority: 3
View all other issues in [re.results].
View all issues with Open status.
Discussion:
The requirement expressed in 28.10 [re.results] p2
The class template match_results shall satisfy the requirements of an allocator-aware container and of a sequence container, as specified in 23.2.3 [sequence.reqmts], except that only operations defined for const-qualified sequence containers are supported.
can be read to require the existence of the described constructors from as well, but they do not exist in the synopsis.
The missing sequence constructors are:match_results(initializer_list<value_type>); match_results(size_type, const value_type&); template<class InputIterator> match_results(InputIterator, InputIterator);
The missing allocator-aware container constructors are:
match_results(const match_results&, const Allocator&); match_results(match_results&&, const Allocator&);
It should be clarified, whether (a) constructors are an exception of above mentioned operations or (b) whether at least some of them (like those accepting a match_results value and an allocator) should be added.
As visible in several places of the standard (including the core language), constructors seem usually to be considered as "operations" and they certainly can be invoked for const-qualified objects. The below given proposed resolution applies only the minimum necessary fix, i.e. it excludes constructors from above requirement.[2013-04-20, Bristol]
Check current implementations to see what they do and, possibly, write a paper.
[2013-09 Chicago]
Ask Daniel to update the proposed wording to include the allocator copy and move constructors.
[2014-01-18 Daniel changes proposed resolution]
Previous resolution from Daniel [SUPERSEDED]:
Change 28.10 [re.results] p2 as indicated:
The class template match_results shall satisfy the requirements of an allocator-aware container and of a sequence container, as specified in 23.2.3 [sequence.reqmts], except that only operations defined for const-qualified sequence containers that are not constructors are supported.
[2015-05-06 Lenexa]
MC passes important knowledge to EF.
VV, RP: Looks good.
TK: Second form should be conditionally noexcept
JY: Sequence constructors are not here, but mentioned in the issue writeup. Why?
TK: That would have been fixed by the superseded wording.
JW: How does this interact with Mike Spertus' allocator-aware regexes? [...] Perhaps it doesn't.
JW: Can't create match_results, want both old and new resolution.
JY: It's problematic that users can't create these, but not this issue.
VV: Why conditional noexcept?
MC: Allocator move might throw.
JW: Update superseded wording to "only non-constructor operations that are"?
MC: Only keep superseded, but append "and the means of constructing match_results are limited to [...]"?
JY: Bullet 4 paragraph 2 needs to address the allocator constructor.
Assigned to JW for drafting.
[2015-10, Kona Saturday afternoon]
STL: I want Mike Spertus to be aware of this issue.
Proposed resolution:
This wording is relative to N3936.
Change 28.10 [re.results] p4, class template match_results synopsis, as indicated:
[…] // 28.10.1, construct/copy/destroy: explicit match_results(const Allocator& a = Allocator()); match_results(const match_results& m); match_results(const match_results& m, const Allocator& a); match_results(match_results&& m) noexcept; match_results(match_results&& m, const Allocator& a) noexcept; […]
Change 28.10.1 [re.results.const] as indicated: [Drafting note: Paragraph 6 as currently written, makes not much sense, because the noexcept does not allow any exception to propagate. Further-on, the allocator requirements do not allow for throwing move constructors. Deleting it seems to be near to editorial — end drafting note]
match_results(const match_results& m); match_results(const match_results& m, const Allocator& a);-4- Effects: Constructs an object of class match_results, as a copy of m.
match_results(match_results&& m) noexcept; match_results(match_results&& m, const Allocator& a) noexcept;-5- Effects: Move-constructs an object of class match_results from m satisfying the same postconditions as Table 142.
AdditionallyFor the first form, the stored Allocator value is move constructed from m.get_allocator().-6- Throws: Nothing if the allocator's move constructor throws nothing.
Section: 23.2.7 [unord.req] Status: Open Submitter: Alisdair Meredith Opened: 2012-10-09 Last modified: 2017-02-03
Priority: 3
View other active issues in [unord.req].
View all other issues in [unord.req].
View all issues with Open status.
Discussion:
The user cannot specify a max_load_factor for their unordered container at construction, it must be supplied after the event, when the container is potentially not empty. The contract for this method is deliberately vague, not guaranteeing to use the value supplied by the user, and any value actually used will be used as a ceiling that the container will attempt to respect.
The only guarantee we have is that, if user requests a max_load_factor that is less than the current load_factor, then the operation will take constant time, thus outlawing an implementation that chooses to rehash and so preserve as a class invariant that load_factor < max_load_factor.
Reasonable options conforming to the standard include ignoring the user's request if the requested value is too low, or deferring the rehash to the next insert operation and allowing the container to have a strange state (wrt max_load_factor) until then - and there is still the question of rehashing if the next insert is for a duplicate key in a unique container.
Given the deliberate vagueness of the current wording, to support a range of reasonable (but not perfect) behaviors, it is not clear why the equally reasonable rehash to restore the constraint should be outlawed. It is not thought that this is a performance critical operation, where users will be repeatedly setting low load factors on populated containers, in a tight or (less unlikely) an instant response scenario.
[2013-03-15 Issues Teleconference]
Moved to Open.
Alisdair to provide wording.
[2016-11-12, Issaquah]
Sat PM: Howard to provide wording
[2016-11-17 Howard provided wording.]
The provided wording is consistent with LWG discussion in Issaquah. An implementation of the proposed wording would be setting max_load_factor() to max(z, load_factor()). This preserves the container invariant:
load_factor() <= max_load_factor()And it preserves the existing behavior that no rehash is done by this operation.
If it is desired to change the max_load_factor() to something smaller than the current load_factor() that can be done by first reducing the current load_factor() by either increasing bucket_count() (via rehash or reserve), or decreasing size() (e.g. erase), and then changing max_load_factor().
This resolution reaffirms that load_factor() <= max_load_factor() is a container invariant which can never be violated.
[2016-11-27, Nico comments]
Current implementations behave differently.
In regard to the sentence"The only guarantee we have is that, if user requests a max_load_factor that is less than the current load_factor, then the operation will take constant time, thus outlawing an implementation that chooses to rehash and so preserve as a class invariant that load_factor < max_load_factor."Note that the current spec says that there is constant complexity without any precondition. So, rehashing to keep the invariant would violate the spec (which is probably not be the intention). This issue is related to LWG 2199.
Proposed resolution:
Modify Table 87 as follows:
| Expression | Return type | Assertion/note pre-/post-condition | Complexity |
|---|---|---|---|
| a.max_load_factor(z) | void |
Pre: z shall be positive. May change the container's maximum load factor, uing z as a hint. Post: a.load_factor() <= a.max_load_factor() Note: a.load_factor() is not modified by this operation. |
Constant |
Section: 30.6.9 [futures.async] Status: Deferred Submitter: Detlef Vollmann Opened: 2012-10-19 Last modified: 2017-02-03
Priority: 4
View other active issues in [futures.async].
View all other issues in [futures.async].
Discussion:
promise, packaged_task, and async are the only places where a shared state is actually supposed to be allocated. Accordingly, promise and packaged_task are "allocator-aware". But function template async provides no way to provide an allocator.
[2013-09 Chicago]
Matt: deprecate async
Nico: read my paper Alisdair: defer issues to wait for polymorphic allocators Alisdair: defer, active topic of research Deferred[2014-02-20 Re-open Deferred issues as Priority 4]
[2015-05 Lenexa, SG1 response]
We want whatever status approximates: "will not fix; we're working on a replacement facility and don't want to add features to a broken one"
Proposed resolution:
Section: 23.2.3 [sequence.reqmts], 23.2.6 [associative.reqmts], 23.2.7 [unord.req], 26.6.1.2 [rand.req.seedseq] Status: Open Submitter: Jeffrey Yasskin Opened: 2012-10-21 Last modified: 2017-02-03
Priority: 3
View other active issues in [sequence.reqmts].
View all other issues in [sequence.reqmts].
View all issues with Open status.
Discussion:
In 23.2.3 [sequence.reqmts] p3, we have "il designates an object of type initializer_list<value_type>", and then several functions that take 'il' as an argument. However, an expression like {1, 2, 'a'} is not an object of type initializer_list<int> unless it's used to initialize an explicitly-typed variable of that type. I believe we want:
std::vector<int> v;
v = {1, 2, 'a'};
to compile portably, so we should say something different when defining 'il'. The same phrasing happens in 23.2.6 [associative.reqmts], 23.2.7 [unord.req], and 26.6.1.2 [rand.req.seedseq].
This may just be an editorial issue because the actual class synopses declare the functions to take initializer_list<exact_type>.[2013-03-15 Issues Teleconference]
Moved to Open.
This is definitely not NAD
Should copy the suggested wording as the proposed resolution.
Proposed resolution:
Section: 27.5.5.2 [basic.ios.cons] Status: Open Submitter: Andrey Semashev Opened: 2012-11-09 Last modified: 2017-02-03
Priority: 4
View all other issues in [basic.ios.cons].
View all issues with Open status.
Discussion:
There is an ambiguity in how std::basic_ios::init method (27.5.5.2 [basic.ios.cons]) can be used in the derived class. The Standard only specify the state of the basic_ios object after the call completes. However, in basic_ios default constructor description (27.5.5.2 [basic.ios.cons]) there is this sentence:
Effects: Constructs an object of class basic_ios (27.5.3.7 [ios.base.cons]) leaving its member objects uninitialized. The object shall be initialized by calling basic_ios::init before its first use or before it is destroyed, whichever comes first; otherwise the behavior is undefined.
This restriction hints that basic_ios::init should be called exactly once before the object can be used or destroyed, because basic_ios::init may not know whether it was called before or not (i.e. whether its members are actually uninitialized or are initialized by the previous call to basic_ios::init). There is no such restriction in the basic_ios::init preconditions so it is not clear whether it is allowed to call basic_ios::init multiple times or not.
This problem has already affected publicly available implementations. For example, Microsoft Visual C++ STL introduces a memory leak if basic_ios::init is called multiple times, while GCC 4.7 and STLPort reinitialize the basic_ios object correctly without memory leak or any other undesired effects. There was a discussion of this issue on Boost developers mailing list, and there is a test case that reproduces the problem. The test case is actually a bug report for my Boost.Log library, which attempts to cache basic_ostream-derived objects internally to avoid expensive construction and destruction. My stream objects allowed resetting the stream buffer pointers the stream is attached to, without requiring to destroy and construct the stream. My personal view of the problem and proposed resolution follows. While apparently the intent of basic_ios::init is to provide a way to initialize basic_ios after default construction, I see no reason to forbid it from being called multiple times to reinitialize the stream. Furthermore, it is possible to implement a conforming basic_ios that does not have this restriction. The quoted above section of the Standard that describes the effects of the default constructor is misleading. The Standard does not mandate any data members of basic_ios or ios_base (27.5.3 [ios.base]), which it derives from. This means that the implementation is allowed to use non-POD data members with default constructors that initialize the members with particular default values. For example, in the case of Microsoft Visual C++ STL the leaked memory is an std::locale instance that is dynamically allocated during basic_ios::init, a raw pointer to which is stored within ios_base. It is possible to store e.g. an unique_ptr instead of a raw pointer as a member of ios_base, the smart pointer will default initialize the underlying raw pointer on default construction and automatically destroy the allocated object upon being reset or destroyed, which would eliminate the leak and allow basic_ios::init to be called multiple times. This leads to conclusion that the default constructor of basic_ios cannot leave "its member objects uninitialized" but instead performs default initialization of the member objects, which would mean the same thing in case of POD types. However, I feel that restricting ios_base and basic_ios members to non-POD types is not acceptable. Since multiple calls to basic_ios::init are not forbidden by the Standard, I propose to correct the basic_ios default constructor description so that it is allowed to destroy basic_ios object without calling basic_ios::init. This would imply that any raw members of basic_ios and ios_base should be initialized to values suitable for destruction (essentially, this means only initializing raw pointers to NULL). The new wording could look like this:Effects: Constructs an object of class basic_ios (27.5.3.7 [ios.base.cons]) initializing its member objects to unspecified state, only suitable for basic_ios destruction. The object shall be initialized by calling basic_ios::init before its first use; otherwise the behavior is undefined.
This would remove the hint that basic_ios::init must be called exactly once. Also, this would remove the requirement for basic_ios::init to be called at all before the destruction. This is also an important issue because the derived stream constructor may throw an exception before it manages to call basic_ios::init (for example, if the streambuf constructor throws), and in this case the basic_ios destructor has undefined behavior.
To my mind, the described modification is sufficient to resolve the issue. But to emphasize the possibility to call basic_ios::init multiple times, a remark or a footnote for basic_ios::init postconditions could be added to explicitly state the semantics of calling it multiple times. The note could read as follows:The function can be called multiple times during the object lifetime. Each subsequent call reinitializes the object to the described in postconditions initial state.
[2013-04-20, Bristol]
Alisdair: The current wording is unclear but the proposed resolution is wrong
Solution: Clarify that init must be called once and only once. Move then to review.Proposed resolution:
This wording is relative to N3485.
Edit 27.5.5.2 [basic.ios.cons] as indicated:
basic_ios();-2- Effects: Constructs an object of class basic_ios (27.5.3.7 [ios.base.cons])
leaving its member objects uninitializedinitializing its member objects to unspecified state, only suitable for basic_ios destruction. The object shall be initialized by calling basic_ios::init before its first useor before it is destroyed, whichever comes first; otherwise the behavior is undefined.void init(basic_streambuf<charT,traits>* sb);Postconditions: The postconditions of this function are indicated in Table 128.
-?- Remarks: The function can be called multiple times during the object lifetime. Each subsequent call reinitializes the object to the described in postconditions initial state.
Section: 23.2.6 [associative.reqmts], 23.2.7 [unord.req] Status: Open Submitter: Alisdair Meredith Opened: 2012-11-14 Last modified: 2017-02-03
Priority: 3
View other active issues in [associative.reqmts].
View all other issues in [associative.reqmts].
View all issues with Open status.
Discussion:
The requirements on the functors used to arrange elements in the various associative and unordered containers are given by a set of expressions in tables 102 — Associative container requirements, and 103 — Unordered associative container requirements. In keeping with Library convention these expressions make the minimal requirements necessary on their types. For example, we have the following 3 row extracts for the unordered containers:
| Expression | Assertion/note pre-/post-condition |
X(n, hf, eq) X a(n, hf, eq) |
Requires: hasher and key_equal are CopyConstructible. |
X(n, hf) X a(n, hf) |
Requires: hasher is CopyConstructible and key_equal is DefaultConstructible. |
X(n) X a(n) |
Requires: hasher and key_equal are DefaultConstructible. |
However, the signature for each class template requires that the functors must effectively be CopyConstructible for each of these expressions:
template <class Key,
class T,
class Hash = hash<Key>,
class Pred = std::equal_to<Key>,
class Allocator = std::allocator<std::pair<const Key, T> > >
class unordered_map
{
...
// construct/destroy/copy
explicit unordered_map(size_type n = see below,
const hasher& hf = hasher(),
const key_equal& eql = key_equal(),
const allocator_type& a = allocator_type());
...
}
The letter of the standard can be honored as long as implementors recognize their freedom to split this one signature into multiple overloads, so that the documented default arguments (requiring a CopyConstructible functor) are not actually passed as default arguments.
As we look into the requirements for the copy constructor and copy-assignment operator, the requirements are even more vague, as the explicit requirements on the functors are not called out, other than saying that the functors are copied.
Must the functors be CopyAssignable? Or is CopyConstructible sufficient in this case? Do we require that the functors be Swappable so that the copy-swap idiom can be deployed here? Note that a type that is both CopyConstructible and CopyAssignable is still not guaranteed to be Swappable as the user may delete the swap function for their type in their own namespace, which would be found via ADL.
Some clean-up of the requirements table looks necessary, to at least document the assignment behavior. In addition, we should have clear guidance on whether these functors should always be CopyConstructible, as suggested by the class template definitions, or if the requirement tables are correct and we should explicitly split up the constructors in the (unordered) associative containers to no longer use default (function) arguments to obtain their defaulted functors.
I recommend the simplest solution would be to always require that the functors for (unordered) associative containers be CopyConstructible, above the requirements tables themselves, so that the issue need not be addressed within the tables. I suggest that the assignment operators for these containers add the requirement that the functors be Swappable, rather than forwarding the corresponding Assignable requirement.
[2013-03-15 Issues Teleconference]
Moved to Open.
Alisdair to propose wording.
[2014-06-08, Daniel comments]
The area of this issue partially overlaps what LWG 2227 addresses.
[2015-10-20, Daniel comments]
The revised resolution of LWG 2227 should resolve this issue as well. It follows the recommendations of the submitter to require CopyConstructible requirements for the function objects owned by containers, but it does not impose any further fundamental requirements.
Proposed resolution:
See the resolution of LWG 2227.
Section: 28.11.4 [re.alg.replace] Status: New Submitter: Jeffrey Yasskin Opened: 2012-11-26 Last modified: 2017-02-03
Priority: 3
View all other issues in [re.alg.replace].
View all issues with New status.
Discussion:
template <class traits, class charT, class ST, class SA>
basic_string<charT, ST, SA>
regex_replace(const basic_string<charT, ST, SA>& s,
const basic_regex<charT, traits>& e,
const charT* fmt,
regex_constants::match_flag_type flags =
regex_constants::match_default);
and friends are documented as
Constructs an empty string result of type basic_string<charT, ST, SA> and calls regex_replace(back_inserter(result), s.begin(), s.end(), e, fmt, flags).
This appears to require the result to have a default-constructed allocator, which isn't even possible for all allocator types. I suspect the allocator should be copied from 's' instead. Possibly there should be an additional defaulted argument to override the allocator of the result.
Proposed resolution:
Section: 28.12.2.2 [re.tokiter.comp] Status: New Submitter: Pete Becker Opened: 2012-11-21 Last modified: 2017-02-03
Priority: 2
View all issues with New status.
Discussion:
Consider the following example:
std::string str0("x");
std::regex rg0("a");
std::regex_token_iterator it0(str0.begin(), str0.end(), rg0, -1); // points at "x" in str0
std::string str1("x");
std::regex rg1("b");
std::regex_token_iterator it1(str1.begin(), str1.end(), rg1, -1); // points at "x" in str1
28.12.2.2 [re.tokiter.comp] p1 says that it0.operator==(it1) returns true "if *this and right are both suffix iterators and suffix == right.suffix"; both conditions are satisfied in this example. It does not say that they must both be iterators into the same sequence, nor does it say (as general iterator requirements do) that they must both be in the domain of == in order for the comparison to be meaningful. It's a simple statement: they're equal if the strings they point at compare equal. Given this being a valid comparison, the obtained result of "true" looks odd.
The problem is that for iterator values prior to the suffix iterator, equality means the same regular expression and the same matched sequence (both uses of "same" refer to identity, not equality); for the suffix iterator, equality means that the matched sequences compare equal.Proposed resolution:
Section: 99 [depr.conversions.string] Status: LEWG Submitter: Glen Fernandes Opened: 2012-12-14 Last modified: 2017-05-31
Priority: Not Prioritized
View other active issues in [depr.conversions.string].
View all other issues in [depr.conversions.string].
View all issues with LEWG status.
Discussion:
The wstring_convert class template, described in 99 [depr.conversions.string], does not support custom stateful allocators. It only supports custom stateless allocators.
The to_bytes member function returns basic_string<char, char_traits<char>, Byte_alloc> but it does not take an instance of Byte_alloc to pass to the constructor of the basic_string. Similarly the from_bytes member function returns basic_string<Elem, char_traits<Elem>, Wide_alloc> but it does not take an instance of Wide_alloc to pass to the constructor of the basic_string. This makes these two member functions and the wstring_convert class template not usable when Wide_alloc or Byte_alloc are stateful allocators.[2013-01-22, Glen provides wording]
[2013-03-15 Issues Teleconference]
Moved to NAD Future.
This is clearly an extension that the LEWG may want to take a look at, once we have more experience with appropriate use of allocators with the C++11 model.
Proposed resolution:
This wording is relative to N3485.
In 99 [depr.conversions.string]/2 and /6 "Class template wstring_convert synopsis" change the overloads of the member function from_bytes() so that all four overloads take an additional parameter which is an instance of Wide_alloc:
wide_string from_bytes(char byte, const Wide_alloc& alloc = Wide_alloc()); wide_string from_bytes(const char *ptr, const Wide_alloc& alloc = Wide_alloc()); wide_string from_bytes(const byte_string& str, const Wide_alloc& alloc = Wide_alloc()); wide_string from_bytes(const char *first, const char *last, const Wide_alloc& alloc = Wide_alloc());
In 99 [depr.conversions.string] /8 specify that this Wide_alloc allocator parameter is used to construct the wide_string object returned from the function:
-7- Effects: The first member function shall convert the single-element sequence byte to a wide string. The second member function shall convert the null-terminated sequence beginning at ptr to a wide string. The third member function shall convert the sequence stored in str to a wide string. The fourth member function shall convert the sequence defined by the range [first, last) to a wide string.
-8- In all cases:If the cvtstate object was not constructed with an explicit value, it shall be set to its default value (the initial conversion state) before the conversion begins. Otherwise it shall be left unchanged.
The number of input elements successfully converted shall be stored in cvtcount.
The Wide_alloc allocator parameter is used to construct the wide_string object returned from the function.
In 99 [depr.conversions.string]/2 and /12 "Class template wstring_convert synopsis" change the overloads of the member function to_bytes() so that all four overloads take an additional parameter which is an instance of Byte_alloc:
byte_string to_bytes(Elem wchar, const Byte_alloc& alloc = Byte_alloc()); byte_string to_bytes(const Elem *wptr, const Byte_alloc& alloc = Byte_alloc()); byte_string to_bytes(const wide_string& wstr, const Byte_alloc& alloc = Byte_alloc()); byte_string to_bytes(const Elem *first, const Elem *last, const Byte_alloc& alloc = Byte_alloc());
In 99 [depr.conversions.string] /13 specify that this Byte_alloc allocator parameter is used to construct the byte_string object returned from the function:
-12- Effects: The first member function shall convert the single-element sequence wchar to a byte string. The second member function shall convert the null-terminated sequence beginning at wptr to a byte string. The third member function shall convert the sequence stored in wstr to a byte string. The fourth member function shall convert the sequence defined by the range [first, last) to a byte string.
-13- In all cases:If the cvtstate object was not constructed with an explicit value, it shall be set to its default value (the initial conversion state) before the conversion begins. Otherwise it shall be left unchanged.
The number of input elements successfully converted shall be stored in cvtcount.
The Byte_alloc allocator parameter is used to construct the byte_string object returned from the function.
Section: 23.2.6 [associative.reqmts] Status: Open Submitter: Juan Soulie Opened: 2012-12-19 Last modified: 2017-02-03
Priority: 3
View other active issues in [associative.reqmts].
View all other issues in [associative.reqmts].
View all issues with Open status.
Discussion:
Table 102 in 23.2.6 [associative.reqmts]/8 states on expression a.key_comp() that it "returns the comparison object out of which a was constructed". At the same time, 23.2.1 [container.requirements.general]/8 states (starting in the third line) that "...Any Compare, Pred, or Hash objects belonging to a and b shall be swappable and shall be exchanged by unqualified calls to non-member swap...". This is problematic for any compliant implementation, since once swapped the container cannot return the comparison object out of which it was constructed unless incurring in storing an otherwise needless object.
The simple solution is to correct that statement in Table 102, but I believe this is part of a larger problem of underspecified behavior: The new standard has made an effort in regards to allocators and now fully specifies what happens to stateful allocator objects. It has even specified what happens to stateful hasher and key_equal members of unordered containers (they propagate), but it says nothing about stateful comparison objects of (ordered) associative containers, except for the statement in 23.2.1 [container.requirements.general]/8 referred above and only related to swap. For example, it is unclear to me what is specified to happen on an assignment: should the comparison object be copied/moved along with the elements, or should the left-hand side object keep its own? Maybe this has been intentionally left unspecified with the purpose of compatibility with C++98, which I understand it specified that comparison objects were kept for the entire life of the container (like allocators) — an unfortunate choice. But anyway, the segment of 23.2.1 [container.requirements.general] quoted above seems to break any possible backwards compatibility with C++98 in this regard. Therefore, taking into consideration consistency with how this is dealed with for unordered associative containers, I propose that Table 102 is modified as follows:The row for expression a.key_comp() is changed so that its "assertion/note pre-/post-condition" reads "Returns a's comparison object."
A new row is added at the appropriate location (which I believe would be after "X(il)" row), with:
Table 102 — Associative container requirements (in addition to container) Expression Return type Assertion/note pre-/post-condition Complexity X(b)
X a(b)X Copy constructor. In addition to
the requirements of Table 96, copies
the comparison object.Linear in b.size() a = b X& Copy assignment operator. In addition to
the requirements of Table 96, copies the
comparison object.Linear in a.size() and b.size()
[2013-03-15 Issues Teleconference]
Moved to Review.
[2013-04-18, Bristol]
STL: can't believe we don't specify this already. this is totally necessary
Alisdair: how does it do this? copy construction? assignment? Also need it for move. STL: we already specify this for constructing from a comparator, not during copy construction though. Jonathan: don't like wording, should say "key_compare is CopyConstructible. Uses b.key_comp() as a comparison object." STL: we get it right for unordered! Jonathan: can't wordsmith this now, but I think implementations do the right thing. Alisdair: not sure what right thing is for moves. Also we say nothing about propagating allocators to functors.Moved to Open.
[2015-02 Cologne]
TK: There's no need for fine-grained propagate/not-propagate control. If you don't want to propagate the predicate, you can simply construct or insert from an iterator range.
VV: libstdc++ already implements the resolution of this issue. GR: There are a couple of other problems. We don't specify move constructor and move assignment for maps. Those are just general. TK: General container requirements already describe the semantics for {copy,move}-{construction,assignment}, so it doesn't seem that there's room for choice in std::map assignments. unordered_map is different, though. [Note: Check what general container requirements say about container equality.] DK will draft wording. The decision is to unambiguously make all {copy,move}-{construction,assignment} operations endow the LHS with the exact state of the RHS, including all predicates and hash function states. Conclusion: Update wording, revisit later.[2015-05-06 Lenexa: Waiting for updated wording]
Previous resolution [SUPERSEDED]:
This wording is relative to N3485.
Change Table 102 as indicated:
Table 102 — Associative container requirements (in addition to container) Expression Return type Assertion/note pre-/post-condition Complexity … X(il) Same as X(il.begin(), il.end()). same as X(il.begin(), il.end()). X(b)
X a(b)Copy constructor. In addition to
the requirements of Table 96, copies
the comparison object.Linear in b.size() a = b X& Copy assignment operator. In addition to
the requirements of Table 96, copies the
comparison object.Linear in a.size() and b.size() … a.key_comp() X::key_compare rReturnsthea's comparison object
out of which a was constructed.constant
[2015-10-19 Daniel comments and provides alternative wording]
The current standard is especially unclear in regard to what effects move operations of unordered/associative containers should have. We have one example that is standardized exactly in this way by looking at 23.6.5.2 [priqueue.cons.alloc] p7:
template <class Alloc> priority_queue(priority_queue&& q, const Alloc& a);-7- Effects: Initializes c with std::move(q.c) as the first argument and a as the second argument, and initializes comp with std::move(q.comp)
A similarly comparable example are the move-operations of std::unique_ptr in regard to the deleter (when this is no a reference), which also respect move-capabilities of that function object.
We have wording from C++98 for associative containers (but not for unordered containers!) that was never adjusted to C++11 move-semantics in 23.2.6 [associative.reqmts] p12:When an associative container is constructed by passing a comparison object the container shall not store a pointer or reference to the passed object, even if that object is passed by reference. When an associative container is copied, either through a copy constructor or an assignment operator, the target container shall then use the comparison object from the container being copied, as if that comparison object had been passed to the target container in its constructor.
The second sentence of this wording is problematic for several reasons:
It only talks about copy operations, not about move operations, except that the term "assignment" without leading "copy" is a bit ambigious (albeit it seems clear in the complete context).
It is not really clear how to interpret "as if that comparison object had been passed to the target container in its constructor" for an assignment operation. A possible but not conclusive interpretation could be that this is wording supporting a "copy-via-swap" idiom.
There does not exist similar wording for unordered containers, except that Table 102 provides entries for copy construction and copy assignment of the containers whose wording just talks of "copies" in either case.
Existing implementations differ already:
Visual Studio 2015 uses copy construction and copy assignment for the two copy operations but uses swap operations for the move operations.
GCC's libstdc++ performs copy construction and copy assignment for the two copy operations and for the two move operations, respectively
clang++'s libc++ performs copy/move construction and copy/move assignment for the corresponding four copy/move operations
The alternative wording provided below attempts to clarify that container copy/move operations perform the corresponding copy/move operations on the owned function objects.
In addition the wording also resolves LWG 2215: I believe that the current wording should require that container function objects should meet the CopyConstructible requirements. Adding this general requirement also fixes the underspecified requirements of the accessor functions key_comp() and value_comp(). I don't think that a general requirement for Swappable is needed, only the member swap function currently requires this. Nonetheless the wording below does support stateful functors that are also moveable or move-assignable, therefore the specified semantics in terms of move operations. I should add the following warning, though: If this proposed wording would be accepted, there is a little chance of code breakage, because the current wording can be read that in general there is no requirement that the container functors are CopyConstructible. The following code example is accepted by gcc + libstd++:
#include <map>
#include <utility>
#include <iostream>
struct Cmp {
Cmp() = default;
Cmp(const Cmp&) = delete;
Cmp(Cmp&&) = delete;
Cmp& operator=(const Cmp&) = delete;
Cmp& operator=(Cmp&&) = delete;
template<class T>
bool operator()(const T& x, const T& y) const
{
return x < y;
}
};
typedef std::map<int, int, Cmp> MyMap;
int main() {
MyMap m;
std::cout << (m.find(12) == m.end()) << std::endl;
}
Previous resolution [SUPERSEDED]:
This wording is relative to N4527.
Change 23.2.6 [associative.reqmts] p8 as indicated:
-8- In Table 101, X denotes an associative container class, a denotes a value of type X, b denotes a possibly const value of type X, rv denotes a non-const rvalue of type X, u denotes the name of a variable being declared, […]
Change Table 101 as indicated:
Table 101 — Associative container requirements (in addition to container) Expression Return type Assertion/note pre-/post-condition Complexity … X::key_compare Compare Requires: Compare is CopyConstructible.
defaults to less<key_type>compile time X(c)
X u(c);Requires: key_compare is CopyConstructible.Effects: Constructs an empty container.
Uses a copy of c as a comparison object.[…] … X(i,j,c)
X u(i,j,c);Requires: key_compare is CopyConstructible.value_type is EmplaceConstructible into X from *i.
Effects: Constructs an empty container and inserts elements
from the range [i, j) into it; uses c as a comparison object.[…] … X(il) Same as X(il.begin(), il.end()). same as X(il.begin(), il.end()). X(b)
X a(b)(In addition to the requirements of Table 95)
Effects: Copy constructs the comparison object of a from
the comparison object of b.Linear in b.size() X(rv)
X a(rv)(In addition to the requirements of Table 95 and Table 98)
Effects: Move constructs the comparison object of a from
the comparison object of rv.constant a = b X& (In addition to the requirements of Table 95 and Table 98)
Requires: key_compare is CopyAssignable.
Effects: Copy assigns the comparison object of b
to the comparison object of a.Linear in a.size() and b.size() a = rv X& (In addition to the requirements of Table 95 and Table 98)
Requires: key_compare is MoveAssignable.
Effects: Move assigns from the comparison object of rv
to the comparison object of a.Linear … a.key_comp() X::key_compare rReturnsthea's comparison object
out of which a was constructed.constant Change 23.2.6 [associative.reqmts] p12 as indicated:
-12- When an associative container is constructed by passing a comparison object the container shall not store a pointer or reference to the passed object, even if that object is passed by reference.
When an associative container is copied, either through a copy constructor or an assignment operator, the target container shall then use the comparison object from the container being copied, as if that comparison object had been passed to the target container in its constructor.Change 23.2.7 [unord.req] p11 as indicated:
-11- In Table 102: X denotes an unordered associative container class, a denotes a value of type X, b denotes a possibly const value of type X, rv denotes a non-const rvalue of type X, […]
Change Table 102 as indicated:
Table 102 — Unordered associative container requirements (in addition to container) Expression Return type Assertion/note pre-/post-condition Complexity … X::hasher Hash Requires: Hash is CopyConstructible.
Hash shall be a unary function object type
such that the expression hf(k) has type std::size_t.compile time X::key_equal Pred Requires: Pred is CopyConstructible.
Pred shall be a binary predicate that takes
two arguments of type Key.
Pred is an equivalence relation.compile time … X(n, hf, eq)
X a(n, hf, eq)X Requires: hasher and key_equal are CopyConstructible.Effects: […][…] X(n, hf)
X a(n, hf)X Requires: hasher is CopyConstructible andkey_equal is DefaultConstructible.
Effects: […][…] … X(i, j, n, hf, eq)
X a(i, j, n, hf, eq)X Requires: hasher and key_equal are CopyConstructible.value_type is EmplaceConstructible into X from *i.
Effects: […][…] X(i, j, n, hf)
X a(i, j, n, hf)X Requires: hasher is CopyConstructible andkey_equal is DefaultConstructible.
value_type is EmplaceConstructible into X from *i.
Effects: […][…] … X(b)
X a(b)X Copy constructor. In addition(In addition to the requirements of Table 95)
to the requirements of Table 95,
copies the hash function,
predicate, and maximum load
factor.
Effects: Copy constructs the hash function, predicate, and maximum load factor
of a from the corresponding objects of b.Average case linear in
b.size(),
worst case quadratic.X(rv)
X a(rv)X (In addition to the requirements of Table 95 and Table 98)
Effects: Move constructs the hash function, predicate, and maximum load factor
of a from the corresponding objects of rv.constant a = b X& Copy assignment operator. In(In addition to the requirements of Table 95 and Table 98)
addition to the requirements of
Table 95, copies the hash
function, predicate, and
maximum load factor.
Requires: hasher and key_equal are CopyAssignable.
Effects: Copy assigns the hash function, predicate, and maximum load factor
of b to the corresponding objects of a.Average case linear in
b.size(),
worst case quadratic.a = rv X& (In addition to the requirements of Table 95 and Table 98)
Requires: hasher and key_equal are MoveAssignable.
Effects: Move assigns the hash function, predicate, and maximum load factor
from rv to the corresponding objects of a.Linear …
[2016-08-07]
Daniel removes the previously proposed wording to work on revised wording.
Proposed resolution:
Section: 29.2 [atomics.syn], 29.3 [atomics.order] Status: SG1 Submitter: Daniel Krügler Opened: 2013-01-19 Last modified: 2017-02-03
Priority: Not Prioritized
View all other issues in [atomics.syn].
View all issues with SG1 status.
Discussion:
The "magic" kill_dependency function is a function without any constraints on the template parameter T and is specified as
template <class T> T kill_dependency(T y) noexcept;-14- Effects: The argument does not carry a dependency to the return value (1.10).
-15- Returns: y.
I wonder whether the unconditional noexcept is really intended here: Assume we have some type U that has a potentially throwing move constructor (or it has a potentially throwing copy constructor and no move constructor), for any "normal" function template with the same signature and the same effects (modulo the dependency magic) this would mean that it cannot safely be declared noexcept because of the return statement being part of the complete function call affected by noexcept (The by-value function argument is irrelevant in this context). In other words it seems that a function call such as
struct S {
...
S(const S& r) { if(some condition) throw Something(); }
...
};
int main() {
S s1 = ...;
S s2 = std::kill_dependency(s1);
}
would be required to call std::terminate if the copy constructor of S throws during the return of std::kill_dependency.
To require copy elision for this already magic function would look like a low-hanging fruit to solve this problem, but this case is not covered by current copy elision rules see 12.8 p31 b1: "— in a return statement in a function with a class return type, when the expression is the name of a non-volatile automatic object (other than a function or catch-clause parameter) with the same cv-unqualified type as the function return type, the copy/move operation can be omitted by constructing the automatic object directly into the function's return value". Some options come into my mind:Make the exception-specification a constrained one in regard via std::is_nothrow_move_constructible:
template <class T> T kill_dependency(T y) noexcept(see below);
This is similar to the approach taken for function templates such as std::swap.
Use perfect forwarding (This needs further wording to correct the effects):
template <class T> T&& kill_dependency(T&& y) noexcept;
Impose constraints on the template arguments in regard to throwing exceptions while copying/moving.
Keep the state as it is but possibly add a note about a call of std::terminate in above scenario.
A second problem is that the current wording is not clear whether it is well-defined to call the function with types that are reference types, such as in the following example:
#include <atomic>
int main()
{
int a = 12;
int& b = std::kill_dependency<int&>(a);
}
It is unclear what kind of dependency is killed here. This is presumably a core language problem, but could affect the possible resolutions of the problem.
[2014-11 Urbana]
Recommend using a revised example:
int lookup(class D* p)
{
class E* q = p->a.load(memory_order_consume);
int y = std::kill_dependency(q->y);
}
[2015-02 Cologne]
Handed over to SG1.
Proposed resolution:
Section: 21.5 [c.strings] Status: New Submitter: Jason Merrill Opened: 2013-01-29 Last modified: 2017-02-03
Priority: 4
View other active issues in [c.strings].
View all other issues in [c.strings].
View all issues with New status.
Discussion:
Apparently C1X changes __STDC_UTF_16__ and __STDC_UTF_32__ from macros defined in uchar.h (and reflected in C++ by Table 79) to be predefined by the compiler. Do we want to do the same?
Proposed resolution:
Section: 21.5 [c.strings] Status: Open Submitter: Johannes Schaub Opened: 2013-02-02 Last modified: 2017-02-03
Priority: 3
View other active issues in [c.strings].
View all other issues in [c.strings].
View all issues with Open status.
Discussion:
The non-explicit nature of the iterator-pair constructor of containers, such a
template <class InputIterator> vector(InputIterator first, InputIterator last, const Allocator& = Allocator());
can be selected in unexpected situations, leading to a hard runtime error, as demonstrated by the following example:
#include <vector>
void f(std::vector<char> v){ /* ... */}
int main() {
f({"A", "B"});
}
The actually intended initializer-list constructor isn't feasible here, so the best match is the constructor template
template <class InputIterator> vector(InputIterator first, InputIterator last, const Allocator& = Allocator());
This compiles, but will result in code running amok. The potential trap (that cannot be easily detected by the library implementation) could be reduced by making this constructor explicit. It would still have the effect to be selected here, but the code would be ill-formed, so the programmer gets a clear message here.
[2014-06 Rapperswil]
JW: can't fix this, don't want to touch this, Do The Right Thing clause has been a source of tricky issues. only really happens with string literals, that's the only way to create an array that isn't obviously an array
GR: want to see paper AM: is it only string literals, or also UDLs? STL: maybe, but we don't need to deal with that. This is only a problem in a very specific case Leave as Open.Proposed resolution:
Section: 27.7.4.3 [istream.unformatted] Status: New Submitter: Juan Soulie Opened: 2013-03-01 Last modified: 2017-02-03
Priority: 3
View all other issues in [istream.unformatted].
View all issues with New status.
Discussion:
In 27.7.4.3 [istream.unformatted] / 34, when describing putback, it says that "rdbuf->sputbackc()" is called. The problem are not the obvious typos in the expression, but the fact that it may lead to different interpretations, since nowhere is specified what the required argument to sputbackc is.
It can be guessed to be "rdbuf()->sputbackc(c)", but "rdbuf()->sputbackc(char_type())" or just anything would be as conforming (or non-confoming) as the first guess.Proposed resolution:
Section: 18.3.4 [numeric.limits] Status: New Submitter: Pete Becker Opened: 2013-03-08 Last modified: 2017-03-05
Priority: 4
View other active issues in [numeric.limits].
View all other issues in [numeric.limits].
View all issues with New status.
Discussion:
This member should probably be named "is_ieee754". Or at least the standard should explain that IEC-559 no longer exists, and that it's been superseded by IEEE-754.
Proposed resolution:
Section: 20.11.1.2 [unique.ptr.single] Status: Open Submitter: Rob Desbois Opened: 2013-05-15 Last modified: 2017-05-31
Priority: 3
View all other issues in [unique.ptr.single].
View all issues with Open status.
Discussion:
N3337 20.11.1.2.5 [unique.ptr.single.modifiers] contains 2 non-normative notes stating:
[para 4]: "The order of these operations is significant because the call to get_deleter() may destroy *this."
[para 5]: "The postcondition does not hold if the call to get_deleter() destroys *this since this->get() is no longer a valid expression."
It seems this wording was created to resolve 998 due to the possibility that a unique_ptr may be destroyed through deletion of its stored pointer where that directly or indirectly refers to the same unique_ptr. If unique_ptr is required to support circular references then it seems this must be normative text: an implementation is currently allowed to operate on *this after the assignment and deletion specified in para 4, since this is only 'disallowed' by the non-normative note.
I propose the following draft rewording:
[para 4]: Effects: assigns p to the stored pointer, and then if the old value of the stored pointer, old_p, was not
equal to nullptr, calls get_deleter()(old_p). No operation shall be performed after the call to
get_deleter()(old_p) that requires *this to be valid, because the deletion may destroy *this if it is
referred to directly or indirectly by the stored pointer. [Note: The order of these operations is significant
because the call to get_deleter() may destroy *this. — end note]
I expect it will also be necessary to amend the requirements for a deleter, so in addition:
20.11.1.2 [unique.ptr.single] [para 1]: The default type for the template parameter D is default_delete. A client-supplied template argument D shall be a function object type (20.10), lvalue-reference to function, or lvalue-reference to function object type for which, given a value d of type D and a value ptr of type unique_ptr<T, D>::pointer, the expression d(ptr) is valid and has the effect of disposing of the pointer as appropriate for that deleter. Where D is not an lvalue reference type, d(ptr) shall be valid if ptr refers directly or indirectly to the invoking unique_ptr object.
[2013-10-05, Stephan T. Lavavej comments and provides alternative wording]
In Chicago, we determined that the original proposed change to 20.11.1.2 [unique.ptr.single]/1 was insufficient, because d might be a reference to a deleter functor that's destroyed during self-destruction.
We believed that 20.11.1.2.5 [unique.ptr.single.modifiers]/4 was already sufficiently clear. The Standard occasionally prevents implementations of X from doing various things, through the principle of "nothing allows X to fail in that situation". For example, v.push_back(v[0]) is required to work for non-empty vectors because nothing allows that to fail. In this case, the intent to allow self-destruction is already clear. Additionally, we did not believe that 20.11.1.2.5 [unique.ptr.single.modifiers]/5 had to be changed. The current note is slightly squirrely but it does not lead to confusion for implementers or users.Previous resolution from Rob Desbois:
Edit 20.11.1.2 [unique.ptr.single] p1 as indicated:
The default type for the template parameter D is default_delete. A client-supplied template argument D shall be a function object type (20.10), lvalue-reference to function, or lvalue-reference to function object type for which, given a value d of type D and a value ptr of type unique_ptr<T, D>::pointer, the expression d(ptr) is valid and has the effect of disposing of the pointer as appropriate for that deleter. Where D is not an lvalue reference type, d(ptr) shall be valid if ptr refers directly or indirectly to the invoking unique_ptr object.
Edit 20.11.1.2.5 [unique.ptr.single.modifiers] p4+5 as indicated:
void reset(pointer p = pointer()) noexcept;-3- Requires: The expression get_deleter()(get()) shall be well formed, shall have well-defined behavior, and shall not throw exceptions.
-4- Effects: assigns p to the stored pointer, and then if the old value of the stored pointer, old_p, was not equal to nullptr, calls get_deleter()(old_p). No operation shall be performed after the call to get_deleter()(old_p) that requires *this to be valid, because the deletion may destroy *this if it is referred to directly or indirectly by the stored pointer.[Note: The order of these operations is significant because the call to get_deleter() may destroy *this. — end note]-5- Postconditions: If the call get_deleter()(old_p) destroyed *this, none. Otherwise, get() == p.[Note: The postcondition does not hold if the call to get_deleter() destroys *this since this->get() is no longer a valid expression. — end note]
Previous resolution [SUPERSEDED]:
This wording is relative to N3691.
Edit 20.11.1.2 [unique.ptr.single] p1 as indicated:
The default type for the template parameter D is default_delete. A client-supplied template argument D shall be a function object type (20.10), lvalue-reference to function, or lvalue-reference to function object type for which, given a value d of type D and a value ptr of type unique_ptr<T, D>::pointer, the expression d(ptr) is valid and has the effect of disposing of the pointer as appropriate for that deleter. d(ptr) shall be valid even if it triggers the destruction of d or (if D is an lvalue reference to function object type) the function object that d refers to.
[2015-05, Lenexa]
After some discussion in Lenexa there was some wavering on if the added sentence is necessary. Here is example code that demonstrates why the extra sentence is necessary. In this example the call to d(ptr) is valid, however the deleter references *this after destructing its element:
#include <cassert>
#include <memory>
#include <iostream>
class Deleter
{
int state_ = 0;
enum
{
destructed = -4,
self_move_assigned = -3,
move_assigned_from = -2,
move_constructed_from = -1
};
public:
~Deleter() {state_ = destructed;}
Deleter() = default;
Deleter(Deleter const&) = default;
Deleter& operator=(Deleter const&) = default;
Deleter(Deleter&& a) noexcept
: state_(a.state_)
{a.state_ = move_constructed_from;}
Deleter& operator=(Deleter&& a) noexcept
{
if (this == &a)
state_ = self_move_assigned;
else
{
state_ = a.state_;
a.state_ = move_assigned_from;
}
return *this;
}
Deleter(int state)
: state_(state)
{
assert(state >= 0);
}
template <class T>
void
operator()(T* t) const
{
std::cout << "Deleter beginning operator()(T*)\n";
std::cout << "The deleter = " << *this << '\n';
std::cout << "Deleter about to destruct the X.\n";
delete t;
std::cout << "Deleter has destructed the X.\n";
std::cout << "The deleter = " << *this << '\n';
std::cout << "Deleter ending operator()(T*)\n";
}
friend
std::ostream&
operator<<(std::ostream& os, const Deleter& a)
{
switch (a.state_)
{
case destructed:
os << "**destructed**";
break;
case self_move_assigned:
os << "self_move_assigned";
break;
case move_assigned_from:
os << "move_assigned_from";
break;
case move_constructed_from:
os << "move_constructed_from";
break;
default:
os << a.state_;
break;
}
return os;
}
};
struct X
{
Deleter deleter_{1};
};
int main()
{
auto xp = new X;
{
std::unique_ptr<X, Deleter&> p(xp, xp->deleter_);
std::cout << "unique_ptr is constructed.\n";
std::cout << "The deleter = " << p.get_deleter() << '\n';
std::cout << "Destructing unique_ptr...\n";
}
std::cout << "unique_ptr is destructed.\n";
}
Which outputs:
unique_ptr is constructed. The deleter = 1 Destructing unique_ptr... Deleter beginning operator()(T*) The deleter = 1 Deleter about to destruct the X. Deleter has destructed the X. The deleter = **destructed** Deleter ending operator()(T*) unique_ptr is destructed.
The line "The deleter = **destructed**" represents the deleter referencing itself after it has been destructed by the d(ptr) expression, but prior to that call returning.
Suggested alternative to the current proposed wording:The expression d(ptr) shall not refer to the object d after it executes ptr->~T().
[2015-07, Telecon]
Geoffrey: Deleter may or may not execute ~T().
Alisdair: After the destructor after the element has run. Say it in words instead of code.
Howard will provide updated wording. Perhaps need both normative and non-normative wording.
[2015-08-03, Howard updates P/R per telecon discussion.]
[2017-03-04, Kona]
This is related to 2751, which has been suggested NAD.
STL wants "Effects equivalent to" here - say it in code. Marshall to research.
Proposed resolution:
This wording is relative to N4431.
Edit 20.11.1.2 [unique.ptr.single] p1 as indicated:
The default type for the template parameter D is default_delete. A client-supplied template argument D shall be a function object type (20.9), lvalue-reference to function, or lvalue-reference to function object type for which, given a value d of type D and a value ptr of type unique_ptr<T, D>::pointer, the expression d(ptr) is valid and has the effect of disposing of the pointer as appropriate for that deleter. The expression d(ptr), if it destructs the object referred to by ptr, shall not refer to the object d after it destructs *ptr. [Note: The object being destructed may control the lifetime of d. — end note]
Section: 29.3 [atomics.order] Status: Open Submitter: Brian Demsky Opened: 2013-06-17 Last modified: 2017-02-03
Priority: 4
View other active issues in [atomics.order].
View all other issues in [atomics.order].
View all issues with Open status.
Discussion:
I believe that the following variation on IRIW should admit executions in which c1 = d1 = 5 and c2 = d2 = 0. If this is allowed, then what is sequence of program evaluations for 29.3 [atomics.order] p9 that justifies the store to z? It seems that 29.3 [atomics.order] p9 should not allow this execution because one of the stores to x or y has to appear earlier in the sequence, each of the fetch_adds reads the previous load in the thread (and thus must appear later in the sequence), and 29.3 [atomics.order] p9 states that each load must read from the last prior assignment in the sequence.
atomic_int x;
atomic_int y;
atomic_int z;
int c1, c2, d1, d2;
static void a(void* obj)
{
atomic_store_explicit(&x, 5, memory_order_relaxed);
}
static void b(void* obj)
{
atomic_store_explicit(&y, 5, memory_order_relaxed);
}
static void c(void* obj)
{
c1 = atomic_load_explicit(&x, memory_order_relaxed);
// this could also be an atomic load if the address depends on c1:
c2 = atomic_fetch_add_explicit(&y, c1, memory_order_relaxed);
}
static void d(void* obj)
{
d1 = atomic_load_explicit(&y, memory_order_relaxed);
d2 = atomic_fetch_add_explicit(&x, d1, memory_order_relaxed);
}
int user_main(int argc, char** argv)
{
thrd_t t1, t2, t3, t4;
atomic_init(&x, 0);
atomic_init(&y, 0);
printf("Main thread: creating 4 threads\n");
thrd_create(&t1, (thrd_start_t)&a, NULL);
thrd_create(&t2, (thrd_start_t)&b, NULL);
thrd_create(&t3, (thrd_start_t)&c, NULL);
thrd_create(&t4, (thrd_start_t)&d, NULL);
thrd_join(t1);
thrd_join(t2);
thrd_join(t3);
thrd_join(t4);
printf("c1=%d c2=%d\n",c1,c2);
printf("d1=%d d2=%d\n",d1,d2);
// Can this store write 1000 (i.e., c1=d1=5, c2=d2=0)?
atomic_store(&z, (c1+d1)*100+c2+d2);
printf("Main thread is finished\n");
return 0;
}
It seems that the easiest fix is to allow a load in 29.3 [atomics.order] p9 to read from any prior store in the evaluation order.
That said, I would personally advocate the following: It seems to me that C/C++ atomics are in a bit of different situation than Java because:People are expected to use relaxed C++ atomics in potentially racy situations, so it isn't clear that semantics as complicated as the JMM's causality would be sane.
People who use C/C++ atomics are likely to be experts and use them in a very controlled fashion. I would be really surprised if compilers would find any real wins by optimizing the use of atomics.
Why not do something like:
There is satisfaction DAG of all program evaluations. Each evaluation observes the values of variables as computed by some prior assignment in the DAG. There is an edge x->y between two evaluations x and y if:the evaluation y observes a value computed by the evaluation x or
the evaluation y is an atomic store, the evaluation x is an atomic load, and there is a condition branch c that may depend (intrathread dependence) on x and x-sb->c and c-sb->y.
This seems to allow reordering of relaxed atomics that processors do without extra fence instructions, allows most reorderings by the compiler, and gets rid of satisfaction cycles.
[2015-02 Cologne]
Handed over to SG1.
[2015-05 Lenexa, SG1 response]
This was partially addressed (weasel-worded) in C++14 (See N3786). The remainder is an open research problem. N3710 outlines a "solution" that doesn't have a consensus behind it because it costs performance. We have no better solution at the moment.
Proposed resolution:
Section: 25.7.1.4 [partial.sort.copy] Status: New Submitter: Matt Austern Opened: 2013-06-26 Last modified: 2017-02-03
Priority: 3
View all issues with New status.
Discussion:
The signature of this function is:
template<class InputIterator, class RandomAccessIterator>
RandomAccessIterator
partial_sort_copy(InputIterator first, InputIterator last,
RandomAccessIterator result_first,
RandomAccessIterator result_last);
(and the usual overload for an explicitly provided comparison function). The standard says nothing about requirements in the case where the input type (iterator_traits<InputIterator>::value_type) and the output type (iterator_traits<RandomAccessIterator>::value_type) are different.
Presumably the input type must be convertible to the output type. What's less clear is what the requirements are on the comparison operator. Does the algorithm only perform comparisons on two values of the output type, or does it also perform comparisons on values of the input type, or might it even perform heterogeneous comparisons?Proposed resolution:
Section: 23.2.1 [container.requirements.general] Status: New Submitter: Matt Austern Opened: 2013-06-26 Last modified: 2017-02-03
Priority: 4
View other active issues in [container.requirements.general].
View all other issues in [container.requirements.general].
View all issues with New status.
Discussion:
Consider the following code snippet:
#include <vector>
#include <algorithm>
int main() {
std::vector<int> v1(100, 3);
std::vector<int> v2(100);
copy(v1.begin(), v1.end(), v2.begin());
}
It compiles without error on my desktop. Is it required to? I can't find evidence from the standard that it is. In my test std::copy was found by argument-dependent lookup because the implementation I used made std::vector<int>::iterator a user-defined type defined in namespace std. But the standard only requires std::vector<int>::iterator to be an implementation specified random access iterator type. I can't find anything requiring it to be a user-defined type at all (and in fact there are reasonable implementation where it isn't), let alone a user defined type defined in a specific namespace.
Since the defining namespace of container iterators is visible to users, should the standard say anything about what that namespace is?
Proposed resolution:
Section: 27.8.2.4 [stringbuf.virtuals] Status: New Submitter: Sergey Zubkov Opened: 2013-08-29 Last modified: 2017-02-03
Priority: 4
View all other issues in [stringbuf.virtuals].
View all issues with New status.
Discussion:
In 27.8.2.4 [stringbuf.virtuals]/1, basic_stringbuf::underflow() is specified to unconditionally return traits::eof() when a read position is not available.
The semantics of basic_stringbuf require, and existing libraries implement it so that this function makes a read position available if possible to do so, e.g. if some characters were inserted into the stream since the last call to overflow(), resulting in pptr() > egptr(). Compare to the conceptually similar D.5.1.3 [depr.strstreambuf.virtuals]/15.Proposed resolution:
This wording is relative to N3691.
Change 27.8.2.4 [stringbuf.virtuals] as indicated:
int_type underflow();-1- Returns: If the input sequence has a read position available or the function makes a read position available (as described below), returns traits::to_int_type(*gptr()). Otherwise, returns traits::eof(). Any character in the underlying buffer which has been initialized is considered to be part of the input sequence.
-?- The function can make a read position available only if (mode & ios_base::in) != 0 and if the write next pointer pptr() is not null and is greater than the current read end pointer egptr(). To make a read position available, the function alters the read end pointer egptr() to equal pptr().
Section: 20.4.2 [pairs.pair], 20.5.3.1 [tuple.cnstr], 20.17.5 [time.duration] Status: Open Submitter: Daniel Krügler Opened: 2013-09-09 Last modified: 2017-02-03
Priority: 3
View other active issues in [pairs.pair].
View all other issues in [pairs.pair].
View all issues with Open status.
Discussion:
During the acceptance of N3471 and some similar constexpr papers, specific wording was added to pair, tuple, and other templates that were intended to impose implementation constraints that ensure that the observable constexpr "character" of a defaulted function template is solely determined by the required expressions of the user-provided types when instantiated, for example:
The defaulted move and copy constructor, respectively, of pair shall be a constexpr function if and only if all required element-wise initializations for copy and move, respectively, would satisfy the requirements for a constexpr function.
This wording doesn't require enough, especially since the core language via CWG 1358 does now support constexpr function template instantiations, even if such function cannot appear in a constant expression (as specified in 5.20 [expr.const]) or as a constant initializer of that object (as specified in [basic.start.init]). The wording should be improved and should require valid uses in constant expressions and as constant initializers instead.
[Lenexa 2015-05-05]
STL : notice order of move/copy and copy/move with "respectively".
General word-smithing; ask for updated wording
Are we happy with this with changes we are suggesting?
unanimous
[2016-12-14, Daniel comments]
LWG 2833 overlaps considerably and both should be resolved together.
Proposed resolution:
This wording is relative to N3691.
Change 20.4.2 [pairs.pair] p2 as indicated:
-2-
The defaulted move and copy constructor, respectively, of pair shall be a constexpr function if and only if all required element-wise initializations for copy and move, respectively, would satisfy the requirements for a constexpr functionAn invocation of the move or copy constructor of pair shall be a constant expression (5.20 [expr.const]) if all required element-wise initializations would be constant expressions. An invocation of the move or copy constructor of pair shall be a constant initializer for that pair object ( [basic.start.init]) if all required element-wise initializations would be constant initializers for the respective subobjects.
Change 20.5.3.1 [tuple.cnstr] p2 as indicated:
-2-
The defaulted move and copy constructor, respectively, of tuple shall be a constexpr function if and only if all required element-wise initializations for copy and move, respectively, would satisfy the requirements for a constexpr function. The defaulted move and copy constructor of tuple<> shall be constexpr functionsAn invocation of the move or copy constructor of tuple shall be a constant expression (5.20 [expr.const]) if all required element-wise initializations would be constant expressions. An invocation of the move or copy constructor of tuple shall be a constant initializer for that tuple object ( [basic.start.init]) if all required element-wise initializations would be constant initializers for the respective subobjects. An invocation of the move or copy constructor of tuple<> shall be a constant expression, or a constant initializer for that tuple<> object, respectively, if the function argument would be constant expression.
Change 20.17.5 [time.duration] p7 as indicated:
-7- Remarks:
The defaulted copy constructor of duration shall be a constexpr function if and only if the required initialization of the member rep_ for copy and move, respectively, would satisfy the requirements for a constexpr function.An invocation of the copy constructor of duration shall be a constant expression (5.20 [expr.const]) if the required initialization of the member rep_ would be a constant expression. An invocation of the copy constructor of duration shall be a constant initializer for that duration object ( [basic.start.init]) if the required initialization of the member rep_ would be constant initializers for this subobject.
Section: 20.15 [meta] Status: Open Submitter: Daniel Krügler Opened: 2013-09-02 Last modified: 2017-02-03
Priority: 3
View other active issues in [meta].
View all other issues in [meta].
View all issues with Open status.
Discussion:
The current library specification uses at several places wording that is intended to refer to core language template deduction failure at the top-level of expressions (aka "SFINAE"), for example:
The expression declval<T>() = declval<U>() is well-formed when treated as an unevaluated operand (Clause 5). Access checking is performed as if in a context unrelated to T and U. Only the validity of the immediate context of the assignment expression is considered. [Note: The compilation of the expression can result in side effects such as the instantiation of class template specializations and function template specializations, the generation of implicitly-defined functions, and so on. Such side effects are not in the "immediate context" and can result in the program being ill-formed. — end note]
Similar wording can be found in the specification of result_of, is_constructible, and is_convertible, being added to resolve an NB comment by LWG 1390 and 1391 through N3142.
This wording is necessary to limit speculative compilations needed to implement these traits, but it is also lengthy and repetitive.[2014-05-19, Daniel suggests a descriptive term]
constrictedly well-formed expression:
An expression e depending on a set of types A1, ..., An which is well-formed when treated as an unevaluated operand (Clause 5). Access checking is performed as if in a context unrelated to A1, ..., An. Only the validity of the immediate context of e is considered. [Note: The compilation of the expression can result in side effects such as the instantiation of class template specializations and function template specializations, the generation of implicitly-defined functions, and so on. Such side effects are not in the "immediate context" and can result in the program being ill-formed. — end note][2014-05-20, Richard and Jonathan suggest better terms]
Richard suggested "locally well-formed"
Jonathan suggested "contextually well-formed" and then "The expression ... is valid in a contrived argument deduction context"[2014-06-07, Daniel comments and improves wording]
The 2014-05-19 suggestion did only apply to expressions, but there are two important examples that are not expressions, but instead are involving an object definition (std::is_constructible) and a function definition (std::is_convertible), respectively, instead. Therefore I suggest to rephrase the usage of "expression" into "program construct" in the definition of Jonathan's suggestion of "valid in a contrived argument deduction context".
I would like to point out that given the new definition of "valid in a contrived argument deduction context", there are several other places of the Library specification that could take advantage of this wording to improve the existing specification, such as 20.14.13.2 [func.wrap.func] p2, most functions in 20.10.8.2 [allocator.traits.members], and the **Insertable, EmplaceConstructible, and Erasable definitions in 23.2.1 [container.requirements.general], but given that these are not fully described in terms of the aforementioned wording yet, I would recommend to fix them by a separate issue once the committee has agreed on following the suggestion presented by this issue.[2015-05-05 Lenexa: Move to Open]
...
MC: I think we like the direction but it isn't quite right: it needs some work
JW: I'm prepared to volunteer to move that further, hopefully with the help of Daniel
Roger Orr: should this be Core wording because it doesn't really have anything to do with libraries - the term could then just be used here
AM: Core has nothing to deal with that, though
HT: it seems there is nothing to imply that allows dropping out with an error - maybe that's a separate issue
MC: I'm not getting what you are getting at: could you write an issue? - any objection to move to Open?
...
Proposed resolution:
This wording is relative to N3936.
Add the following new definition to 17.3 [definitions] as indicated:
valid in a contrived argument deduction context [defns.valid.contr.context]
A program construct c depending on a set of types A1, ..., An, and treated as an unevaluated operand (Clause 5) when c is an expression, which is well-formed. Access checking is performed as if in a context unrelated to A1, ..., An. Only the validity of the immediate context (14.8.2 [temp.deduct]) of c is considered. [Note: The compilation of c can result in side effects such as the instantiation of class template specializations and function template specializations, the generation of implicitly-defined functions, and so on. Such side effects are not in the "immediate context" and can result in the program being ill-formed. — end note].Change Table 49 ("Type property predicates") as indicated:
Table 49 — Type property predicates Template Condition Preconditions … template <class T, class U>
struct is_assignable;The expression declval<T>() =
declval<U>() is valid in a
contrived argument deduction context
([defns.valid.contr.context]) for types
T and U.well-formed when treated
as an unevaluated operand
(Clause 5). Access
checking is performed as if
in a context unrelated to T
and U. Only the validity of
the immediate context of
the assignment expression
is considered. [Note: The
compilation of the
expression can result in
side effects such as the
instantiation of class
template specializations
and function template
specializations, the
generation of
implicitly-defined
functions, and so on. Such
side effects are not in the
"immediate context" and
can result in the program
being ill-formed. — end
note][…] …
Change 20.15.4.3 [meta.unary.prop] p7 as indicated:
-7- Given the following function prototype:
template <class T> add_rvalue_reference_t<T> create() noexcept;the predicate condition for a template specialization is_constructible<T, Args...> shall be satisfied if and only if the following variable definition
would be well-formedfor some invented variable t would be valid in a contrived argument deduction context ([defns.valid.contr.context]) for types T and Args...:T t(create<Args>()...);[Note: These tokens are never interpreted as a function declaration. — end note]
Access checking is performed as if in a context unrelated to T and any of the Args. Only the validity of the immediate context of the variable initialization is considered. [Note: The evaluation of the initialization can result in side effects such as the instantiation of class template specializations and function template specializations, the generation of implicitly-defined functions, and so on. Such side effects are not in the "immediate context" and can result in the program being ill-formed. — end note]
Change Table 57 ("Other transformations") as indicated:
Table 57 — Other transformations Template Condition Comments … template <class Fn, class... ArgTypes>
struct result_of<Fn(ArgTypes...)>;[…] If the expression
INVOKE(declval<Fn>(),
declval<ArgTypes>()...) is
valid in a contrived argument deduction
context ([defns.valid.contr.context]) for types
Fn and ArgTypes...well, the
formed when treated as an
unevaluated operand (Clause 5)
member typedef type shall name the
type
decltype(INVOKE(declval<Fn>(),
declval<ArgTypes>()...));
otherwise, there shall be no member
type.Access checking is performed as
if in a context unrelated to Fn and
ArgTypes. Only the validity of the
immediate context of the expression is
considered. [Note: The compilation of
the expression can result in side
effects such as the instantiation of
class template specializations and
function template specializations, the
generation of implicitly-defined
functions, and so on. Such side effects
are not in the "immediate context"
and can result in the program being
ill-formed. — end note]…
Change 20.15.6 [meta.rel] p4 as indicated:
-4- Given the following function prototype:
template <class T> add_rvalue_reference_t<T> create() noexcept;the predicate condition for a template specialization is_convertible<From, To> shall be satisfied if and only if the return expression in the following code would be
well-formedvalid in a contrived argument deduction context ([defns.valid.contr.context]) for types To and From, including any implicit conversions to the return type of the function:To test() { return create<From>(); }[Note: This requirement gives well defined results for reference types, void types, array types, and function types. — end note]
Access checking is performed as if in a context unrelated to To and From. Only the validity of the immediate context of the expression of the return-statement (including conversions to the return type) is considered. [Note: The evaluation of the conversion can result in side effects such as the instantiation of class template specializations and function template specializations, the generation of implicitly-defined functions, and so on. Such side effects are not in the "immediate context" and can result in the program being ill-formed. — end note]
Section: 17.4.1.4 [structure.specifications] Status: New Submitter: Jeffrey Yasskin Opened: 2013-09-03 Last modified: 2017-02-03
Priority: 3
View all other issues in [structure.specifications].
View all issues with New status.
Discussion:
The C++14 CD has 25 sections including the phrase "X shall not participate in overload resolution ...". Most of these uses are double negatives, which are hard to interpret. "shall not ... unless" tends to be the easiest to read, since the condition is true when the function is available, but we also have a lot of "if X is not Y, then Z shall not participate", which actually means "You can call Z if X is Y." The current wording is also clumsy and long-winded. We should find a better and more concise phrasing.
As an initial proposal, I'd suggest using "X is enabled if and only if Y" in prose and adding an "Enabled If: ..." element to 17.4.1.4 [structure.specifications]. Daniel: I suggest to name this new specification element for 17.4.1.4 [structure.specifications] as "Template Constraints:" instead, because the mentioned wording form was intentionally provided starting with LWG 1237 to give implementations more freedom to realize the concrete constraints. Instead of the original std::enable_if-based specifications we can use better forms of "SFINAE" constraints today and it eases the path to possible language-based constraints in the future.Proposed resolution:
Section: 22.3.1.2 [locale.cons] Status: New Submitter: Juan Soulie Opened: 2013-09-04 Last modified: 2017-02-03
Priority: 3
View all issues with New status.
Discussion:
22.3.1.2 [locale.cons] p14 ends with:
"[…] If f is null, the resulting object is a copy of other."
but the next line p15 says:
"Remarks: The resulting locale has no name."
But both can't be true when other has a name and f is null.
I've tried it on two implementations (MSVC,GCC) and they are inconsistent with each other on this.Daniel Krügler:
As currently written, the Remarks element applies unconditionally for all cases and thus should "win". The question arises whether the introduction of this element by LWG 424 had actually intended to change the previous Note to a Remarks element. In either case the wording should be improved to clarify this special case.Proposed resolution:
Section: 18.6.2.3 [new.delete.placement] Status: New Submitter: Daniel Krügler Opened: 2013-09-18 Last modified: 2017-02-03
Priority: 3
View all other issues in [new.delete.placement].
View all issues with New status.
Discussion:
The library gives explicit permission in 17.5.4.2.1 [namespace.std] p2 that user code may explicitly instantiate a library template provided that the instantiations depend on at least one user-defined type:
A program may explicitly instantiate a template defined in the standard library only if the declaration depends on the name of a user-defined type and the instantiation meets the standard library requirements for the original template.
But it seems that the C++11 library is not specified in a way that guarantees such an instantiation to be well-formed if the minimum requirements of the library is not satisfied.
For example, in general, the first template parameter of std::vector is not required to be DefaultConstructible in general, but due to the split of the single C++03 member function with default argumentvoid resize(size_type sz, T c = T());
into
void resize(size_type sz); void resize(size_type sz, const T& c);
the effect is now that for a type ND that is not DefaultConstructible, such as
struct NP {
NP(int);
};
the explicit instantiation of std::vector<ND> is no longer well-formed, because the attempt to instantiate the single-argument overload of resize cannot not succeed, because this function imposes the DefaultInsertable requirements and given the default allocator this effectively requires DefaultConstructible.
But DefaultConstructible is not the only point, what about CopyConstructible versus MoveConstructible alone? It turns out that currently the second resize overload would fail during an explicit instantiation for a type like
struct MO {
MO() = default;
MO(MO&&) = default;
};
because it imposes CopyInsertable requirements that end up being equivalent to the CopyConstructible requirements for the default allocator.
Technically a library can solve these issues: For special member functions by defining them in some base class, for others by transforming them effectively into a function template due to the great feature of default template arguments for function templates (At the very moment the validity of the latter approach depends on a resolution of core language issue CWG 1635, though). E.g. the here mentioned resize functions of std::vector could be prevented from instantiation by defining them like this with an implementation:
template<class = void>
void resize(size_type sz) { […] }
template<class = void>
void resize(size_type sz, const T& c) { […] }
In this case, these functions could also be defined in a base class, but the latter approach won't work in all cases.
Basically such an implementation is required to constrain all member functions that are not covered by the general requirements imposed on the actual library template parameters. I tested three different C++11 library implementations and but none could instantiate for example std::list, std::vector, or std::deque with value types that are not DefaultConstructible or only MoveConstructible. This issue is raised to clarify the current situation in regard to the actual requirements imposed on user-provided types that are used to explicitly instantiate Library-provided templates. For example, the current Container requirements impose very little requirements on the actual value type and it is unclear to which extend library implementations have to respect that. The minimum solution of this issue should be to at least realize that there is no fundamental requirement on DefaultConstructible for value types of library containers, because we have since C++03 the general statement of 17.5.3.1 [utility.arg.requirements] ("In general, a default constructor is not required."). It is unclear whether CopyConstructible should be required for an explicit instantiation request, but given the careful introduction of move operations in the library it would seem astonishing that a MoveConstructible type wouldn't suffice for value types of the container types. In any case I can envision at least two approaches to solve this issue:As indicated in LWG 2292, those function could get an explicit "Template Constraints:" element, albeit this promises more than needed to solve this issue.
The library could introduce a completely new element form, such as "Instantiation Constraints:" that would handle this situation for explicit instantiation situations. This would allow for simpler techniques to solve the issue when explicit instantiation is required compared to the first bullet, because it would not (necessarily) guarantee SFINAE-friendly expression-wellformedness, such as inspecting the expression std::declval<std::vector<ND>&>.resize(0) in an unevaluated context.
It should be noted that the 2013-08-27 comment to LWG 2193 could be resolved by a similar solution as indicated in this issue here.
Proposed resolution:
Section: 23 [containers] Status: Open Submitter: Zhihao Yuan Opened: 2013-09-26 Last modified: 2017-02-03
Priority: 2
View other active issues in [containers].
View all other issues in [containers].
View all issues with Open status.
Discussion:
LWG 2193 yields explicit for default ctors to allow {}, but not for all cases of uniform initialization. For example:
explicit vector(size_type count, const Allocator& alloc = Allocator());
This prevents {n, alloc()}. Although this use is relatively rare, but the behavior is inconsistent with that of
vector(size_type count, const T& value, const Allocator& alloc = Allocator());
[Urbana 2014-11-07: Move to Open]
Proposed resolution:
Section: 21.3.2 [basic.string] Status: New Submitter: Stephan T. Lavavej Opened: 2013-09-21 Last modified: 2017-02-03
Priority: 4
View other active issues in [basic.string].
View all other issues in [basic.string].
View all issues with New status.
Discussion:
21.3.2.4 [string.capacity]/8 specifies basic_string::resize(n, c) with:
Effects: Alters the length of the string designated by *this as follows:
If n <= size(), the function replaces the string designated by *this with a string of length n whose elements are a copy of the initial elements of the original string designated by *this.
If n > size(), the function replaces the string designated by *this with a string of length n whose first size() elements are a copy of the original string designated by *this, and whose remaining elements are all initialized to c.
This wording is a relic of the copy-on-write era. In addition to being extremely confusing, it has undesirable implications. Saying "replaces the string designated by *this with a string of length n whose elements are a copy" suggests that the trimming case can reallocate. Reallocation during trimming should be forbidden, like vector.
At least 7 paragraphs are affected: 21.3.2.4 [string.capacity]/8, 21.3.2.6.2 [string.append]/9, 21.3.2.6.3 [string.assign]/3 and /10, 21.3.2.6.4 [string.insert]/11, 21.3.2.6.5 [string.erase]/4, and 21.3.2.6.6 [string.replace]/11 say "replaces the string [designated/controlled] by *this". (21.3.2.6.7 [string.copy]/3 is different — it "replaces the string designated by s".) Of the affected paragraphs, resize() and erase() are the most important to fix because they should forbid reallocation during trimming.Proposed resolution:
Section: 23.2.1 [container.requirements.general] Status: Open Submitter: Stephan T. Lavavej Opened: 2013-09-21 Last modified: 2017-02-03
Priority: 2
View other active issues in [container.requirements.general].
View all other issues in [container.requirements.general].
View all issues with Open status.
Discussion:
23.2.1 [container.requirements.general]/10 says that unless otherwise specified, "no swap() function invalidates any references, pointers, or iterators referring to the elements of the containers being swapped. [Note: The end() iterator does not refer to any element, so it may be invalidated. — end note]". However, move constructors and move assignment operators aren't given similar invalidation guarantees. The guarantees need several exceptions, so I do not believe that blanket language like /11 "Unless otherwise specified (either explicitly or by defining a function in terms of other functions), invoking a container member function or passing a container as an argument to a library function shall not invalidate iterators to, or change the values of, objects within that container." is applicable.
[2014-02-13 Issaquah]
General agreeement on intent, several wording nits and additional paragraphs to hit.
STL to provide updated wording. Move to Open.
[2015-02 Cologne]
AM: in the proposed wording, I'd like to mention that the iterators now refer to elements of a different container. I think we're saying something like this somewhere. JY: There's some wording like that for swap I think. TK: It's also in list::splice(). DK to JY: 23.2.1p9.
VV: The issue says that STL was going to propose new wording. Has he done that? AM: I believe we're looking at that. GR: The request touches on multiple paragraphs, and this PR has only one new paragraph, so this looks like it's not up-to-date. MC: This was last updated a year ago in Issaquah. Conclusion: Skip, not up to date.[2015-06, Telecon]
Still waiting for updated wording
[2015-08 Chicago]
Still waiting for updated wording
Proposed resolution:
This wording is relative to N3691.
In 23.2.1 [container.requirements.general]/10 change as indicated:
-10- Unless otherwise specified (see 23.2.4.1, 23.2.5.1, 23.3.3.4, and 23.3.7.5) all container types defined in this Clause meet the following additional requirements:
[…]
no copy constructor or assignment operator of a returned iterator throws an exception.
no move constructor (or move assignment operator when allocator_traits<allocator_type>::propagate_on_container_move_assignment::value is true) of a container (except for array) invalidates any references, pointers, or iterators referring to the elements of the source container. [Note: The end() iterator does not refer to any element, so it may be invalidated. — end note]
no swap() function throws an exception.
no swap() function invalidates any references, pointers, or iterators referring to the elements of the containers being swapped. [Note: The end() iterator does not refer to any element, so it may be invalidated. — end note]
Section: 28.5.1 [re.synopt] Status: Open Submitter: Stephan T. Lavavej Opened: 2013-09-21 Last modified: 2017-02-03
Priority: 3
View all other issues in [re.synopt].
View all issues with Open status.
Discussion:
The table in 28.5.1 [re.synopt]/1 says that regex_constants::collate "Specifies that character ranges of the form "[a-b]" shall be locale sensitive.", but 28.13 [re.grammar]/14 says that it affects individual character comparisons too.
[2012-02-12 Issaquah : recategorize as P3]
Marshall Clow: 28.13/14 only applies to ECMAScript
All: we're unsure
Jonathan Wakely: we should ask John Maddock
Move to P3
[2014-5-14, John Maddock response]
The original intention was the original wording: namely that collate only made character ranges locale sensitive. To be frank it's a feature that's probably hardly ever used (though I have no real hard data on that), and is a leftover from early POSIX standards which required locale sensitive collation for character ranges, and then later changed to implementation defined if I remember correctly (basically nobody implemented locale-dependent collation).
So I guess the question is do we gain anything by requiring all character-comparisons to go through the locale when this bit is set? Certainly it adds a great deal to the implementation effort (it's not what Boost.Regex has ever done). I guess the question is are differing code-points that collate identically an important use case? I guess there might be a few Unicode code points that do that, but I don't know how to go about verifying that. STL: If this was unintentional, then 28.5.1 [re.synopt]/1's table should be left alone, while 28.13 [re.grammar]/14 should be changed instead. Jeffrey Yasskin: This page mentions that [V] in Swedish should match "W" in a perfect world. However, the most recent version of TR18 retracts both language-specific loose matches and language-specific ranges because "for most full-featured regular expression engines, it is quite difficult to match under code point equivalences that are not 1:1" and "tailored ranges can be quite difficult to implement properly, and can have very unexpected results in practice. For example, languages may also vary whether they consider lowercase below uppercase or the reverse. This can have some surprising results: [a-Z] may not match anything if Z < a in that locale." ECMAScript doesn't include collation at all. IMO, +1 to changing 28.13 instead of 28.5.1. It seems like we'd be on fairly solid ground if we wanted to remove regex_constants::collate entirely, in favor of named character classes, but of course that's not for this issue.Proposed resolution:
This wording is relative to N3691.
In 28.5.1 [re.synopt]/1, Table 138 — "syntax_option_type effects", change as indicated:
Table 138 — syntax_option_type effects Element Effect(s) if set … collate Specifies that character ranges of the form "[a-b]"comparisons and character range comparisons shall be locale sensitive.…
Section: 29.6 [atomics.types.operations] Status: SG1 Submitter: Daniel Krügler Opened: 2013-10-03 Last modified: 2017-05-31
Priority: Not Prioritized
View all other issues in [atomics.types.operations].
View all issues with SG1 status.
Discussion:
According to 29.6.5 [atomics.types.operations.req] p4,
A ::A () noexcept = default;Effects: leaves the atomic object in an uninitialized state. [Note: These semantics ensure compatibility with C. — end note]
This implementation requirement is OK for POD types, like int, but 29.5 [atomics.types.generic] p1 intentionally allows template arguments of atomic with a non-trivial default constructor ("The type of the template argument T shall be trivially copyable (3.9)"), so this wording can be read in a way that makes the behaviour of the following code undefined:
#include <atomic>
#include <iostream>
struct S {
S() noexcept : v(42) {}
int v;
};
int main() {
std::atomic<S> as; // Default-initialization
std::cout << as.load().v << std::endl; // ?
}
For a user-defined emulation of atomic the expected outcome would be defined and the program would output "42", but existing implementations differ and the result value is a "random number" for at least one implementation. This seems very surprising to me.
To realize that seemingly existing requirement, an implementation is either required to violate normal language rules internally or to perform specific bit-randomization-techniques after the normal default-initialization that called the default constructor of S. According to my understanding, the non-normative note in 29.6.5 [atomics.types.operations.req] p4 is intended to refer to types that are valid C-types, but the example type S is not such a type. To make the mental model of atomic's default constructor more intuitive for user-code, I suggest to clarify the wording to have the effects of default-initialization instead. The current state seems more like an unintended effect of imprecise language used here and has some similarities to wording that was incorrectly used to specify atomic_flag initialization as described by LWG 2159.[2014-05-17, Daniel comments and provides alternative wording]
The current wording was considered controversial as expressed by reflector discussions. To me, the actual problem is not newly introduced by that wording, but instead is already present in basically all paragraphs specifying semantics of atomic types, since the wording never clearly distinguishes the value of the actual atomic type A and the value of the "underlying", corresponding non-atomic type C. The revised proposed wording attempts to improve the current ambiguity of these two kinds of values.
Previous resolution from Daniel [SUPERSEDED]:This wording is relative to N3691.
Modify 29.6.5 [atomics.types.operations.req] p4 as indicated: [Editorial note: There is no exposition-only member in atomic, which makes it a bit hard to specify what actually is initialized, but the usage of the term "value" seems consistent with similar wording used to specify the effects of the atomic load functions]
A ::A () noexcept = default;-4- Effects:
leaves the atomic object in an uninitialized stateThe value of the atomic object is default-initialized (8.6 [dcl.init]). [Note: These semantics ensure compatibility with C. — end note]
[2015-02 Cologne]
Handed over to SG1.
Proposed resolution:
This wording is relative to N3936.
Modify 29.6.5 [atomics.types.operations.req] p2 as indicated: [Editorial note: This is a near-to editorial change not directly affecting this issue, but atomic_address does no longer exist and the pointed to definition is relevant in the context of this issue resolution.]
-2- In the following operation definitions:
an A refers to one of the atomic types.
a C refers to its corresponding non-atomic type.
The atomic_address atomic type corresponds to the void* non-atomic type.[…]
Modify 29.6.5 [atomics.types.operations.req] p4 and the following as indicated: [Editorial note: There is no exposition-only member in atomic, which makes it a bit hard to specify what actually is initialized, but the introductory wording of 29.6.5 [atomics.types.operations.req] p2 b2 defines: "a C refers to its corresponding non-atomic type." which helps to specify the semantics in terms of "the C value referred to by the atomic object"]
A::A() noexcept = default;-4- Effects:
leaves the atomic object in an uninitialized stateDefault-initializes (8.6 [dcl.init]) the C value referred to by the atomic object. [Note: These semantics ensure compatibility with C. — end note]constexpr A::A(C desired) noexcept;-5- Effects: Direct-i
[…]Initializes the C value referred to by the atomic object with the value desired. Initialization is not an atomic operation (1.10). […]void atomic_init(volatile A* object, C desired) noexcept; void atomic_init(A* object, C desired) noexcept;-8- Effects: Non-atomically initializes the C value referred to by *object with value desired. […]
void atomic_store(volatile A* object, C desired) noexcept; […] void A::store(C desired, memory_order order = memory_order_seq_cst) noexcept;-9- […]
-10- Effects: Atomically replaces the C value pointed to by object or by this with the value of desired. […] […]C atomic_load(const volatile A* object) noexcept; […] C A::load(memory_order order = memory_order_seq_cst) const noexcept;-13- […]
-14- […] -15- Returns: Atomically returns the C value pointed to by object or by this. […]C atomic_exchange(volatile A* object, C desired) noexcept; […] C A::exchange(C desired, memory_order order = memory_order_seq_cst) noexcept;-18- Effects: Atomically replaces the C value pointed to by object or by this with desired. […]
-19- Returns: Atomically returns the C value pointed to by object or by this immediately before the effects. […]C atomic_fetch_key(volatile A* object, M operand) noexcept; […] C A::fetch_key(M operand, memory_order order = memory_order_seq_cst) noexcept;-28- Effects: Atomically replaces the C value pointed to by object or by this with the result of the computation applied to the C value pointed to by object or by this and the given operand. […]
-29- Returns: Atomically,returns the C value pointed to by object or by this immediately before the effects. […]
Modify 29.7 [atomics.flag] p5 and the following as indicated:
bool atomic_flag_test_and_set(volatile atomic_flag* object) noexcept; […] bool atomic_flag::test_and_set(memory_order order = memory_order_seq_cst) noexcept;-5- Effects: Atomically sets the bool value pointed to by object or by this to true. […]
-6- Returns: Atomically,returns the bool valueof thepointed to by object or by this immediately before the effects.void atomic_flag_clear(volatile atomic_flag* object) noexcept; […] void atomic_flag::clear(memory_order order = memory_order_seq_cst) noexcept;-7- […]
-8- Effects: Atomically sets the bool value pointed to by object or by this to false. […]
Section: 23.3.7 [array] Status: New Submitter: Jeffrey Yasskin Opened: 2013-10-04 Last modified: 2017-02-03
Priority: 3
View all other issues in [array].
View all issues with New status.
Discussion:
In order to replace some uses of C arrays with std::array, we need it to be possible to cast from a std::array<> to an equivalent C array. Core wording doesn't appear to be in quite the right state to allow casting, but if we specify that appropriate types are layout-compatible, we can at least write:
union {
array<array<array<int, 2>, 3>, 4> arr;
int carr[4][3][2];
};
to view memory as the other type: C++14 CD [class.mem]p18.
I believe it's sufficient to add "array<T, N> shall be layout-compatible (3.9 [basic.types]) with T[N]." to 23.3.7.1 [array.overview], but we might also need some extension to 9.2 [class.mem] to address the possibility of layout-compatibility between struct and array types.I checked that libc++ on MacOS already implements this, although it would be good for someone else to double-check; I haven't checked any other standard libraries.
Proposed resolution:
Section: 20.11.2.2.5 [util.smartptr.shared.obs] Status: Tentatively NAD Submitter: Stephan T. Lavavej Opened: 2013-10-05 Last modified: 2017-02-03
Priority: 2
View other active issues in [util.smartptr.shared.obs].
View all other issues in [util.smartptr.shared.obs].
View all issues with Tentatively NAD status.
Discussion:
20.11.1.2.4 [unique.ptr.single.observers]/3: "pointer operator->() const noexcept; Requires: get() != nullptr."
20.11.2.2.5 [util.smartptr.shared.obs]/2: "T& operator*() const noexcept; Requires: get() != 0." 20.11.2.2.5 [util.smartptr.shared.obs]/5: "T* operator->() const noexcept; Requires: get() != 0." Narrow-contract functions should not be noexcept.[2014-02-15 Issaquah]
Issue is contentious, raise to P2.
[2015-02 Cologne]
AM: This ship has sailed. JM: What's the issue? AM: operator-> has narrow contract and should never have had noexcept. DK: Not quite. We explicitly called out that for shared_ptr this is fine. You said so in your "narrow contract" paper. GR: This would be a fairly major regression in the design of {unique,shared}_ptr over raw pointers; raw pointer dereferencing is noexcept. It's not a performance regression but a usability regression. AM: Do we expect users to query noexpect on dereference expressions? Room: Yes. VV: We don't just expect it, we have seen it. JM: Yes, users may be querying something like noexcept(x->y) and expect to be checking y, but silently end up checking x->.
Close as NAD, with explanation from GR. Previous resolution [SUPERSEDED]:This wording is relative to N3691.
In 20.11.1.2 [unique.ptr.single]/1, class template unique_ptr synopsis for single objects, change as indicated:
pointer operator->() constnoexcept;In 20.11.1.2.4 [unique.ptr.single.observers] change as indicated:
pointer operator->() constnoexcept;-3- Requires: get() != nullptr.
-4- Returns: get(). -?- Throws: Nothing. -5- Note: use typically requires that T be a complete type.In 20.11.2.2 [util.smartptr.shared]/1, class template shared_ptr synopsis, change as indicated:
T& operator*() constnoexcept; T* operator->() constnoexcept;In 20.11.2.2.5 [util.smartptr.shared.obs] change as indicated:
T& operator*() constnoexcept;-2- Requires: get() != 0.
-3- Returns: *get(). -?- Throws: Nothing. -4- Remarks: When T is void, it is unspecified whether this member function is declared. If it is declared, it is unspecified what its return type is, except that the declaration (although not necessarily the definition) of the function shall be well formed.T* operator->() constnoexcept;-5- Requires: get() != 0.
-6- Returns: get(). -?- Throws: Nothing.
[2015-03-03, Geoffrey provides rationale]
Rationale:
It is by design that these members are noexcept, and changing that now would be a substantial regression in functionality. These classes were designed to substitute for plain pointers as transparently as possible, so since those operations are effectively noexcept on plain pointers, they should be noexcept on unique_ptr and shared_ptr as well. This matters in practice because we expect these members to be used fairly often inside the noexcept operator, and such code could be broken by this change. These design considerations override our general policy against noexcept for narrow-contract functions.
It is notable that N3279, which proposed this policy, did not propose striking noexcept from these operations. It's not clear if the omission of operator* and operator-> was an oversight, or an intentional reflection of the above considerations. N3279 was based on N3248 by the same authors, which states that:"Most applications of noexcept for unique_ptr and shared_ptr are on functions with wide contracts. However, there are preconditions on the atomic access functions, so these should lose the specification."
Proposed resolution:
Section: 28.7 [re.traits], 22.3.1.1.2 [locale.facet] Status: Open Submitter: Sergey Zubkov Opened: 2013-10-15 Last modified: 2017-02-03
Priority: 3
View all other issues in [re.traits].
View all issues with Open status.
Discussion:
28.7 [re.traits]/7, begins with "if typeid(use_facet<collate<charT> >) == typeid(collate_byname<charT>)", which appears to be pseudocode with the intention to convey that the collate facet has not been replaced by the user. Cf. the wording in N1429 "there is no portable way to implement transform_primary in terms of std::locale, since even if the sort key format returned by std::collate_byname<>::transform is known and can be converted into a primary sort key, the user can still install their own custom std::collate implementation into the locale object used, and that can use any sort key format they see fit.".
Taken literally, 28.7 [re.traits]/7 appears to imply that named locales are required to hold their collate facets with dynamic type std::collate_byname<charT>, which is in fact true in some implementations (e.g libc++), but not others (e.g. libstdc++). This does not follow from the description of _byname in 22.3.1.1.2 [locale.facet]/4, which is only required to provide equivalent semantics, to the named locale's facet, not to actually be one.[2015-05-06 Lenexa: Move to Open]
MC, RP: Consequence of failing to follow the rule is UB.
MC: Tightening of requirements.
RP: It should be this way, we just didn't impose it before.
MC: Second change is a bug fix, original code didn't work.
TK: Doesn't seem to make things worse.
Bring up in larger group tomorrow.
JW arrives.
JW: libstdc++ violates this due to two std::string ABIs.
JW: This prevents installing a type derived from Facet_byname, constrains the implementor from using a smarter derived class version.
JW: Can't look at facet id to detect replacement, because replacements have the same id.
RP: Can you give it multiple ids through multiple inheritance?
JW: No, the facet mechanism wouldn't like that.
JW: We should also ask Martin Sebor, he's implemented this stuff recently.
MC: Sounds like this resolution doesn't work, need a better solution.
JW: Write in words "if the facet has not been replaced by the user", the implementation knows how to detect that, but not like this.
RP: User RE traits need to detect this too.
JW: =(
Move to Open, JW will invite Martin Sebor to join LWG for discussion.
Later ...
JW: This is not needed for user specializations after all.
MC: Agree, [re.traits]/7 only applies to the stdlib traits.
NM: Effects: doesn't make sense.
JW, NM, Martin Sebor to come up with new wording.
Proposed resolution:
This wording is relative to N3691.
Modify 22.3.1.1.2 [locale.facet]/4 as indicated:
For some standard facets a standard "..._byname" class, derived from it, implements the virtual function semantics
equivalent toprovided by that facet of the locale constructed by locale(const char*) with the same name. Each such facet provides a constructor that takes a const char* argument, which names the locale, and a refs argument, which is passed to the base class constructor. Each such facet also provides a constructor that takes a string argument str and a refs argument, which has the same effect as calling the first constructor with the two arguments str.c_str() and refs. If there is no "..._byname" version of a facet, the base class implements named locale semantics itself by reference to other facets. For any locale loc constructed by locale(const char*) and facet Facet that has a corresponding standard Facet_byname class, typeid(use_facet<Facet>(loc)) == typeid(Facet_byname).
Modify 28.7 [re.traits]/7 as indicated:
template <class ForwardIterator> string_type transform_primary(ForwardIterator first, ForwardIterator last) const;-7- Effects: if typeid(use_facet<collate<charT> >(getloc())) == typeid(collate_byname<charT>) and the form of the sort key returned by collate_byname<charT>::transform(first, last) is known and can be converted into a primary sort key then returns that key, otherwise returns an empty string.
Section: 27.7.5.1 [ostream] Status: New Submitter: Alf P. Steinbach Opened: 2013-10-29 Last modified: 2017-02-03
Priority: 4
View all other issues in [ostream].
View all issues with New status.
Discussion:
For wide streams argument types wchar_t const* and wchar_t are supported only as template parameters. User defined conversions are not considered for template parameter matching. Hence inappropriate overloads of operator<< are selected when an implicit conversion is required for the argument, which is inconsistent with the behavior for char const* and char, is unexpected, and is a useless result.
Demonstration:
#include <iostream>
struct Byte_string
{
operator char const*() const { return "Hurray, it works!"; }
};
struct Wide_string
{
operator wchar_t const*() const { return L"Hurray, it works!"; }
};
struct Byte_ch
{
operator char() const { return 'X'; }
};
struct Wide_ch
{
operator wchar_t() const { return L'X'; }
};
auto main() -> int
{
using namespace std;
wcout << "'X' as char value : " << Byte_ch() << endl;
wcout << "'X' as wchar_t value: " << Wide_ch() << endl;
wcout << "Byte string pointer : " << Byte_string() << endl;
wcout << "Wide string pointer : " << Wide_string() << endl;
}
Example output:
'X' as char value : X 'X' as wchar_t value: 88 Byte string pointer : Hurray, it works! Wide string pointer : 000803C8
Proposed resolution:
This wording is relative to N3797.
Modify 27.7.5.1 [ostream], class template basic_ostream synopsis, as indicated:
namespace std {
[…]
// 27.7.3.6.4 character inserters
template<class charT, class traits>
basic_ostream<charT,traits>& operator<<(basic_ostream<charT,traits>&,
charT);
template<class charT, class traits>
basic_ostream<charT,traits>& operator<<(basic_ostream<charT,traits>&,
char);
template<class traits>
basic_ostream<char,traits>& operator<<(basic_ostream<char,traits>&,
char);
template<class traits>
basic_ostream<wchar_t,traits>& operator<<(basic_ostream<wchar_t,traits>&,
wchar_t);
[…]
template<class charT, class traits>
basic_ostream<charT,traits>& operator<<(basic_ostream<charT,traits>&,
const charT*);
template<class charT, class traits>
basic_ostream<charT,traits>& operator<<(basic_ostream<charT,traits>&,
const char*);
template<class traits>
basic_ostream<char,traits>& operator<<(basic_ostream<char,traits>&,
const char*);
template<class traits>
basic_ostream<wchar_t,traits>& operator<<(basic_ostream<wchar_t,traits>&,
const wchar_t*);
[…]
}
Modify 27.7.5.2.4 [ostream.inserters.character] as indicated: [Drafting note: The replacement of os by out in p1 and the insertion of "out." in p4 just fix two obvious typos — end drafting note]
template<class charT, class traits> basic_ostream<charT,traits>& operator<<(basic_ostream<charT,traits>& out, charT c); template<class charT, class traits> basic_ostream<charT,traits>& operator<<(basic_ostream<charT,traits>& out, char c); // specialization template<class traits> basic_ostream<char,traits>& operator<<(basic_ostream<char,traits>& out, char c); template<class traits> basic_ostream<wchar_t,traits>& operator<<(basic_ostream<wchar_t,traits>& out, wchar_t c); // signed and unsigned template<class traits> basic_ostream<char,traits>& operator<<(basic_ostream<char,traits>& out, signed char c); template<class traits> basic_ostream<char,traits>& operator<<(basic_ostream<char,traits>& out, unsigned char c);-1- Effects: Behaves as a formatted output function (27.7.5.2.1 [ostream.formatted.reqmts]) of out. Constructs a character sequence seq. If c has type char and the character type of the stream is not char, then seq consists of out.widen(c); otherwise seq consists of c. Determines padding for seq as described in 27.7.5.2.1 [ostream.formatted.reqmts]. Inserts seq into out. Calls
-2- Returns: out.osout.width(0).template<class charT, class traits> basic_ostream<charT,traits>& operator<<(basic_ostream<charT,traits>& out, const charT* s); template<class charT, class traits> basic_ostream<charT,traits>& operator<<(basic_ostream<charT,traits>& out, const char* s); template<class traits> basic_ostream<char,traits>& operator<<(basic_ostream<char,traits>& out, const char* s); template<class traits> basic_ostream<wchar_t,traits>& operator<<(basic_ostream<wchar_t,traits>& out, const wchar_t* s); template<class traits> basic_ostream<char,traits>& operator<<(basic_ostream<char,traits>& out, const signed char* s); template<class traits> basic_ostream<char,traits>& operator<<(basic_ostream<char,traits>& out, const unsigned char* s);-3- Requires: s shall not be a null pointer.
-4- Effects: Behaves like a formatted inserter (as described in 27.7.5.2.1 [ostream.formatted.reqmts]) of out. Creates a character sequence seq of n characters starting at s, each widened using out.widen() (27.5.5.3), where n is the number that would be computed as if by:
traits::length(s) for the following overloads:
where the first argument is of type basic_ostream<charT, traits>& and the second is of type const charT*,
and also for the overloadwhere the first argument is of type basic_ostream<char, traits>& and the second is of type const char*,where the first argument is of type basic_ostream<wchar_t, traits>& and the second is of type const wchar_t*,
std::char_traits<char>::length(s) for the overload where the first argument is of type basic_ostream<charT, traits>& and the second is of type const char*,
traits::length(reinterpret_cast<const char*>(s)) for the other two overloads.
Determines padding for seq as described in 27.7.5.2.1 [ostream.formatted.reqmts]. Inserts seq into out. Calls out.width(0).
-5- Returns: out.
Section: 20.9.2 [template.bitset], 27.7.8 [quoted.manip] Status: Open Submitter: Zhihao Yuan Opened: 2013-12-02 Last modified: 2017-02-03
Priority: 3
View all other issues in [template.bitset].
View all issues with Open status.
Discussion:
Example: char16_t('1') != u'1' is possible.
The numeric value of char16_t is defined to be Unicode code point, which is same to the ASCII value and UTF-8 for 7-bit chars. However, char is not guaranteed to have an encoding which is compatible with ASCII. For example, '1' in EBCDIC is 241. I found three places in the standard casting narrow char literals: bitset::bitset, bitset::to_string and quoted. PJ confirmed this issue and says he has a solution used in their <filesystem> implementation, and he may want to propose it to the standard. The solution in my mind, for now, is to make those default arguments magical, where the "magic" can be implemented with a C11 _Generic selection (works in clang):
#define _G(T, literal) _Generic(T{}, \
char: literal, \
wchar_t: L ## literal, \
char16_t: u ## literal, \
char32_t: U ## literal)
_G(char16_t, '1') == u'1'
[Lenexa 2015-05-05: Move to Open]
Ask for complete PR (need quoted, to string, et al.)
Will then take it up again
Expectation is that this is correct way to fix this
Proposed resolution:
This wording is relative to N3797.
[Drafting note: This is a sample wording fixing only one case; I'm just too lazy to copy-paste it before we discussed whether the solution is worth and sufficient (for example, should the other `charT`s like `unsigned char` just don't compile without supplying those arguments? I hope so). — end drafting note]Modify 20.9.2 [template.bitset] p1, class template bitset synopsis, as indicated:
namespace std {
template <size_t N> class bitset {
public:
[…]
template<class charT, class traits, class Allocator>
explicit bitset(
const basic_string<charT,traits,Allocator>& str,
typename basic_string<charT,traits,Allocator>::size_type pos = 0,
typename basic_string<charT,traits,Allocator>::size_type n =
basic_string<charT,traits,Allocator>::npos,
charT zero = charT('0')see below, charT one = charT('1')see below);
[…]
};
[…]
}
Modify 20.9.2.1 [bitset.cons] as indicated:
template<class charT, class traits, class Allocator> explicit bitset(const basic_string<charT, traits, Allocator>& str, typename basic_string<charT, traits, Allocator>::size_type pos = 0, typename basic_string<charT, traits, Allocator>::size_type n = basic_string<charT, traits, Allocator>::npos, charT zero =charT('0')see below, charT one =charT('1')see below);-?- The default values of zero and one compare equal to the character literals 0 and 1 of type charT, respectively.
-3- Requires:: pos <= str.size(). […]
Section: 27.7.4.2.1 [istream.formatted.reqmts] Status: Open Submitter: Zhihao Yuan Opened: 2013-12-06 Last modified: 2017-02-03
Priority: 3
View all other issues in [istream.formatted.reqmts].
View all issues with Open status.
Discussion:
The formatted input function requirement says in 27.7.4.2.1 [istream.formatted.reqmts]:
"If an exception is thrown during input then ios::badbit is turned on in *this's error state. If (exceptions()&badbit) != 0 then the exception is rethrown."
while some formatted function may throw an exception from basic_ios::clear, for example in 20.9.4 [bitset.operators] p6:
"If no characters are stored in str, calls is.setstate(ios_base::failbit) (which may throw ios_base::failure)"
So should this exception be considered as "an exception [...] thrown during input"? And here is an implementation divergence (or you can read the following as "a bug libc++ only has" :)
cin.exceptions(ios_base::failbit);
bitset<N> b;
try {
cin >> b; // type 'a' and return
} catch (...)
{}
Now cin.rdstate() is just failbit in libstdc++ (and Dinkumware, by PJ), but failbit & badbit libc++. Similar difference found in other places, like eofbit & badbid after std::getline.
PJ and Matt both agree that the intention (of badbit + rethrow) is "to signify an exception arising in user code, not the iostreams package". In addition, I found the following words in unformatted input function's requirements (27.7.4.3 [istream.unformatted]):If an exception is thrown during input then ios::badbit is turned on in *this's error state. (Exceptions thrown from basic_ios<>::clear() are not caught or rethrown.) If (exceptions()&badbit) != 0 then the exception is rethrown.
The content within the parenthesis is added by LWG defect 61, and does fix the ambiguity. However, it only fixed the 1 of 4 requirements, and it lost some context (the word "rethrown" is not seen before this sentence within this section).
[Lenexa 2015-05-07: Marshall to research and report]
Proposed resolution:
This wording is relative to N3797.
[Drafting note: The editor is kindly asked to introduce additional spaces at the following marked occurrences of operator& — end drafting note]Modify 27.7.4.2.1 [istream.formatted.reqmts] p1 as indicated:
-1- Each formatted input function begins execution by constructing an object of class sentry with the noskipws (second) argument false. If the sentry object returns true, when converted to a value of type bool, the function endeavors to obtain the requested input. If an exception, other than the ones thrown from clear(), if any, is thrown during input then ios::badbit is turned on[Footnote 314] in *this's error state. If (exceptions() & badbit) != 0 then the exception is rethrown. In any case, the formatted input function destroys the sentry object. If no exception has been thrown, it returns *this.
Modify 27.7.5.2.1 [ostream.formatted.reqmts] p1 as indicated:
-1- Each formatted output function begins execution by constructing an object of class sentry. If this object returns true when converted to a value of type bool, the function endeavors to generate the requested output. If the generation fails, then the formatted output function does setstate(ios_base::failbit), which might throw an exception. If an exception, other than the ones thrown from clear(), if any, is thrown during output, then ios::badbit is turned on[Footnote 327] in *this's error state. If (exceptions() & badbit) != 0 then the exception is rethrown. Whether or not an exception is thrown, the sentry object is destroyed before leaving the formatted output function. If no exception is thrown, the result of the formatted output function is *this.
Modify 27.7.5.3 [ostream.unformatted] p1 as indicated:
-1- Each unformatted output function begins execution by constructing an object of class sentry. If this object returns true, while converting to a value of type bool, the function endeavors to generate the requested output. If an exception, other than the ones thrown from clear(), if any, is thrown during output, then ios::badbit is turned on[Footnote 330] in *this's error state. If (exceptions() & badbit) != 0 then the exception is rethrown. In any case, the unformatted output function ends by destroying the sentry object, then, if no exception was thrown, returning the value specified for the unformatted output function.
Modify 27.7.4.3 [istream.unformatted] p1 as indicated:
-1- Each unformatted input function begins execution by constructing an object of class sentry with the default argument noskipws (second) argument true. If the sentry object returns true, when converted to a value of type bool, the function endeavors to obtain the requested input. Otherwise, if the sentry constructor exits by throwing an exception or if the sentry object returns false, when converted to a value of type bool, the function returns without attempting to obtain any input. In either case the number of extracted characters is set to 0; unformatted input functions taking a character array of non-zero size as an argument shall also store a null character (using charT()) in the first location of the array. If an exception, other than the ones thrown from clear(), if any, is thrown during input then ios::badbit is turned on[Footnote 317] in *this's error state.
(Exceptions thrown from basic_ios<>::clear() are not caught or rethrown.)If (exceptions() & badbit) != 0 then the exception is rethrown. It also counts the number of characters extracted. If no exception has been thrown it ends by storing the count in a member object and returning the value specified. In any event the sentry object is destroyed before leaving the unformatted input function.
Section: 26.6.7.1 [rand.util.seedseq] Status: New Submitter: Thomas Plum Opened: 2013-12-02 Last modified: 2017-02-03
Priority: 2
View all other issues in [rand.util.seedseq].
View all issues with New status.
Discussion:
With respect to class seed_seq 26.6.7.1 [rand.util.seedseq], is a default-constructed std::seed_seq intended to produce a predictable .generate() sequence?
Implementations differ.Proposed resolution:
Section: 20.15.4.3 [meta.unary.prop] Status: Open Submitter: Richard Smith Opened: 2014-02-01 Last modified: 2017-02-03
Priority: 3
View other active issues in [meta.unary.prop].
View all other issues in [meta.unary.prop].
View all issues with Open status.
Discussion:
The 'Condition' for std::is_empty is listed as:
"T is a class type, but not a union type, with no non-static data members other than bit-fields of length 0, no virtual member functions, no virtual base classes, and no base class B for which is_empty<B>::value is false."
This is incorrect: there is no such thing as a non-static data member that is a bit-field of length 0, since bit-fields of length 0 must be unnamed, and unnamed bit-fields are not members (see 9.2.4 [class.bit] p2).
It also means that classes such as:
struct S {
int : 3;
};
are empty (because they have no non-static data members). There's implementation divergence on the value of is_empty<S>::value.
I'm not sure what the purpose of is_empty is (or how it could be useful), but if it's desirable for the above type to not be treated as empty, something like this could work:"T is a class type, but not a union type, with no non-static data members
other than, no unnamed bit-fields of non-zero length0, no virtual member functions, no virtual base classes, and no base class B for which is_empty<B>::value is false."
and if the above type should be treated as empty, then this might be appropriate:
"T is a class type, but not a union type, with no (named) non-static data members
other than bit-fields of length 0, no virtual member functions, no virtual base classes, and no base class B for which is_empty<B>::value is false."
[2016-08 Chicago]
Walter says: We want is_empty_v<S> to produce false as a result. Therefore, we recommend adoption of the first of the issue's suggestions.
Tuesday AM: Moved to Tentatively Ready
Previous resolution [SUPERSEDED]:
[2016-10 by Marshall - this PR incorrectly highlighted changed portions]
Modify Table 38 — Type property predicates for is_empty as follows:
T is a non-union class type with no non-static data members
other than, no unnamed bit-fields of non-zero length0, no virtual member functions, no virtual base classes, and no base class B for which is_empty_v<B> is false.
[2016-10 Telecon]
Should probably point at section 1.8 for some of this. Status back to 'Open'
Proposed resolution:
Modify Table 38 — Type property predicates for is_empty as follows:
T
is a class type, but not a union type,is a non-union class type with no non-static data membersother than, no unnamed bit-fields of non-zero length0, no virtual member functions, no virtual base classes, and no base class B for which is_empty_v<B> is false.
Section: 23.2.6 [associative.reqmts], 23.2.7 [unord.req] Status: New Submitter: Jeffrey Yasskin Opened: 2014-02-15 Last modified: 2017-02-03
Priority: 3
View other active issues in [associative.reqmts].
View all other issues in [associative.reqmts].
View all issues with New status.
Discussion:
a_uniq.emplace(args) is specified as:
Effects: Inserts a value_type object t constructed with
std::forward<Args>(args)... if and only if there is no element in the
container with key equivalent to the key of t. The bool component of
the returned pair is true if and only if the insertion takes place,
and the iterator component of the pair points to the element with key
equivalent to the key of t.
However, we occasionally find code of the form:
std::unique_ptr<Foo> p(new Foo);
auto res = m.emplace("foo", std::move(p));
where we'd like to avoid destroying the Foo if the insertion doesn't take place (if the container already had an element with the specified key).
N3873 includes a partial solution to this in the form of a new emplace_stable member function, but LEWG's discussion strongly agreed that we'd rather have emplace() Just Work: Should map::emplace() be guaranteed not to move/copy its arguments if the insertion doesn't happen? SF: 8 F: 3 N: 0 A: 0 SA: 0 This poll was marred by the fact that we didn't notice or call out that emplace() must construct the key before doing the lookup, and it must not then move the key after it determines whether an insert is going to happen, and the mapped_type instance must live next to the key. The very similar issue 2006 was previously marked NAD, with N3178 as discussion. However, given LEWG's interest in the alternate behavior, we should reopen the question in this issue. We will need a paper that describes how to implement this before we can make more progress.Proposed resolution:
Section: 30.4.3.5.1 [thread.sharedtimedmutex.class] Status: Open Submitter: Richard Smith Opened: 2014-02-16 Last modified: 2017-02-03
Priority: 2
View all issues with Open status.
Discussion:
30.4.3.5.1 [thread.sharedtimedmutex.class] paragraph 2:
The class shared_timed_mutex shall satisfy all of the SharedTimedMutex requirements (30.4.1.4). It shall be a standard layout class (Clause 9).
There's no SharedTimedMutex requirements; this name doesn't appear anywhere else in the standard. (Prior to N3891, this was SharedMutex, which was equally undefined.)
I assume this concept should be defined somewhere? Also, n3891 changes 30.4.3.5 [thread.sharedtimedmutex.requirements] from defining "shared mutex type" to defining "shared timed mutex type", but its paragraph 2 still talks about "shared mutex type". Is that OK? I think you could argue that it's clear enough what it means, but presumably it should use the term that paragraph 1 defined. 30.4.4.3 [thread.lock.shared] paragraph 1 talks about the "shared mutex requirements", which again is a term that isn't defined, and presumably means "the requirements on a shared timed mutex type" or similar (maybe if SharedMutex or SharedTimedMutex were defined it could be reused here).[2014-05-22, Daniel comments]
As for SharedTimedMutex, there exists a similar problem in regard to TimedMutex referred to in 30.4.3.3.1 [thread.timedmutex.class] p2 and in 30.4.3.3.2 [thread.timedmutex.recursive] p2, but nowhere defined.
Another problem is, that according to 30.4.3.2.1 [thread.mutex.class] p3, "The class mutex shall satisfy all the Mutex requirements (30.4.3 [thread.mutex.requirements]).", but there are no concrete Mutex requirements, 30.4.3 [thread.mutex.requirements] — titled as "Mutex requirements" — describes mutex types, timed mutex types, and shared timed mutex types.[2014-06-08, Daniel comments and provides wording]
The presented wording adds to the existing mutex types, timed mutex types, and shared timed mutex types terms a new set of corresponding MutexType, TimedMutexType, and SharedTimedMutexType requirements.
The reason for the change of requirement names is two-fold: First, the new name better matches the intention to have a concrete name for the requirements imposed on the corresponding mutex types (This kind of requirement deviate from the more general Lockable requirements, which are not restricted to a explicitly enumerated set of library types). Second, using **MutexType over **Mutex provides the additional advantage that it reduces the chances of confusing named requirements from template parameters named Mutex (such as for unique_lock or shared_lock). Nonetheless the here presented wording has one unfortunate side-effect: Once applied it would have the effect that types used to instantiate std::shared_lock cannot be user-defined shared mutex types due to 30.4.4.3 [thread.lock.shared]. The reason is based on the currently lack of an existing SharedLockable requirement set, which would complete the existing BasicLockable and Lockable requirements (which are "real" requirements). This restriction is not actually a problem introduced by the provided resolution but instead one that existed before but becomes more obvious now.[2015-02 Cologne]
Handed over to SG1.
[2015-05 Lenexa, SG1 response]
Thanks to Daniel, and please put it in SG1-OK status. Perhaps open another issue for the remaining problem Daniel points out?
[2015-10 pre-Kona]
SG1 hands this over to LWG for wording review
[2015-10-21 Kona, Daniel comments and adjusts wording to to untimed shared mutex types]
The new wording reflects the addition of the new shared mutex types. The approach used for shared_lock is similar to the one used for unique_lock: The template argument Mutex has a reduced requirement set that is not sufficient for all operations. Only those members that require stronger requirements of SharedTimedMutexType specify that additionally in the Requires element of the corresponding prototype specifications.
The proposed wording could be more general if we would introduce more fundamental requirements set for SharedLockable and SharedTimedLockable types which could be satisfied by user-provided types as well, because the SharedMutexType and SharedTimedMutexType requirements are essentially restricted to an enumerated set of types provided by the Standard Library. But this extension seemed too large for this issue and can be easily fixed later without any harm.Previous resolution [SUPERSEDED]:
This wording is relative to N3936.
Change 30.4.3.2 [thread.mutex.requirements.mutex] as indicated:
-1- The mutex types are the standard library types std::mutex, std::recursive_mutex, std::timed_mutex, std::recursive_timed_mutex, and std::shared_timed_mutex. They shall meet the MutexType requirements set out in this section. In this description, m denotes an object of a mutex type.
Change 30.4.3.2.1 [thread.mutex.class] as indicated:
-3- The class mutex shall satisfy all the MutexType requirements (30.4.3.2 [thread.mutex.requirements.mutex]
30.4.3 [thread.mutex.requirements]). It shall be a standard-layout class (Clause 9).Change 30.4.3.2.2 [thread.mutex.recursive] as indicated:
-2- The class recursive_mutex shall satisfy all the
MutexMutexType requirements (30.4.3.2 [thread.mutex.requirements.mutex]30.4.3 [thread.mutex.requirements]). It shall be a standard-layout class (Clause 9).Change 30.4.3.3 [thread.timedmutex.requirements] as indicated:
-1- The timed mutex types are the standard library types std::timed_mutex, std::recursive_timed_mutex, and std::shared_timed_mutex. They shall meet the TimedMutexType requirements set out below. In this description, m denotes an object of a mutex type, rel_time denotes an object of an instantiation of duration (20.12.5), and abs_time denotes an object of an instantiation of time_point (20.12.6).
Change 30.4.3.3.1 [thread.timedmutex.class] as indicated:
-2- The class timed_mutex shall satisfy all of the TimedMutexType requirements (30.4.3.3 [thread.timedmutex.requirements]). It shall be a standard-layout class (Clause 9).
Change 30.4.3.3.2 [thread.timedmutex.recursive] as indicated:
-2- The class recursive_timed_mutex shall satisfy all of the TimedMutexType requirements (30.4.3.3 [thread.timedmutex.requirements]). It shall be a standard-layout class (Clause 9).
Change 30.4.3.5 [thread.sharedtimedmutex.requirements] as indicated: [Drafting note: The reference to the timed mutex types requirements has been moved after introducing the new requirement set to ensure that SharedTimedMutexType refine TimedMutexType.]
-1- The standard library type std::shared_timed_mutex is a shared timed mutex type. Shared timed mutex types shall meet the SharedTimedMutexType requirements
-?- The shared timed mutex types shall meet the TimedMutexType requirements (30.4.3.3 [thread.timedmutex.requirements]).of timed mutex types (30.4.3.3 [thread.timedmutex.requirements]), and additionally shall meet the requirementsset out below. In this description, m denotes an object of a mutex type, rel_type denotes an object of an instantiation of duration (20.12.5), and abs_time denotes an object of an instantiation of time_point (20.12.6).Change 30.4.3.5.1 [thread.sharedtimedmutex.class] as indicated:
-2- The class shared_timed_mutex shall satisfy all of the SharedTimedMutexType requirements (30.4.3.5 [thread.sharedtimedmutex.requirements]). It shall be a standard-layout class (Clause 9).
Change 30.4.4.3 [thread.lock.shared] as indicated: [Drafting note: Once N3995 has been applied, the following reference should be changed to the new SharedMutexType requirements ([thread.sharedmutex.requirements]) or even better to some new SharedLockable requirements (to be defined) — end drafting note]
-1- […] The supplied Mutex type shall meet the
-2- [Note: shared_lock<Mutex> meets the TimedLockable requirements (30.2.5.4). — end note]shared mutexSharedTimedMutexType requirements (30.4.3.5 [thread.sharedtimedmutex.requirements]).
[2016-02 Jacksonville]
Marshall to review wording.
Proposed resolution:
This wording is relative to N4527.
Change 30.4.3.2 [thread.mutex.requirements.mutex] as indicated:
-1- The mutex types are the standard library types std::mutex, std::recursive_mutex, std::timed_mutex, std::recursive_timed_mutex, std::shared_mutex, and std::shared_timed_mutex. They shall meet the MutexType requirements set out in this section. In this description, m denotes an object of a mutex type.
-2- The mutex types shall meet the Lockable requirements (30.2.5.3 [thread.req.lockable.req]).
Change 30.4.3.2.1 [thread.mutex.class] as indicated:
-3- The class mutex shall satisfy all the MutexType requirements (30.4.3.2 [thread.mutex.requirements.mutex]
30.4.3 [thread.mutex.requirements]). It shall be a standard-layout class (Clause 9).
Change 30.4.3.2.2 [thread.mutex.recursive] as indicated:
-2- The class recursive_mutex shall satisfy all the
MutexMutexType requirements (30.4.3.2 [thread.mutex.requirements.mutex]30.4.3 [thread.mutex.requirements]). It shall be a standard-layout class (Clause 9).
Change 30.4.3.3 [thread.timedmutex.requirements] as indicated:
-1- The timed mutex types are the standard library types std::timed_mutex, std::recursive_timed_mutex, and std::shared_timed_mutex. They shall meet the TimedMutexType requirements set out below. In this description, m denotes an object of a mutex type, rel_time denotes an object of an instantiation of duration (20.12.5), and abs_time denotes an object of an instantiation of time_point (20.12.6).
-2- The timed mutex types shall meet the TimedLockable requirements (30.2.5.4 [thread.req.lockable.timed]).
Change 30.4.3.3.1 [thread.timedmutex.class] as indicated:
-2- The class timed_mutex shall satisfy all of the TimedMutexType requirements (30.4.3.3 [thread.timedmutex.requirements]). It shall be a standard-layout class (Clause 9).
Change 30.4.3.3.2 [thread.timedmutex.recursive] as indicated:
-2- The class recursive_timed_mutex shall satisfy all of the TimedMutexType requirements (30.4.3.3 [thread.timedmutex.requirements]). It shall be a standard-layout class (Clause 9).
Change 30.4.3.4 [thread.sharedmutex.requirements] as indicated: [Drafting note: The reference to the mutex types requirements has been moved after introducing the new requirement set to ensure that SharedMutexType refines MutexType.]
-1- The standard library types std::shared_mutex and std::shared_timed_mutex are shared mutex types. Shared mutex types shall meet the SharedMutexType requirements
-?- The shared mutex types shall meet the MutexType requirements (30.4.3.2 [thread.mutex.requirements.mutex]).of mutex types (30.4.3.2 [thread.mutex.requirements.mutex]), and additionally shall meet the requirementsset out below. In this description, m denotes an object of a shared mutex type.
Change 30.4.3.4.1 [thread.sharedmutex.class] as indicated:
-2- The class shared_mutex shall satisfy all of the SharedMutexType requirements
for shared mutexes(30.4.3.4 [thread.sharedmutex.requirements]). It shall be a standard-layout class (Clause 9).
Change 30.4.3.5 [thread.sharedtimedmutex.requirements] as indicated: [Drafting note: The reference to the timed mutex types requirements has been moved after introducing the new requirement set to ensure that SharedTimedMutexType refines TimedMutexType and SharedMutexType.]
-1- The standard library type std::shared_timed_mutex is a shared timed mutex type. Shared timed mutex types shall meet the SharedTimedMutexType requirements
-?- The shared timed mutex types shall meet the TimedMutexType requirements (30.4.3.3 [thread.timedmutex.requirements]) and the SharedMutexType requirements (30.4.3.4 [thread.sharedmutex.requirements]).of timed mutex types (30.4.3.3 [thread.timedmutex.requirements]), shared mutex types (30.4.3.4 [thread.sharedmutex.requirements]), and additionally shall meet the requirementsset out below. In this description, m denotes an object of a shared timed mutex type, rel_type denotes an object of an instantiation of duration (20.12.5), and abs_time denotes an object of an instantiation of time_point (20.12.6).
Change 30.4.3.5.1 [thread.sharedtimedmutex.class] as indicated:
-2- The class shared_timed_mutex shall satisfy all of the SharedTimedMutexType requirements
for shared timed mutexes(30.4.3.5 [thread.sharedtimedmutex.requirements]). It shall be a standard-layout class (Clause 9).
Change 30.4.4.3 [thread.lock.shared] as indicated:
-1- […] The supplied Mutex type shall meet the
-2- [Note: shared_lock<Mutex> meets the TimedLockable requirements (30.2.5.4). — end note]shared mutexSharedMutexType requirements (30.4.3.5 [thread.sharedtimedmutex.requirements]30.4.3.4 [thread.sharedmutex.requirements]).
Change 30.4.4.3.1 [thread.lock.shared.cons] as indicated:
template <class Clock, class Duration> shared_lock(mutex_type& m, const chrono::time_point<Clock, Duration>& abs_time);-14- Requires: The supplied Mutex type shall meet the SharedTimedMutexType requirements (30.4.3.5 [thread.sharedtimedmutex.requirements]). The calling thread does not own the mutex for any ownership mode.
-15- Effects: Constructs an object of type shared_lock and calls m.try_lock_shared_until(abs_time). […]template <class Rep, class Period> shared_lock(mutex_type& m, const chrono::duration<Rep, Period>& rel_time);-17- Requires: The supplied Mutex type shall meet the SharedTimedMutexType requirements (30.4.3.5 [thread.sharedtimedmutex.requirements]). The calling thread does not own the mutex for any ownership mode.
-18- Effects: Constructs an object of type shared_lock and calls m.try_lock_shared_for(rel_time). […]
Change 30.4.4.3.2 [thread.lock.shared.locking] as indicated:
template <class Clock, class Duration> bool try_lock_until(const chrono::time_point<Clock, Duration>& abs_time);-?- Requires: The supplied Mutex type shall meet the SharedTimedMutexType requirements (30.4.3.5 [thread.sharedtimedmutex.requirements]).
-8- Effects: pm->try_lock_shared_until(abs_time). […]template <class Rep, class Period> bool try_lock_for(const chrono::duration<Rep, Period>& rel_time);-?- Requires: The supplied Mutex type shall meet the SharedTimedMutexType requirements (30.4.3.5 [thread.sharedtimedmutex.requirements]).
-12- Effects: pm->try_lock_shared_for(rel_time). […]
Section: 24.6.3 [istreambuf.iterator] Status: New Submitter: Hyman Rosen Opened: 2014-02-19 Last modified: 2017-02-03
Priority: 3
View all other issues in [istreambuf.iterator].
View all issues with New status.
Discussion:
Given the following code,
#include <sstream> std::stringbuf buf; std::istreambuf_iterator<char> begin(&buf); std::istreambuf_iterator<char> end;
it is not clear from the wording of the Standard whether begin.equal(end) must be true. In at least one implementation it is not (CC: Sun C++ 5.10 SunOS_sparc Patch 128228-25 2013/02/20) and in at least one implementation it is (gcc version 4.3.2 x86_64-unknown-linux-gnu).
24.6.3 [istreambuf.iterator] says that end is an end-of-stream iterator since it was default constructed. It also says that an iterator becomes equal to an end-of-stream iterator when end of stream is reached by sgetc() having returned eof(). 24.6.3.5 [istreambuf.iterator::equal] says that equal() returns true iff both iterators are end of stream or not end of stream. But there seems to be no requirement that equal check for end-of-stream by calling sgetc(). Jiahan Zi at BloombergLP discovered this issue through his code failing to work correctly. Dietmar Kühl has opined in a private communication that the iterators should compare equal.Proposed resolution:
Section: 18.6.2 [new.delete] Status: Open Submitter: Stephen Clamage Opened: 2014-02-20 Last modified: 2017-02-03
Priority: 2
View all other issues in [new.delete].
View all issues with Open status.
Discussion:
Section 18.6.2 [new.delete] and subsections shows:
void* operator new(std::size_t size); void* operator new[](std::size_t size);
That is, without exception-specifications. (Recall that C++03 specified these functions with throw(std::bad_alloc).)
Section 17.5.5.12 [res.on.exception.handling] the end of paragraph 4 says:Any other functions defined in the C++ standard library that do not have an exception-specification may throw implementation-defined exceptions unless otherwise specified. An implementation may strengthen this implicit exception-specification by adding an explicit one.
For example, an implementation could provide C++03-compatible declarations of operator new.
Programmers are allowed to replace these operator new functions. But how can you write the definition of these functions when the exception specification can vary among implementations? For example, the declarationsvoid* operator new(std::size_t size) throw(std::bad_alloc); void* operator new(std::size_t size);
are not compatible.
From what I have been able to determine, gcc has a hack for the special case of operator new to ignore the differences in (at least) the two cases I show above. But can users expect all compilers to quietly ignore the incompatibility? The blanket permission to add any explicit exception specification could cause a problem for any user-overridable function. Different implementations could provide incompatible specifications, making portable code impossible to write.[2016-03, Jacksonville]
STL: Core changes to remove dynamic exception specs would make this moot
Room: This is on track to be resolved by P0003, or may be moot.
Proposed resolution:
Section: 24.2.1 [iterator.requirements.general] Status: New Submitter: Marshall Clow Opened: 2014-03-25 Last modified: 2017-02-03
Priority: 3
View all issues with New status.
Discussion:
24.2.1 [iterator.requirements.general] p9 says:
Destruction of an iterator may invalidate pointers and references previously obtained from that iterator.
But the resolution of LWG issue 2360 specifically advocates returning *--temp; where temp is a local variable.
And 24.2.5 [forward.iterators] p6 says:If a and b are both dereferenceable, then a == b if and only if *a and *b are bound to the same object.
which disallows "stashing" iterators (i.e, iterators that refer to data inside themselves).
So, I suspect that the restriction in p9 should only apply to input iterators, and can probably be moved into 24.2.3 [input.iterators] instead of 24.2.1 [iterator.requirements.general].[2014-05-22, Daniel comments]
Given that forward iterators (and beyond) are refinements of input iterator, moving this constraint to input iterators won't help much because it would still hold for all refined forms.
Proposed resolution:
Section: 22.4.2.1.2 [facet.num.get.virtuals] Status: Open Submitter: Marshall Clow Opened: 2014-04-30 Last modified: 2017-02-03
Priority: 2
View all other issues in [facet.num.get.virtuals].
View all issues with Open status.
Discussion:
In 22.4.2.1.2 [facet.num.get.virtuals] we have:
Stage 3: The sequence of chars accumulated in stage 2 (the field) is converted to a numeric value by the rules of one of the functions declared in the header <cstdlib>:
For a signed integer value, the function strtoll.
For an unsigned integer value, the function strtoull.
For a floating-point value, the function strtold.
This implies that for many cases, this routine should return true:
bool is_same(const char* p)
{
std::string str{p};
double val1 = std::strtod(str.c_str(), nullptr);
std::stringstream ss(str);
double val2;
ss >> val2;
return std::isinf(val1) == std::isinf(val2) && // either they're both infinity
std::isnan(val1) == std::isnan(val2) && // or they're both NaN
(std::isinf(val1) || std::isnan(val1) || val1 == val2); // or they're equal
}
and this is indeed true, for many strings:
assert(is_same("0"));
assert(is_same("1.0"));
assert(is_same("-1.0"));
assert(is_same("100.123"));
assert(is_same("1234.456e89"));
but not for others
assert(is_same("0xABp-4")); // hex float
assert(is_same("inf"));
assert(is_same("+inf"));
assert(is_same("-inf"));
assert(is_same("nan"));
assert(is_same("+nan"));
assert(is_same("-nan"));
assert(is_same("infinity"));
assert(is_same("+infinity"));
assert(is_same("-infinity"));
These are all strings that are correctly parsed by std::strtod, but not by the stream extraction operators. They contain characters that are deemed invalid in stage 2 of parsing.
If we're going to say that we're converting by the rules of strtold, then we should accept all the things that strtold accepts.[2016-04, Issues Telecon]
People are much more interested in round-tripping hex floats than handling inf and nan. Priority changed to P2.
Marshall says he'll try to write some wording, noting that this is a very closely specified part of the standard, and has remained unchanged for a long time. Also, there will need to be a sample implementation.
[2016-08, Chicago]
Zhihao provides wording
The src array in Stage 2 does narrowing only. The actual input validation is delegated to strtold (independent from the parsing in Stage 3 which is again being delegated to strtold) by saying:
[...] If it is not discarded, then a check is made to determine if c is allowed as the next character of an input field of the conversion specifier returned by Stage 1.
So a conforming C++11 num_get is supposed to magically accept an hexfloat without an exponent
0x3.AB
because we refers to C99, and the fix to this issue should be just expanding the src array.
Support for Infs and NaNs are not proposed because of the complexity of nan(n-chars).
[2016-08, Chicago]
Tues PM: Move to Open
[2016-09-08, Zhihao Yuan comments and updates proposed wording]
Examples added.
Proposed resolution:
This wording is relative to N4606.
Change 22.4.2.1.2 [facet.num.get.virtuals]/3 Stage 2 as indicated:
static const char src[] = "0123456789abcdefpxABCDEFPX+-";
Append the following examples to 22.4.2.1.2 [facet.num.get.virtuals]/3 Stage 2 as indicated:
[Example:
Given an input sequence of "0x1a.bp+07p",
if Stage 1 returns %d, "0" is accumulated;
if Stage 1 returns %i, "0x1a" are accumulated;
if Stage 1 returns %g, "0x1a.bp+07" are accumulated.
In all cases, leaving the rest in the input.
— end example]
Section: 20.17.5.8 [time.duration.literals] Status: Open Submitter: Jonathan Wakely Opened: 2014-05-16 Last modified: 2017-02-03
Priority: 3
View all issues with Open status.
Discussion:
20.17.5.8 [time.duration.literals] p3 says:
If any of these suffixes are applied to an integer literal and the resulting chrono::duration value cannot be represented in the result type because of overflow, the program is ill-formed.
Ill-formed requires a diagnostic at compile-time, but there is no way to detect the overflow from unsigned long long to the signed duration<>::rep type.
Overflow could be detected if the duration integer literals were literal operator templates, otherwise overflow can either be undefined or a run-time error, not ill-formed.[Urbana 2014-11-07: Move to Open]
Proposed resolution:
Section: 17.3.15 [defns.ntcts], 22.3.1.1.1 [locale.category], 27.2.2 [iostreams.limits.pos], 27.7.5.2.1 [ostream.formatted.reqmts], 27.7.5.2.4 [ostream.inserters.character] Status: New Submitter: Jeffrey Yasskin Opened: 2014-06-01 Last modified: 2017-05-31
Priority: 3
View all issues with New status.
Discussion:
The term "character type" is used in 17.3.15 [defns.ntcts], 22.3.1.1.1 [locale.category], 27.2.2 [iostreams.limits.pos], 27.7.5.2.1 [ostream.formatted.reqmts], and 27.7.5.2.4 [ostream.inserters.character], but the core language only defines "narrow character types" (3.9.1 [basic.fundamental]).
"wide-character type" is used in 99 [depr.locale.stdcvt], but the core language only defines a "wide-character set" and "wide-character literal".Proposed resolution:
Section: 18.7.2 [type.info] Status: Open Submitter: Stephan T. Lavavej Opened: 2014-06-14 Last modified: 2017-02-03
Priority: 3
View all other issues in [type.info].
View all issues with Open status.
Discussion:
type_info's destructor is depicted as being virtual, which is nearly unobservable to users (since they can't construct or copy this class, they can't usefully derive from it). However, it's technically observable (via is_polymorphic and has_virtual_destructor). It also imposes real costs on implementations, requiring them to store one vptr per type_info object, when RTTI space consumption is a significant concern.
Making this implementation-defined wouldn't affect users (who can observe this only if they're specifically looking for it) and wouldn't affect implementations who need virtual here, but it would allow other implementations to drop virtual and improve their RTTI space consumption. Richard Smith: It's observable in a few other ways.std::map<void*, something> m; m[dynamic_cast<void*>(&typeid(blah))] = stuff;
... is broken by this change, because you can't dynamic_cast a non-polymorphic class type to void*.
type_info& f(); typeid(f());
... evaluates f() at runtime without this change, and might not do so with this change.
These are probably rare things, but I can imagine at least some forms of the latter being used in SFINAE tricks.[Lenexa 2015-05-05: Move to Open]
Marshall to poll LEWG for their opinion
[2016-06]
On the reflector, STL wrote:
We'll prototype this change and report back with data in the future.
[2016-08 Chicago]
No update from STL. Set priority to P3
Proposed resolution:
This wording is relative to N3936.
Change 18.7.2 [type.info] as indicated:
namespace std { class type_info { public:virtualsee below ~type_info(); […] }; }-1- The class type_info describes type information generated by the implementation. Objects of this class effectively store a pointer to a name for the type, and an encoded value suitable for comparing two types for equality or collating order. The names, encoding rule, and collating sequence for types are all unspecified and may differ between programs. Whether ~type_info() is virtual is implementation-defined.
Section: 30.6.6 [futures.promise], 30.6.10.1 [futures.task.members] Status: Review Submitter: Jonathan Wakely Opened: 2014-06-23 Last modified: 2017-03-05
Priority: Not Prioritized
View other active issues in [futures.promise].
View all other issues in [futures.promise].
View all issues with Review status.
Discussion:
The following code has a data race according to the standard:
std::promise<void> p;
std::thread t{ []{
p.get_future().wait();
}};
p.set_value();
t.join();
The problem is that both promise::set_value() and promise::get_future() are non-const member functions which modify the same object, and we only have wording saying that the set_value() and wait() calls (i.e. calls setting and reading the shared state) are synchronized.
The calls don't actually access the same memory locations, so the standard should allow it. My suggestion is to state that calling get_future() does not conflict with calling the various functions that make the shared state ready, but clarify with a note that this does not imply any synchronization or "happens before", only being free from data races.[2015-02 Cologne]
Handed over to SG1.
[2016-10-21, Nico comments]
After creating a promise or packaged task one thread can call get_future() while another thread can set values/exceptions (either directly or via function call). This happens very easily.
Consider:promise<string> p; thread t(doSomething, ref(p)); cout << "result: " << p.get_future().get() << endl;
AFAIK, this is currently UB due to a data race (calling get_future() for the promise might happen while setting the value in the promise).
Yes, a fix is pretty easy:promise<string> p; future<string> f(p.get_future()); thread t(doSomething, ref(p)); cout << "result: " << f.get() << endl;
but I would like to have get_future() and setters be synchronized to avoid this UB.
This would especially make the use of packaged tasks a lot easier. Consider:
vector<packaged_task<int(char)>> tasks;
packaged_task<int(char)> t1(func);
// start separate thread to run all tasks:
auto futCallTasks = async(launch::async, callTasks, ref(tasks));
for (auto& fut : tasksResults) {
cout << "result: " << fut.get_future().get() << endl; // OOPS: UB
}
Again, AFAIK, this program currently is UB due to a data race. Instead, currently I'd have to program, which is a lot less convenient:
vector<packaged_task<int(char)>> tasks;
vector<future<int>> tasksResults;
packaged_task<int(char)> t1(func);
tasksResults.push_back(t1.getFuture()));
tasks.push_back(move(t1));
// start separate thread to run all tasks:
auto futCallTasks = async(launch::async, callTasks, ref(tasks));
for (auto& fut : tasksResults) {
cout << "result: " << fut.get() << endl;
}
With my naive thinking I see not reason not to guarantee that these calls synchronize (as get_future returns an "address/reference" while all setters set the values there).
Previous resolution [SUPERSEDED]:
This wording is relative to N3936.
Change 30.6.6 [futures.promise] around p12 as indicated:
future<R> get_future();-12- Returns: A future<R> object with the same shared state as *this.
-?- Synchronization: Calls to this function do not conflict (1.10 [intro.multithread]) with calls to set_value, set_exception, set_value_at_thread_exit, or set_exception_at_thread_exit. [Note: Such calls need not be synchronized, but implementations must ensure they do not introduce data races. — end note] -13- Throws: future_error if *this has no shared state or if get_future has already been called on a promise with the same shared state as *this. -14- Error conditions: […]Change 30.6.10.1 [futures.task.members] around p13 as indicated:
future<R> get_future();-13- Returns: A future<R> object that shares the same shared state as *this.
-?- Synchronization: Calls to this function do not conflict (1.10 [intro.multithread]) with calls to operator() or make_ready_at_thread_exit. [Note: Such calls need not be synchronized, but implementations must ensure they do not introduce data races. — end note] -14- Throws: a future_error object if an error occurs. -15- Error conditions: […]
[2017-02-28, Kona]
SG1 has updated wording for LWG 2412. SG1 voted to move this to Ready status by unanimous consent.
[2017-03-01, Kona, SG1]
GeoffR to forward revised wording.
Proposed resolution:
This wording is relative to N4640.
Change 30.6.6 [futures.promise] around p12 as indicated:
future<R> get_future();-12- Returns: A future<R> object with the same shared state as *this.
-?- Synchronization: Calls to this function do not introduce data races (1.10 [intro.multithread]) with calls to set_value, set_exception, set_value_at_thread_exit, or set_exception_at_thread_exit. [Note: Such calls need not synchronize with each other. — end note] -13- Throws: future_error if *this has no shared state or if get_future has already been called on a promise with the same shared state as *this. -14- Error conditions: […]
Change 30.6.10.1 [futures.task.members] around p13 as indicated:
future<R> get_future();-13- Returns: A future object that shares the same shared state as *this.
-?- Synchronization: Calls to this function do not introduce data races (1.10 [intro.multithread]) with calls to operator() or make_ready_at_thread_exit. [Note: Such calls need not synchronize with each other. — end note] -14- Throws: a future_error object if an error occurs. -15- Error conditions: […]
Section: 19.3 [assertions] Status: New Submitter: David Krauss Opened: 2014-06-25 Last modified: 2017-02-03
Priority: 4
View all other issues in [assertions].
View all issues with New status.
Discussion:
When NDEBUG is defined, assert must expand exactly to the token sequence ((void)0), with no whitespace (C99 §7.2/1 and also C11 §7.2/1). This is a lost opportunity to pass the condition along to the optimizer.
The user may observe the token sequence using the stringize operator or discriminate it by making a matching #define directive. There is little chance of practical code doing such things. It's reasonable to allow any expansion that is a void expression with no side effects or semantic requirements, for example, an extension keyword or an attribute-specifier finagled into the context. Conforming optimizations would still be limited to treating the condition as hint, not a requirement. Nonconformance on this point is quite reasonable though, given user preferences. Anyway, it shouldn't depend on preprocessor quirks. As for current practice, Darwin OS <assert.h> provides a GCC-style compiler hint __builtin_expect but only in debug mode. Shouldn't release mode preserve hints? Daniel: The corresponding resolution should take care not to conflict with the intention behind LWG 2234.Proposed resolution:
Section: 17.5.5.8 [reentrancy] Status: Open Submitter: Stephan T. Lavavej Opened: 2014-07-01 Last modified: 2017-02-03
Priority: 3
View all other issues in [reentrancy].
View all issues with Open status.
Discussion:
N3936 17.5.5.8 [reentrancy]/1 talks about "functions", but that doesn't address the scenario of calling different member functions of a single object. Member functions often have to violate and then re-establish invariants. For example, vectors often have "holes" during insertion, and element constructors/destructors/etc. shouldn't be allowed to observe the vector while it's in this invariant-violating state. The [reentrancy] Standardese should be extended to cover member functions, so that implementers can either say that member function reentrancy is universally prohibited, or selectively allowed for very specific scenarios.
(For clarity, this issue has been split off from LWG 2382.)[2014-11-03 Urbana]
AJM confirmed with SG1 that they had no special concerns with this issue, and LWG should retain ownership.
AM: this is too overly broad as it also covers calling the exact same member function on a different objectMove to Open
[2015-07 Telecon Urbana]
Marshall to ping STL for updated wording.
Proposed resolution:
This wording is relative to N3936.
Change 17.5.5.8 [reentrancy] p1 as indicated:
-1- Except where explicitly specified in this standard, it is implementation-defined which functions (including different member functions called on a single object) in the Standard C++ library may be recursively reentered.
Section: 99 [fund.ts.v2::optional.relops], 99 [fund.ts.v2::optional.comp_with_t] Status: LEWG Submitter: Daniel Krügler Opened: 2014-06-20 Last modified: 2017-02-03
Priority: Not Prioritized
View all issues with LEWG status.
Discussion:
Addresses: fund.ts.v2
Currently, std::experimental::optional::operator== imposes the EqualityComparable requirement which provides two guarantees: It ensures that operator!= can rely on the equivalence-relation property and more importantly, that the BooleanTestable requirements suggested by issue 2114 are automatically implied.
std::experimental::optional::operator< doesn't provide a LessThanComparable requirement, but there was quite an historic set of changes involved with that family of types: As of N3527 this operator was defined in terms of operator< of the contained type T and imposed the LessThanComparable requirement. In the final acceptance step of optional by the committee, the definition was expressed in terms of std::less and the LessThanComparable requirement had been removed. The inconsistency between operator== and operator< should be removed. One possible course of action would be to add the LessThanComparable to std::experimental::optional::operator<. The EqualityComparable requirement of operator== could also be removed, but in that case both operators would at least need to require the BooleanTestable requirements (see 2114) for the result type of T's operator== and operator<. Arguably, corresponding operators for pair and tuple do not impose LessThanComparable (nor EqualityComparable), albeit the definition of the "derived" relation functions depend on properties ensured by LessThanComparable. According to the SGI definition, the intention was to imposed both EqualityComparable and LessThanComparable. If this is not intended, the standard should clarify this position.[2015-02 Cologne]
VV, DK, JY discuss why and when LessThanComparable was removed. AM: Move to LEWG. Please tell LWG when you look at it.
[2016-11-08, Issaquah]
Not adopted during NB comment resolution
Proposed resolution:
Section: 20.5.3.1 [tuple.cnstr] Status: LEWG Submitter: Akim Demaille Opened: 2014-07-11 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [tuple.cnstr].
View all other issues in [tuple.cnstr].
View all issues with LEWG status.
Discussion:
The issue has been submitted after exchanges with the clang++ team as a consequence of two PR I sent:
Issue 20174 Issue 20175 The short version is shown in the program below:
#include <iostream>
#include <tuple>
struct base
{
void out(const std::tuple<char, char>& w) const
{
std::cerr << "Tuple: " << std::get<0>(w) << std::get<1>(w) << '\n';
}
};
struct decorator
{
base b_;
template <typename... Args>
auto
out(Args&&... args)
-> decltype(b_.out(args...))
{
return b_.out(args...);
}
void out(const char& w)
{
std::cerr << "char: " << w << '\n';
}
};
int main()
{
decorator d{base{}};
char l = 'a';
d.out(l);
}
This is a stripped down version of a real world case where I wrap objects in decorators. These decorators contributes some functions, and forward all the rest of the API to the wrapped object using perfect forwarding. There can be overloaded names.
Here the inner object provides anout(const std::tuple<char, char>&) -> void
function, and the wrappers, in addition to perfect forwarding, provides
out(const char&) -> void
The main function then call out(l) where l is a char lvalue.
With (GCC's) libstdc++ I get the expected result: the char overload is run. With (clang++'s) libc++ it is the tuple version which is run.$ g++-mp-4.9 -std=c++11 bar.cc && ./a.out char: a $ clang++-mp-3.5 -std=c++11 bar.cc -Wall && ./a.out Tuple: a
It turns out that this is the result of an extension of std::tuple in libc++ where they accept constructors with fewer values that tuple elements.
The purpose of this issue is to ask the standard to forbid that this extension be allowed to participate in overload resolution.[2014-10-05, Daniel comments]
This issue is closely related to LWG 2312.
[2014-11 Urbana]
Moved to LEWG.
Extensions to tuple's design are initially a question for LEWG.
Proposed resolution:
Section: 20.10.5 [ptr.align] Status: New Submitter: Melissa Mears Opened: 2014-08-06 Last modified: 2017-02-03
Priority: 3
View all other issues in [ptr.align].
View all issues with New status.
Discussion:
The specification of std::align does not appear to specify what happens when the value of the size parameter is 0. (The question of what happens when alignment is 0 is mentioned in another Defect Report, 2377; it would change the behavior to be undefined rather than potentially implementation-defined.)
The case of size being 0 is interesting because the result is ambiguous. Consider the following code's output:
#include <cstdio>
#include <memory>
int main()
{
alignas(8) char buffer[8];
void *ptr = &buffer[1];
std::size_t space = sizeof(buffer) - sizeof(char[1]);
void *result = std::align(8, 0, ptr, space);
std::printf("%d %td\n", !!result, result ? (static_cast<char*>(result) - buffer) : std::ptrdiff_t(-1));
}
There are four straightforward answers as to what the behavior of std::align with size 0 should be:
The behavior is undefined because the size is invalid.
The behavior is implementation-defined. This seems to be the status quo, with current implementations using #3.
Act the same as size == 1, except that if size == 1 would fail but would be defined and succeed if space were exactly 1 larger, the result is a pointer to the byte past the end of the ptr buffer. That is, the "aligned" version of a 0-byte object can be one past the end of an allocation. Such pointers are, of course, valid when not dereferenced (and a "0-byte object" shouldn't be), but whether that is desired is not specified in the Standard's definition of std::align, it appears. The output of the code sample is "1 8" in this case.
Act the same as size == 1; this means that returning "one past the end" is not a possible result. In this case, the code sample's output is "0 -1".
The two compilers I could get working with std::align, Visual Studio 2013 and Clang 3.4, implement #3. (Change %td to %Id on Visual Studio 2013 and earlier. 2014 and later will have %td.)
Proposed resolution:
Section: 26.7.5 [template.slice.array], 26.7.7 [template.gslice.array], 26.7.8 [template.mask.array], 26.7.9 [template.indirect.array] Status: New Submitter: Akira Takahashi Opened: 2014-08-12 Last modified: 2017-02-03
Priority: 4
View all other issues in [template.slice.array].
View all issues with New status.
Discussion:
I found a missing specification of the copy constructor of the following class templates:
slice_array (26.7.5 [template.slice.array])
gslice_array (26.7.7 [template.gslice.array])
mask_array (26.7.8 [template.mask.array])
indirect_array (26.7.9 [template.indirect.array])
Proposed resolution:
Before 26.7.5.2 [slice.arr.assign] insert a new sub-clause as indicated:
-?- slice_array constructors [slice.arr.cons]
slice_array(const slice_array&);-?- Effects: The constructed slice refers to the same valarray<T> object to which the argument slice refers.
Before 26.7.7.2 [gslice.array.assign] insert a new sub-clause as indicated:
-?- gslice_array constructors [gslice.array.cons]
gslice_array(const gslice_array&);-?- Effects: The constructed slice refers to the same valarray<T> object to which the argument slice refers.
Before 26.7.8.2 [mask.array.assign] insert a new sub-clause as indicated:
-?- mask_array constructors [mask.array.cons]
mask_array(const mask_array&);-?- Effects: The constructed slice refers to the same valarray<T> object to which the argument slice refers.
Before 26.7.9.2 [indirect.array.assign] insert a new sub-clause as indicated:
-?- indirect_array constructors [indirect.array.cons]
indirect_array(const indirect_array&);-?- Effects: The constructed slice refers to the same valarray<T> object to which the argument slice refers.
Section: 28.3 [re.req] Status: New Submitter: Jonathan Wakely Opened: 2014-09-30 Last modified: 2017-02-03
Priority: 3
View all issues with New status.
Discussion:
The requirements on the traits class in 28.3 [re.req] do not say whether a regular expression traits class is required to be DefaultConstructible, CopyConstructible, CopyAssignable etc.
The std::regex_traits class appears to be all of the above, but can basic_regex assume that for user-defined traits classes? Should the following statements all leave u in equivalent states?
X u{v};
X u; u = v;
X u; u.imbue(v.getloc();
Whether they are equivalent has implications for basic_regex copy construction and assignment.
Proposed resolution:
Section: 18.9 [support.initlist] Status: EWG Submitter: David Krauss Opened: 2014-09-30 Last modified: 2017-02-03
Priority: 2
View other active issues in [support.initlist].
View all other issues in [support.initlist].
View all issues with EWG status.
Discussion:
std::initializer_list::operator= 18.9 [support.initlist] is horribly broken and it needs deprecation:
std::initializer_list<foo> a = {{1}, {2}, {3}};
a = {{4}, {5}, {6}};
// New sequence is already destroyed.
Assignability of initializer_list isn't explicitly specified, but most implementations supply a default assignment operator. I'm not sure what 17.4 [description] says, but it probably doesn't matter.
[Lenexa 2015-05-05: Send to EWG as discussed in Telecon]
Proposed resolution:
Edit 18.9 [support.initlist] p1, class template initializer_list synopsis, as indicated:
namespace std {
template<class E> class initializer_list {
public:
[…]
constexpr initializer_list() noexcept;
initializer_list(const initializer_list&) = default;
initializer_list(initializer_list&&) = default;
initializer_list& operator=(const initializer_list&) = delete;
initializer_list& operator=(initializer_list&&) = delete;
constexpr size_t size() const noexcept;
[…]
};
[…]
}
Section: 25.7.7.4 [sort.heap] Status: Tentatively Ready Submitter: François Dumont Opened: 2014-10-07 Last modified: 2017-05-31
Priority: 3
View all issues with Tentatively Ready status.
Discussion:
While creating complexity tests for the GNU libstdc++ implementation I stumbled across a surprising requirement for the std::sort_heap algorithm.
In 25.7.7.4 [sort.heap] p3 the Standard states:Complexity: At most N log(N) comparisons (where N == last - first).
As stated on the libstdc++ mailing list by Marc Glisse sort_heap can be implemented by N calls to pop_heap. As max number of comparisons of pop_heap is 2 * log(N) then sort_heap max limit should be 2 * log(1) + 2 * log(2) + .... + 2 * log(N) that is to say 2 * log(N!). In terms of log(N) we can also consider that this limit is also cap by 2 * N * log(N) which is surely what the Standard wanted to set as a limit.
This is why I would like to propose to replace paragraph 3 by:Complexity: At most 2N log(N) comparisons (where N == last - first).
[2015-02 Cologne]
Marshall will research the maths and report back in Lenexa.
[2015-05-06 Lenexa]
STL: I dislike exact complexity requirements, they prevent one or two extra checks in debug mode. Would it be better to say O(N log(N)) not at most?
[2017-03-04, Kona]
Move to Tentatively Ready. STL may write a paper (with Thomas & Robert) offering guidance about Big-O notation vs. exact requirements.
Proposed resolution:
This wording is relative to N3936.
In 25.7.7.4 [sort.heap] p3 the Standard states:
template<class RandomAccessIterator> void sort_heap(RandomAccessIterator first, RandomAccessIterator last); template<class RandomAccessIterator, class Compare> void sort_heap(RandomAccessIterator first, RandomAccessIterator last, Compare comp);[…]
-3- Complexity: At most 2N log(N) comparisons (where N == last - first).
Section: 20.15 [meta] Status: Core Submitter: Hubert Tong Opened: 2014-11-04 Last modified: 2017-02-03
Priority: 3
View other active issues in [meta].
View all other issues in [meta].
View all issues with Core status.
Discussion:
The BaseCharacteristic for is_constructible is defined in terms of the well-formedness of a declaration for an invented variable. The well-formedness of the described declaration itself may change for the same set of arguments because of the introduction of default arguments.
In the following program, there appears to be conflicting definitions of a specialization of std::is_constructible; however, it seems that this situation is caused without a user violation of the library requirements or the ODR. There is a similar issue with is_convertible, result_of and others. a.cc:
#include <type_traits>
struct A { A(int, int); };
const std::false_type& x1 = std::is_constructible<A, int>();
int main() { }
b.cc:
#include <type_traits>
struct A { A(int, int); };
inline A::A(int, int = 0) { }
const std::true_type& x2 = std::is_constructible<A, int>();
Presumably this program should invoke undefined behaviour, but the Library specification doesn't say that.
[2015-02 Cologne]
Core wording should say "this kind of thing is ill-formed, no diagnostic required"
Proposed resolution:
Section: 18.9 [support.initlist], 24.7 [iterator.range], 24.8 [iterator.container] Status: New Submitter: Richard Smith Opened: 2014-11-11 Last modified: 2017-02-03
Priority: 3
View other active issues in [support.initlist].
View all other issues in [support.initlist].
View all issues with New status.
Discussion:
These sections define helper functions, some of which apply to initializer_list<T>. And they're available if you include one of a long list of header files, many of which include <initializer_list>. But they are not available if you include <initializer_list>. This seems very odd.
#include <initializer_list>
auto x = {1, 2, 3};
const int *p = data(x); // error, undeclared
#include <vector>
const int *q = data(x); // ok
Proposed resolution:
Section: 24.7 [iterator.range] Status: New Submitter: Janez Žemva Opened: 2014-11-16 Last modified: 2017-02-03
Priority: 3
View all other issues in [iterator.range].
View all issues with New status.
Discussion:
The following code:
#include <algorithm>
#include <iterator>
#include <iostream>
#include <cassert>
int main()
{
int a[2][3][4] = { { { 1, 2, 3, 4}, { 5, 6, 7, 8}, { 9, 10, 11, 12} },
{ {13, 14, 15, 16}, {17, 18, 19, 20}, {21, 22, 23, 24} } };
int b[2][3][4];
assert(std::distance(std::begin(a), std::end(a)) == 2 * 3 * 4);
std::copy(std::begin(a), std::end(a), std::begin(b));
std::copy(std::begin(b), std::end(b), std::ostream_iterator<int>(std::cout, ","));
}
does not compile.
A possible way to remedy this would be to add the following overloads of begin, end, rbegin, and rend to 24.7 [iterator.range], relying on recursive evaluation:
namespace std {
template <typename T, size_t M, size_t N>
constexpr remove_all_extents_t<T>*
begin(T (&array)[M][N])
{
return begin(*array);
}
template <typename T, size_t M, size_t N>
constexpr remove_all_extents_t<T>*
end(T (&array)[M][N])
{
return end(array[M - 1]);
}
template <typename T, size_t M, size_t N>
reverse_iterator<remove_all_extents_t<T>*>
rbegin(T (&array)[M][N])
{
return decltype(rbegin(array))(end(array[M - 1]));
}
template <typename T, size_t M, size_t N>
reverse_iterator<remove_all_extents_t<T>*>
rend(T (&array)[M][N])
{
return decltype(rend(array))(begin(*array));
}
}
Proposed resolution:
Section: 17.5.3.5 [allocator.requirements], 23.3.11.3 [vector.capacity], 23.3.11.5 [vector.modifiers] Status: New Submitter: dyp Opened: 2014-12-06 Last modified: 2017-02-03
Priority: 3
View other active issues in [allocator.requirements].
View all other issues in [allocator.requirements].
View all issues with New status.
Discussion:
When resizing a vector, the accessibility and exception specification of the value type's constructors determines whether the elements are copied or moved to the new buffer. However, the copy/move is performed via the allocator's construct member function, which is assumed, but not required, to call the copy/move constructor and propagate only exceptions from the value type's copy/move constructor. The issue might also affect other classes.
The current wording in N4296 relevant here is from Table 28 — "Allocator requirements" in 17.5.3.5 [allocator.requirements]:
Table 28 — Allocator requirements Expression Return type Assertion/note
pre-/post-conditionDefault … a.construct(c, args) (not used) Effect: Constructs an object of type C at c ::new ((void*)c) C(forward<Args>(args)...) …
and from 17.5.3.5 [allocator.requirements] p9:
An allocator may constrain the types on which it can be instantiated and the arguments for which its construct member may be called. If a type cannot be used with a particular allocator, the allocator class or the call to construct may fail to instantiate.
I conclude the following from the wording:
The allocator is not required to call the copy constructor if the arguments (args) is a single (potentially const) lvalue of the value type. Similarly for a non-const rvalue + move constructor. See also 23.2.1 [container.requirements.general] p15 which seems to try to require this, but is not sufficient: That paragraph specifies the semantics of the allocator's operations, but not which constructors of the value type are used, if any.
The allocator may throw exceptions in addition to the exceptions propagated by the constructors of the value type; it can also propagate exceptions from constructors other than a copy/move constructor.
This leads to an issue with the wording of the exception safety guarantees for vector modifiers in 23.3.11.5 [vector.modifiers] p1:
[…]
void push_back(const T& x); void push_back(T&& x);Remarks: Causes reallocation if the new size is greater than the old capacity. If no reallocation happens, all the iterators and references before the insertion point remain valid. If an exception is thrown other than by the copy constructor, move constructor, assignment operator, or move assignment operator of T or by any InputIterator operation there are no effects. If an exception is thrown while inserting a single element at the end and T is CopyInsertable or is_nothrow_move_constructible<T>::value is true, there are no effects. Otherwise, if an exception is thrown by the move constructor of a non-CopyInsertable T, the effects are unspecified.
The wording leads to the following problem: Copy and move assignment are invoked directly from vector. For intermediary objects (see 2164), vector also directly invokes the copy and move constructor of the value type. However, construction of the actual element within the buffer is invoked via the allocator abstraction. As discussed above, the allocator currently is not required to call a copy/move constructor. If is_nothrow_move_constructible<T>::value is true for some value type T, but the allocator uses modifying operations for MoveInsertion that do throw, the implementation is required to ensure that "there are no effects", even if the source buffer has been modified.
Similarly, the vector capacity functions specify exception safety guarantees referring to the move constructor of the value type. For example, vector::resize in 23.3.11.3 [vector.capacity] p14:Remarks: If an exception is thrown other than by the move constructor of a non-CopyInsertable T there are no effects.
The wording leads to the same issue as described above.
Code example:
template<class T>
class allocator;
class pot_reg_type // a type which creates
// potentially registered instances
{
private:
friend class allocator<pot_reg_type>;
struct register_t {};
static std::set<pot_reg_type*>& get_registry()
{
static std::set<pot_reg_type*> registry;
return registry;
}
void enregister() noexcept(false)
{
get_registry().insert(this);
}
void deregister()
{
get_registry().erase(this);
}
public:
pot_reg_type(void ) noexcept(true) {}
pot_reg_type(pot_reg_type const&) noexcept(true) {}
pot_reg_type(pot_reg_type&& ) noexcept(true) {}
private:
pot_reg_type(register_t ) noexcept(false)
{ enregister(); }
pot_reg_type(register_t, pot_reg_type const&) noexcept(false)
{ enregister(); }
pot_reg_type(register_t, pot_reg_type&& ) noexcept(false)
{ enregister(); }
};
template<class T>
class allocator
{
public:
using value_type = T;
value_type* allocate(std::size_t p)
{ return (value_type*) ::operator new(p); }
void deallocate(value_type* p, std::size_t)
{ ::operator delete(p); }
void construct(pot_reg_type* pos)
{
new((void*)pos) pot_reg_type((pot_reg_type::register_t()));
}
void construct(pot_reg_type* pos, pot_reg_type const& source)
{
new((void*)pos) pot_reg_type(pot_reg_type::register_t(), source);
}
template<class... Args>
void construct(T* p, Args&&... args)
{
new((void*)p) T(std::forward<Args>(args)...);
}
};
The construct member function template is only required for rebinding, which can be required e.g. to store additional debug information in the allocated memory (e.g. VS2013).
Even though the value type has an accessible and noexcept(true) move constructor, this allocator won't call that constructor for rvalue arguments. In any case, it does not call a constructor for which vector has formulated its requirements. An exception thrown by a constructor called by this allocator is not covered by the specification in 23.3.11.5 [vector.modifiers] and therefore is guaranteed not to have any effect on the vector object when resizing. For an example how this might invalidate the exception safety guarantees, see this post on the std-discussion mailing list. Another problem arises for value types whose constructors are private, but may be called by the allocator e.g. via friendship. Those value types are not MoveConstructible (is_move_constructible is false), yet they can be MoveInsertable. It is not possible for vector to create intermediary objects (see 2164) of such a type by directly using the move constructor. Current implementations of the single-element forms of vector::insert and vector::emplace do create intermediary objects by directly calling one of the value type's constructors, probably to allow inserting objects from references that alias other elements of the container. As far as I can see, Table 100 — "Sequence container requirements" in 23.2.3 [sequence.reqmts] does not require that the creation of such intermediare objects can be performed by containers using the value type's constructor directly. It is unclear to me if the allocator's construct function could be used to create those intermediary objects, given that they have not been allocated by the allocator. Two possible solutions:Add the following requirement to the allocator_traits::construct function: If the parameter pack args consists of a single parameter of the type value_type&&, the function may only propagate exceptions if is_nothrow_move_constructible<value_type>::value is false.
Requiring alloctor_traits::construct to call a true copy/move constructor of the value type breaks std::scoped_allocator_adapter, as pointed out by Casey Carter in a post on the std-discussion mailing list.Change vector's criterion whether to move or copy when resizing:
Instead of testing the value type's constructors via is_move_constructible, check the value of noexcept( allocator_traits<Allocator>::construct(alloc, ptr, rval) ) where alloc is an lvalue of type Allocator, ptr is an expression of type allocator_traits<Allocator>::pointer and rval is a non-const rvalue of type value_type.A short discussion of the two solutions:
Solution 1 allows keeping is_nothrow_move_constructible<value_type> as the criterion for vector to decide between copying and moving when resizing. It restricts what can be done inside the construct member function of allocators, and requires implementers of allocators to pay attention to the value types used. One could conceive allocators checking the following with a static_assert: If the value type is_nothrow_move_constructible, then the constructor actually called for MoveInsertion within the construct member function is also declared as noexcept. Solution 2 requires changing both the implementation of the default allocator (add a conditional noexcept) and vector (replace is_move_constructible with an allocator-targeted check). It does not impose additional restrictions on the allocator (other than 23.2.1 [container.requirements.general] p15), and works nicely even if the move constructor of a MoveInsertable type is private or deleted (the allocator might be a friend of the value type). In both cases, an addition might be required to provide the basic exception safety guarantee. A short discussion on this topic can be found in the std-discussion mailing list. Essentially, if allocator_traits<Allocator>::construct throws an exception, the object may or may not have been constructed. Two solutions are mentioned in that discussion:allocator_traits<Allocator>::construct needs to tell its caller whether or not the construction was successful, in case of an exception.
If allocator_traits<Allocator>::construct propagates an exception, it shall either not have constructed an object at the specified location, or that object shall have been destroyed (or it shall ensure otherwise that no resources are leaked).
[2015-05-23, Tomasz Kamiński comments]
Solution 1 discussed in this issue also breaks support for the polymorphic_allocator proposed in the part of the Library Fundamentals TS v1, in addition to already mentioned std::scoped_allocator_adapter. Furthermore there is unknown impact on the other user-defined state-full allocators code written in the C++11.
In addition the library resolution proposed in the LWG issues 2089 and N4462, will break the relation between the std::allocator_trait::construct method and copy/move constructor even for the standard std::allocator. As example please consider following class:
struct NonCopyable
{
NonCopyable() = default;
NonCopyable(NonCopyable const&) = delete;
NonCopyable(NonCopyable&&) = delete;
};
struct InitListConstructor : NonCopyable
{
InitListConstructor() = default;
InitListConstructor(std::initializer_list<int>);
operator int() const;
};
For the above declarations following expression are ill-formed:
InitListConstructor copy(std::declval<InitListConstructor const&>()); InitListConstructor move(std::declval<InitListConstructor&&>());
So the class is not CopyConstructible nor MoveConstructible. However the following are well formed:
InitListConstructor copy{std::declval<InitListConstructor const&>()};
InitListConstructor move{std::declval<InitListConstructor&&>()};
And will be used by std::allocator<InitListConstructor>::construct in case of move-insertion and copy-insertion, after appliance of the resolution proposed in mentioned papers:
The gist of the proposed library fix is simple:
if is_constructible_v<TargetType, Args...>, use direct-nonlist-initialization
otherwise, use brace-initialization.
As consequence the requirement proposed in the Solution 1:
If the parameter pack args consists of a single parameter of the type value_type&&, the function may only propagate exceptions if is_nothrow_move_constructible<value_type>::value is false.
Will no longer hold for the std::allocator.
Proposed resolution:
Section: 25.6.1 [alg.copy] Status: LEWG Submitter: Jonathan Wakely Opened: 2015-01-28 Last modified: 2017-02-03
Priority: 3
View all other issues in [alg.copy].
View all issues with LEWG status.
Discussion:
It's unspecified how many times copy_n increments the InputIterator. uninitialized_copy_n is specified to increment it exactly n times, which means if an istream_iterator is used then the next character after those copied is read from the stream and then discarded, losing data.
I believe all three of Dinkumware, libc++ and libstdc++ implement copy_n with n - 1 increments of the InputIterator, which avoids reading and discarding a character when used with istream_iterator, but is inconsistent with uninitialized_copy_n and causes surprising behaviour with istreambuf_iterator instead, because copy_n(in, 2, copy_n(in, 2, out)) is not equivalent to copy_n(in, 4, out)[2016-08 Chicago]
Tues PM: refer to LEWG
Proposed resolution:
Section: 20.5.3.8 [tuple.rel], 20.10.9.2 [allocator.globals], 20.11.1.5 [unique.ptr.special], 20.11.2.2.7 [util.smartptr.shared.cmp], 20.17.5.6 [time.duration.comparisons], 20.17.6.6 [time.point.comparisons], 20.13.5 [scoped.adaptor.operators], 24.5.1.3.13 [reverse.iter.op==], 24.5.3.3.13 [move.iter.op.comp] Status: New Submitter: Richard Smith Opened: 2015-02-07 Last modified: 2017-02-03
Priority: 3
View other active issues in [tuple.rel].
View all other issues in [tuple.rel].
View all issues with New status.
Discussion:
The standard library specifies a lot of heterogeneous comparison operators. For instance:
template<class... TTypes, class... UTypes> constexpr bool operator!=(const tuple<TTypes...>&, const tuple<UTypes...>&);
This has an unfortunate consequence:
#include <tuple> #include <utility> using namespace std::rel_ops; std::tuple<int> a(0); bool b = a != a;
The last line here is ill-formed due to ambiguity: it might be rel_ops::operator!=, and it might be the heterogeneous tuple operator!=. These are not partially ordered, because they have different constraints: rel_ops requires the types to match, whereas the tuple comparison requires both types to be tuples (but not to match). The same thing happens for user code that defines its own unconstrained 'template<typename T> operator!=(const T&, const T&)' rather than using rel_ops.
One straightforward fix would be to add a homogeneous overload for each heterogeneous comparison:template<class... TTypes> constexpr bool operator!=(const tuple<TTypes...>&, const tuple<TTypes...>&);This is then unambiguously chosen over the other options in the preceding case. FWIW, libstdc++ already does this in some cases.
Proposed resolution:
Section: 99 [depr.conversions.string] Status: New Submitter: Jonathan Wakely Opened: 2015-03-04 Last modified: 2017-05-31
Priority: 4
View other active issues in [depr.conversions.string].
View all other issues in [depr.conversions.string].
View all issues with New status.
Discussion:
How do wstring_convert::from_bytes and wstring_convert::to_bytes use the cvtstate member?
Is it passed to the codecvt member functions? Is a copy of it passed to the member functions? "Otherwise it shall be left unchanged" implies a copy is used, but if that's really what's intended there are simpler ways to say so.Proposed resolution:
Section: 99 [depr.conversions.buffer] Status: New Submitter: Jonathan Wakely Opened: 2015-03-04 Last modified: 2017-05-31
Priority: 4
View other active issues in [depr.conversions.buffer].
View all other issues in [depr.conversions.buffer].
View all issues with New status.
Discussion:
How does wbuffer_convert use the cvtstate member?
Is the same conversion state object used for converting both the get and put areas? That means a read which runs out of bytes halfway through a multibyte character will leave some shift state in cvtstate, which would then be used by a following write, even though the shift state of the get area is unrelated to the put area.Proposed resolution:
Section: 99 [depr.conversions.buffer] Status: New Submitter: Jonathan Wakely Opened: 2015-03-04 Last modified: 2017-05-31
Priority: 4
View other active issues in [depr.conversions.buffer].
View all other issues in [depr.conversions.buffer].
View all issues with New status.
Discussion:
If a codecvt conversion returns codecvt_base::error should that be treated as EOF? An exception? Should all the successfully converted characters before a conversion error be available to the users of the wbuffer_convert and/or the internal streambuf, or does a conversion error lose information?
Proposed resolution:
Section: 99 [depr.conversions.string] Status: New Submitter: Jonathan Wakely Opened: 2015-03-04 Last modified: 2017-05-31
Priority: 4
View other active issues in [depr.conversions.string].
View all other issues in [depr.conversions.string].
View all issues with New status.
Discussion:
Paragraph 4 of 99 [depr.conversions.string] introduces byte_err_string as "a byte string to display on errors". What does display mean? The string is returned on error, it's not displayed anywhere.
Paragraph 14 says "Otherwise, if the object was constructed with a byte-error string, the member function shall return the byte-error string." The term byte-error string is not used anywhere else. Paragraph 17 talks about storing "default values in byte_err_string". What default value? Is "Hello, world!" allowed? If it means default-construction it should say so. If paragraph 14 says it won't be used what does it matter how it's initialized? The end of the paragraph refers to storing "byte_err in byte_err_string". This should be more clearly related to the wording in paragraph 14. It might help if the constructor (and destructor) was specified before the other member functions, so it can more formally define the difference between being "constructed with a byte-error string" and not. All the same issues apply to the wide_err_string member.Proposed resolution:
Section: 28 [re] Status: New Submitter: Stephan T. Lavavej Opened: 2015-03-27 Last modified: 2017-02-03
Priority: 3
View other active issues in [re].
View all other issues in [re].
View all issues with New status.
Discussion:
Only 4 functions are marked noexcept in all of Clause 28. Many more need to be marked — for example, regex_error::code(), basic_regex::swap(), and sub_match::length().
Proposed resolution:
Section: 20.14.7 [comparisons] Status: New Submitter: Agustín K-ballo Bergé Opened: 2015-04-01 Last modified: 2017-02-03
Priority: 3
View other active issues in [comparisons].
View all other issues in [comparisons].
View all issues with New status.
Discussion:
It is not entirely clear if and when the specializations of std::less (and friends) for pointer types can be used in a constant expression. Consider the following code:
#include <functional>
struct foo {};
foo x, y;
constexpr bool b = std::less<foo*>{}(&x, &y); // [1]
foo z[] = {{}, {}};
constexpr bool ba = std::less<foo*>{}(&z[0], &z[1]); // [2]
Comparing the address of unrelated objects is not a constant expression since the result is unspecified, so it could be expected for [1] to fail and [2] to succeed. However, std::less specialization for pointer types is well-defined and yields a total order, so it could just as well be expected for [1] to succeed. Finally, since the implementation of such specializations is not mandated, [2] could fail as well (This could happen, if an implementation would provide such a specialization and if that would use built-in functions that would not be allowed in constant expressions, for example). In any case, the standard should be clear so as to avoid implementation-defined constexpr-ness.
[2017-01-22, Jens provides rationale and proposed wording]
std::less<T*> is required to deliver a total order on pointers. However, the layout of global objects is typically determined by the linker, not the compiler, so requiring std::less<T*> to provide an ordering at compile-time that is consistent with run-time would need results from linking to feed back to the compiler, something that C++ has traditionally not required.
Proposed resolution:
This wording is relative to N4618.
Add at the end of 20.14.7 [comparisons]:
-2- For templates less, greater, less_equal, and greater_equal, […], if the call operator calls a built-in operator comparing pointers, the call operator yields a strict total order that is consistent among those specializations and is also consistent with the partial order imposed by those built-in operators. Relational comparisons of pointer values are not required to be usable as constant expressions.
Section: 18.9 [support.initlist] Status: New Submitter: David Krauss Opened: 2015-04-27 Last modified: 2017-02-03
Priority: 4
View other active issues in [support.initlist].
View all other issues in [support.initlist].
View all issues with New status.
Discussion:
The typical use-case of std::initializer_list<T> is for a pass-by-value parameter of T's constructor. However, this contravenes 17.5.4.8 [res.on.functions]/2.5 because initializer_list doesn't specifically allow incomplete types (as do for example std::unique_ptr (20.11.1 [unique.ptr]/5) and std::enable_shared_from_this (20.11.2.5 [util.smartptr.enab]/2)).
A resolution would be to copy-paste the relevant text from such a paragraph.Proposed resolution:
Section: 20.15.4.3 [meta.unary.prop] Status: New Submitter: Hubert Tong Opened: 2015-05-07 Last modified: 2017-02-03
Priority: 3
View other active issues in [meta.unary.prop].
View all other issues in [meta.unary.prop].
View all issues with New status.
Discussion:
I do not believe that the wording in 20.15.4.3 [meta.unary.prop] paragraph 3 allows for the following program to be ill-formed:
#include <type_traits>
template <typename T> struct B : T { };
template <typename T> struct A { A& operator=(const B<T>&); };
std::is_assignable<A<int>, int> q;
In particular, I do not see where the wording allows for the "compilation of the expression" declval<T>() = declval<U>() to occur as a consequence of instantiating std::is_assignable<T, U> (where T and U are, respectively, A<int> and int in the example code).
Instantiating A<int> as a result of requiring it to be a complete type does not trigger the instantiation of B<int>; however, the "compilation of the expression" in question does.Proposed resolution:
Section: 27.7.5.1.3 [ostream::sentry] Status: New Submitter: Roger Orr Opened: 2015-05-08 Last modified: 2017-02-03
Priority: 3
View all other issues in [ostream::sentry].
View all issues with New status.
Discussion:
In the current 27.7.5.1.3 [ostream::sentry], p4 refers to the now deprecated std::uncaught_exception(): D.9 [depr.uncaught].
If ((os.flags() & ios_base::unitbuf) && !uncaught_exception() && os.good()) is true, calls os.rdbuf()->pubsync().
This needs to be changed, for example to use std::uncaught_exceptions() and to capture the value on entry and compare with the saved value on exit.
[2015-06, Telecon]
JW: I already added an 's' here to make it use the new function, but that doesn't resolve Roger's suggestion to capture the value on entry and check it.
Proposed resolution:
Section: 27.7.4.5 [istream.rvalue] Status: New Submitter: Richard Smith Opened: 2015-05-08 Last modified: 2017-02-03
Priority: 3
View all other issues in [istream.rvalue].
View all issues with New status.
Discussion:
Consider:
auto& is = make_istream() >> x; // oops, istream object is already gone
With a basic_istream&& return type, the above would be ill-formed, and generally we'd preserve the value category properly.
[2015-06, Telecon]
JW: think this needs proper consideration, it would make
stream() >> x >> y >> zgo from 3 operator>> calls to 6 operator>> calls, and wouldn't prevent dangling references (change the example to auto&&)
Proposed resolution:
Section: 27.7.4.2.3 [istream.extractors] Status: Open Submitter: Richard Smith Opened: 2015-05-08 Last modified: 2017-02-03
Priority: 2
View all other issues in [istream.extractors].
View all issues with Open status.
Discussion:
We removed gets() (due to an NB comment and C11 — bastion of backwards compatibility — doing the same). Should we remove this too?
Unlike gets(), there are legitimate uses:char buffer[32]; char text[32] = // ... ostream_for_buffer(text) >> buffer; // ok, can't overrun buffer
… but the risk from constructs like "std::cin >> buffer" seems to outweigh the benefit.
The issue had been discussed on the library reflector starting around c++std-lib-35541.[2015-06, Telecon]
VV: Request a paper to deprecate / remove anything
[2015-10, Kona Saturday afternoon]
STL: This overload is evil and should probably die.
VV: I agree with that, even though I don't care.
STL: Say that we either remove it outright following the gets() rationale, or at least deprecate it.
Move to Open; needs a paper.
[2016-08, Chicago: Zhihao Yuan comments and provides wording]
Rationale:
I would like to keep some reasonable code working;
Reasonable code includes two cases:
width() > 0, any pointer argument
width() >= 0, array argument
For a), banning bad code will become a silent behavior change at runtime; for b), it breaks at compile time.
I propose to replace these signatures with references to arrays. An implementation may want to ship the old instantiatations in the binary without exposing the old signatures.
[2016-08, Chicago]
Tues PM: General agreement on deprecating the unsafe call, but no consensus for the P/R.
General feeling that implementation experience would be useful.
Proposed resolution:
This wording is relative to N4606.
Modify 27.7.4.2.3 [istream.extractors] as indicated:
template<class charT, class traits, size_t N> basic_istream<charT, traits>& operator>>(basic_istream<charT, traits>& in,charT* scharT (&s)[N]); template<class traits, size_t N> basic_istream<char, traits>& operator>>(basic_istream<char, traits>& in,unsigned char* sunsigned char (&s)[N]); template<class traits, size_t N> basic_istream<char, traits>& operator>>(basic_istream<char, traits>& in,signed char* ssigned char (&s)[N]);-7- Effects: Behaves like a formatted input member (as described in 27.7.4.2.1 [istream.formatted.reqmts]) of in. After a sentry object is constructed, operator>> extracts characters and stores them into
successive locations of an array whose first element is designated bys. If width() is greater than zero, n iswidth()min(size_t(width()), N). Otherwise n isthe number of elements of the largest array of char_type that can store a terminating charT()N. n is the maximum number of characters stored.
Section: 27.6.3 [streambuf] Status: New Submitter: Jonathan Wakely Opened: 2015-05-28 Last modified: 2017-02-03
Priority: 3
View all other issues in [streambuf].
View all issues with New status.
Discussion:
27.6.3 [streambuf] p1 says:
The class template basic_streambuf<charT, traits> serves as an abstract base class for deriving various stream buffers whose objects each control two character sequences: […]
The term "abstract base class" is not defined in the standard, but "abstract class" is (10.4 [class.abstract]).
According to the synopsis basic_streambuf has no pure virtual functions so is not an abstract class and none of libstdc++, libc++, or dinkumware implement it as an abstract class. I don't believe the wording was ever intended to require it to be an abstract class, but it could be read that way. I suggest the wording be changed to "polymorphic base class" or something else that can't be seen to imply a normative requirement to make it an abstract class.Proposed resolution:
Section: 1.10 [intro.multithread], 29.5 [atomics.types.generic], 18.10 [support.runtime] Status: New Submitter: Geoffrey Romer Opened: 2015-05-29 Last modified: 2017-02-03
Priority: 3
View all other issues in [intro.multithread].
View all issues with New status.
Discussion:
The concurrency libraries specified in clauses 29 and 30 do not adequately specify how they relate to the concurrency model specified in 1.10 [intro.multithread]. In particular:
1.10 [intro.multithread] specifies "atomic objects" as having certain properties. I can only assume that instances of the classes defined in Clause 29 are intended to be "atomic objects" in this sense, but I can't find any wording to specify that, and it's genuinely unclear whether Clause 30 objects are atomic objects. In fact, on a literal reading the C++ Standard doesn't appear to provide any portable way to create an atomic object, or even determine whether an object is an atomic object. (It's not clear if the term "atomic object" is actually needed, given that atomic objects can have non-atomic operations, and non-atomic objects can have atomic operations. But even if the term itself goes away, there still needs to be some indication that Clause 29 objects have the properties currently attributed to atomic objects). Similarly, 1.10 [intro.multithread] uses "atomic operation" as a term of art, but the standard never unambiguously identifies any operation as an "atomic operation" (although in one case it unambiguously identifies an operation that is not atomic). It does come close in a few cases, but not close enough:1.10 [intro.multithread]/p7 could be read to imply that "synchronization operations" in Clauses 29 and 30 are also atomic operations. However, that's vague and indirect, and somewhat belied by 30.4.3.2 [thread.mutex.requirements.mutex]/p5, which specifies that mutex lock and unlock operations "behave as atomic operations", but only "for purposes of determining the existence of a data race". Furthermore, not a single operation in Clause 29 explicitly identifies itself as a "synchronization operation".
29.5 [atomics.types.generic]/p4 states in part that "There shall be a specialization atomic<bool> which provides the general atomic operations as specified in 29.6.1", but read in context, "general atomic operations" appears to be a loose synonym for "general operations on atomic types" as defined in 29.6.1 [atomics.types.operations.general], rather than a use of "atomic object" as Words of Power. Incidentally, "atomic type" is never satisfactorily defined either (although the <atomic> synopsis comes close).
18.10 [support.runtime]/p10 specifies exactly which operations are "plain lock-free atomic operations", but in a standard where an "integral constant expression" isn't necessarily a "constant expression", I do not feel safe assuming that a "plain lock-free atomic operation" is an "atomic operation".
Hans Boehm tells me the operations with "atomically" in the Effects element are intended to be atomic operations, but since "atomic operation" is a term of art (e.g. in 1.10 [intro.multithread]/p27.4), I think this needs to be spelled out rather than assumed. Furthermore, this does not help with 29.8 [atomics.fences], or anything in Clause 30.
Proposed resolution:
Section: 99 [depr.locale.stdcvt] Status: New Submitter: Jonathan Wakely Opened: 2015-06-08 Last modified: 2017-05-31
Priority: 3
View all other issues in [depr.locale.stdcvt].
View all issues with New status.
Discussion:
The enumeration type codecvt_mode is effectively a bitmask type (17.4.2.1.4 [bitmask.types]) with three elements, but isn't defined as such.
This harms usability because bitmask types are required to work well with bitwise operators, but codecvt_mode doesn't have overloaded operators, making it very inconvenient to combine values:std::codecvt_utf16<char32_t, 0x10FFFF, static_cast<std::codecvt_mode>(std::little_endian|std::generate_header)> cvt;
The static_cast harms readability and should not be necessary.
I suggest that either codecvt_mode is specified to be a bitmask type, or as a minimal fix we provide an overloaded operator| that returns the right type.Proposed resolution:
Section: 18.6.2.4 [new.delete.dataraces] Status: New Submitter: Hans Boehm Opened: 2015-06-09 Last modified: 2017-02-03
Priority: 3
View all other issues in [new.delete.dataraces].
View all issues with New status.
Discussion:
18.6.2.4 [new.delete.dataraces] uses obsolete wording.
It should introduce a "synchronizes with" relationship. "Happens before" is too weak, since that may not composes with sequenced before. The "shall not introduce a data race" wording is probably not technically correct either. These may race with other (non-allocation/deallocation) concurrent accesses to the object being allocated or deallocated.Proposed resolution:
Section: 20.13.4 [allocator.adaptor.members] Status: New Submitter: David Krauss Opened: 2015-06-16 Last modified: 2017-02-03
Priority: 3
View other active issues in [allocator.adaptor.members].
View all other issues in [allocator.adaptor.members].
View all issues with New status.
Discussion:
20.13.4 [allocator.adaptor.members]/10 requires that the argument types in the piecewise-construction tuples all be CopyConstructible. These tuples are typically created by std::forward_as_tuple, such as in ¶13. So they will be a mix of lvalue and rvalue references, the latter of which are not CopyConstructible.
My guess is that CopyConstructible was specified to feed the tuple_cat, before that function could handle rvalues. Since the argument tuple is already moved in ¶11, the requirement is obsolete. It should either be changed to MoveConstructible, or perhaps better, convert the whole tuple to references (i.e. form tuple<Args1&&...>) so nothing needs to be moved. After all, this is a facility for handling non-movable types. It appears that the resolution of DR 2203, which added std::move to ¶11, simply omitted the change to ¶10.[2016-11-08, Jonathan comments]
My paper P0475R0 provides a proposed resolution.
Proposed resolution:
Section: 22.4.5.1.2 [locale.time.get.virtuals] Status: Open Submitter: Hubert Tong Opened: 2015-06-19 Last modified: 2017-02-03
Priority: 4
View all other issues in [locale.time.get.virtuals].
View all issues with Open status.
Discussion:
I recently encountered a failure related to questionable use of do_get_year. The platform where the code happened to work had an implementation which handled certain three-digit "year identifiers" as the number of years since 1900 (this article describes such an implementation).
22.4.5.1.2 [locale.time.get.virtuals] makes it implementation defined whether two-digit years are accepted, etc., but does not say anything specifically about three-digit years. The implementation freedom to not report errors in 22.4.5.1 [locale.time.get] paragraph 1 also seems to be too broad. See also the discussion following c++std-lib-38042.[2016-08 Chicago]
Wed PM: This has been this way since C++98. Don't think it's a P2.
Change to P4, and move to Open.
Proposed resolution:
Section: 21.1 [strings.general] Status: New Submitter: Jonathan Wakely Opened: 2015-06-26 Last modified: 2017-02-03
Priority: 4
View all other issues in [strings.general].
View all issues with New status.
Discussion:
The allocator-aware container requirements in Table 98 impose no MoveAssignable requirements on the value_type when propagate_on_container_move_assignment is true, because typically the container's storage would be moved by just exchanging some pointers.
However for a basic_string using the small string optimization move assignment may need to assign individual characters into the small string buffer, even when the allocator propagates. The only requirement on the char-like objects stored in a basic_string are that they are non-array POD types and Destructible, which means that a POD type with a deleted move assignment operator should be usable in a basic_string, despite it being impossible to move assign:
#include <string>
struct odd_pod
{
odd_pod() = default;
odd_pod& operator=(odd_pod&&) = delete;
};
static_assert(std::is_pod<odd_pod>::value, "POD");
int main()
{
using S = std::basic_string<odd_pod>;
S s;
s = S{}; // fails
}
Using libstdc++ basic_string<odd_pod> cannot even be default-constructed because the constructor attempts to assign the null terminator to the first element of the small string buffer.
Similar problems exist with POD types with a deleted default constructor. I believe that basic_string should require its value_type to be at least DefaultConstructible and MoveAssignable.[2016-06, Oulu]
This should be resolved by P0178
Note: P0178 was sent back to LEWG in Oulu.
Proposed resolution:
Section: 26.6.8.4.2 [rand.dist.pois.exp] Status: Open Submitter: Michael Prähofer Opened: 2015-08-20 Last modified: 2017-02-03
Priority: 2
View all issues with Open status.
Discussion:
Original title was: exponential_distribution<float> sometimes returns inf.
The random number distribution class template exponential_distribution<float> may return "inf" as can be seen from the following example program:
// compiled with
// g++ -std=c++11 Error_exp_distr.cpp
#include <iostream>
#include <random>
#include <bitset>
int main(){
unsigned long long h;
std::mt19937_64 mt1(1);
std::mt19937_64 mt2(1);
mt1.discard(517517);
mt2.discard(517517);
std::exponential_distribution<float> dis(1.0);
h = mt2();
std::cout << std::bitset<64>(h) << " " << (float) -log(1 - h/pow(2, 64)) << " "
<< -log(1 - (float) h/pow(2, 64)) << " " << dis(mt1) << std::endl;
h = mt2();
std::cout << std::bitset<64>(h) << " " << (float) -log(1 - h/pow(2, 64)) << " "
<< -log(1 - (float) h/pow(2, 64)) << " " << dis(mt1) << std::endl;
}
output:
0110010110001001010011000111000101001100111110100001110011100001 0.505218 0.505218 0.505218 1111111111111111111111111101010011000110011110011000110101100110 18.4143 inf inf
The reason seems to be that converting a double x in the range [0, 1) to float may result in 1.0f if x is close enough to 1. I see two possibilities to fix that:
use internally double (or long double?) and then convert the result at the very end to float.
take only 24 random bits and convert them to a float x in the range [0, 1) and then return -log(1 - x).
I have not checked if std::exponential_distribution<double> has the same problem: For float on the average 1 out of 224 (~107) draws returns "inf", which is easily confirmed. For double on the average 1 out of 253 (~1016) draws might return "inf", which I have not tested.
Marshall:
I don't think the problem is in std::exponential_distribution; but rather in generate_canonical.
Consider:
which outputs:std::mt19937_64 mt2(1); mt2.discard(517517); std::cout << std::hexfloat << std::generate_canonical<float, std::numeric_limits<float>::digits>(mt2) << std::endl; std::cout << std::hexfloat << std::generate_canonical<float, std::numeric_limits<float>::digits>(mt2) << std::endl; std::cout << std::hexfloat << std::generate_canonical<float, std::numeric_limits<float>::digits>(mt2) << std::endl;
but generate_canonical is defined to return a result in the range [0, 1).0x1.962532p-2 0x1p+0 0x1.20d0cap-3
[2015-10, Kona Saturday afternoon]
Options:
WEB: The one thing we cannot tolerate is any output range other than [0, 1).
WEB: I believe there may be a documented algorithm for the generator, and perhaps it's possible to discover en-route that the algorithm produces the wrong result and fix it.
MC: No. I analyzed this once, and here it is: the algorithm is in [rand.util.canonical], and it's all fine until p5. The expression S/R^k is mathematically less than one, but it may round to one.
GR: Could we change the rounding mode for the computation?
HH: No, because the rounding mode is global, not thread-local.
AM: SG1 wants to get rid of the floating point environment.
STL: The problem is that the standard specifies the implementation, and the implementation doesn't work.
MC: I'm not sure if nudging it down will introduce a subtle bias.
EF: I worry about how the user's choice of floating point environment affects the behaviour.
MS offers to run the topic past colleagues.
MC: Will set the status to open. STL wants to rename the issue. WEB wants to be able to find the issue by its original name still.
Mike Spertus to run the options past his mathematical colleagues, and report back.
Proposed resolution:
Section: 20.5.3.1 [tuple.cnstr] Status: New Submitter: Brian Rodriguez Opened: 2015-08-25 Last modified: 2017-02-19
Priority: 3
View other active issues in [tuple.cnstr].
View all other issues in [tuple.cnstr].
View all issues with New status.
Discussion:
The std::tuple order of element construction is unspecified. It is either in the same order of the type list or in reverse.
Consider the following program:
#include <iostream>
#include <tuple>
struct X
{
X(int) { std::cout << "X constructor\n"; }
};
struct Y
{
Y(int) { std::cout << "Y constructor\n"; }
};
int main()
{
std::tuple<X, Y> t(1, 2);
}
Here is a link to two sample compilations. The first uses libstdc++ and constructs in reverse order, and the second uses libc++ and constructs in in-order.
A std::tuple mimics both a struct and type-generic container and should thus follow their standards. Construction is fundamentally different from a function call, and it has been historically important for a specific order to be guaranteed; namely: whichever the developer may decide. Mandating construction order will allow developers to reference younger elements later on in the chain as well, much like a struct allows you to do with its members. There are implementation issues as well. Reversed lists will require unnecessary overhead for braced-initializer-list initialization. Since lists are evaluated from left to right, the initializers must be placed onto the stack to respect the construction order. This issue could be significant for large tuples, deeply nested tuples, or tuples with elements that require many constructor arguments. I propose that the std::tuple<A, B, ..., Y, Z>'s constructor implementation be standardized, and made to construct in the same order as its type list e.g. A{}, B{}, ..., Y{}, Z{}.Daniel:
When N3140 became accepted, wording had been added that gives at least an indication of requiring element initialization in the order of the declaration of the template parameters. This argumentation can be based on 20.5.3.1 [tuple.cnstr] p3 (emphasize mine):-3- In the constructor descriptions that follow, let i be in the range [0,sizeof...(Types)) in order, Ti be the ith type in Types, and Ui be the ith type in a template parameter pack named UTypes, where indexing is zero-based.
But the current wording needs to be improved to make that intention clearer and an issue like this one is necessary to be sure that the committee is agreeing (or disagreeing) with that intention, especially because N3140 didn't really point out the relevance of the element construction order in the discussion, and because not all constructors explicitly refer to the ordered sequence of numbers generated by the variable i (The move constructor does it right, but most other don't do that).
[2017-02-12, Alisdair comments]
Note that this issue should not be extended to cover the assignment operators, as implementations may want the freedom to re-order member-wise assignment so that, for example, all potentially-throwing assignments are performed before non-throwing assignments (as indicated by the noexcept operator).
Proposed resolution:
Section: 30.6.5 [futures.state] Status: Open Submitter: Agustín K-ballo Bergé Opened: 2015-09-03 Last modified: 2017-02-03
Priority: 3
View all other issues in [futures.state].
View all issues with Open status.
Discussion:
When a shared-state is released, it may be necessary to execute user defined code for the destructor of a stored value or exception. It is unclear whether the execution of said destructor constitutes an observable side effect.
While discussing N4445 in Lenexa, Nat Goodspeed pointed out that 30.6.5 [futures.state]/5.1 does not explicitly mention the destruction of the result, so implementations should be allowed to release (or reuse) a shared state ahead of time under the "as-if" rule.
The standard should clarify whether the execution of destructors is a visible side effect of releasing a shared state.[2016-08-03 Chicago]
This is related to 2532
Fri AM: Moved to Open
Proposed resolution:
Section: 30.6.6 [futures.promise] Status: Open Submitter: Agustín K-ballo Bergé Opened: 2015-09-03 Last modified: 2017-02-03
Priority: 3
View other active issues in [futures.promise].
View all other issues in [futures.promise].
View all issues with Open status.
Discussion:
promise::set_value_at_thread_exit and promise::set_exception_at_thread_exit operate on a shared state at thread exit, without making the thread participate in the ownership of such shared state.
Consider the following snippet:
std::promise<int>{}.set_value_at_thread_exit(42);
Arguably, since the promise abandons its shared state without actually making it ready, a broken_promise error condition should be stored in the shared state. Implementations diverge, they either crash at thread exit by dereferencing an invalid pointer, or keep the shared state around until thread exit.
[2016-08-03 Chicago]
This is related to 2530
[2016-08-03, Billy O'Neal suggests concrete wording]
Fri AM: Moved to Open
Proposed resolution:
This wording is relative to N4606.
Change 30.6.5 [futures.state] p7 as indicated:
-7- When an asynchronous provider is said to abandon its shared state, it means:
(7.1) — first, if that state is not ready or scheduled to be made ready at thread exit, the provider
(7.1.1) — stores an exception object of type future_error with an error condition of broken_promise within its shared state; and then
(7.1.2) — makes its shared state ready;
Change 30.6.5 [futures.state] p10 as indicated:
-10- Some functions (e.g., promise::set_value_at_thread_exit)
delay making the shared state ready untilschedule the shared state to be made ready when the calling thread exits. This associates a reference to the shared state with the calling thread. The destruction of each of that thread's objects with thread storage duration (3.7.2 [basic.stc.thread]) is sequenced before making that shared state ready. When the calling thread makes the shared state ready, if the thread holds the last reference to the shared state, the shared state is destroyed. [Note: This means that the shared state may not become ready until after the asynchronous provider has been destroyed. — end note]
Section: 99 [concurr.ts::futures.unique_future] Status: SG1 Submitter: Agustín K-ballo Bergé Opened: 2015-09-03 Last modified: 2017-03-05
Priority: Not Prioritized
View other active issues in [concurr.ts::futures.unique_future].
View all other issues in [concurr.ts::futures.unique_future].
View all issues with SG1 status.
Discussion:
Addresses: concurr.ts
In N4538, the continuation given to future::then can be run "on an unspecified thread of execution". This is too broad, as it allows the continuation to be run on the main thread, a UI thread, or any other thread. In comparison, functions given to async run "as if in a new thread of execution", while the Parallelism TS gives less guarantees by running "in either the invoking thread or in a thread implicitly created by the library to support parallel algorithm execution". The threads on which the continuation given to future::then can run should be similarly constrained.
[2017-03-01, Kona, SG1]
Agreement that this is a problem. Suggested addition to the issue is below. We have no immediate delivery vehicle for a fix at the moment, but we would like to make the intended direction clear.
There is SG1 consensus that .then continuations should, by default, and in the absence of executors, be run only in the following ways:If the future is not ready when .then() is called, the .then argument may be run on the execution agent that fulfills the promise.
In all cases, the .then argument may be run on an implementation-provided thread, i.e. a thread that is neither the main thread nor explicitly created by the user.
In the absence of an executor argument (which currently cannot be supplied), running of the .then() continuation will not block the thread calling .then(), even if the future is ready at the time.
Straw polls: SF | F | N | A | SA For the default behaviour: "1. Run on completed task or new execution agent" 0 | 7 | 5 | 1 | 0 "2. Run on completed task or .then caller" 0 | 0 | 5 | 5 | 3 "3. Leave as implementation defined" 1 | 2 | 4 | 3 | 3 "4. Always new execution agent" 2 | 3 | 6 | 2 | 0 The actual conclusion was to allow either (1) or (4) for now, since they are quite close, but present a very different programming mode from (2).Proposed resolution:
Section: 99 [parallel.ts::parallel.alg.overloads] Status: New Submitter: Tim Song Opened: 2015-09-26 Last modified: 2017-02-03
Priority: 1
View all issues with New status.
Discussion:
Addresses: parallel.ts
99 [parallel.alg.overloads] provides parallel algorithm overloads for many algorithms in the standard library, but I can't find any normative wording specifying which headers these new overloads live in. Presumably, if the original algorithm is in <meow>, the new overloads should be in <experimental/meow>.
Proposed resolution:
Section: 28.13 [re.grammar] Status: New Submitter: Hubert Tong Opened: 2015-10-08 Last modified: 2017-02-03
Priority: 4
View all other issues in [re.grammar].
View all issues with New status.
Discussion:
In 28.13 [re.grammar] paragraph 2:
basic_regex member functions shall not call any locale dependent C or C++ API, including the formatted string input functions. Instead they shall call the appropriate traits member function to achieve the required effect.
Yet, the required interface for a regular expression traits class (28.3 [re.req]) does not appear to have any reliable method for determining whether a character as encoded for the locale associated with the traits instance is the same as a character represented by a UnicodeEscapeSequence, e.g., assuming a sane ru_RU.koi8r locale:
#include <stdio.h>
#include <stdlib.h>
#include <regex>
const char data[] = "\xB3";
const char matchCyrillicCaptialLetterYo[] = R"(\u0401)";
int main(void)
{
try {
std::regex myRegex;
myRegex.imbue(std::locale("ru_RU.koi8r"));
myRegex.assign(matchCyrillicCaptialLetterYo, std::regex_constants::ECMAScript);
printf("(%s)\n", std::regex_replace(std::string(data), myRegex, std::string("E")).c_str());
myRegex.assign("[[:alpha:]]", std::regex_constants::ECMAScript);
printf("(%s)\n", std::regex_replace(std::string(data), myRegex, std::string("E")).c_str());
} catch (std::regex_error& e) {
abort();
}
return 0;
}
The implementation I tried prints:
(Ё) (E)
Which means that the character class matching worked, but not the matching to the UnicodeEscapeSequence.
Proposed resolution:
Section: 20.14.7 [comparisons], 23.2.1 [container.requirements.general], 30.3.2.1 [thread.thread.id] Status: New Submitter: Matt Austern Opened: 2015-10-08 Last modified: 2017-02-03
Priority: 3
View other active issues in [comparisons].
View all other issues in [comparisons].
View all issues with New status.
Discussion:
A number of places in the library, including 20.14.7 [comparisons]/14, the Optional container requirements in 23.2.1 [container.requirements.general], and 30.3.2.1 [thread.thread.id]/8, use the phrase "total order". Unfortunately, that phrase is ambiguous. In mathematics, the most common definition is that a relation ≤ is a total order if it's total, transitive, and antisymmetric in the sense that x≤y ∧ y≤x ⇒ x=y. What we really want is a strict total order: a relation < is a strict total order if it's total, transitive, and antisymmetric in the sense that exactly one of x<y, y<x, and x=y holds.
The non-normative note in 25.7 [alg.sorting]/4 correctly uses the phrase "strict total ordering" rather than simply "total ordering".
We could address this issue by replacing "total order" with "strict total order" everywhere it appears, since I think there are no cases where we actually want a non-strict total order, or we could add something in Clause 17 saying that we always mean strict total order whenever we say total order.
Proposed resolution:
Section: 99 [fund.ts.v2::optional.object.swap], 99 [fund.ts.v2::propagate_const.modifiers] Status: New Submitter: Daniel Krügler Opened: 2015-11-14 Last modified: 2017-02-03
Priority: 3
View all issues with New status.
Discussion:
Addresses: fund.ts.v2
As pointed out in N4511, the Library fundamentals are affected by a similar problem as described in LWG 2456. First, it is caused by optional's member swap (99 [optional.object.swap]):
void swap(optional<T>& rhs) noexcept(see below);
with
The expression inside noexcept is equivalent to:
is_nothrow_move_constructible_v<T> && noexcept(swap(declval<T&>(), declval<T&>()))
Again, the unqualified lookup for swap finds the member swap instead of the result of a normal argument-depending lookup, making this ill-formed.
A second example of such a problem recently entered the arena with the addition of the propagate_const template with another member swap (99 [propagate_const.modifiers]):constexpr void swap(propagate_const& pt) noexcept(see below);-2- The constant-expression in the exception-specification is noexcept(swap(t_, pt.t_)).
A working approach is presented in N4511. By adding a new trait to the standard library and referencing this by the library fundamentals (A similar approach had been applied in the file system specification where the quoted manipulator from C++14 had been referred to, albeit the file system specification is generally based on the C++11 standard), optional's member swap exception specification could be rephrased as follows:
The expression inside noexcept is equivalent to:
is_nothrow_move_constructible_v<T> && is_nothrow_swappable_v<T>noexcept(swap(declval<T&>(), declval<T&>()))
and propagate_const's member swap exception specification could be rephrased as follows:
constexpr void swap(propagate_const& pt) noexcept(see below);-2- The constant-expression in the exception-specification is is_nothrow_swappable_v<T>
noexcept(swap(t_, pt.t_)).
[2016-02-20, Ville comments]
Feedback from an implementation:
libstdc++ already applies the proposed resolution for propagate_const, but not for optional.[2016-02-20, Daniel comments]
A recent paper update has been provided: P0185R0.
[2016-03, Jacksonville]
Add a link to 2456[2016-11-08, Issaquah]
Not adopted during NB comment resolution
Proposed resolution:
Section: 17.5.5.5 [member.functions] Status: New Submitter: Ville Voutilainen Opened: 2015-11-29 Last modified: 2017-02-03
Priority: 2
View other active issues in [member.functions].
View all other issues in [member.functions].
View all issues with New status.
Discussion:
The combination of 17.5.5.5 [member.functions], paragraphs 2 and 3 that LWG 2259 does seems to drop a requirement that any call behaves as if no overloads were added. Paragraph 3 used to say "A call to a member function signature described in the C ++ standard library behaves as if the implementation declares no additional member function signatures." whereas the new wording says "provided that any call to the member function that would select an overload from the set of declarations described in this standard behaves as if that overload were selected."
This can be read as meaning that if there's no default constructor specified, like for instance for std::ostream, an implementation is free to add it. It can also be read as meaning that an implementation is free to add any overloads that wouldn't change the overload resolution result of any call expression that would select a specified overload. That's vastly different from allowing extensions that add new functions rather than new overloads. Was this relaxation intentional?[2016-04, Issues Telecon]
Ville provides a motivating example.
#include <iostream>
class my_stream : std::ostream
{
};
int main()
{
my_stream ms;
}
This example is accepted by libstdc++, msvc rejects it, and clang+libc++ segfault on melpon.org/wandbox o_O. An earlier clang+libc++ just accepts it. I don't think the implementation divergence is caused by the acceptance of the referred-to 2259, but it certainly seems to increasingly bless the implementation divergence.
[2016-05 Issues Telecon]
This is related to issue 2695.
Proposed resolution:
Section: 99 [fund.ts.v2::func.wrap.func] Status: New Submitter: Tim Song Opened: 2015-12-05 Last modified: 2017-02-03
Priority: 3
View all other issues in [fund.ts.v2::func.wrap.func].
View all issues with New status.
Discussion:
Addresses: fund.ts.v2
[This is essentially LWG 2370, but deals with the fundamentals TS version rather than the one in the standard]
In 99 [func.wrap.func] of library fundamentals TS, the constructorstemplate<class A> function(allocator_arg_t, const A&) noexcept; template<class A> function(allocator_arg_t, const A&, nullptr_t) noexcept;
must type-erase and store the provided allocator, since the operator= specification requires using the "allocator specified in the construction of" the std::experimental::function object. This may require a dynamic allocation and so cannot be noexcept. Similarly, the following constructors
template<class A> function(allocator_arg_t, const A&, const function&); template<class A> function(allocator_arg_t, const A&, function&&); template<class F, class A> function(allocator_arg_t, const A&, F);
cannot satisfy the C++14 requirement that they "shall not throw exceptions if [the function object to be stored] is a callable object passed via reference_wrapper or a function pointer" if they need to type-erase and store the allocator.
[2016-11-08, Issaquah]
Not adopted during NB comment resolution
Proposed resolution:
This wording is relative to N4562.
Edit 99 [func.wrap.func], class template function synopsis, as follows:
namespace std {
namespace experimental {
inline namespace fundamentals_v2 {
[…]
template<class R, class... ArgTypes>
class function<R(ArgTypes...)> {
public:
[…]
template<class A> function(allocator_arg_t, const A&) noexcept;
template<class A> function(allocator_arg_t, const A&,
nullptr_t) noexcept;
[…]
};
[…]
} // namespace fundamentals_v2
} // namespace experimental
[…]
} // namespace std
Insert the following paragraphs after 99 [func.wrap.func.con]/1:
[Drafting note: This just reproduces the wording from C++14 with the "shall not throw exceptions for reference_wrapper/function pointer" provision deleted. — end drafting note]
-1- When a function constructor that takes a first argument of type allocator_arg_t is invoked, the second argument is treated as a type-erased allocator (8.3). If the constructor moves or makes a copy of a function object (C++14 §20.9), including an instance of the experimental::function class template, then that move or copy is performed by using-allocator construction with allocator get_memory_resource().
template <class A> function(allocator_arg_t, const A& a); template <class A> function(allocator_arg_t, const A& a, nullptr_t);-?- Postconditions: !*this.
template <class A> function(allocator_arg_t, const A& a, const function& f);-?- Postconditions: !*this if !f; otherwise, *this targets a copy of f.target().
-?- Throws: May throw bad_alloc or any exception thrown by the copy constructor of the stored callable object. [Note: Implementations are encouraged to avoid the use of dynamically allocated memory for small callable objects, for example, where f's target is an object holding only a pointer or reference to an object and a member function pointer. — end note]template <class A> function(allocator_arg_t, const A& a, function&& f);-?- Effects: If !f, *this has no target; otherwise, move-constructs the target of f into the target of *this, leaving f in a valid state with an unspecified value.
template <class F, class A> function(allocator_arg_t, const A& a, F f);-?- Requires: F shall be CopyConstructible.
-?- Remarks: This constructor shall not participate in overload resolution unless f is Callable (C++14 §20.9.11.2) for argument types ArgTypes... and return type R. -?- Postconditions: !*this if any of the following hold:-?- Otherwise, *this targets a copy of f initialized with std::move(f). [Note: Implementations are encouraged to avoid the use of dynamically allocated memory for small callable objects, for example, where f's target is an object holding only a pointer or reference to an object and a member function pointer. — end note] -?- Throws: May throw bad_alloc or any exception thrown by F's copy or move constructor.
f is a null function pointer value.
f is a null member pointer value.
F is an instance of the function class template, and !f.
-2- In the following descriptions, let ALLOCATOR_OF(f) be the allocator specified in the construction of function f, or allocator<char>() if no allocator was specified.
[…]
Section: 20.15.8 [meta.logical] Status: Open Submitter: Tim Song Opened: 2015-12-11 Last modified: 2017-02-03
Priority: 2
View other active issues in [meta.logical].
View all other issues in [meta.logical].
View all issues with Open status.
Discussion:
20.15.8 [meta.logical]/2 and /5 impose the following requirement on the arguments of conjunction and disjunction:
Every template type argument shall be usable as a base class and shall have a static data member value which is convertible to bool, is not hidden, and is unambiguously available in the type.
Since the requirement is unconditional, it applies even to type arguments whose instantiation is not required due to short circuiting. This seems contrary to the design intent, expressed in P0013R1, that it is valid to write conjunction_v<is_class<T>, is_foo<T>> even if instantiating is_foo<T>::value is ill-formed for non-class types.
[2016-08 Chicago]
Ville provided wording for both 2569 and 2570.
Tuesday AM: Move to Tentatively Ready
[2016-11-15, Reopen upon request of Dawn Perchik ]
The proposed wording requires an update, because the referenced issue LWG 2568 is still open.
Proposed resolution:
[We recommend applying the proposed resolution for LWG issues 2567 and 2568 before this proposed resolution, lest the poor editor gets confused.]
In [meta.logical],
- insert a new paragraph before paragraph 2:
The class template conjunction forms the logical conjunction of its template type arguments.
- move paragraph 4 before paragraph 2, and edit paragraph 2 as follows:
The class template conjunction forms the logical conjunction of its
template type arguments.
Every template type argument for which Bi::value is instantiated
shall be usable as a base class and shall have a member value which is
convertible to bool, is not hidden, and is unambiguously available in the type.
- insert a new paragraph before paragraph 5:
The class template disjunction forms the logical disjunction of its template type arguments.
- move paragraph 7 before paragraph 5, and edit paragraph 5 as follows:
The class template disjunction forms the logical disjunction
of its template type arguments.
Every template type argument for which Bi::value is instantiated
shall be usable as a base class and shall have a member value which is
convertible to bool, is not hidden, and is unambiguously available in the type.
Section: 20.15.8 [meta.logical] Status: Open Submitter: Tim Song Opened: 2016-01-18 Last modified: 2017-02-03
Priority: 3
View other active issues in [meta.logical].
View all other issues in [meta.logical].
View all issues with Open status.
Discussion:
The specification of conjunction and disjunction in 20.15.8 [meta.logical] p2 and p5 requires Bi::value to be convertible to bool, but nothing in the specification of the actual behavior of the templates, which instead uses the expressions Bi::value == false and Bi::value != false instead, actually requires this conversion.
If the intention of this requirement is to allow implementations to pass Bi::value directly to std::conditional, like the sample implementation in P0013R1:
template<class B1, class B2>
struct and_<B1, B2> : conditional_t<B1::value, B2, B1> { };
then it's insufficient in at least two ways:
Nothing in the specification requires the result of comparing Bi::value with false to be consistent with the result of the implicit conversion. This is similar to LWG 2114, though I don't think the BooleanTestable requirements in that issue's P/R covers Bi::value == false and Bi::value != false.
More importantly, the above implementation is ill-formed for, e.g., std::conjunction<std::integral_constant<int, 2>, std::integral_constant<int, 4>>, because converting 2 to bool is a narrowing conversion that is not allowed for non-type template arguments (see 5.20 [expr.const]/4). (Note that GCC currently doesn't diagnose this error at all, and Clang doesn't diagnose it inside system headers.) It's not clear whether such constructs are intended to be supported, but if they are not, the current wording doesn't prohibit it.
[2016-08-03 Chicago LWG]
Walter, Nevin, and Jason provide initial Proposed Resolution.
Previous resolution [SUPERSEDED]:
This wording is relative to N4606.
Change 20.15.8 [meta.logical] as indicated:
template<class... B> struct conjunction : see below { };[…]
-3- The BaseCharacteristic of a specialization conjunction<B1, ..., BN> is the first type Bi in the list true_type, B1, ..., BN for whichBi::value == false! bool(Bi::value), or if everyBi::value != falsebool(Bi::value), the BaseCharacteristic is the last type in the list. […] -4- For a specialization conjunction<B1, ..., BN>, if there is a template type argument Bi withBi::value == false! bool(Bi::value), then instantiating […]template<class... B> struct disjunction : see below { };[…]
-6- The BaseCharacteristic of a specialization disjunction<B1, ..., BN> is the first type Bi in the list false_type, B1, ..., BN for whichBi::value != falsebool(Bi::value), or if everyBi::value == false! bool(Bi::value), the BaseCharacteristic is the last type in the list. […] -7- For a specialization disjunction<B1, ..., BN>, if there is a template type argument Bi withBi::value != falsebool(Bi::value), then instantiating […]template<class B> struct negation : bool_constant<!bool(B::value)> { };-8- The class template negation forms the logical negation of its template type argument. The type negation<B> is a UnaryTypeTrait with a BaseCharacteristic of bool_constant<!bool(B::value)>.
Proposed resolution:
The resolution for this issue was combined with the resolution for LWG 2567, so 2567 resolves this issue here as well.
Section: 20.17.2 [time.syn] Status: New Submitter: Andy Giese Opened: 2016-02-05 Last modified: 2017-02-03
Priority: 4
View all other issues in [time.syn].
View all issues with New status.
Discussion:
Currently 20.17.2 [time.syn] states
// convenience typedefs typedef duration<signed integer type of at least 64 bits, nano> nanoseconds; typedef duration<signed integer type of at least 55 bits, micro> microseconds; typedef duration<signed integer type of at least 45 bits, milli> milliseconds; typedef duration<signed integer type of at least 35 bits > seconds; typedef duration<signed integer type of at least 29 bits, ratio< 60>> minutes; typedef duration<signed integer type of at least 23 bits, ratio<3600>> hours;
However, a duration_cast<minutes>(seconds::max()) would cause overflow if the underlying signed integers only met the minimums specified.
The standard should specify that implementations guarantee that a duration_cast from any smaller duration in these "convenience typedefs" will not overflow any larger duration. That is, hours should be able to hold the maximum of minutes, which should be able to hold the maximum of seconds and so on. More formally, if the ratio typedef A and typedef B is 1:Y where Y > 1 (e.g., 1 : 60 in case of minutes : seconds), then #bitsA-1 must be at least ceil(log2(2#bitsB-1)/Y)). In the case of minutes : seconds, X = 1, Y = 60. Let #bitsseconds = 32. Therefore:2(#bitsseconds - 1) = 231 = 2147483648
ceil(log2(231 / 60) = 26
#bitsminutes - 1 = 26
#bitsminutes = 27
Therefore, a minimum of 27 bits would be needed to store minutes if 32 were used to store seconds.
I propose to change the definitions of the convenience typedefs as follows:// convenience typedefs typedef duration<signed integer type of at least 64 bits, nano> nanoseconds; typedef duration<signed integer type of at least 55 bits, micro> microseconds; typedef duration<signed integer type of at least 46 bits, milli> milliseconds; typedef duration<signed integer type of at least 37 bits > seconds; typedef duration<signed integer type of at least 32 bits, ratio< 60>> minutes; typedef duration<signed integer type of at least 27 bits, ratio<3600>> hours;
These bits were chosen to satisfy the above formula. Note that minimums only increased, so larger ranges could be held. A nice outcome of this choice is that minutes does not go above 32 bits.
[2016-04-23, Tim Song comments]
The P/R of LWG 2592 doesn't fix the issue it wants to solve, because the actual underlying type will likely have more bits than the specified minimum.
Consider seconds, which the P/R requires to have at least 37 bits. On a typical system this implies using a 64-bit integer. To ensure that casting from seconds::max() to minutes doesn't overflow in such a system, it is necessary for the latter to have at least 59 bits (which means, in practice, 64 bits too), not just 32 bits. Thus, just changing the minimum number of bits will not be able to provide the desired guarantee that casting from a smaller unit to a larger one never overflow. If such a guarantee is to be provided, it needs to be spelled out directly. Note that the difference here is 9 bits (for the 1000-fold case) and 5 bits (for the 60-fold case), which is less than the size difference between integer types on common systems, so such a requirement would effectively require those convenience typedefs to use the same underlying integer type.Proposed resolution:
This wording is relative to N4567.
Change 20.17.2 [time.syn], header <chrono> synopsis, as indicated
[…] // convenience typedefs typedef duration<signed integer type of at least 64 bits, nano> nanoseconds; typedef duration<signed integer type of at least 55 bits, micro> microseconds; typedef duration<signed integer type of at least 4645bits, milli> milliseconds; typedef duration<signed integer type of at least 3735bits > seconds; typedef duration<signed integer type of at least 3229bits, ratio< 60>> minutes; typedef duration<signed integer type of at least 2723bits, ratio<3600>> hours; […]
Section: 17.5.3.5 [allocator.requirements] Status: Tentatively Ready Submitter: David Krauss Opened: 2016-02-19 Last modified: 2017-06-15
Priority: 4
View other active issues in [allocator.requirements].
View all other issues in [allocator.requirements].
View all issues with Tentatively Ready status.
Discussion:
17.5.3.5 [allocator.requirements] suggests that the moved-from state of an allocator may be unequal to its previous state. Such a move constructor would break most container implementations, which move-construct the embedded allocator along with a compressed pair. Even if a moved-from container is empty, it should still subsequently allocate from the same resource pool as it did before.
std::vector<int, pool> a(500, my_pool); auto b = std::move(a); // b uses my_pool too. a.resize(500); // should still use my_pool.
[2016-02, Jacksonville]
Marshall will see if this can be resolved editorially.
After discussion, the editors and I decided that this could not be handled editorially. The bit about a moved-from state of an allocator being the same as the original state is a normative change. I submitted a pull request to handle the mismatched variables in the table.
Previous resolution [SUPERSEDED]:
This wording is relative to N4567.
Change 17.5.3.5 [allocator.requirements], Table 28 — "Allocator requirements" as indicated:
Note there's an editorial error in Table 28 in that line and the surrounding ones. The left column was apparently updated to use u and the right column is still using a/a1/b.
Table 28 — Allocator requirements Expression Return type Assertion/note
pre-/post-conditionDefault … X u(move(a));
X u = move(a);Shall not exit via an exception. post: u is equal to a and equal to the prior value of a a1 equals the prior value of a.…
[2016-06-20, Oulu, Daniel comments]
According to the current working draft, the situation has changed due to changes performed by the project editor, the revised resolution has been adjusted to N4594.
[2016-08 - Chicago]
Thurs AM: Moved to LEWG, as this decision (should allocators only be copyable, not movable) is design.
[2017-02 in Kona, LEWG responds]
Alisdair Meredith says that if you have a do-not-propagate-on-move-assignment, then the move of the allocator must compare equal to the original.
Have a data structure where all allocators are equal. Construct an element somewhere else moving from inside the container (which doesn't have a POCMA trait); you don't want the allocator of the moved-from element to now be different. So in that case, the allocator's move constructor must behave the same as copy.
We don't need to go as far as this issue, but going that far is ok for Bloomberg.
[2017-06-02 Issues Telecon]
We discussed containers that have sentinel nodes, etc, and so might have to allocate/deallocate using a moved-from allocator - and decided that we didn't want any part of that
Adjusted the wording slightly, and moved to Tentatively Ready
Previous resolution [SUPERSEDED]:
This wording is relative to N4594.
Change 17.5.3.5 [allocator.requirements], Table 28 — "Allocator requirements" as indicated:
Table 28 — Allocator requirements Expression Return type Assertion/note
pre-/post-conditionDefault … X u(std::move(a));
X u = std::move(a);Shall not exit via an exception. post: u is equal to a and equal to the prior value of a u is equal to the prior value of a..…
Proposed resolution:
This wording is relative to N4594.
Change 17.5.3.5 [allocator.requirements], Table 28 — "Allocator requirements" as indicated:
Table 28 — Allocator requirements Expression Return type Assertion/note
pre-/post-conditionDefault … X u(std::move(a));
X u = std::move(a);Shall not exit via an exception. post: The value of a is unchanged and is equal to u u is equal to the prior value of a.…
Section: 20.11.2.2 [util.smartptr.shared] Status: New Submitter: Kazutoshi Satoda Opened: 2016-02-20 Last modified: 2017-02-03
Priority: 3
View other active issues in [util.smartptr.shared].
View all other issues in [util.smartptr.shared].
View all issues with New status.
Discussion:
Latest draft (N4567) 20.11.2.2 [util.smartptr.shared] p1 says:
A shared_ptr object is empty if it does not own a pointer.
Please note that it says "own a pointer". This definition was added as the resolution for LWG defect 813.
20.11.2.2.1 [util.smartptr.shared.const] p8 says about the effect of shared_ptr(nullptr_t p, D d):Effects: Constructs a shared_ptr object that owns the object p and the deleter d.
Please note that it says "owns the object". This was intentionally changed from "the pointer" as a part of resolution for LWG defect 758, to cover nullptr_t case.
Since shared_ptr(nullptr, d) owns an object of type nullptr_t, but does not own a pointer, it is said as "empty" by a strict reading of the above mentioned definition in 20.11.2.2 [util.smartptr.shared] p1. These cause a contradiction: 20.11.2.2.1 [util.smartptr.shared.const] p9 sets a postcondition use_count() == 1 on shared_ptr(nullptr, d). But 20.11.2.2.5 [util.smartptr.shared.obs] p7 says that the return value of use_count() is "0 when *this is empty".Proposed wording changes:
Replace the last 2 words in 20.11.2.2 [util.smartptr.shared] p1 from[…] empty if it does not own a pointer.
to
[…] empty if it does not own an object.
Note that shared_ptr(nullptr_t) is defined to be empty in synopsis in 20.11.2.2 [util.smartptr.shared].
constexpr shared_ptr(nullptr_t) noexcept : shared_ptr() { }
It could be less confusing if shared_ptr(nullptr, d) could be defined to be empty. But it seems too late to change that (which means changing whether the deleter is called or not, see this Stackoverflow article). Then I'm proposing just fix the contradiction.
Proposed resolution:
This wording is relative to N4594.
Change 20.11.2.2 [util.smartptr.shared] p1 as indicated:
-1- The shared_ptr class template stores a pointer, usually obtained via new. shared_ptr implements semantics of shared ownership; the last remaining owner of the pointer is responsible for destroying the object, or otherwise releasing the resources associated with the stored pointer. A shared_ptr object is empty if it does not own an object
a pointer.
Section: 24.5.1.1 [reverse.iterator], 24.5.1.3.12 [reverse.iter.opindex] Status: New Submitter: Robert Haberlach Opened: 2016-02-28 Last modified: 2017-02-03
Priority: 3
View all other issues in [reverse.iterator].
View all issues with New status.
Discussion:
Issue 386 changed the return type of reverse_iterator::operator[] to unspecified. However, as of N3066, the return type of a random access iterator's operator[] shall be convertible to reference; thus the return type of reverse_iterator::operator[] should be reference (and it is in all common implementations).
Suggested resolution: Adjust 24.5.1.1 [reverse.iterator]'s synopsis and 24.5.1.3.12 [reverse.iter.opindex] to use reference instead of unspecified.Proposed resolution:
This wording is relative to N4582.
Edit 24.5.1.1 [reverse.iterator], class template synopsis, as indicated:
namespace std {
template <class Iterator>
class reverse_iterator {
public:
[…]
typedef typename iterator_traits<Iterator>::reference reference;
[…]
constexpr referenceunspecified operator[](difference_type n) const;
[…]
};
}
Change 24.5.1.3.12 [reverse.iter.opindex] before p1 as indicated:
constexpr referenceunspecifiedoperator[]( typename reverse_iterator<Iterator>::difference_type n) const;
Section: 26.5.8 [complex.transcendentals] Status: Tentatively Ready Submitter: Thomas Koeppe Opened: 2016-03-01 Last modified: 2017-03-13
Priority: 3
View all other issues in [complex.transcendentals].
View all issues with Tentatively Ready status.
Discussion:
The current specification of std::log is inconsistent for complex numbers, specifically, the Returns clause (26.5.8 [complex.transcendentals]). On the one hand, it states that the imaginary part of the return value lies in the closed interval [-i π, +i π]. On the other hand, it says that "the branch cuts are along the negative real axis" and "the imaginary part of log(x) is +π when x is a negative real number".
The inconsistency lies in the difference between the mathematical concept of a branch cut and the nature of floating point numbers in C++. The corresponding specification in the C standard makes it clearer that if x is a real number, then log(x + 0i) = +π, but log(x - 0i) = -π, i.e. they consider positive and negative zero to represent the two different limits of approaching the branch cut from opposite directions. In other words, the term "negative real number" is misleading, and in fact there are two distinct real numbers, x + 0i and x - 0i, that compare equal but whose logarithms differ by 2 π i.
The resolution should consist of two parts:
Double-check that our usage and definition of "branch cut" is sufficiently unambiguous. The C standard contains a lot more wording around this that we don't have in C++.
Change the Returns clause of log appropriately. For example: "When x is a negative real number, imag(log(x + 0i)) is π, and imag(log(x - 0i)) is -π."
Current implementations seem to behave as described in (2). (Try-it-at-home link)
[2016-11-12, Issaquah]
Move to Open - Thomas to provide wording
[2016-11-15, Thomas comments and provides wording]
Following LWG discussion in Issaquah, I now propose to resolve this issue by removing the normative requirement on the function limits, and instead adding a note that the intention is to match the behaviour of C. This allows implementations to use the behaviour of C without having to specify what floating point numbers really are.
The change applies to both std::log and std::sqrt. Updated try-at-home link, see here.[2017-03-04, Kona]
Minor wording update and status to Tentatively Ready.
Previous resolution [SUPERSEDED]:
This wording is relative to N4606.
Change the "returns" element for std::log (26.5.8 [complex.transcendentals] p17):
template<class T> complex<T> log(const complex<T>& x);-16- Remarks: The branch cuts are along the negative real axis.
-17- Returns: The complex natural (base-ℯ) logarithm of x. For all x, imag(log(x)) lies in the interval [-π, π], and when x is a negative real number, imag(log(x)) is π. [Note: The semantics of std::log are intended to be the same in C++ as they are for clog in C. — end note]Change the "returns" element for std::sqrt (26.5.8 [complex.transcendentals] p25):
template<class T> complex<T> sqrt(const complex<T>& x);-24- Remarks: The branch cuts are along the negative real axis.
-25- Returns: The complex square root of x, in the range of the right half-plane.If the argument is a negative real number, the value returned lies on the positive imaginary axis.[Note: The semantics of std::sqrt are intended to be the same in C++ as they are for csqrt in C. — end note]
Proposed resolution:
This wording is relative to N4606.
Change the "returns" element for std::log (26.5.8 [complex.transcendentals] p17):
template<class T> complex<T> log(const complex<T>& x);-16- Remarks: The branch cuts are along the negative real axis.
-17- Returns: The complex natural (base-ℯ) logarithm of x. For all x, imag(log(x)) lies in the interval [-π, π], and when x is a negative real number, imag(log(x)) is π. [Note: the semantics of this function are intended to be the same in C++ as they are for clog in C. — end note]
Change the "returns" element for std::sqrt (26.5.8 [complex.transcendentals] p25):
template<class T> complex<T> sqrt(const complex<T>& x);-24- Remarks: The branch cuts are along the negative real axis.
-25- Returns: The complex square root of x, in the range of the right half-plane.If the argument is a negative real number, the value returned lies on the positive imaginary axis.[Note: The semantics of this function are intended to be the same in C++ as they are for csqrt in C. — end note]
Section: 20.2.7 [declval], 20.11.1 [unique.ptr], 20.11.1.1.1 [unique.ptr.dltr.general], 20.11.2.2 [util.smartptr.shared], 20.11.2.3 [util.smartptr.weak], 20.11.2.5 [util.smartptr.enab] Status: New Submitter: Zhihao Yuan Opened: 2016-03-08 Last modified: 2017-02-03
Priority: 3
View all issues with New status.
Discussion:
Currently the phrase to grant this permission is:
The template parameter T of LibraryTemplate may be an incomplete type.
Two problems:
The timing is unclear. We always allow specializations like LibraryTemplate<Incomp>* p;
To the users of a template, the correct terminology should be "argument" rather than "parameter".
Suggested resolution:
In an instantiation of LibraryTemplate, an incomplete type may be used as the template argument for the template parameter T.
as shown here.
Or, to copy N4510's wording:An incomplete type T may be used when instantiating LibraryTemplate.
Proposed resolution:
Section: 27.5.3.5 [ios.base.storage] Status: LEWG Submitter: David Krauss Opened: 2016-03-14 Last modified: 2017-02-03
Priority: Not Prioritized
View all other issues in [ios.base.storage].
View all issues with LEWG status.
Discussion:
DR 41, "Ios_base needs clear(), exceptions()" stopped short of providing the interface suggested in its title, but it did require the underlying state to be stored in ios_base. Because rdstate() is also missing, ios_base manipulators relying on iword and pword cannot detect failure. The only safe alternative is to manipulate a derived class, which must be a template.
libc++ already provides the interface as a nonconforming extension. libstdc++ implements the internal state but leaves it frustratingly inaccessible, as specified. Any conforming implementation should be able to provide the interface without ABI problems.[2016-04, Issues Telecon]
This is really a request for an (feature) API. Passing to LEWG.
Proposed resolution:
remove_filename() post condition is incorrectSection: 99 [fs.path.modifiers] Status: Open Submitter: Eric Fiselier Opened: 2014-06-07 Last modified: 2017-03-19
Priority: 1
View all issues with Open status.
Discussion:
remove_filename() specifies !has_filename() as the post condition.
This post condition is not correct. For example the path "/foo"
has a filename of "foo". If we remove the filename we get "/",
and "/" has a filename of "/".
[2014-06-08 Beman supplies an Effects: element.]
[2014-06-17 Rapperswil LWG will investigate issue at a subsequent meeting.]
[2016-04 Issue updated to address the C++ Working Paper. Previously addressed File System TS]
[2016-04, Issues Telecon]
There was concern that the effects wording is not right. Jonathan provided updated wording.
Previous resolution [SUPERSEDED]:
path& remove_filename();
Postcondition:!has_filename().Effects:
*this = parent_path(), except that ifparent_path() == root_path(),clear().Returns:
*this.[Example:
std::cout << path("/foo").remove_filename();// outputs "/" std::cout << path("/").remove_filename(); // outputs ""—end example]
[2016-08 Chicago]
Wed AM: Move to Tentatively Ready
Previous resolution [SUPERSEDED]:
path& remove_filename();
Postcondition: !has_filename().Effects: If
*this == root_path(), thenclear(). Otherwise,*this = parent_path().Returns:
*this.[Example:
std::cout << path("/foo").remove_filename();// outputs "/" std::cout << path("/").remove_filename(); // outputs ""
—end example]
[2016-10-16, Eric reopens and provides improved wording]
The suggested PR is incorrect. root_path() removes redundant directory separators.
Therefore the condition *this == root_path() will fail for "//foo///" because root_path() returns "//foo/". However using path::compare instead would solve this problem.[2016-11-21, Beman comments]
This issue is closely related to CD NB comments US 25, US 37, US 51, US 52, US 53, US 54, and US 60. The Filesystem SG in Issaquah recommended that these all be handled together, as they all revolve around the exact meaning of "filename" and path decomposition.
Proposed resolution:
This wording is relative to N4606.
Modify 99 [fs.path.modifiers] as indicated:
path& remove_filename();-5-
-6- Returns: *this. -7- [Example:PostconditionEffects:!has_filename()If this->compare(root_path()) == 0, then clear(). Otherwise, *this = parent_path()..std::cout << path("/foo").remove_filename(); // outputs "/" std::cout << path("/").remove_filename(); // outputs ""— end example]
Section: 27.5.3.6 [ios.base.callback] Status: New Submitter: David Krauss Opened: 2016-03-14 Last modified: 2017-02-03
Priority: 3
View all issues with New status.
Discussion:
register_callback allocates memory and so it can fail, but the case is unspecified. libc++ sets badbit, which is consistent with iword and pword. libstdc++ throws std::bad_alloc.
Proposed resolution:
filesystem::copy() won't create a symlink to a directorySection: 27.10.15.3 [fs.op.copy] Status: Open Submitter: Jonathan Wakely Opened: 2016-04-19 Last modified: 2017-02-03
Priority: 2
View all other issues in [fs.op.copy].
View all issues with Open status.
Discussion:
(First raised in c++std-lib-38544)
filesystem::copy doesn't create a symlink to a directory in this case:
copy("/", "root", copy_options::create_symlinks);
If the first path is a file then a symlink is created, but I think my
implementation is correct to do nothing for a directory. We get to
bullet 27.10.15.3 [fs.op.copy] (3.6) where is_directory(f) is true, but options
== create_symlinks, so we go to the next bullet (3.7) which says
"Otherwise, no effects."
create_symlink to create a symlink to a directory.
[2016-05 Issues Telecon]
This is related to 2681; and should be considered together.
[2016-08 Chicago]
Wed AM: Move to Tentatively Ready
Previous resolution [SUPERSEDED]:
Add a new bullet following (3.6) in 27.10.15.3 [fs.op.copy] as shown:
If
!exists(t), thencreate_directory(to, from).Then, iterate over the files in
from, as if byfor (directory_entry& x : directory_iterator(from)), and for each iterationcopy(x.path(), to/x.path().filename(), options | copy_options::unspecified )Otherwise, if
is_directory(f) && (options & copy_options::create_symlinks) != copy_options::none, then report an error with anerror_codeargument equal tomake_error_code(errc::is_a_directory).Otherwise, no effects.
[2016-10-16, Eric reopens and provides improved wording]
The current PR makes using copy(...) to copy/create a directory symlink an error. For example, the following is now an error:
copy("/", "root", copy_options::create_symlinks);
However the current PR doesn't handle the case where both copy_options::create_symlinks and copy_options::recursive are specified. This case is still incorrectly handled by bullet (3.6) [fs.op.copy].
I suggest we move the PR before this bullet so that it catches the recursive copy case, since currently the conditions are ordered:3.6 Otherwise if is_directory(f) && (bool(options & copy_options::recursive) || ...)
3.X Otherwise if is_directory(f) && bool(options & copy_options::create_symlinks)
So 3.6 catches create_symlinks | recursive but I believe we want 3.X to handle it instead.
Proposed resolution:
This wording is relative to N4606.
Add a new bullet before (3.6) in 27.10.15.3 [fs.op.copy] as shown:
(3.5) — Otherwise, if is_regular_file(f), then:
[…](3.?) — Otherwise, if
is_directory(f) && (options & copy_options::create_symlinks) != copy_options::nonethen report an error with an
error_codeargument equal tomake_error_code(errc::is_a_directory).(3.6) — Otherwise, if
is_directory(f) && ((options & copy_options::recursive) != copy_options::none || options == copy_options::none)then:
(3.6.1) — If
!exists(t), thencreate_directory(to, from).(3.6.2) — Then, iterate over the files in
from, as if byfor (directory_entry& x : directory_iterator(from)), and for each iterationcopy(x.path(), to/x.path().filename(), options | copy_options::unspecified)(3.7) — Otherwise, for the signature with argument ec, ec.clear().
(3.8) — Otherwise, no effects.
Section: 20.14.4 [func.invoke] Status: LEWG Submitter: Zhihao Yuan Opened: 2016-03-25 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [func.invoke].
View all other issues in [func.invoke].
View all issues with LEWG status.
Discussion:
In N4169 the author dropped the invoke<R> support by claiming that it's an unnecessary cruft in TR1, obsoleted by C++11 type inference. But now we have some new business went to *INVOKE*(f, t1, t2, ..., tN, R), that is to discard the return type when R is void. This form is very useful, or possible even more useful than the basic form when implementing a call wrapper. Also note that the optional R support is already in std::is_callable and std::is_nothrow_callable.
[2016-07-31, Tomasz Kamiński comments]
The lack of invoke<R> was basically a result of the concurrent publication of the never revision of the paper and additional special semantics of INVOKE(f, args..., void).
In contrast to existing std::invoke function, the proposed invoke<R> version is not SFINAE friendly, as elimination of the standard version of invoke is guaranteed by std::result_of_t in the result type that is missing for proposed invoke<R> version. To provide this guarantee, following remarks shall be added to the specification:Remarks: This function shall not participate in overload resolution unless is_callable_v<F(Args...), R> is true.
[2016-08-01, Tomasz Kamiński and Zhihao Yuan update the proposed wording]
Previous resolution [SUPERSEDED]:
This wording is relative to N4606.
Modify 20.14 [function.objects]/2, header <functional> synopsis, as indicated:
namespace std { // 20.12.3, invoke: template <class F, class... Args> result_of_t<F&&(Args&&...)> invoke(F&& f, Args&&... args); template <class R, class F, class... Args> R invoke(F&& f, Args&&... args);Add the following sequence of paragraphs after 20.14.4 [func.invoke]/1 as indicated:
template <class R, class F, class... Args> R invoke(F&& f, Args&&... args);-?- Returns: INVOKE(std::forward<F>(f), std::forward<Args>(args)..., R) (20.14.3 [func.require]).
-?- Remarks: This function shall not participate in overload resolution unless is_callable_v<F(Args...), R> is true.
[2016-09-04, Tomasz Kamiński comments and improves wording]
The usage of is_callable_v<F(Args...), R> causes problem in situation when either F or Args is an abstract type and the function type F(Args...) cannot be formed or when one of the args is cv-qualified, as top-level cv-qualification for function parameters is dropped by language rules. It should use is_callable_v<F&&(Args&&...), R> instead.
Proposed resolution:
This wording is relative to N4606.
Modify 20.14 [function.objects]/2, header <functional> synopsis, as indicated:
namespace std {
// 20.12.3, invoke:
template <class F, class... Args> result_of_t<F&&(Args&&...)> invoke(F&& f, Args&&... args);
template <class R, class F, class... Args> R invoke(F&& f, Args&&... args);
Add the following sequence of paragraphs after 20.14.4 [func.invoke]/1 as indicated:
template <class R, class F, class... Args> R invoke(F&& f, Args&&... args);-?- Returns: INVOKE(std::forward<F>(f), std::forward<Args>(args)..., R) (20.14.3 [func.require]).
-?- Remarks: This function shall not participate in overload resolution unless is_callable_v<F&&(Args&&...), R> is true.
Section: 22.4.6.3 [locale.moneypunct] Status: New Submitter: Hubert Tong Opened: 2016-04-12 Last modified: 2017-02-03
Priority: 3
View all other issues in [locale.moneypunct].
View all issues with New status.
Discussion:
The description of money_base::space is that "at least one space is required at that position." (N4582 subclause 22.4.6.3 [locale.moneypunct] paragraph 2)
When formatting for output (22.4.6.2.2 [locale.money.put.virtuals]), it is not clear that"the number of characters generated for the specified format" (excluding fill padding) includes exactly one character for money_base::space (if present), and
all characters corresponding to money_base::space (excluding fill padding) are copies of fill.
In particular, there is implementation divergence over point (b) as to whether U+0020 or fill should be used. Further, should a character other than fill be used, it is unclear when "the fill characters are placed where none or space appears in the formatting pattern", whether the fill characters are placed at the beginning or the end of the "space field".
I believe that a strict interpretation of the current wording supports U+0020; however, fill is more likely to be the pragmatic choice.Proposed resolution:
This wording is relative to N4582.
Change 22.4.6.3 [locale.moneypunct] paragraph 2 as indicated:
-2- Where none or space appears, white space is permitted in the format, except where none appears at the end, in which case no white space is permitted. For input, the value space indicates that at least one space is required at that position. For output, the value space indicates one instance of the fill character (22.4.6.2.2 [locale.money.put.virtuals]).
The value space indicates that at least one space is required at that position. Where symbol appears, the sequence of characters returned by curr_symbol() is permitted, and can be required. Where sign appears, the first (if any) of the sequence of characters returned by positive_sign() or negative_sign() (respectively as the monetary value is non-negative or negative) is required. Any remaining characters of the sign sequence are required after all other format components. Where value appears, the absolute numeric monetary value is required.
Section: 17.4.2.1.4 [bitmask.types] Status: Tentatively NAD Submitter: Hubert Tong Opened: 2016-04-14 Last modified: 2017-02-03
Priority: 3
View all other issues in [bitmask.types].
View all issues with Tentatively NAD status.
Discussion:
The usual pattern now used for identifying where bitmask elements are declared, namely, as variables, preclude declaring them as enumerators.
Compare: ctype_base::space in C++03 subclause 22.2.1 [lib.category.ctype] versus the same in N4582 subclause 22.4.1 [category.ctype]. It is unclear whether this is intentional. Further it is unclear if odr-use of bitmask elements is intended to be allowed.[2016-05 Issues Telecon]
Jonathan believes that this was intentional, and was done by N3110. Jonathan will provide more precise references.
Proposed resolution:
Section: 26.5 [complex.numbers] Status: New Submitter: Oliver Rosten Opened: 2016-04-14 Last modified: 2017-02-03
Priority: 3
View other active issues in [complex.numbers].
View all other issues in [complex.numbers].
View all issues with New status.
Discussion:
This modification will allow complex-number arithmetic to be performed at compile time. From a mathematical standpoint, it is natural (and desirable) to treat complex numbers on the same footing as the reals. From a programming perspective, this change will broaden the scope in which std::complex can be used, allowing it to be smoothly incorporated into classes exploiting constexpr.
Suggested resolution: The following functions in the std::complex namespace should be made constexpr:Section 26.5.5 [complex.member.ops]: The member (arithmetic) operators {+=, -=, /=, *=}
Section 26.5.6 [complex.ops]: The arithmetic operators unary operators {+, -} and binary operators {+, -, /, *};
Section 26.5.7 [complex.value.ops]: The 'value' operators abs, norm, and conj.
In terms of modification of the standard, all that is required is the insertion of the constexpr specifier in the relevant parts of:
Section 26.5.1 [complex.syn] (the Header synopsis), together with the corresponding entries in sections 26.5.6 [complex.ops] and 26.5.7 [complex.value.ops].
Sections 26.5.2 [complex], 26.5.3 [complex.special] (class template and specializations), together with the corresponding entries in section 26.5.5 [complex.member.ops].
[2016-05 Issues Telecon]
This kind of work (new feature) has been being done via papers rather than via the issues list.
Walter believes that he knows someone who would be willing to write such a paper.
Proposed resolution:
Section: 17.5.5.5 [member.functions] Status: New Submitter: Hubert Tong Opened: 2016-04-15 Last modified: 2017-02-03
Priority: 3
View other active issues in [member.functions].
View all other issues in [member.functions].
View all issues with New status.
Discussion:
In N4582 subclause 17.6.5.5 [member.functions], the requirement that:
any call to the member function that would select an overload from the set of declarations described in this standard behaves as if that overload were selected
is unclear in the extent of the "as if". For example, in providing:
basic_string(const charT* s);
for a one-argument call to:
basic_string(const charT* s, const Allocator& a = Allocator());
it can be read that an implementation may be required to call the copy constructor for the allocator since the core language rules for copy elision would not allow the "a" argument to be constructed directly into the member used to store the allocator.
Clarification (even if just a note) would be appreciated.[2016-05 Issues Telecon]
This is related to issue 2563.
Proposed resolution:
Section: 99 [concurr.ts::futures.unique_future], 99 [concurr.ts::futures.shared_future] Status: Review Submitter: Tim Song Opened: 2016-04-22 Last modified: 2017-03-05
Priority: 2
View other active issues in [concurr.ts::futures.unique_future].
View all other issues in [concurr.ts::futures.unique_future].
View all issues with Review status.
Discussion:
Addresses: concurr.ts
In the concurrency TS, the future/shared_future unwrapping constructors
future(future<future<R>>&&) noexcept; shared_future(future<shared_future<R>>&& rhs) noexcept;
appear to implicitly require rhs be valid (e.g., by referring to its shared state, and by requiring a valid() == true postcondition). However, they are also marked noexcept, suggesting that they are wide-contract, and also makes the usual suggested handling for invalid futures, throwing a future_error, impossible.
Either the noexcept should be removed, or the behavior with an invalid future should be specified.Original resolution alternative #1 [NOT CHOSEN]:
This wording is relative to N4577.
Strike the noexcept on these constructors in 99 [futures.unique_future]/1-2 and 99 [futures.shared_future]/1-2, and optionally add a Requires: rhs.valid() == true paragraph.
[2016-11-12, Issaquah]
Sat PM: We prefer alternative #2 - Move to review
Proposed resolution:
This wording is relative to N4577.
Alternative #2: Specify that an empty (shared_)future object is constructed if rhs is invalid, and adjust the postcondition accordingly.
Edit 99 [futures.unique_future] as indicated:
future(future<future<R>>&& rhs) noexcept;-3- Effects: If rhs.valid() == false, constructs an empty future object that does not refer to a shared state. Otherwise, c
Constructs a future object from the shared state referred to by rhs. The future becomes ready when one of the following occurs:
Both the rhs and rhs.get() are ready. The value or the exception from rhs.get() is stored in the future's shared state.
rhs is ready but rhs.get() is invalid. An exception of type std::future_error, with an error condition of std::future_errc::broken_promise is stored in the future's shared state.
-4- Postconditions:
valid() == truevalid() returns the same value as rhs.valid() prior to the constructor invocation..rhs.valid() == false.
Edit 99 [futures.shared_future] as indicated:
shared_future(future<shared_future<R>>&& rhs) noexcept;-3- Effects: If rhs.valid() == false, constructs an empty shared_future object that does not refer to a shared state. Otherwise, c
Constructs a shared_future object from the shared state referred to by rhs. The shared_future becomes ready when one of the following occurs:
Both the rhs and rhs.get() are ready. The value or the exception from rhs.get() is stored in the shared_future's shared state.
rhs is ready but rhs.get() is invalid. The shared_future stores an exception of type std::future_error, with an error condition of std::future_errc::broken_promise.
-4- Postconditions:
valid() == truevalid() returns the same value as rhs.valid() prior to the constructor invocation..rhs.valid() == false.
Section: 22.4.2.2.2 [facet.num.put.virtuals] Status: New Submitter: Hubert Tong Opened: 2016-05-07 Last modified: 2017-02-03
Priority: 3
View other active issues in [facet.num.put.virtuals].
View all other issues in [facet.num.put.virtuals].
View all issues with New status.
Discussion:
The call to do_put(out, str, fill, (int)val) in N4582 subclause 22.4.2.2.2 [facet.num.put.virtuals] paragraph 6 cannot select a best viable function in overload resolution given the overloads listed for do_put in 22.4.2.2 [locale.nm.put].
There is implementation divergence:Some implementations call the long overload (as overriden);
some implementations call the unsigned long overload (as overriden);
some implementations call something else.
It appears that the resolution to DR 359 attempted a fix; however, the relevant portion of the change was not applied to the WP.
Proposed resolution:
Section: 22.4.2.2.2 [facet.num.put.virtuals] Status: New Submitter: Hubert Tong Opened: 2016-05-07 Last modified: 2017-02-03
Priority: 3
View other active issues in [facet.num.put.virtuals].
View all other issues in [facet.num.put.virtuals].
View all issues with New status.
Discussion:
N4582 subclause 22.4.2.2.2 [facet.num.put.virtuals] paragraph 6 makes no provision for fill-padding in its specification of the behaviour when (str.flags() & ios_base::boolalpha) != 0.
Proposed resolution:
Section: 23.2.3 [sequence.reqmts] Status: New Submitter: Kazutoshi Satoda Opened: 2016-05-08 Last modified: 2017-02-03
Priority: 3
View other active issues in [sequence.reqmts].
View all other issues in [sequence.reqmts].
View all issues with New status.
Discussion:
Please look through the following modifications on a value v of type vector<T>:
assert(v.size() > 0); v.push_back(v[0]); v.insert(v.begin(), v[0]); v.resize(v.size() * 2, v[0]); v.assign(v.size() * 2, v[0]);
All of these use an element of itself which may be moved or destroyed by the modification.
From what I see so far, the first three are required to work. Please see library issue 526 for validity of them. But only the last one is undefined because it violates a precondition of a sequence container operation. I think this is too subtle. Should it be like that, really? The precondition is in Table 107 "Sequence container requirements" at the next of 23.2.3 [sequence.reqmts] p3.In Tables 107 and 108, X denotes a sequence container class, a denotes a value of X containing elements of type T, […] n denotes a value of X::size_type, […] t denotes an lvalue or a const rvalue of X::value_type, […]
[…]
Table 107 — Sequence container requirements (in addition to container) Expression Return type Assertion/note
pre-/post-condition[…] a.assign(n, t) void Requires: T shall be CopyInsertable into X and CopyAssignable.
pre: t is not a reference into a.
Replaces elements in a with n copies of t.
I looked into the following implementations:
libc++ relies on the precondition.
It deallocates first on n > capacity() case, see here.libstdc++ doesn't rely on the precondition.
It creates temporary vector(n, t) and swap() on n > capacity() case, see here.MSVC relies on the precondition.
It unconditionally does clear() and then insert(begin(), n, t). I looked into my local "%PROGRAMFILES(X86)%/Microsoft Visual Studio 14.0/VC/include/vector".One drawback of libstdc++ implementation, I could find so far, is possibly increased peek memory usage (both old and new buffer exist at the same time). But, because the same can happen on the most other modifications, it seems a reasonable trade-off to remove the precondition to fill the subtle gap. Users who really needs less memory usage can do clear() and insert() by themselves.
I also found that basic_string::assign(n, c) is safe on this point. At 21.3.2.6.3 [string.assign] p17:basic_string& assign(size_type n, charT c);Effects: Equivalent to assign(basic_string(n, c)).
Returns: *this.
This can be seen as another gap.
Looking back on the history, I found that the definition of assign(n, t) was changed at C++14 for library issue 2209. There were more restricting definitions like this:void assign(size_type n, const T& t);Effects:
erase(begin(), end()); insert(begin(), n, t);
I think the precondition was probably set to accept this old definition and is not required inherently. And if the less memory usage was really intended, the standard is now underspecifying about that.
[2016-05 Issues Telecon]
Howard believes this should be NAD, but we tabled the discussion.
Proposed resolution:
This wording is relative to N4582.
In 23.2.3 [sequence.reqmts], edit Table 107 (Sequence container requirements) as indicated:
Table 107 — Sequence container requirements (in addition to container) Expression Return type Assertion/note
pre-/post-condition[…] a.assign(n, t) void Requires: T shall be CopyInsertable into X and CopyAssignable.
pre: t is not a reference into a.
Replaces elements in a with n copies of t.
Section: 99 [fs.rec.dir.itr.members] Status: Open Submitter: Eric Fiselier Opened: 2016-05-09 Last modified: 2017-03-19
Priority: 2
View all other issues in [fs.rec.dir.itr.members].
View all issues with Open status.
Discussion:
The current specification of recursion_pending() says (99 [fs.rec.dir.itr.members]/24):
Returns: true if disable_recursion_pending() has not been called subsequent to the prior construction or increment operation, otherwise false.
This language does not take into account cases where the prior construction was a copy construction from a iterator, it, where it.recursion_pending() == false.
[2016-08 Chicago]
Wed AM: Move to Open
Proposed resolution:
This wording is relative to N4582.
Change 99 [fs.rec.dir.itr.members] as indicated:
explicit recursive_directory_iterator(const path& p); recursive_directory_iterator(const path& p, directory_options options); recursive_directory_iterator(const path& p, directory_options options, error_code& ec) noexcept; recursive_directory_iterator(const path& p, error_code& ec) noexcept;[…]
-3- Postcondition:options() == options for the signatures with a directory_options argument, otherwise options() == directory_options::none.
options() == options for the signatures with a directory_options argument, otherwise options() == directory_options::none.
recursion_pending() == true.
[…]
[Drafting note: The following changes the specification of recursion_pending() seemingly recursive. Perhaps it would be easier to specify recursion_pending() in terms of a exposition only member in recursive_directory_iterator.]
bool recursion_pending() const;[…]
-24- Returns:true if disable_recursion_pending() has not been called subsequent to the prior construction or increment operation, otherwise falsefalse if disable_recursion_pending() has been called subsequent to the prior construction or increment operation, otherwise the value of recursion_pending() set by that operation. […]recursive_directory_iterator& operator++(); recursive_directory_iterator& increment(error_code& ec) noexcept;[…]
-27- Effects: As specified by Input iterators (24.2.3), except that: […] -?- Postcondition: recursion_pending() == true.
Section: 23.5 [unord] Status: New Submitter: Billy Robert O'Neal III Opened: 2016-05-20 Last modified: 2017-02-03
Priority: 3
View all other issues in [unord].
View all issues with New status.
Discussion:
The resolution of LWG 2210 missed constructors accepting a range or initializer list and allocator.
Previous resolution [SUPERSEDED]:
This wording is relative to N4582.
Add to the synopsis in 23.5.4.1 [unord.map.overview] p3:
namespace std { template <class Key, class T, class Hash = hash<Key>, class Pred = std::equal_to<Key>, class Allocator = std::allocator<std::pair<const Key, T> > > { class unordered_map { public: […] unordered_map(size_type n, const hasher& hf, const allocator_type& a) : unordered_map(n, hf, key_equal(), a) { } template <class InputIterator> unordered_map(InputIterator f, InputIterator l, const allocator_type& a) : unordered_map(f, l, see below, hasher(), key_equal(), a) { } template <class InputIterator> unordered_map(InputIterator f, InputIterator l, size_type n, const allocator_type& a) : unordered_map(f, l, n, hasher(), key_equal(), a) { } template <class InputIterator> unordered_map(InputIterator f, InputIterator l, size_type n, const hasher& hf, const allocator_type& a) : unordered_map(f, l, n, hf, key_equal(), a) { } unordered_map(initializer_list<value_type> il, const allocator_type& a) : unordered_map(il, see below, hasher(), key_equal(), a) { } unordered_map(initializer_list<value_type> il, size_type n, const allocator_type& a) : unordered_map(il, n, hasher(), key_equal(), a) { } […] }; }Add to the synopsis in 23.5.5.1 [unord.multimap.overview] p3:
namespace std { template <class Key, class T, class Hash = hash<Key>, class Pred = std::equal_to<Key>, class Allocator = std::allocator<std::pair<const Key, T> > > { class unordered_multimap { public: […] unordered_multimap(size_type n, const hasher& hf, const allocator_type& a) : unordered_multimap(n, hf, key_equal(), a) { } template <class InputIterator> unordered_multimap(InputIterator f, InputIterator l, const allocator_type& a) : unordered_multimap(f, l, see below, hasher(), key_equal(), a) { } template <class InputIterator> unordered_multimap(InputIterator f, InputIterator l, size_type n, const allocator_type& a) : unordered_multimap(f, l, n, hasher(), key_equal(), a) { } template <class InputIterator> unordered_multimap(InputIterator f, InputIterator l, size_type n, const hasher& hf, const allocator_type& a) : unordered_multimap(f, l, n, hf, key_equal(), a) { } unordered_multimap(initializer_list<value_type> il, const allocator_type& a) : unordered_multimap(il, see below, hasher(), key_equal(), a) { } unordered_multimap(initializer_list<value_type> il, size_type n, const allocator_type& a) : unordered_multimap(il, n, hasher(), key_equal(), a) { } […] }; }Add to the synopsis in 23.5.6.1 [unord.set.overview] p3:
namespace std { template <class Key, class Hash = hash<Key>, class Pred = std::equal_to<Key>, class Allocator = std::allocator<Key> > { class unordered_set { public: […] unordered_set(size_type n, const hasher& hf, const allocator_type& a) : unordered_set(n, hf, key_equal(), a) { } template <class InputIterator> unordered_set(InputIterator f, InputIterator l, const allocator_type& a) : unordered_set(f, l, see below, hasher(), key_equal(), a) { } template <class InputIterator> unordered_set(InputIterator f, InputIterator l, size_type n, const allocator_type& a) : unordered_set(f, l, n, hasher(), key_equal(), a) { } template <class InputIterator> unordered_set(InputIterator f, InputIterator l, size_type n, const hasher& hf, const allocator_type& a) : unordered_set(f, l, n, hf, key_equal(), a) { } unordered_set(initializer_list<value_type> il, const allocator_type& a) : unordered_set(il, see below, hasher(), key_equal(), a) { } unordered_set(initializer_list<value_type> il, size_type n, const allocator_type& a) : unordered_set(il, n, hasher(), key_equal(), a) { } […] }; }Add to the synopsis in 23.5.7.1 [unord.multiset.overview] p3:
namespace std { template <class Key, class Hash = hash<Key>, class Pred = std::equal_to<Key>, class Allocator = std::allocator<Key> > { class unordered_multiset { public: […] unordered_multiset(size_type n, const hasher& hf, const allocator_type& a) : unordered_multiset(n, hf, key_equal(), a) { } template <class InputIterator> unordered_multiset(InputIterator f, InputIterator l, const allocator_type& a) : unordered_multiset(f, l, see below, hasher(), key_equal(), a) { } template <class InputIterator> unordered_multiset(InputIterator f, InputIterator l, size_type n, const allocator_type& a) : unordered_multiset(f, l, n, hasher(), key_equal(), a) { } template <class InputIterator> unordered_multiset(InputIterator f, InputIterator l, size_type n, const hasher& hf, const allocator_type& a) : unordered_multiset(f, l, n, hf, key_equal(), a) { } unordered_multiset(initializer_list<value_type> il, const allocator_type& a) : unordered_multiset(il, see below, hasher(), key_equal(), a) { } unordered_multiset(initializer_list<value_type> il, size_type n, const allocator_type& a) : unordered_multiset(il, n, hasher(), key_equal(), a) { } […] }; }
[2016-06, Oulu — Daniel comments and provides new wording]
During the LWG discussion of this issue it has been observed, that the interpretation of the embedded see below is not really clear and that we should split declaration and definition of the new overloads, so that we have a place that allows us to specify what "see below" stands for. In addition, the new wording wraps the "see below" as "size_type(see below)" to clarify the provided expression type, similar as we did for the default constructor of unordered_map.
[Oulu, 2016-06]
Alisdair to review wording.
Proposed resolution:
This wording is relative to N4594.
Add to the synopsis in 23.5.4.1 [unord.map.overview] p3:
namespace std {
template <class Key, class T,
class Hash = hash<Key>,
class Pred = std::equal_to<Key>,
class Allocator = std::allocator<std::pair<const Key, T> > > {
class unordered_map {
public:
[…]
unordered_map(size_type n, const hasher& hf, const allocator_type& a)
: unordered_map(n, hf, key_equal(), a) { }
template <class InputIterator>
unordered_map(InputIterator f, InputIterator l, const allocator_type& a);
template <class InputIterator>
unordered_map(InputIterator f, InputIterator l, size_type n, const allocator_type& a)
: unordered_map(f, l, n, hasher(), key_equal(), a) { }
template <class InputIterator>
unordered_map(InputIterator f, InputIterator l, size_type n, const hasher& hf,
const allocator_type& a)
: unordered_map(f, l, n, hf, key_equal(), a) { }
unordered_map(initializer_list<value_type> il, const allocator_type& a);
unordered_map(initializer_list<value_type> il, size_type n, const allocator_type& a)
: unordered_map(il, n, hasher(), key_equal(), a) { }
[…]
};
}
Insert the following new prototype specification just after 23.5.4.2 [unord.map.cnstr] p2
template <class InputIterator> unordered_map(InputIterator f, InputIterator l, const allocator_type& a) : unordered_map(f, l, size_type(see below), hasher(), key_equal(), a) { } unordered_map(initializer_list<value_type> il, const allocator_type& a) : unordered_map(il, size_type(see below), hasher(), key_equal(), a) { }-?- Remarks: The number of buckets is implementation-defined.
Add to the synopsis in 23.5.5.1 [unord.multimap.overview] p3:
namespace std {
template <class Key, class T,
class Hash = hash<Key>,
class Pred = std::equal_to<Key>,
class Allocator = std::allocator<std::pair<const Key, T> > > {
class unordered_multimap {
public:
[…]
unordered_multimap(size_type n, const hasher& hf, const allocator_type& a)
: unordered_multimap(n, hf, key_equal(), a) { }
template <class InputIterator>
unordered_multimap(InputIterator f, InputIterator l, const allocator_type& a);
template <class InputIterator>
unordered_multimap(InputIterator f, InputIterator l, size_type n, const allocator_type& a)
: unordered_multimap(f, l, n, hasher(), key_equal(), a) { }
template <class InputIterator>
unordered_multimap(InputIterator f, InputIterator l, size_type n, const hasher& hf,
const allocator_type& a)
: unordered_multimap(f, l, n, hf, key_equal(), a) { }
unordered_multimap(initializer_list<value_type> il, const allocator_type& a);
unordered_multimap(initializer_list<value_type> il, size_type n, const allocator_type& a)
: unordered_multimap(il, n, hasher(), key_equal(), a) { }
[…]
};
}
Insert the following new prototype specification just after 23.5.5.2 [unord.multimap.cnstr] p2
template <class InputIterator> unordered_multimap(InputIterator f, InputIterator l, const allocator_type& a) : unordered_multimap(f, l, size_type(see below), hasher(), key_equal(), a) { } unordered_multimap(initializer_list<value_type> il, const allocator_type& a) : unordered_multimap(il, size_type(see below), hasher(), key_equal(), a) { }-?- Remarks: The number of buckets is implementation-defined.
Add to the synopsis in 23.5.6.1 [unord.set.overview] p3:
namespace std {
template <class Key,
class Hash = hash<Key>,
class Pred = std::equal_to<Key>,
class Allocator = std::allocator<Key> > {
class unordered_set {
public:
[…]
unordered_set(size_type n, const hasher& hf, const allocator_type& a)
: unordered_set(n, hf, key_equal(), a) { }
template <class InputIterator>
unordered_set(InputIterator f, InputIterator l, const allocator_type& a);
template <class InputIterator>
unordered_set(InputIterator f, InputIterator l, size_type n, const allocator_type& a)
: unordered_set(f, l, n, hasher(), key_equal(), a) { }
template <class InputIterator>
unordered_set(InputIterator f, InputIterator l, size_type n, const hasher& hf,
const allocator_type& a)
: unordered_set(f, l, n, hf, key_equal(), a) { }
unordered_set(initializer_list<value_type> il, const allocator_type& a);
unordered_set(initializer_list<value_type> il, size_type n, const allocator_type& a)
: unordered_set(il, n, hasher(), key_equal(), a) { }
[…]
};
}
Insert the following new prototype specification just after 23.5.6.2 [unord.set.cnstr] p2
template <class InputIterator> unordered_set(InputIterator f, InputIterator l, const allocator_type& a) : unordered_set(f, l, size_type(see below), hasher(), key_equal(), a) { } unordered_set(initializer_list<value_type> il, const allocator_type& a) : unordered_set(il, size_type(see below), hasher(), key_equal(), a) { }-?- Remarks: The number of buckets is implementation-defined.
Add to the synopsis in 23.5.7.1 [unord.multiset.overview] p3:
namespace std {
template <class Key,
class Hash = hash<Key>,
class Pred = std::equal_to<Key>,
class Allocator = std::allocator<Key> > {
class unordered_multiset {
public:
[…]
unordered_multiset(size_type n, const hasher& hf, const allocator_type& a)
: unordered_multiset(n, hf, key_equal(), a) { }
template <class InputIterator>
unordered_multiset(InputIterator f, InputIterator l, const allocator_type& a);
template <class InputIterator>
unordered_multiset(InputIterator f, InputIterator l, size_type n, const allocator_type& a)
: unordered_multiset(f, l, n, hasher(), key_equal(), a) { }
template <class InputIterator>
unordered_multiset(InputIterator f, InputIterator l, size_type n, const hasher& hf,
const allocator_type& a)
: unordered_multiset(f, l, n, hf, key_equal(), a) { }
unordered_multiset(initializer_list<value_type> il, const allocator_type& a);
unordered_multiset(initializer_list<value_type> il, size_type n, const allocator_type& a)
: unordered_multiset(il, n, hasher(), key_equal(), a) { }
[…]
};
}
Insert the following new prototype specification just after 23.5.7.2 [unord.multiset.cnstr] p2
template <class InputIterator> unordered_multiset(InputIterator f, InputIterator l, const allocator_type& a) : unordered_multiset(f, l, size_type(see below), hasher(), key_equal(), a) { } unordered_multiset(initializer_list<value_type> il, const allocator_type& a) : unordered_multiset(il, size_type(see below), hasher(), key_equal(), a) { }-?- Remarks: The number of buckets is implementation-defined.
Section: 26.5.6 [complex.ops] Status: New Submitter: Tim Song Opened: 2016-05-23 Last modified: 2017-05-31
Priority: 3
View all other issues in [complex.ops].
View all issues with New status.
Discussion:
The specification of operator>>(istream&, complex<T>&) is extremely short on details. It currently reads, in its entirety (26.5.6 [complex.ops]/12-15):
template<class T, class charT, class traits> basic_istream<charT, traits>& operator>>(basic_istream<charT, traits>& is, complex<T>& x);Effects: Extracts a complex number x of the form: u, (u), or (u,v), where u is the real part and v is the imaginary part (27.7.4.2 [istream.formatted]).
Requires: The input values shall be convertible to T. If bad input is encountered, calls is.setstate(ios_base::failbit) (which may throw ios::failure (27.5.5.4 [iostate.flags])). Returns: is. Remarks: This extraction is performed as a series of simpler extractions. Therefore, the skipping of whitespace is specified to be the same for each of the simpler extractions.
It is completely unclear:
Proposed resolution:
Drafting note: the following wording is based on:
- Characters are extracted using operator>> and compared using traits::eq.
- Mismatched characters are returned to the stream.
This wording is relative to N4582.
Replace 26.5.6 [complex.ops]/12-15 with the following paragraphs:
template<class T, class charT, class traits> basic_istream<charT, traits>& operator>>(basic_istream<charT, traits>& is, complex<T>& x);-?- Effects: First, extracts a character from is.
In the description above, characters are extracted from is as if by operator>> (27.7.4.2.3 [istream.extractors]), and returned to the stream as if by basic_istream::putback (27.7.4.3 [istream.unformatted]). Character equality is determined using traits::eq. An object t of type T is extracted from is as if by is >> t. If any extraction operation fails, no further operation is performed and the whole extraction fails. On failure, calls is.setstate(ios_base::failbit) (which may throw ios::failure (27.5.5.4 [iostate.flags])). -?- Returns: is. -?- [Note: This extraction is performed as a series of simpler extractions. Therefore, the skipping of whitespace is specified to be the same for each of the simpler extractions. — end note]
- If the character extracted is equal to is.widen('('), extracts an object u of type T from is, then extracts a character from is.
- If this character is equal to is.widen(')'), then assigns complex<T>(u) to x.
- Otherwise, if this character is equal to is.widen(','), extracts an object v of type T from is, then extracts a character from is. If this character is equal to is.widen(')'), then assigns complex<T>(u, v) to x; otherwise returns the character to is and the extraction fails.
- Otherwise, returns the character to is and the extraction fails.
- Otherwise, returns the character to is, extracts an object u of type T from is, and assigns complex<T>(u) to x.
Section: 20.13.4 [allocator.adaptor.members] Status: Tentatively NAD Submitter: Billy Robert O'Neal III Opened: 2016-05-24 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [allocator.adaptor.members].
View all other issues in [allocator.adaptor.members].
View all issues with Tentatively NAD status.
Discussion:
scoped_allocator_adaptor is specified to use forward when what it is really doing is moving elements. It should use move.
Previous resolution [SUPERSEDED]:
This wording is relative to N4582.
Edit 20.13.4 [allocator.adaptor.members] p15 as indicated:
template <class T1, class T2, class U, class V> void construct(pair<T1, T2>* p, pair<U, V>&& x);Effects: Equivalent to this->construct(p, piecewise_construct, forward_as_tuple(std::
forwardmove<U>(x.first)), forward_as_tuple(std::forwardmove<V>(x.second))).
Proposed resolution:
Withdrawn by the submitter, since the prerequisites were incorrect.
Section: 18.3.4 [numeric.limits] Status: Open Submitter: Richard Smith Opened: 2016-06-09 Last modified: 2017-03-05
Priority: 3
View other active issues in [numeric.limits].
View all other issues in [numeric.limits].
View all issues with Open status.
Discussion:
I've received this report at the project editor mail alias, and it seems like it may be worthy of a LWG issue:
I recently had this problem:
- I was storing data in a vector of __uint128_ts
- I used a sorting library which used numeric_limits<T>::max() as a sentinel value
- GCC's libstdc++ provides a numeric_limits specialisation for that type, but
- Clang's libc++ does not.
This broke the sorting for me on different platforms, and it was quite difficult to determine why. If the default numeric_limits didn't default to 0s and false values (18.3.2.4 of N4582), and instead static_asserted, causing my code to not compile, I would have found the solution immediately.
I know that __uint128_t is non-standard, so neither GCC nor Clang is doing the wrong thing nor the right thing here. I could just submit a patch to libc++ providing the specialisations, but it doesn't fix the problem at its core. I am wondering, what is the rationale behind the defaults being 0 and false? It seems like it is inviting a problem for any future numeric types, whether part of a library, compiler extension, and possibly even future updates to C++'s numeric types. I think it would be much better to prevent code that tries to use unspecified numeric_limits from compiling.
An alternative to this suggestion would be to still define the primary template, but not provide any of the members except is_specialized. Either way, this would make numeric_limits members SFINAEable.
Along the same lines, one might wonder why the members that only make sense for floating-point types are required to be defined to nonsense values for integer types.[2016-11-12, Issaquah]
Sat PM: This looks like a good idea. Jonathan and Marshall will do post C++17 implementations and report back.
Proposed resolution:
Section: 30.4.4.1 [thread.lock.guard] Status: New Submitter: Eric Fiselier Opened: 2016-06-13 Last modified: 2017-02-03
Priority: 3
View all other issues in [thread.lock.guard].
View all issues with New status.
Discussion:
In the synopsis of 30.4.4.1 [thread.lock.guard] the mutex_type typedef is specified as follows:
template <class... MutexTypes>
class lock_guard {
public:
typedef Mutex mutex_type; // If MutexTypes... consists of the single type Mutex
[…]
};
The comment seems ambiguous as it could mean either:
I originally took the language to mean (2), but upon further review it seems that (1) is the intended interpretation, as suggested in the LEWG discussion in Lenexa.
I think the language should be clarified to prevent implementation divergence.Proposed resolution:
This wording is relative to N4594.
Edit 30.4.4.1 [thread.lock.guard]/1, class template lock_guard synopsis, as indicated:
template <class... MutexTypes>
class lock_guard {
public:
typedef Mutex mutex_type; // Only iIf MutexTypes... consists of theexpands to a single type Mutex
[…]
};
Section: 99 [fs.path.concat] Status: New Submitter: Tim Song Opened: 2016-06-16 Last modified: 2017-03-19
Priority: 2
View all issues with New status.
Discussion:
99 [fs.path.concat] specifies that the postcondition for
path& operator+=(const path& x); path& operator+=(const string_type& x); path& operator+=(const value_type* x); path& operator+=(value_type x); template<class Source> path& operator+=(const Source& x); template<class EcharT> path& operator+=(EcharT x); template<class Source> path& concat(const Source& x); template<class InputIterator> path& concat(InputIterator first, InputIterator last);
is
native() == prior_native + effective-argument
where effective-argument is
It also says that
If the value type of effective-argument would not be path::value_type, the actual argument or argument range is first converted ( [fs.path.type.cvt]) so that effective-argument has value type path::value_type.
There are several problems with this specification:
First, there is no overload taking "source" (note the lower case); all single-argument overloads take "x". Second, there's nothing that defines what it means to use operator+ on a string and an iterator range; clearly concatentation is intended but there is no wording to that effect. Third, the final portion uses "value type", but the "value type" of a single character is not a defined concept. Also, the reference only to [fs.path.type.cvt] seems to imply that any format conversion specified in [fs.path.fmt.cvt] will not be performed, in seeming contradiction to the rule that native() is to return the native pathname format (99 [fs.path.native.obs]/1). Is that intended?[2016-11-10, Billy suggests wording]
The wording for LWG 2798 resolves this issue as well.
Proposed resolution:
The wording for LWG 2798 resolves this issue as well.
Section: 18.6.2.1 [new.delete.single] Status: New Submitter: Clark Nelson Opened: 2016-06-21 Last modified: 2017-02-03
Priority: 3
View all other issues in [new.delete.single].
View all issues with New status.
Discussion:
It should be considered whether the description of the single-object allocation functions should say "or smaller", like the array allocation functions. For example, according to 18.6.2.1 [new.delete.single] p1 (emphasis mine):
The allocation function (3.7.4.1) called by a new-expression (5.3.4) to allocate size bytes of storage suitably aligned to represent any object of that size.
In contrast to this, 18.6.2.2 [new.delete.array] p1 says (emphasis mine):
The allocation function (3.7.4.1) called by the array form of a new-expression (5.3.4) to allocate size bytes of storage suitably aligned to represent any array object of that size or smaller. (footnote: It is not the direct responsibility of operator new[](std::size_t) or operator delete[](void*) to note the repetition count or element size of the array. Those operations are performed elsewhere in the array new and delete expressions. The array new expression, may, however, increase the size argument to operator new[](std::size_t) to obtain space to store supplemental information.)
Proposed resolution:
Section: 25.7.4 [alg.partitions] Status: New Submitter: Jonathan Wakely Opened: 2016-07-06 Last modified: 2017-02-03
Priority: 3
View all other issues in [alg.partitions].
View all issues with New status.
Discussion:
Requires: InputIterator's value type shall be convertible to Predicate's argument type.
This seems to date from the days of adaptable function objects with an argument_type typedef, but in modern C++ the predicate might not have an argument type. It could have a function template that accepts various arguments, so it doesn't make sense to state requirements in terms of a type that isn't well defined.
Proposed resolution:
This wording is relative to N4594.
Edit 25.7.4 [alg.partitions] as indicated:
template <class InputIterator, class Predicate> bool is_partitioned(InputIterator first, InputIterator last, Predicate pred);-1- Requires:
[…]InputIterator's value type shall be convertible to Predicate's argument typeThe expression pred(*i) shall be well-formed for all i in [first, last).template <class InputIterator, class OutputIterator1, class OutputIterator2, class Predicate> pair<OutputIterator1, OutputIterator2> partition_copy(InputIterator first, InputIterator last, OutputIterator1 out_true, OutputIterator2 out_false, Predicate pred);-12- Requires: InputIterator's value type shall be CopyAssignable, and shall be writable (24.2.1 [iterator.requirements.general]) to the out_true and out_false OutputIterators, and
[…]shall be convertible to Predicate's argument typethe expression pred(*i) shall be well-formed for all i in [first, last). The input range shall not overlap with either of the output ranges.template<class ForwardIterator, class Predicate> ForwardIterator partition_point(ForwardIterator first, ForwardIterator last, Predicate pred);-16- Requires:
[…]ForwardIterator's value type shall be convertible to Predicate's argument typeThe expression pred(*i) shall be well-formed for all i in [first, last). [first, last) shall be partitioned by pred, i.e. all elements that satisfy pred shall appear before those that do not.
Section: 23.2.4.1 [container.node.overview] Status: New Submitter: Richard Smith Opened: 2016-07-08 Last modified: 2017-02-03
Priority: 3
View all issues with New status.
Discussion:
The private members of node_handle are missing the usual "exposition only" comment. As a consequence, ptr_ and alloc_ now appear to be names defined by the library (so programs defining these names as macros before including a library header have undefined behavior).
Presumably this is unintentional and these members should be considered to be for exposition only. It's also not clear whether the name node_handle is reserved for library usage or not; 23.2.4.1 [container.node.overview]/3 says the implementation need not provide a type with this name, but doesn't seem to rule out the possibility that an implementation will choose to do so regardless.Daniel:
A similar problem seems to exist for the exposition-only type call_wrapper from p0358r1, which exposes a private data member named fd and a typedef FD.[2016-07 Chicago]
Jonathan says that we need to make clear that the name node_handle is not reserved
Proposed resolution:
Section: 20.6.3 [optional.optional] Status: New Submitter: Richard Smith Opened: 2016-07-13 Last modified: 2017-02-03
Priority: 3
View other active issues in [optional.optional].
View all other issues in [optional.optional].
View all issues with New status.
Discussion:
Referring to N4604:
In [optional.object.assign]: emplace (normal form) has a Requires that the construction works.
Requires: is_constructible_v<T, Args&&...> is true.
emplace (initializer_list form) has a SFINAE condition:
Remarks: […] unless is_constructible_v<T, initializer_list<U>&, Args&&...> is true.
In 20.8.3.3 [any.modifiers]: emplace (normal form) has a Requires that the construction works:
Requires: is_constructible_v<T, Args...> is true.
emplace (initializer_list form) has a SFINAE condition:
Remarks: […] unless is_constructible_v<T, initializer_list<U>&, Args...> is true.
In 20.7.3.4 [variant.mod]: emplace (T, normal form) has a SFINAE condition:
Remarks: […] unless is_constructible_v<T, Args...> is true, and T occurs exactly once in Types....
emplace (Idx, normal form) has a both a Requires and a SFINAE condition:
Requires: I < sizeof...(Types)
Remarks: […] unless is_constructible_v<T, Args...> is true, and T occurs exactly once in Types....
emplace (T, initializer_list form) has a SFINAE condition:
Remarks: […] unless is_constructible_v<T, initializer_list<U>&, Args...> is true, and T occurs exactly once in Types....
emplace (Idx, initializer_list form) has a both a Requires and a SFINAE condition:
Requires: I < sizeof...(Types)
Remarks: […] unless is_constructible_v<T, Args...> is true, and T occurs exactly once in Types....
Why the inconsistency? Should all the cases have a SFINAE requirement?
I see that variant has an additional requirement (T occurs exactly once in Types...), but that only agues that it must be a SFINAE condition — doesn't say that the other cases (any/variant) should not. map/multimap/unordered_map/unordered_multimap have SFINAE'd versions of emplace that don't take initializer_lists, but they don't have any emplace versions that take ILs. Suggested resolution: Add SFINAE requirements to optional::emplace(Args&&... args) and any::emplace(Args&&... args);[2016-08 Chicago]
During issue prioritization, people suggested that this might apply to any as well.
Ville notes that 2746, 2754 and 2756 all go together.
Proposed resolution:
Section: 20.11.2.2.2 [util.smartptr.shared.dest] Status: New Submitter: Aaron Jacobs Opened: 2016-07-21 Last modified: 2017-05-31
Priority: 4
View other active issues in [util.smartptr.shared.dest].
View all other issues in [util.smartptr.shared.dest].
View all issues with New status.
Discussion:
The C++14 standard contains no language that guarantees the deleter run by a shared_ptr will see all associated weak_ptr instances as expired. For example, the standard doesn't appear to guarantee that the assertion in the following snippet won't fire:
std::weak_ptr<Foo> weak;
std::shared_ptr<Foo> strong{
new Foo,
[&weak] (Foo* f) {
assert(weak.expired());
delete f;
},
};
weak = strong;
strong.reset();
It seems clear that the intent is that associated weak_ptrs are expired, because otherwise shared_ptr deleters could resurrect a reference to an object that is being deleted.
Suggested fix: 20.11.2.2.2 [util.smartptr.shared.dest] should specify that the decrease in use_count() caused by the destructor is sequenced before the call to the deleter or the call to delete p.[2016-11-08, Jonathan and STL suggest NAD]
STL and Jonathan feel that the example has unspecified behaviour, and the assertion is allowed to fire, and that's OK (the program's expectation is not reasonable). Otherwise it's necessary to move-construct a copy of the deleter and use that copy to destroy the owned pointer. We do not want to be required to do that.
See also 2262.
Proposed resolution:
Section: 20.11.1.2.4 [unique.ptr.single.observers] Status: LEWG Submitter: Ville Voutilainen Opened: 2016-08-04 Last modified: 2017-02-03
Priority: 3
View all other issues in [unique.ptr.single.observers].
View all issues with LEWG status.
Discussion:
See LWG 2337. Since we aren't removing noexcept from shared_ptr's operator*, we should consider adding noexcept to unique_ptr's operator*.
[2016-08 — Chicago]
Thurs PM: P3, and status to 'LEWG'
[2016-08-05 Chicago]
Ville provides an initial proposed wording.
Proposed resolution:
This wording is relative to N4606.
[Drafting note: since this issue is all about consistency, optional's pointer-like operators are additionally included.]
In 20.11.1.2 [unique.ptr.single] synopsis, edit as follows:
add_lvalue_reference_t<T> operator*() const noexcept;
Before 20.11.1.2.4 [unique.ptr.single.observers]/1, edit as follows:
add_lvalue_reference_t<T> operator*() const noexcept;
In 20.6.3 [optional.optional] synopsis, edit as follows:
constexpr T const *operator->() const noexcept; constexpr T *operator->() noexcept; constexpr T const &operator*() const & noexcept; constexpr T &operator*() & noexcept; constexpr T &&operator*() && noexcept; constexpr const T &&operator*() const && noexcept;
Before [optional.object.observe]/1, edit as follows:
constexpr T const* operator->() const noexcept; constexpr T* operator->() noexcept;
Before [optional.object.observe]/5, edit as follows:
constexpr T const& operator*() const & noexcept; constexpr T& operator*() & noexcept;
Before [optional.object.observe]/9, edit as follows:
constexpr T&& operator*() && noexcept; constexpr const T&& operator*() const && noexcept;
Section: 20.4.3 [pairs.spec], 20.5.3.10 [tuple.special], 20.6.9 [optional.specalg], 20.7.10 [variant.specalg], 20.11.1.5 [unique.ptr.special], 23.3.7.3 [array.special], 23.6.4.5 [queue.special], 23.6.5.4 [priqueue.special], 23.6.6.5 [stack.special] Status: New Submitter: Agustín K-ballo Bergé Opened: 2016-08-15 Last modified: 2017-02-03
Priority: 3
View all issues with New status.
Discussion:
Related: 2748 swappable traits for optionals, 2749 swappable traits for variants.
The adoption of P0185R1 "Adding [nothrow-]swappable traits" makes certain non-swappable types indirectly swappable. Consider a type defined as follows:
struct non_swappable {
friend void swap(non_swappable&, non_swappable&) = delete;
};
non_swappable ns1, ns2;
using std::swap;
swap(ns1, ns2); // ill-formed
static_assert(std::is_swappable_v<non_swappable> == false); // holds
Lvalues of type non_swappable are not swappable, as defined by 17.5.3.2 [swappable.requirements], overload resolution selects the deleted function. Consistently, is_swappable_v<non_swappable> yields false. It should be noted that since non_swappable is move constructible and move assignable, a qualified call to std::swap would be well-formed, even under P0185. Now consider the following snippet:
std::tuple<non_swappable> tns1, tns2; using std::swap; swap(tns1, tns2); // previously ill-formed, now well-formed static_assert(std::is_swappable_v<std::tuple<non_swappable>> == false); // fires
Before P0185, this snippet would violate the implicit requirement of specialized swap for tuples that each tuple element be swappable. After P0185, this specialized swap overload for tuples would be SFINAEd away, resulting in overload resolution selecting the base swap overload, and performing the exchange via move construction and move assignment of tuples.
This issue affects all of pair, tuple, unique_ptr, array, queue, priority_queue, stack, and should eventually also apply to optional and variant.Proposed resolution:
This wording is relative to N4606, except when otherwise noted.
Modify 20.4.3 [pairs.spec] as indicated:
template<class T1, class T2> void swap(pair<T1, T2>& x, pair<T1, T2>& y) noexcept(noexcept(x.swap(y)));-7- Effects: As if by x.swap(y).
-8- Remarks: This function shallnot participate in overload resolutionbe defined as deleted unless is_swappable_v<T1> is true and is_swappable_v<T2> is true.
Modify 20.5.3.10 [tuple.special] as indicated:
template <class... Types> void swap(tuple<Types...>& x, tuple<Types...>& y) noexcept(see below);-1- Remarks: This function shall
not participate in overload resolutionbe defined as deleted unless is_swappable_v<Ti> is true for all i, where 0 <= i and i < sizeof...(Types). The expression inside noexcept is equivalent to:noexcept(x.swap(y))-2- Effects: As if by x.swap(y).
Modify 20.11.1.5 [unique.ptr.special] as indicated:
template <class T, class D> void swap(unique_ptr<T, D>& x, unique_ptr<T, D>& y) noexcept;-1- Remarks: This function shall
-2- Effects: Calls x.swap(y).not participate in overload resolutionbe defined as deleted unless is_swappable_v<D> is true.
Modify 23.3.7.3 [array.special] as indicated:
template <class T, size_t N> void swap(array<T, N>& x, array<T, N>& y) noexcept(noexcept(x.swap(y)));-1- Remarks: This function shall
-2- Effects: As if by x.swap(y). […]not participate in overload resolutionbe defined as deleted unless N == 0 or is_swappable_v<T> is true.
Modify 23.6.4.5 [queue.special] as indicated:
template <class T, class Container> void swap(queue<T, Container>& x, queue<T, Container>& y) noexcept(noexcept(x.swap(y)));-1- Remarks: This function shall
-2- Effects: As if by x.swap(y).not participate in overload resolutionbe defined as deleted unless is_swappable_v<Container> is true.
Modify 23.6.5.4 [priqueue.special] as indicated:
template <class T, class Container, class Compare> void swap(priority_queue<T, Container, Compare>& x, priority_queue<T, Container, Compare>& y) noexcept(noexcept(x.swap(y)));-1- Remarks: This function shall
-2- Effects: As if by x.swap(y).not participate in overload resolutionbe defined as deleted unless is_swappable_v<Container> is true and is_swappable_v<Compare> is true.
Modify 23.6.6.5 [stack.special] as indicated:
template <class T, class Container> void swap(stack<T, Container>& x, stack<T, Container>& y) noexcept(noexcept(x.swap(y)));-1- Remarks: This function shall
-2- Effects: As if by x.swap(y).not participate in overload resolutionbe defined as deleted unless is_swappable_v<Container> is true.
Modify 20.6.9 [optional.specalg] as indicated:
This change should be performed if and only if LWG 2748 is accepted and is against the wording of 2748:
template <class T> void swap(optional<T>& x, optional<T>& y) noexcept(noexcept(x.swap(y)));-1- Effects: Calls x.swap(y).
-2- Remarks: This function shallnot participate in overload resolutionbe defined as deleted unless is_move_constructible_v<T> is true and is_swappable_v<T> is true.
Modify 20.7.10 [variant.specalg] as indicated:
This change should be performed if and only if LWG 2749 is accepted and is against the wording of 2749:
template <class... Types> void swap(variant<Types...>& v, variant<Types...>& w) noexcept(see below);-1- Effects: Equivalent to v.swap(w).
-2- Remarks: This function shallnot participate in overload resolutionbe defined as deleted unless is_move_constructible_v<Ti> && is_swappable_v<Ti> is true for all i. The expression inside noexcept is equivalent to noexcept(v.swap(w)).
Section: 23.2.6 [associative.reqmts], 23.2.7 [unord.req], 23.4.2 [associative.map.syn], 23.5.2 [unord.map.syn] Status: New Submitter: Tomasz Kamiński Opened: 2016-09-06 Last modified: 2017-02-03
Priority: 2
View other active issues in [associative.reqmts].
View all other issues in [associative.reqmts].
View all issues with New status.
Discussion:
In C++17 the interface of the unique map was extended to include following function:
pair<iterator, bool> try_emplace(const key_type& k, Args&&... args); //and move version iterator try_emplace(const_iterator hint, const key_type& k, Args&&... args); //and move version iterator insert(const_iterator hint, node_type&& nh) insert_return_type insert(node_type&& nh);
All of the functions share a common property, that they are performing basically no-operation in case when element with given key (part of node) is already stored in the map. However there is major difference in their interface. The first three functions:
pair<iterator, bool> try_emplace(key_type&& k, Args&&... args); //and copy version iterator try_emplace(const_iterator hint, key_type&& k, Args&&... args); //and copy version iterator insert(const_iterator hint, node_type&& nh)
are guaranteeing that the value of the arguments (k, nh, args...) will not be changed if the map already contains a key with given value, so the programmer is free to reuse it for their own purpose.
However, the interface of the fourth function is a bit different:insert_return_type insert(node_type&& nh);
The insert_return_type is an unspecified type that contains:
bool inserted; X::iterator position; X::node_type node;
As we can see, the insert function is returning a node. This difference is actually misleading, as the programmer may start to wonder, why the function returns a node handle, instead of being guaranteed that the argument will not be modified (as other functions do). Most reasonable explanation is that, this function actually return a handle to a different node, that one passed as the argument, i.e. this function replaces an existing node with the nh argument and returns the handle to the old node. However, this function actually has the same semantics as the other insert function and returns a node that was passed as argument.
In addition, this design makes the interface of the insert function for the map inconsistent. Value inserting functions are returning pair<iterator, bool> while node inserting function is returning an unspecified type with guaranteed set of members. The only potential benefit of this signature is that it could potentially allow programmer to use decomposition declaration, so instead of:auto nh = node_provider(); if (map.insert(std::move(nh)).second) handle_node_in_other_way(std::move(nh));
The user would be able to write:
if (auto [it, ins, nh] = map.insert(node_provider); ins) handle_node_in_other_way(std::move(nh));
However, the insert_return_type is not currently required to work with decomposition declaration, so this is only "potential" benefit that could be added in future.
Furthermore, this change is preventing a user to use structured binding with combination with insert in generic code:
template<typename UniqMap, typename Elem>
void log_duplicate_insertion(UniqMap& map, Elem&& elem)
{
if (auto [it, ins] = map.insert(std::forward<Elem>(elem)); !ins)
std::cout << "attempt to insert duplicate for " << *it;
}
Currently, log_duplicate_insertion will not work with node_handle_type.
So, I am proposing to change the interface of the insert(node_handle) function for associative containers with unique keys, to be consistent with the other insert operation and try_emplace function. I.e. change the signature to:std::pair<iterator, bool> insert(node_type&& nh);
and provide the guarantee that nh will be unchanged if an element was not inserted.
[2016-09-06, Howard comments]
This is related to LWG 839.
Proposed resolution:
Section: 20.14.13.2.1 [func.wrap.func.con] Status: New Submitter: Barry Revzin Opened: 2016-09-14 Last modified: 2017-02-03
Priority: 3
View other active issues in [func.wrap.func.con].
View all other issues in [func.wrap.func.con].
View all issues with New status.
Discussion:
I think there's a minor defect in the std::function interface. The constructor template is:
template <class F> function(F f);
while the assignment operator template is
template <class F> function& operator=(F&& f);
The latter came about as a result of LWG 1288, but that one was dealing with a specific issue that wouldn't have affected the constructor. I think the constructor should also take f by forwarding reference, this saves a move in the lvalue/xvalue cases and is also just generally more consistent. Should just make sure that it's stored as std::decay_t<F> instead of F.
Is there any reason to favor a by-value constructor over a forwarding-reference constructor?Proposed resolution:
Section: 20.11.2.2.5 [util.smartptr.shared.obs] Status: New Submitter: Hans Boehm Opened: 2016-09-22 Last modified: 2017-02-03
Priority: 2
View other active issues in [util.smartptr.shared.obs].
View all other issues in [util.smartptr.shared.obs].
View all issues with New status.
Discussion:
The removal of the "debug only" restriction for use_count() and unique() in shared_ptr by LWG 2434 introduced a bug. In order for unique() to produce a useful and reliable value, it needs a synchronize clause to ensure that prior accesses through another reference are visible to the successful caller of unique(). Many current implementations use a relaxed load, and do not provide this guarantee, since it's not stated in the standard. For debug/hint usage that was OK. Without it the specification is unclear and probably misleading.
I would vote for making unique() use memory_order_acquire, and specifying that reference count decrement operations synchronize with unique(). That still doesn't give us sequential consistency by default, like we're supposed to have. But the violations seem sufficiently obscure that I think it's OK. All uses that anybody should care about will work correctly, and the bad uses are clearly bad. I agree with Peter that this version of unique() may be quite useful. I would prefer to specify use_count() as only providing an unreliable hint of the actual count (another way of saying debug only). Or deprecate it, as JF suggested. We can't make use_count() reliable without adding substantially more fencing. We really don't want someone waiting for use_count() == 2 to determine that another thread got that far. And unfortunately, I don't think we currently say anything to make it clear that's a mistake. This would imply that use_count() normally uses memory_order_relaxed, and unique is neither specified nor implemented in terms of use_count().Proposed resolution:
Section: 99 [networking.ts::buffer.reqmts.mutablebuffersequence] Status: LEWG Submitter: Vinnie Falco Opened: 2016-10-05 Last modified: 2017-03-05
Priority: Not Prioritized
View all issues with LEWG status.
Discussion:
We propose to relax the ForwardIterator requirements of buffer sequences in [networking.ts] by allowing buffer sequence iterators to return rvalues when dereferenced, and skip providing operator->. A paraphrased explanation of why the referential equality rules of ForwardIterator are harmful to the buffer sequence requirements comes from N4128, 3.3.7 "Ranges For The Standard Library":
The [networking.ts] dependence on ForwardIterator in the buffer sequence requirements ties together the traversal and access properties of iterators. For instance, no forward iterator may return an rvalue proxy when it is dereferenced; the ForwardIterator concept requires that unary operator* return an lvalue. This problem has serious consequences for lazy evaluation that applies transformations to buffer sequence elements on the fly. If the transformation function does not return an lvalue, the range's iterator can model no concept stronger than InputIterator, even if the resulting iterator could in theory support BidirectionalIterator. The result in practice is that most range adaptors today will not be compatible with [networking.ts], thereby limiting the types that [networking.ts] can be passed, for no good reason.
Consider a user defined function trim which lazily adapts a ConstBufferSequence, such that when iterating the buffers in the new sequence, each buffer appears one byte shorter than in the underlying sequence:
#include <boost/range/adaptor/transformed.hpp>
struct trim
{
using result_type = const_buffer;
result_type operator()(const_buffer b)
{
return const_buffer{b.data(), b.size() - 1};
}
};
template <ConstBufferSequence>
auto
trim(ConstBufferSequence const& buffers)
{
using namespace boost::adaptors;
return buffers | transformed(trim{});
}
trim returns a BidirectionalRange, whose const_iterator returns an rvalue when dereferenced. This breaks the requirements of ForwardIterator. A solution that meets the referential equality rules of ForwardIterator, would be to evaluate the transformed sequence upon construction (for example, by storing each transformed const_buffer in a vector). Unfortunately this work-around is more expensive since it would add heap allocation which the original example avoids.
The requirement of InputIterator operator-> is also unnecessary for buffer sequence iterators, and should be removed. Because [networking.ts] only requires that a buffer sequence iterator's value_type be convertible to const_buffer or mutable_buffer, implementations of [networking.ts] cannot assume the existence of any particular member functions or data members other than an implicit conversion to const_buffer or mutable_buffer. Removing the requirement for operator-> to be present, provides additional relief from the referential equality requirements of ForwardIterator and allows transformations of buffer sequences to meet the requirements of buffer sequences.
This proposal imposes no changes on existing implementations of [networking.ts]. It does not change anything in the standard. The proposal is precise, minimal, and allows range adapters to transform buffer sequences in optimized, compatible ways.
[Issues processing Telecon 2016-10-07]
Status set to LEWG
[2017-02-21, Jonathan comments]
The use of the term "strict aliasing" in the issue discussion is misleading as that refers to type-based alias analysis in compilers, but the rule for ForwardIterators is related to referential equality and not strict aliasing.
[2017-02-22, Vinnie Falco comments]
We have eliminated the use of the term "strict aliasing" from the discussion.
Proposed resolution:
This wording is relative to N4588.
Modify 16.2.1 [buffer.reqmts.mutablebuffersequence] as indicated:
An iterator type meeting the requirements for bidirectional iterators (C++Std [bidirectional.iterators]) whose value type is convertible to mutable_bufferAn iterator type whose reference type is convertible to mutable_buffer, and which satisfies all the requirements for bidirectional iterators (C++Std [bidirectional.iterators]) except that:
- there is no requirement that operator-> is provided, and
- there is no requirement that reference be a reference type.
Modify 16.2.2 [buffer.reqmts.constbuffersequence] as indicated:
An iterator type meeting the requirements for bidirectional iterators (C++Std [bidirectional.iterators]) whose value type is convertible to const_buffer.An iterator type whose reference type is convertible to const_buffer, and which satisfies all the requirements for bidirectional iterators (C++Std [bidirectional.iterators]) except that:
- there is no requirement that operator-> is provided, and
- there is no requirement that reference be a reference type.
Section: 21.4 [string.view] Status: LEWG Submitter: Billy Robert O'Neal III Opened: 2016-10-07 Last modified: 2017-02-03
Priority: 2
View other active issues in [string.view].
View all other issues in [string.view].
View all issues with LEWG status.
Discussion:
Now that we have C++14 relaxed constexpr, we can also take care of the assignment operator and copy.
[Issues processing Telecon 2016-10-07]
Split off from 2778
[2016-11-12, Issaquah]
Sat PM: Status LEWG
Proposed resolution:
This wording is relative to N4606.
In 21.4.2 [string.view.template], add constexpr to copy:
constexpr size_type copy(charT* s, size_type n, size_type pos = 0) const;
In 21.4.2.6 [string.view.ops], add constexpr to copy:
constexpr size_type copy(charT* s, size_type n, size_type pos = 0) const;
Section: 23.6.4.1 [queue.defn], 23.6.6.1 [stack.defn] Status: Tentatively Ready Submitter: Jonathan Wakely Opened: 2016-10-14 Last modified: 2017-03-13
Priority: 2
View all issues with Tentatively Ready status.
Discussion:
The stack and queue adaptors are now defined as:
template <class... Args>
reference emplace(Args&&... args) { return c.emplace_back(std::forward<Args>(args)...); }
This breaks any code using queue<UserDefinedSequence> or stack<UserDefinedSequence> until the user-defined containers are updated to meet the new C++17 requirements.
If we defined them as returning decltype(auto) then we don't break any code. When used with std::vector or std::deque they will return reference, as required, but when used with C++14-conforming containers they will return void, as before.[2016-11-12, Issaquah]
Sat AM: P2
[2017-03-04, Kona]
Status to Tentatively Ready.
Proposed resolution:
This wording is relative to N4606.
Change return type of emplace in class definition in 23.6.4.1 [queue.defn]:
template <class... Args>referencedecltype(auto) emplace(Args&&... args) { return c.emplace_back(std::forward<Args>(args)...); }
Change return type of emplace in class definition in 23.6.6.1 [stack.defn]:
template <class... Args>referencedecltype(auto) emplace(Args&&... args) { return c.emplace_back(std::forward<Args>(args)...); }
Section: 20.15.2 [meta.type.synop] Status: Open Submitter: Russia Opened: 2016-11-09 Last modified: 2017-06-15
Priority: 2
View other active issues in [meta.type.synop].
View all other issues in [meta.type.synop].
View all issues with Open status.
Discussion:
Addresses RU 2
Failed prerequirement for the type trait must result in ill-formed program. Otherwise hard detectable errors will happen:
#include <type_traits>
struct foo;
void damage_type_trait() {
// must be ill-formed
std::is_constructible<foo, foo>::value;
}
struct foo{};
int main() {
static_assert(
// produces invalid result
std::is_constructible<foo, foo>::value,
"foo must be constructible from foo"
);
}
Suggested resolution:
Add to the end of the [meta.type.synop] section:Program is ill-formed if precondition for the type trait is violated.
[2016-11-09, Jonathan provides wording]
[Issues Telecon 16-Dec-2016]
Priority 2
[2017-03-02, Daniel comments]
LWG 2939 has been created to signal that some of our current type trait constraints are not quite correct and I recommend not to enforce the required diagnostics for traits that are sensitive to mismatches of the current approximate rules.
[2017-03-03, Kona Friday morning]
Unanimous consent to adopt this for C++17, but due to a misunderstanding, it wasn't on the ballot
Setting status to 'Ready' so we'll get it in immediately post-C++17
[2017-06-15 request from Daniel]
I don't believe that this should be "Ready"; I added the extra note to LWG 2797 *and* added the new issue 2939 exactly to *prevent* 2797 being accepted for C++17
Setting status back to 'Open'
Proposed resolution:
This wording is relative to N4606.
Add a new paragraph after 20.15.4.3 [meta.unary.prop] paragraph 3:
-?- If an instantiation of any template declared in this subclause fails to meet the preconditions, the program is ill-formed.
Change the specification for alignment_of in Table 39 in 20.15.5 [meta.unary.prop.query]:
Table 39 — Type property queries template <class T> struct alignment_of; alignof(T).
Requires: alignof(T) shall be a valid expression (5.3.6), otherwise the program is ill-formed
Change the specification for is_base_of, is_convertible, is_callable, and is_nothrow_callable in Table 40 in 20.15.6 [meta.rel]:
Table 40 — Type relationship predicates […] template <class Base, class
Derived>
struct is_base_of;[…] If Base and Derived are
non-union class types and are
different types (ignoring possible
cv-qualifiers) then Derived shall
be a complete type, otherwise the program is ill-formed.
[Note: Base classes that
are private, protected, or ambiguous are,
nonetheless, base classes. — end note]template <class From, class To>
struct is_convertible;see below From and To shall be complete
types, arrays of unknown bound,
or (possibly cv-qualified) void
types, otherwise the program is
ill-formed.template <class Fn, class...
ArgTypes, class R>
struct is_callable<
Fn(ArgTypes...), R>;[…] Fn, R, and all types in the
parameter pack ArgTypes shall
be complete types, (possibly
cv-qualified) void, or arrays of
unknown bound,
otherwise the program is ill-formed.template <class Fn, class...
ArgTypes, class R>
struct is_nothrow_callable<
Fn(ArgTypes...), R>;[…] Fn, R, and all types in the
parameter pack ArgTypes shall
be complete types, (possibly
cv-qualified) void, or arrays of
unknown bound,
otherwise the program is ill-formed.
Add a new paragraph after 20.15.7 [meta.trans] paragraph 2:
-2- Each of the templates in this subclause shall be a TransformationTrait (20.15.1).
-?- If an instantiation of any template declared in this subclause fails to meet the Requires: preconditions, the program is ill-formed.
Section: 30.6.8 [futures.shared_future] Status: New Submitter: Zhihao Yuan Opened: 2016-11-11 Last modified: 2017-02-03
Priority: 2
View all other issues in [futures.shared_future].
View all issues with New status.
Discussion:
Addresses GB 62
There is an implicit precondition on most shared_future operations that valid() == true, 30.6.7p3. The list of exempted functions seems copied directly from class future, and would also include copy operations for shared_futures, which are copyable. Similarly, this would be a wide contract that cannot throw, so those members would be marked noexcept.Suggested resolution:
Revise p3: "The effect of calling any member function other than the move constructor, the copy constructor, the destructor, the move-assignment operator, the copy-assignment operator, or valid() on a shared_future object for which valid() == false is undefined." […] Add noexcept specification to the copy constructor and copy-assignment operator, in the class definition and where those members are specified.[2016-11-10, Zhihao Yuan comments and provides wording]
This resolution should close LWG 2697 as well, but that one was filed against concurrent TS.
[Issues Telecon 16-Dec-2016]
This is also addressed (but in a slightly different way by P0516R0
Proposed resolution:
This wording is relative to N4606.
[Drafting note: This change supersedes the 1st among 3 edits in LWG 2556, please omit that edit and apply the one below.]
Modify 30.6.7 [futures.unique_future]/3 as follows:
-3- The effect of calling
any member function other than the destructor, the move-assignment operator, or validthe member functions get, wait, wait_for, or wait_until on a future object for which valid() == false is undefined. [Note: Implementations are encouraged to detect this case and throw an object of type future_error with an error condition of future_errc::no_state. — end note]
Change 30.6.8 [futures.shared_future] as indicated:
-3- The effect of calling
any member function other than the destructor, the move-assignment operator, or valid()the member functions get, wait, wait_for, or wait_until on a shared_future object for which valid() == false is undefined. [Note: Implementations are encouraged to detect this case and throw an object of type future_error with an error condition of future_errc::no_state. — end note][…]namespace std { template <class R> class shared_future { public: shared_future() noexcept; shared_future(const shared_future& rhs) noexcept; shared_future(future<R>&&) noexcept; shared_future(shared_future&& rhs) noexcept; ~shared_future(); shared_future& operator=(const shared_future& rhs) noexcept; shared_future& operator=(shared_future&& rhs) noexcept; […] }; […] }shared_future(const shared_future& rhs) noexcept;[…]-7- Effects: […]
shared_future& operator=(const shared_future& rhs) noexcept;-14- Effects: […]
Section: 20.2.3 [utility.swap] Status: New Submitter: United States Opened: 2016-11-09 Last modified: 2017-02-20
Priority: 3
View other active issues in [utility.swap].
View all other issues in [utility.swap].
View all issues with New status.
Discussion:
Addresses US 108
swap is a critical function in the standard library, and should be declared constexpr to support more widespread support for constexpr in libraries. This was proposed in P0202R1 which was reviewed favourably at Oulu, but the widespread changes to the <algorithm> header were too risky and unproven for C++17. We should not lose constexpr support for the much simpler (and more important) <utility> functions because they were attached to a larger paper. Similarly, the fundamental value wrappers, pair and tuple, should have constexpr swap functions, and the same should be considered for optional and variant. It is not possible to mark swap for std::array as constexpr without adopting the rest of the P0202R1 though, or rewriting the specification for array swap to not use swap_ranges.Suggested resolution:
Adopt the changes to the <utility> header proposed in P0202R1, i.e., only bullets C, D, and E. In addition, mark the swap functions of pair and tuple as constexpr, and consider doing the same for optional and variant.[Issues Telecon 16-Dec-2016]
Priority 3
Proposed resolution:
Section: 27.5.4.2 [fpos.operations] Status: New Submitter: Great Britain Opened: 2016-11-09 Last modified: 2017-02-20
Priority: 4
View other active issues in [fpos.operations].
View all other issues in [fpos.operations].
View all issues with New status.
Discussion:
Addresses GB 60
The requirements on the stateT type used to instantiate class template fpos are not clear, and the following Table 108 — "Position type requirements" is a bit of a mess. This is old wording, and should be cleaned up with better terminology from the Clause 17 Requirements. For example, stateT might be require CopyConstructible?, CopyAssignable?, and Destructible. Several entries in the final column of the table appear to be post-conditions, but without the post markup to clarify they are not assertions or preconditions. They frequently refer to identifiers that do not apply to all entries in their corresponding Expression column, leaving some expressions without a clearly defined semantic. If stateT is a trivial type, is fpos also a trivial type, or is a default constructor not required/supported?
Proposed change:
Clarify the requirements and the table.[Issues Telecon 16-Dec-2016]
Priority 4; no consensus for any concrete change
Proposed resolution:
Section: 20.6.3.1 [optional.ctor], 20.6.3.3 [optional.assign], 20.7.3.1 [variant.ctor], 20.8.3.1 [any.cons], 20.8.3.3 [any.modifiers] Status: New Submitter: Tim Song Opened: 2016-10-29 Last modified: 2017-02-03
Priority: 3
View all other issues in [optional.ctor].
View all issues with New status.
Discussion:
Throughout optional/variant/any's specification references are made to "the selected constructor of T". For example, 20.6.3.1 [optional.ctor]/16 says of the constructor from const T&:
-16- Remarks: If T's selected constructor is a constexpr constructor, this constructor shall be a constexpr constructor.
Similarly, the in-place constructor has this wording (20.6.3.1 [optional.ctor]/25-26):
-25- Throws: Any exception thrown by the selected constructor of T.
-26- Remarks: If T's constructor selected for the initialization is a constexpr constructor, this constructor shall be a constexpr constructor.
If T is a scalar type, it has no constructor at all. Moreover, even for class types, the in-place constructor wording ignores any implicit conversion done on the argument before the selected constructor is called, which 1) may not be valid in constant expressions and 2) may throw an exception; such exceptions aren't thrown "by the selected constructor of T" but outside it.
The wording should probably be recast to refer to the entire initialization.[Issues Telecon 16-Dec-2016]
Priority 3; Jonathan to provide wording.
Proposed resolution:
Section: 20.14.13.2.1 [func.wrap.func.con] Status: New Submitter: Brian Bi Opened: 2016-11-03 Last modified: 2017-02-03
Priority: 2
View other active issues in [func.wrap.func.con].
View all other issues in [func.wrap.func.con].
View all issues with New status.
Discussion:
If a std::function has a reference as a return type, and that reference binds to a prvalue returned by the callable that it wraps, then the reference is always dangling. Because any use of such a reference results in undefined behaviour, the std::function should not be allowed to be initialized with such a callable. Instead, the program should be ill-formed.
A minimal example of well-formed code under the current standard that exhibits this issue:
#include <functional>
int main()
{
std::function<const int&()> F([]{ return 42; });
int x = F(); // oops!
}
[2016-11-22, David Krauss comments and suggests wording]
Indirect bindings may also introduce temporaries inside std::function, e.g.:
void f(std::function<long const&()>); // Retains an observer to a long.
void g() {
int v;
f([&]()->int& { return v; } ); // int lvalue binds to long const& through a temporary.
}
A fix has been implemented. Conversions that may be conversion operators are allowed, though, because those can produce legitimate glvalues. Before adopting this, it need to be considered considered whether there should be SFINAE or a hard error.
[Issues Telecon 16-Dec-2016]
Priority 2
Proposed resolution:
This wording is relative to N4618.
Add a second paragraph to the remarks section of 20.14.13.2.1 [func.wrap.func.con]:
template<class F> function(F f);-7- Requires: F shall be CopyConstructible.
-8- Remarks: This constructor template shall not participate in overload resolution unless
F is Lvalue-Callable (20.14.13.2 [func.wrap.func]) for argument types ArgTypes... and return type R, and
If R is type "reference to T" and INVOKE(ArgTypes...) has value category V and type U:
V is a prvalue, U is a class type, and T is not reference-related (8.6.3 [dcl.init.ref]) to U, and
V is an lvalue or xvalue, and either U is a class type or T is reference-related to U.
[…]
Section: 99 [fund.ts.v2::func.wrap.func.con] Status: LEWG Submitter: Zhihao Yuan Opened: 2016-11-07 Last modified: 2017-06-15
Priority: 3
View all other issues in [fund.ts.v2::func.wrap.func.con].
View all issues with LEWG status.
Discussion:
Addresses: fund.ts.v2
C++ doesn't have a prvalue expression of array type, but rvalue arrays can still come from different kinds of sources:
C99 compound literals (int[]) {2, 4},
std::move(arr),
Deduction to_array<int const>({ 2, 4 }).
See also CWG 1591: Deducing array bound and element type from initializer list.
For 3), users are "abusing" to_array to get access to uniform initialization to benefit from initializing elements through braced-init-list and/or better narrowing conversion support.
We should just add rvalue reference support to to_array.[Issues Telecon 16-Dec-2016]
Status to LEWG
[2017-02 in Kona, LEWG responds]
Would like a small paper; see examples before and after. How does this affect overload resolution?
[2017-06-02 Issues Telecon]
Leave status as LEWG; priority 3
Proposed resolution:
This wording is relative to N4600.
Add the following signature to [header.array.synop]:
// 9.2.2, Array creation functions template <class D = void, class... Types> constexpr array<VT, sizeof...(Types)> make_array(Types&&... t); template <class T, size_t N> constexpr array<remove_cv_t<T>, N> to_array(T (&a)[N]); template <class T, size_t N> constexpr array<remove_cv_t<T>, N> to_array(T (&&a)[N]);
Modify [container.array.creation] as follows:
template <class T, size_t N> constexpr array<remove_cv_t<T>, N> to_array(T (&a)[N]); template <class T, size_t N> constexpr array<remove_cv_t<T>, N> to_array(T (&&a)[N]);-6- Returns: For all i, 0 ≤ i < N, a
An array<remove_cv_t<T>, N>such that each element is copy-initialized with the corresponding element of ainitialized with { a[i]... } for the first form, or { std::move(a[i])... } for the second form..
Section: 18.5 [support.start.term] Status: New Submitter: Jean-François Bastien Opened: 2016-11-07 Last modified: 2017-05-31
Priority: 3
View all other issues in [support.start.term].
View all issues with New status.
Discussion:
While SG1 was processing NB comments CA1 and LATE2 regarding P0270R1, we decided to remove the proposed guarantee that quick_exit be made signal safe.
Our reasoning is that functions registered with at_quick_exit aren't forbidden from calling quick_exit, but quick_exit implementations likely acquire some form of a lock before processing all registered functions (because a note forbids the implementation from introducing data races). The following code can therefore deadlock:
#include <cstdlib>
int main()
{
std::at_quick_exit([] () { std::quick_exit(0); });
std::quick_exit(1);
return 0;
}
The same applies if a function registered in at_quick_exit handles a signal, and that signal calls quick_exit. SG1 believes that both issues (same thread deadlock, and signal deadlock) can be resolved in the same manner. Either:
Option 2. seems preferable, and can be implemented along the lines of:
#include <array>
#include <atomic>
#include <cstddef>
namespace {
typedef void (*func)();
std::array<func, 32> quick_exit_functions;
const auto* quick_exit_functions_ptr = &quick_exit_functions;
std::atomic_flag lock = ATOMIC_FLAG_INIT;
struct scope
{
scope() { while (lock.test_and_set(std::memory_order_acquire)) ; }
~scope() { lock.clear(std::memory_order_release); }
};
}
namespace std {
extern "C" void quick_exit(int status) noexcept
{
decltype(quick_exit_functions_ptr) f;
{
scope s;
f = quick_exit_functions_ptr;
quick_exit_functions_ptr = nullptr;
}
if (f) {
size_t pos = f->size();
while (pos > 0)
(*f)[--pos]();
}
_Exit(status);
}
extern "C++" int at_quick_exit(func f) noexcept
{
scope s;
if (!quick_exit_functions_ptr || quick_exit_functions.size() == quick_exit_functions.max_size())
return -1;
quick_exit_functions[quick_exit_functions.size()] = f;
return 0;
}
}
Ideally, the resolution would also add back the wording which SG1 dropped from P0270R1:
Add at new element to the end of 18.5 [support.start.term] p13 (quick_exit()):
Remarks: The function quick_exit() is signal-safe (18.10.5 [csignal.syn]). [Note: It might still be unsafe to call quick_exit() from a handler, because the functions registered with at_quick_exit() might not be signal-safe. — end note]
[Issues Telecon 16-Dec-2016]
Priority 3
Proposed resolution:
This wording is relative to N4606.
Add at new element to the end of 18.5 [support.start.term] p13 (quick_exit()):
[[noreturn]] void quick_exit(int status) noexcept;-13- Effects: Functions registered by calls to at_quick_exit are called in the reverse order of their registration, except that a function shall be called after any previously registered functions that had already been called at the time it was registered. Objects shall not be destroyed as a result of calling quick_exit. If control leaves a registered function called by quick_exit because the function does not provide a handler for a thrown exception, std::terminate() shall be called. [Note: at_quick_exit may call a registered function from a different thread than the one that registered it, so registered functions should not rely on the identity of objects with thread storage duration. — end note] After calling registered functions, quick_exit shall call _Exit(status). [Note: The standard file buffers are not flushed. See: ISO C 7.22.4.5. — end note]
-?- Remarks: The function quick_exit() is signal-safe (18.10.5 [csignal.syn]). [Note: It might still be unsafe to call quick_exit() from a handler, because the functions registered with at_quick_exit() might not be signal-safe. — end note]
Section: 27.10.15.33 [fs.op.resize_file] Status: New Submitter: Richard Smith Opened: 2016-11-07 Last modified: 2017-02-03
Priority: 3
View all issues with New status.
Discussion:
resize_file has this postcondition (after resolving late comment 42, see P0489R0):
Postcondition: file_size(p) == new_size.
This is impossible for an implementation to satisfy, due to the possibility of file system races. This is not actually a postcondition; rather, it is an effect that need no longer hold when the function returns.
[Issues Telecon 16-Dec-2016]
Priority 3
Proposed resolution:
Section: 17.5.1.1 [contents] Status: New Submitter: Tim Song Opened: 2016-11-11 Last modified: 2017-02-03
Priority: 2
View all other issues in [contents].
View all issues with New status.
Discussion:
[contents]/3 says
Whenever a name x defined in the standard library is mentioned, the name x is assumed to be fully qualified as ::std::x, unless explicitly described otherwise. For example, if the Effects section for library function F is described as calling library function G, the function ::std::G is meant.
With the introduction of nested namespaces inside std, this rule needs tweaking. For instance, time_point_cast's Returns clause says "time_point<Clock, ToDuration>(duration_cast<ToDuration>(t.time_since_epoch()))"; that reference to duration_cast obviously means ::std::chrono::duration_cast, not ::std::duration_cast, which doesn't exist.
[Issues Telecon 16-Dec-2016]
Priority 2; Jonathan to provide wording
Proposed resolution:
Section: 30.2.5 [thread.req.lockable], 30.4.3 [thread.mutex.requirements] Status: New Submitter: Agustín K-ballo Bergé Opened: 2016-11-12 Last modified: 2017-02-03
Priority: 3
View all issues with New status.
Discussion:
The current draft contains 14 occurrences of a Return type: clause. That clause is not covered by 17.4.1.4 [structure.specifications] p3. This was reported as editorial request #266.
[Issues Telecon 16-Dec-2016]
Priority 3; Jonathan to provide wording.
Proposed resolution:
Section: 18.4 [cstdint] Status: New Submitter: Thomas Koeppe Opened: 2016-11-12 Last modified: 2017-05-31
Priority: 3
View all other issues in [cstdint].
View all issues with New status.
Discussion:
I would like clarification from LWG regarding the various limit macros like INT_8_MIN in <cstdint>, in pursuit of editorial cleanup of this header's synopsis. I have two questions:
At present, macros like INT_8_MIN that correspond to the optional type int8_t are required (unconditionally), whereas the underlying type to which they appertain is only optional. Is this deliberate? Should the macros also be optional?
Is it deliberate that C++ only specifies sized aliases for the sizes 8, 16, 32 and 64, whereas the corresponding C header allows type aliases and macros for arbitrary sizes for implementations that choose to provide extended integer types? Is the C++ wording more restrictive by accident?
[2017-01-27 Telecon]
Priority 3
[2017-03-04, Kona]
C11 ties the macro names to the existence of the types. Marshall to research the second question.
Close 2764 as a duplicate of this issue.
[2017-03-18, Thomas comments and provides wording]
This is as close as I can get to the C wording without resolving part (a) of the issue (whether we deliberately don't allow sized type aliases for sizes other than 8, 16, 32, 64, a departure from C). Once we resolve part (a), we need to revisit <cinttypes> and fix up the synopsis (perhaps to get rid of N) and add similar wording as the one below to make the formatting macros for the fixed-width types optional. For historical interest, this issue is related to LWG 553 and LWG 841.
Proposed resolution:
This wording is relative to N4640.
Append the following content to 18.4.1 [cstdint.syn] p2:
-2- The header defines all types and macros the same as the C standard library header <stdint.h>. In particular, for each of the fixed-width types (int8_t, int16_t, int32_t, int64_t, uint8_t, uint16_t, uint32_t, uint64_t) the type alias and the corresponding limit macros are defined if and only if the implementation provides the corresponding type.
Section: 18.6.4 [ptr.launder] Status: New Submitter: Tony van Eerd Opened: 2016-11-13 Last modified: 2017-03-13
Priority: 3
View other active issues in [ptr.launder].
View all other issues in [ptr.launder].
View all issues with New status.
Discussion:
As pointed out by Nevin: A use of std::launder that does not make use of its return value is always pointless; the function has no side effects.
[2017-01-27 Telecon]
Priority 3; Nico's upcoming paper P0532 should address this and other issues around launder.
[2017-03-04, Kona]
This should be handled post-C++17 by Nico's paper P0600.
Proposed resolution:
Section: 23.3.7.1 [array.overview] Status: New Submitter: Robert Haberlach Opened: 2016-11-16 Last modified: 2017-02-03
Priority: 2
View other active issues in [array.overview].
View all other issues in [array.overview].
View all issues with New status.
Discussion:
LWG 2590's resolution is incomplete:
std::array<int, 1> arr{{0}};
should be fine, but isn't guaranteed, since {0} has no type. We should rather go for implicit conversion:
An array is an aggregate (8.6.1 [dcl.init.aggr]) that can be list-initialized with up to N elements
whose types are convertible to Tthat can be implicitly converted to T.
[2016-11-26, Tim Song comments]
This is not possible as written, because due to the brace elision rules for aggregate initialization, std::array<int, 2> arr{{0}, {1}}; will never work: the {0} is taken as initializing the inner array, and the {1} causes an error.
[2017-01-27 Telecon]
Priority 2; consensus is that the P/R is not quite right.
Proposed resolution:
This wording is relative to N4606.
Change 23.3.7.1 [array.overview] p2 as indicated:
-2- An array is an aggregate (8.6.1 [dcl.init.aggr]) that can be list-initialized with up to N elements
whose types are convertiblethat can be implicitly converted to T.
Section: 20.6.3 [optional.optional] Status: LEWG Submitter: Richard Smith Opened: 2016-11-24 Last modified: 2017-02-03
Priority: 2
View other active issues in [optional.optional].
View all other issues in [optional.optional].
View all issues with LEWG status.
Discussion:
LWG 2756 applies these changes:
constexpr optional(const T&); constexpr optional(T&&);template <class... Args> constexpr explicit optional(in_place_t, Args&&...); template <class U, class... Args> constexpr explicit optional(in_place_t, initializer_list<U>, Args&&...); template <class U = T> EXPLICIT constexpr optional(U&&); template <class U> EXPLICIT optional(const optional<U>&); template <class U> EXPLICIT optional(optional<U>&&);
These break the ability for optional to perform class template argument deduction, as there is now no way to map from optional's argument to the template parameter T.
[2017-01-27 Telecon]
Priority 2
[2017-01-30 Ville comments:]
Seems like the problem is resolved by a simple deduction guide:
template <class T> optional(T) -> optional<T>;
The paper p0433r0 seems to suggest a different guide,
template<class T> optional(T&& t) -> optional<remove_reference_t<T>>;
but I don't think the paper is up to speed with LWG 2756. There's no reason to use such an universal reference in the guide and remove_reference in its target, just guide with T, with the target optional<T>, optional's constructors do the right thing once the type has been deduced.
Proposed resolution:
Section: 20.15.4.3 [meta.unary.prop] Status: New Submitter: Richard Smith Opened: 2016-11-17 Last modified: 2017-02-03
Priority: 3
View other active issues in [meta.unary.prop].
View all other issues in [meta.unary.prop].
View all issues with New status.
Discussion:
struct S
{
~S(); // non-trivial
};
static_assert(std::is_trivially_constructible<S>::value, "");
Should the assert pass? Implementations disagree.
Per 20.15.4.3 [meta.unary.prop]'s Table 38, this trait looks at whether the following variable definition is known to call no operation that is not trivial:S t(create<Args>()...);
... where Args is an empty pack in this case. That variable definition results in a call to the S destructor. Should that call be considered by the trait?
[2017-01-27 Telecon]
Priority 3
This issue interacts with 2116
Proposed resolution:
Section: 20.6.3.5 [optional.observe] Status: New Submitter: Richard Smith Opened: 2016-11-24 Last modified: 2017-02-03
Priority: 2
View all other issues in [optional.observe].
View all issues with New status.
Discussion:
After applying LWG 2740, we have:
constexpr const T* operator->() const; constexpr T* operator->();-1- Requires: *this contains a value.
-2- Returns: val. -3- Throws: Nothing. -4- Remarks: These functions shall be constexpr functions.
Paragraph 4 is completely superfluous. We already said these functions were constexpr in the synopsis. Can it be removed?
[Issues Telecon 16-Dec-2016]
Priority 2
Jonathan notes: Although Richard is correct, I suggest we don't strike the paragraph, so that we remember to fix it as part of 2833, when we know how to say this properly.
Proposed resolution:
Section: 23.2.7 [unord.req] Status: LEWG Submitter: Daniel James Opened: 2016-11-24 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [unord.req].
View all other issues in [unord.req].
View all issues with LEWG status.
Discussion:
In 23.2.7 [unord.req] paragraph 12, it says that the behaviour of operator== is undefined unless the Hash and Pred function objects respectively have the same behaviour. This makes comparing containers with randomized hashes with different seeds undefined behaviour, but I think that's a valid use case. It's not much more difficult to support it when the Hash function objects behave differently. I did a little testing and both libstdc++ and libc++ appear to support this correctly.
I suggest changing the appropriate sentence in 23.2.7 [unord.req] paragraph 12: "The behavior of a program that uses operator== or operator!= on unordered containers is undefined unless the[2017-01-27 Telecon]
This is a design issue; send to LEWG
Proposed resolution:
This wording is relative to N4618.
Change 23.2.7 [unord.req] as indicated:
-12- Two unordered containers a and b compare equal if a.size() == b.size() and, for every equivalent-key group [Ea1, Ea2) obtained from a.equal_range(Ea1), there exists an equivalent-key group [Eb1, Eb2) obtained from b.equal_range(Ea1), such that is_permutation(Ea1, Ea2, Eb1, Eb2) returns true. For unordered_set and unordered_map, the complexity of operator== (i.e., the number of calls to the == operator of the value_type, to the predicate returned by key_eq(), and to the hasher returned by hash_function()) is proportional to N in the average case and to N2 in the worst case, where N is a.size(). For unordered_multiset and unordered_multimap, the complexity of operator== is proportional to ∑ Ei2 in the average case and to N2 in the worst case, where N is a.size(), and Ei is the size of the ith equivalent-key group in a. However, if the respective elements of each corresponding pair of equivalent-key groups Eai and Ebi are arranged in the same order (as is commonly the case, e.g., if a and b are unmodified copies of the same container), then the average-case complexity for unordered_multiset and unordered_multimap becomes proportional to N (but worst-case complexity remains 𝒪(N2), e.g., for a pathologically bad hash function). The behavior of a program that uses operator== or operator!= on unordered containers is undefined unless the
Hash andPred function objects respectively havehas the same behavior for both containers and the equality comparison operator for Key is a refinement(footnote 258) of the partition into equivalent-key groups produced by Pred.
Section: 27.5.4.2 [fpos.operations] Status: New Submitter: Jens Maurer Opened: 2016-11-24 Last modified: 2017-02-03
Priority: 3
View other active issues in [fpos.operations].
View all other issues in [fpos.operations].
View all issues with New status.
Discussion:
This is from editorial issue #1031.
The first row in Table 112 "Position type requirements" talks about the expression P(i) and then has an assertion p == P(i). However, there are no constraints on p other than being of type P, so (on the face of it) this seems to require that operator== on P always returns true, which is non-sensical.[2017-01-27 Telecon]
Priority 3
Proposed resolution:
Section: 20.7.3.1 [variant.ctor] Status: New Submitter: Richard Smith Opened: 2016-11-28 Last modified: 2017-03-05
Priority: 2
View all other issues in [variant.ctor].
View all issues with New status.
Discussion:
The library has lots of functions declared constexpr, but it's not clear what that means. The constexpr keyword implies that there needs to be some invocation of the function, for some set of template arguments and function arguments, that is valid in a constant expression (otherwise the program would be ill-formed, with no diagnostic required), along with a few side conditions. I suspect the library intends to require something a lot stronger than that from implementations (something along the lines of "all calls that could reasonably be constant subexpressions are in fact constant subexpressions, unless otherwise stated").
[variant.ctor]/1 contains this, which should also be fixed:"This function shall be constexpr if and only if the value-initialization of the alternative type T0 would satisfy the requirements for a constexpr function."
This is the wrong constraint: instead of constraining whether the function is constexpr, we should constrain whether a call to it is a constant subexpression.
Daniel:
This is has some considerable overlap with LWG 2289 but is phrased in a more general way.[2016-12-16, Issues Telecon]
Priority 2; this is also the general case of 2829.
[2017-02-20, Alisdair comments and suggests concrete wording]
Below is is draft wording I was working on at Issaquah to try to address both issues.
Proposed resolution:
This wording is relative to N4640.
Modify 17.5.5.6 [constexpr.functions] as indicated:
17.6.5.6 constexpr functions and constructors [constexpr.functions]
-1- This International Standard explicitly requires that certain standard library functions are constexpr (7.1.5 [dcl.constexpr]). If the specification for a templated entity requires that it shall be a constexpr templated entity, then that templated entity shall be usable in a constant expression.. An implementation
shall notmay declareanyadditional standard library function signature as constexprexcept for those where it is explicitly required. Within any header that provides any non-defining declarations of constexpr functions or constructors an implementation shall provide corresponding definitions.
Section: 21.3.2 [basic.string], 21.3.2.7.2 [string.find], 21.3.2.7.3 [string.rfind], 21.3.2.7.4 [string.find.first.of], 21.3.2.7.5 [string.find.last.of], 21.3.2.7.6 [string.find.first.not.of], 21.3.2.7.7 [string.find.last.not.of], 21.3.2.7.9 [string.compare] Status: New Submitter: Jonathan Wakely Opened: 2016-12-05 Last modified: 2017-02-03
Priority: 2
View other active issues in [basic.string].
View all other issues in [basic.string].
View all issues with New status.
Discussion:
Currently some overloads of basic_string::find are noexcept and some are not. Historically this was because some were specified in terms of constructing a temporary basic_string, which could throw. In practice creating a temporary (and potentially allocating memory) is a silly implementation, and so they could be noexcept. In the C++17 draft most of them have been changed to create a temporary basic_string_view instead, which can't throw anyway (P0254R2 made those changes).
This is confusing for users, as they need to carefully check which overload their code will resolve to, and consider whether that overload can throw. Refactoring code can change whether it calls a throwing or non-throwing overload. This is an unnecessary burden on users when realistically none of the functions will ever throw. The find, rfind, find_first_of, find_last_of, find_first_not_of and find_last_not_of overloads that are defined in terms of basic_string_view should be noexcept, or "Throws: Nothing." for the ones with narrow contracts (even though those narrow contracts are not enforcable or testable). The remaining overloads that are still specified in terms of a temporary string could also be noexcept. They construct basic_string of length one (which won't throw for an SSO implementation anyway), but can easily be defined in terms of basic_string_view instead. There's one basic_string::compare overload that is still defined in terms of a temporary basic_string, which should be basic_string_view and so can also be noexcept (the other compare overloads can throw out_of_range).[2016-12-15, Tim Song comments]
The following overloads invoking basic_string_view<charT>(s, n) are implicitly narrow-contract (the basic_string_view constructor requires [s, s+n) to be a valid range) and should be "Throws: Nothing" rather than noexcept:
size_type find(const charT* s, size_type pos, size_type n) const; size_type rfind(const charT* s, size_type pos, size_type n) const; size_type find_first_of(const charT* s, size_type pos, size_type n) const; size_type find_last_of(const charT* s, size_type pos, size_type n) const; size_type find_first_not_of(const charT* s, size_type pos, size_type n) const; size_type find_last_not_of(const charT* s, size_type pos, size_type n) const;
Similarly, the basic_string_view constructor invoked by this overload of compare requires [s, s + traits::length(s)) to be a valid range, and so it should be "Throws: Nothing" rather than noexcept:
int compare(const charT* s) const;
[2017-01-27 Telecon]
Priority 2; Jonathan to provide updated wording
Proposed resolution:
This wording is relative to N4618.
Change class template synopsis std::basic_string, 21.3.2 [basic.string], as indicated:
size_type find (basic_string_view<charT, traits> sv,
size_type pos = 0) const noexcept;
size_type find (const basic_string& str, size_type pos = 0) const noexcept;
size_type find (const charT* s, size_type pos, size_type n) const noexcept;
size_type find (const charT* s, size_type pos = 0) const;
size_type find (charT c, size_type pos = 0) const noexcept;
size_type rfind(basic_string_view<charT, traits> sv,
size_type pos = npos) const noexcept;
size_type rfind(const basic_string& str, size_type pos = npos) const noexcept;
size_type rfind(const charT* s, size_type pos, size_type n) const noexcept;
size_type rfind(const charT* s, size_type pos = npos) const;
size_type rfind(charT c, size_type pos = npos) const noexcept;
size_type find_first_of(basic_string_view<charT, traits> sv,
size_type pos = 0) const noexcept;
size_type find_first_of(const basic_string& str,
size_type pos = 0) const noexcept;
size_type find_first_of(const charT* s,
size_type pos, size_type n) const noexcept;
size_type find_first_of(const charT* s, size_type pos = 0) const;
size_type find_first_of(charT c, size_type pos = 0) const noexcept;
size_type find_last_of (basic_string_view<charT, traits> sv,
size_type pos = npos) const noexcept;
size_type find_last_of (const basic_string& str,
size_type pos = npos) const noexcept;
size_type find_last_of (const charT* s,
size_type pos, size_type n) const noexcept;
size_type find_last_of (const charT* s, size_type pos = npos) const;
size_type find_last_of (charT c, size_type pos = npos) const noexcept;
size_type find_first_not_of(basic_string_view<charT, traits> sv,
size_type pos = 0) const noexcept;
size_type find_first_not_of(const basic_string& str,
size_type pos = 0) const noexcept;
size_type find_first_not_of(const charT* s, size_type pos,
size_type n) const noexcept;
size_type find_first_not_of(const charT* s, size_type pos = 0) const;
size_type find_first_not_of(charT c, size_type pos = 0) const noexcept;
size_type find_last_not_of (basic_string_view<charT, traits> sv,
size_type pos = npos) const noexcept;
size_type find_last_not_of (const basic_string& str,
size_type pos = npos) const noexcept;
size_type find_last_not_of (const charT* s, size_type pos,
size_type n) const noexcept;
size_type find_last_not_of (const charT* s,
size_type pos = npos) const;
size_type find_last_not_of (charT c, size_type pos = npos) const noexcept;
basic_string substr(size_type pos = 0, size_type n = npos) const;
int compare(basic_string_view<charT, traits> sv) const noexcept;
int compare(size_type pos1, size_type n1,
basic_string_view<charT, traits> sv) const;
template<class T>
int compare(size_type pos1, size_type n1, const T& t,
size_type pos2, size_type n2 = npos) const;
int compare(const basic_string& str) const noexcept;
int compare(size_type pos1, size_type n1,
const basic_string& str) const;
int compare(size_type pos1, size_type n1,
const basic_string& str,
size_type pos2, size_type n2 = npos) const;
int compare(const charT* s) const noexcept;
int compare(size_type pos1, size_type n1,
const charT* s) const;
int compare(size_type pos1, size_type n1,
const charT* s, size_type n2) const;
Change 21.3.2.7.2 [string.find] as indicated:
size_type find(const charT* s, size_type pos, size_type n) const noexcept;-5- Returns: find(basic_string_view<charT, traits>(s, n), pos).
size_type find(const charT* s, size_type pos = 0) const;-6- Requires: s points to an array of at least traits::length(s) + 1 elements of charT.
-7- Returns: find(basic_string_view<charT, traits>(s), pos). -?- Throws: Nothing.size_type find(charT c, size_type pos = 0) const noexcept;-8- Returns: find(basic_string_view<charT, traits>(addressof(c), 1)
(1, c), pos).
Change 21.3.2.7.3 [string.rfind] as indicated:
size_type rfind(const charT* s, size_type pos, size_type n) const noexcept;-5- Returns: rfind(basic_string_view<charT, traits>(s, n), pos).
size_type rfind(const charT* s, size_type pos = npos) const;-6- Requires: s points to an array of at least traits::length(s) + 1 elements of charT.
-7- Returns: rfind(basic_string_view<charT, traits>(s), pos). -?- Throws: Nothing.size_type rfind(charT c, size_type pos = npos) const noexcept;-8- Returns: rfind(basic_string_view<charT, traits>(addressof(c), 1)
(1, c), pos).
Change 21.3.2.7.4 [string.find.first.of] as indicated:
size_type find_first_of(const charT* s, size_type pos, size_type n) const noexcept;-5- Returns: find_first_of(basic_string_view<charT, traits>(s, n), pos).
size_type find_first_of(const charT* s, size_type pos = 0) const;-6- Requires: s points to an array of at least traits::length(s) + 1 elements of charT.
-7- Returns: find_first_of(basic_string_view<charT, traits>(s), pos). -?- Throws: Nothing.size_type find_first_of(charT c, size_type pos = 0) const noexcept;-8- Returns: find_first_of(basic_string_view<charT, traits>(addressof(c), 1)
(1, c), pos).
Change 21.3.2.7.5 [string.find.last.of] as indicated:
size_type find_last_of(const charT* s, size_type pos, size_type n) const noexcept;-5- Returns: find_last_of(basic_string_view<charT, traits>(s, n), pos).
size_type find_last_of(const charT* s, size_type pos = npos) const;-6- Requires: s points to an array of at least traits::length(s) + 1 elements of charT.
-7- Returns: find_last_of(basic_string_view<charT, traits>(s), pos). -?- Throws: Nothing.size_type find_last_of(charT c, size_type pos = npos) const noexcept;-8- Returns: find_last_of(basic_string_view<charT, traits>(addressof(c), 1)
(1, c), pos).
Change 21.3.2.7.6 [string.find.first.not.of] as indicated:
size_type find_first_not_of(const charT* s, size_type pos, size_type n) const noexcept;-5- Returns: find_first_not_of(basic_string_view<charT, traits>(s, n), pos).
size_type find_first_not_of(const charT* s, size_type pos = 0) const;-6- Requires: s points to an array of at least traits::length(s) + 1 elements of charT.
-7- Returns: find_first_not_of(basic_string_view<charT, traits>(s), pos). -?- Throws: Nothing.size_type find_first_not_of(charT c, size_type pos = 0) const noexcept;-8- Returns: find_first_not_of(basic_string_view<charT, traits>(addressof(c), 1)
(1, c), pos).
Change 21.3.2.7.7 [string.find.last.not.of] as indicated:
size_type find_last_not_of(const charT* s, size_type pos, size_type n) const noexcept;-5- Returns: find_last_not_of(basic_string_view<charT, traits>(s, n), pos).
size_type find_last_not_of(const charT* s, size_type pos = npos) const;-6- Requires: s points to an array of at least traits::length(s) + 1 elements of charT.
-7- Returns: find_last_not_of(basic_string_view<charT, traits>(s), pos). -?- Throws: Nothing.size_type find_last_not_of(charT c, size_type pos = npos) const noexcept;-8- Returns: find_last_not_of(basic_string_view<charT, traits>(addressof(c), 1)
(1, c), pos).
Change 21.3.2.7.9 [string.compare] as indicated:
int compare(const charT* s) const noexcept;-9- Returns: compare(basic_string_view<charT, traits>(s)).
Section: 17.5.5.15 [lib.types.movedfrom], 17.5.4.9 [res.on.arguments], 23.2.1 [container.requirements.general] Status: New Submitter: Tim Song Opened: 2016-12-09 Last modified: 2017-02-03
Priority: 2
View all issues with New status.
Discussion:
LWG 2468's resolution added to MoveAssignable the requirement to tolerate self-move-assignment, but that does nothing for library types that aren't explicitly specified to meet MoveAssignable other than make those types not meet MoveAssignable any longer.
To realize the intent here, we need to carve out an exception to 17.5.4.9 [res.on.arguments]'s restriction for move assignment operators and specify that self-move-assignment results in valid but unspecified state unless otherwise specified. The proposed wording below adds that to 17.5.5.15 [lib.types.movedfrom] since it seems to fit well with the theme of the current paragraph in that section. In addition, to address the issue with 23.2.1 [container.requirements.general] noted in LWG 2468's discussion, the requirement tables in that subclause will need to be edited in a way similar to LWG 2468.[2017-01-27 Telecon]
Priority 2
Proposed resolution:
This wording is relative to N4618.
Add a new paragraph at the end of 17.5.5.15 [lib.types.movedfrom]:
-1- Objects of types defined in the C++ standard library may be moved from (12.8). Move operations may be explicitly specified or implicitly generated. Unless otherwise specified, such moved-from objects shall be placed in a valid but unspecified state.
-?- An object of a type defined in the C++ standard library may be move-assigned (12.8.2 [class.copy.assign]) to itself. Such an assignment places the object in a valid but unspecified state unless otherwise specified.
Add a note at the end of 17.5.4.9 [res.on.arguments]/1, bullet 3, as indicated:
-1- Each of the following applies to all arguments to functions defined in the C++ standard library, unless explicitly stated otherwise.
(1.1) — […]
(1.2) — […]
(1.3) — If a function argument binds to an rvalue reference parameter, the implementation may assume that this parameter is a unique reference to this argument. [Note: If the parameter is a generic parameter of the form T&& and an lvalue of type A is bound, the argument binds to an lvalue reference (14.8.2.1) and thus is not covered by the previous sentence. — end note] [Note: If a program casts an lvalue to an xvalue while passing that lvalue to a library function (e.g. by calling the function with the argument std::move(x)), the program is effectively asking that function to treat that lvalue as a temporary. The implementation is free to optimize away aliasing checks which might be needed if the argument was an lvalue. — end note] [Note: This does not apply to the argument passed to a move assignment operator (17.5.5.15 [lib.types.movedfrom]). — end note]
Edit Table 83 "Container requirements" in 23.2.1 [container.requirements.general] as indicated:
Table 83 — Container requirements Expression Return type Operational
semanticsAssertion/note
pre-/post-conditionComplexity … a = rv T& All existing elements of a
are either move
assigned to or
destroyedpost: If a and rv do not refer to the same object,
a shall be equal to the value that
rv had before this assignmentlinear …
Edit Table 86 "Allocator-aware container requirements" in 23.2.1 [container.requirements.general] as indicated:
Table 86 — Allocator-aware container requirements Expression Return type Assertion/note
pre-/post-conditionComplexity … a = rv T& Requires: If allocator_traits<allocator_type
>::propagate_on_container_move_assignment::value
is false, T is MoveInsertable
into X and MoveAssignable.
All existing elements of a are either
move assigned to or destroyed.
post: If a and rv do not refer
to the same object, a shall be equal
to the value that rv had before this assignmentlinear …
Section: 99 [fs.dir.itr.members], 99 [fs.rec.dir.itr.members] Status: New Submitter: Tim Song Opened: 2016-12-09 Last modified: 2017-03-19
Priority: 2
View all other issues in [fs.dir.itr.members].
View all issues with New status.
Discussion:
directory_iterator::increment and recursive_directory_iterator::increment are specified by reference to the input iterator requirements, which is narrow-contract: it has a precondition that the iterator is dereferenceable. Yet they are marked as noexcept.
Either the noexcept (one of which was added by LWG 2637) should be removed, or the behavior of increment when given a nondereferenceable iterator should be specified. Currently, libstdc++ and MSVC report an error via the error_code argument for a nondereferenceable directory_iterator, while libc++ and boost::filesystem assert.[2017-01-27 Telecon]
Priority 2; there are some problems with the wording in alternative B.
Proposed resolution:
This wording is relative to N4618.
Alternative A (remove the noexcept):
Edit 99 [fs.class.directory_iterator], class directory_iterator synopsis, as indicated:
directory_iterator& operator++(); directory_iterator& increment(error_code& ec)noexcept;
Edit 99 [fs.dir.itr.members] before p10 as indicated:
directory_iterator& operator++(); directory_iterator& increment(error_code& ec)noexcept;-10- Effects: As specified for the prefix increment operation of Input iterators (24.2.3).
-11- Returns: *this. -12- Throws: As specified in 27.10.7.
Edit 99 [fs.class.rec.dir.itr], class recursive_directory_iterator synopsis, as indicated:
recursive_directory_iterator& operator++(); recursive_directory_iterator& increment(error_code& ec)noexcept;
Edit 99 [fs.rec.dir.itr.members] before p23 as indicated:
recursive_directory_iterator& operator++(); recursive_directory_iterator& increment(error_code& ec)noexcept;-23- Effects: As specified for the prefix increment operation of Input iterators (24.2.3), except that: […]
-24- Returns: *this. -25- Throws: As specified in 27.10.7.
Alternative B (specify that increment reports an error if the iterator is not dereferenceable):
Edit 99 [fs.dir.itr.members] p10 as indicated:
directory_iterator& operator++(); directory_iterator& increment(error_code& ec) noexcept;-10- Effects: As specified for the prefix increment operation of Input iterators (24.2.3), except that increment reports an error if *this is not dereferenceable.
-11- Returns: *this. -12- Throws: As specified in 27.10.7.
Edit 99 [fs.rec.dir.itr.members] p23 as indicated:
recursive_directory_iterator& operator++(); recursive_directory_iterator& increment(error_code& ec) noexcept;-23- Effects: As specified for the prefix increment operation of Input iterators (24.2.3), except that:
increment reports an error if *this is not dereferenceable.
If there are no more entries at the current depth, […]
Otherwise if […]
-24- Returns: *this.
-25- Throws: As specified in 27.10.7.
Section: 21 [strings] Status: New Submitter: Jeffrey Yasskin Opened: 2016-12-09 Last modified: 2017-02-03
Priority: 3
View all other issues in [strings].
View all issues with New status.
Discussion:
Reported via editorial issue #165:
The following places should probably use the "Equivalent to" wording introduced in [structure.specifications]/4.
[string.cons]/18
Effects: Same as basic_string(il.begin(), il.end(), a).
[string.capacity]/3
Returns: size().
[string.capacity]/8
Effects: As if by resize(n, charT()).
[string.capacity]/16
Effects: Behaves as if the function calls: erase(begin(), end());
[string.capacity]/17
Returns: size() == 0.
[string.insert]/27
Effects: As if by insert(p, il.begin(), il.end()).
Returns: An iterator which refers to the copy of the first inserted character, or p if i1 is empty.
[2017-01-27 Telecon]
Priority 3; Marshall to provide wording.
Proposed resolution:
Section: 20.12.2.2 [mem.res.private] Status: New Submitter: Jens Maurer Opened: 2016-12-13 Last modified: 2017-03-05
Priority: 3
View all other issues in [mem.res.private].
View all issues with New status.
Discussion:
The specification of do_allocate() (20.12.2.2 [mem.res.private] p2+p3) says:
Returns: A derived class shall implement this function to return a pointer to allocated storage (3.7.4.2) with a size of at least bytes. The returned storage is aligned to the specified alignment, if such alignment is supported (3.11); otherwise it is aligned to max_align.
Throws: A derived class implementation shall throw an appropriate exception if it is unable to allocate memory with the requested size and alignment.
It is unclear whether a request for an unsupported alignment (e.g. larger than max_align) yields an exception or the returned storage is silently aligned to max_align.
This is editorial issue #966.[2017-01-27 Telecon]
Priority 3; Marshall to ping Pablo for intent and provide wording.
[2017-02-12 Pablo responds and provides wording]
The original intent was:
However, the description of do_allocate might have gone stale as the aligned-allocation proposal made its way into the standard.
The understanding I take from the definition of extended alignment in (the current text of) 3.11/3 [basic.align] and "assembling an argument list" in 5.3.4/14 [expr.new] is that it is intended that, when allocating space for an object with extended alignment in a well-formed program, the alignment will be honored and will not be truncated to max_align. I think this is a change from earlier drafts of the extended-alignment proposal, where silent truncation to max_align was permitted (I could be wrong). Anyway, it seems wrong to ever ignore the alignment parameter in do_allocate().
Proposed resolution:
Change the specification of do_allocate() (20.12.2.2 [mem.res.private] p2+p3) as follows:
Returns: A derived class shall implement this function to return a pointer to allocated storage (3.7.4.2) with a size of at least bytes, aligned to the specified alignment.
The returned storage is aligned to the specified alignment, if such alignment is supported; otherwise it is aligned to max_align.Throws: A derived class implementation shall throw an appropriate exception if it is unable to allocate memory with the requested size and alignment.
Section: 23.2.6 [associative.reqmts], 23.2.7 [unord.req] Status: New Submitter: Matt Austern Opened: 2016-12-14 Last modified: 2017-02-03
Priority: 3
View other active issues in [associative.reqmts].
View all other issues in [associative.reqmts].
View all issues with New status.
Discussion:
If we write a_uniq.insert(i, j) and [i, j) has multiple elements with keys that compare equivalent, which ones get inserted? Consider, for example, inserting into a map<string, int> with
m.insert({{"red", 5}, {"green", 3}, {"red", 7}, {"blue", 2}, {"pink", 6}});
Which value for "red" will the map have?
On my implementation we got "red" -> 5, and I suspect that's true on most or all implementations, but I don't believe that's guaranteed by anything in the requirements. The wording in Table 90 just says that it "inserts each element from the range [i, j) if and only if there is no element with key equivalent to the key of that element", but that doesn't tell us what happens if [i, j) contains duplicate keys because it doesn't say what order the insertions are performed in. The standard should either guarantee that the first value is the one that gets inserted, or explicitly say that this is unspecified. The same issue applies to the range constructor, and to the unordered associative containers.[2017-01-27 Telecon]
Priority 3; Nico to provide wording.
Proposed resolution:
Section: 20.15.1 [meta.rqmts] Status: New Submitter: Tim Song Opened: 2016-12-14 Last modified: 2017-02-03
Priority: 3
View all other issues in [meta.rqmts].
View all issues with New status.
Discussion:
[meta.rqmts]/3 defines TransformationTrait as follows:
A TransformationTrait modifies a property of a type. It shall be a class template that takes one template type argument and, optionally, additional arguments that help define the modification. It shall define a publicly accessible nested type named type, which shall be a synonym for the modified type.
enable_if, result_of and common_type do not necessarily "define a publicly accessible nested type named type". aligned_storage takes no template type argument (it only has two non-type parameters). Yet [meta.trans]/2 says that they are all TransformationTraits.
Incidentally, with the exception of decay, it's not clear that any of the traits in [meta.trans.other] could really be described as "modify[ing] a property of a type".[2017-01-27 Telecon]
Priority 3
Proposed resolution:
Section: 26.5.9 [cmplx.over], 26.9.1 [cmath.syn] Status: New Submitter: Jens Maurer Opened: 2016-12-15 Last modified: 2017-02-03
Priority: 3
View all other issues in [cmplx.over].
View all issues with New status.
Discussion:
In [cmplx.over] and [cmath.syn], when talking about "sufficient additional overloads", we use the phrase "effectively cast", but that is not a defined term.
A hostile interpretation could read "reinterpret_cast" here. Likely we mean "apply floating-point promotions, floating-integral conversions, and floating-point conversions", but that should be spelled out somewhere, e.g. in the library definitions section. (Source: Editorial issue #1248)[2017-01-27 Telecon]
Priority 3
Proposed resolution:
Section: 26.9.1 [cmath.syn] Status: New Submitter: Jens Maurer Opened: 2016-12-15 Last modified: 2017-02-03
Priority: 3
View other active issues in [cmath.syn].
View all other issues in [cmath.syn].
View all issues with New status.
Discussion:
With P0175R1, we now show in [cmath.syn] three overloads for the sin function: One taking a float, one taking a double, and one taking a long double. However, there is no statement that sin(long double) should actually invoke sinl, presumably delivering extra precision.
An implementation like
inline long double sin(long double x)
{ return sinf(x); }
seems to satisfy the "effectively cast" requirement, but is certainly unintentional.
The same issue arises for all math functions inherited from C. (Source: Editorial issue #1247)[2017-01-27 Telecon]
Priority 3
Proposed resolution:
Section: 20.12.5.2 [mem.res.pool.options] Status: New Submitter: Jens Maurer Opened: 2016-12-15 Last modified: 2017-02-03
Priority: 3
View all issues with New status.
Discussion:
20.12.5.2 [mem.res.pool.options] p3 talks about a "pass-through-threshold".
First, the phrase is not defined and it seems it could be easily avoided given the context. Second, given the phrasing here, it seems the implementation is essentially allowed to ignore the value largest_required_pool_block as it sees fit. It is unclear whether that is the intention.[2017-01-27 Telecon]
Priority 3; Jonathan will ask Alisdair for wording.
Proposed resolution:
Section: 27.10.15.4 [fs.op.copy_file] Status: New Submitter: Tim Song Opened: 2016-12-17 Last modified: 2017-02-03
Priority: 2
View all other issues in [fs.op.copy_file].
View all issues with New status.
Discussion:
27.10.15.4 [fs.op.copy_file]/4 says that copy_file reports "a file already exists error as specified in [fs.err.report] if" any of several error conditions exist.
It's not clear how some of those error conditions, such as !is_regular_file(from), can be sensibly described as "file already exists". Pretty much everywhere else in the filesystem specification just says "an error" without further elaboration.[2017-01-27 Telecon]
Priority 2; Jonathan to provide updated wording.
Proposed resolution:
This wording is relative to N4618.
Edit [fs.op.copy_file]/4 as indicated:
bool copy_file(const path& from, const path& to, copy_options options); bool copy_file(const path& from, const path& to, copy_options options, error_code& ec) noexcept;-4- Effects: As follows:
(4.1) — Report
a file already existsan error as specified in 27.10.7 [fs.err.report] if:
(4.1.1) —[…]
(4.2) —[…]
Section: 99 [fs.enum.file_type], 99 [fs.enum.copy.opts], 99 [fs.enum.dir.opts] Status: New Submitter: Tim Song Opened: 2016-12-18 Last modified: 2017-03-19
Priority: 2
View all other issues in [fs.enum.file_type].
View all issues with New status.
Discussion:
LWG 2678 stripped the numerical values of the enumerators from three enum classes in 27.10.10 [fs.enum]; in doing so it also removed the implicit specification 1) of the bitmask elements for the two bitmask types (copy_options and directory_options) and 2) that the file_type constants are distinct.
[2017-01-27 Telecon]
Priority 2; Jonathan to work with Tim to tweak wording.
Proposed resolution:
This wording is relative to N4618.
Edit 99 [fs.enum.file_type]/1 as indicated:
This enum class specifies constants used to identify file types, with the meanings listed in Table 122. The values of the constants are distinct.
Edit 99 [fs.enum.copy.opts]/1 as indicated:
The enum class type copy_options is a bitmask type (17.4.2.1.4 [bitmask.types]) that specifies bitmask constants used to control the semantics of copy operations. The constants are specified in option groups with the meanings listed in Table 123. The constant none represents the empty bitmask, and
Constant noneis shown in each option group for purposes of exposition; implementations shall provide only a single definition. Every other constant in the table represents a distinct bitmask element. Calling a library function with more than a single constant for an option group results in undefined behavior.
Edit 99 [fs.enum.dir.opts]/1 as indicated:
The enum class type directory_options is a bitmask type (17.4.2.1.4 [bitmask.types]) that specifies bitmask constants used to identify directory traversal options, with the meanings listed in Table 125. The constant none represents the empty bitmask; every other constant in the table represents a distinct bitmask element.
Section: 21.3.3.2 [string.operator==], 21.3.2.7.9 [string.compare] Status: New Submitter: Ahti Leppänen Opened: 2017-01-09 Last modified: 2017-02-05
Priority: 2
View all issues with New status.
Discussion:
Currently (N4618, 2016-11-28) the specification of operator== for std::basic_string and std::basic_string_view objects is clearly defined, but when interpreted as written, it may lead to comparison of strings of different sizes being a 𝒪(n) operation instead of a simple size check. Actual implementations in standard libraries vary so that in practice the programmers can't rely neither on having the literal version of the standard specification nor reasonable performance characteristics.
The definition for basic_string operator== in N4618 is as follows:21.3.3.2 [string.operator==]
bool operator==(const basic_string<charT, traits, Allocator>& lhs, const basic_string<charT, traits, Allocator>& rhs) noexcept;-1- Returns: lhs.compare(rhs) == 0.
21.3.2.7.9 [string.compare]
int compare(const basic_string& str) const noexcept;-6- Effects: Equivalent to: return compare(basic_string_view<charT, traits>(str));
21.3.2.7.9 [string.compare]
int compare(basic_string_view<charT, traits> sv) const noexcept;-1- Effects: Determines the effective length rlen of the strings to compare as the smaller of size() and sv.size(). The function then compares the two strings by calling traits::compare(data(), sv.data(), rlen).
-2- Returns: The nonzero result if the result of the comparison is nonzero. Otherwise, returns a value as indicated in Table 63.
Table 63 — compare() results Condition Return Value size() < sv.size() < 0 size() == sv.size() 0 size() > sv.size() > 0
From these it seems that compare() of strings of different sizes can't return zero and operator== will return false. However some implementations do not seem to call traits::compare() for basic_strings of different sizes even when the traits and it's compare() are user-defined. And those that call, make the operator== a worst case 𝒪(n) operation even for strings of different sizes.
This defect report does not propose a wording, but on a general level the wording should allow standard library implementers to write a standard conforming operator== for basic_string and basic_string_view (others?) in such a way that it's performance characteristics are reasonable and the programmers can rely on having a consistent behaviour across implementations. Perhaps the key issue here is that operator== is defined through compare() == 0: while it returns the intended result, for some inputs it does computations that are not needed by operator==. There are also related specifications that may need to be revised, for example operator!= for basic_string_views is defined in 21.4.3 [string.view.comparison] asReturns: lhs.compare(rhs) != 0
[2017-01-26, Jonathan Wakely comments and provides proposed resolution]
As mentioned above, some implementations do not make a call to Traits::compare if the string lengths are not equal, even though in general this is an observable side effect. Some implementations only perform that optimisation for std::string and std::wstring, where we know that calls to std::char_traits<char>::compare and std::char_traits<wchar_t>::compare are not observable.
My reading is that the Returns: element describes the value that must be returned, not the precise steps that must be taken to calculate that value. If we intended to specify the precise steps that must be taken then we could say that using "Effects: Equivalent to […]", but we don't do that. I would prefer this issue to be closed NAD with the rationale that my reading is correct and comparing the lengths to avoid calling Traits::compare is already permitted. But if my reading is wrong we need to permit this obvious optimisation.[2017-01-27 Telecon]
Priority 2
[2017-02-04, Ahti Leppänen comments and recommends NAD]
While there seems to be varying interpretations of the standards wording, given the comments in this defect report and definitions in [structure.specification] (N4618):
Effects: the actions performed by the function
Returns: a description of the value(s) returned by the function
I fail to see that the specification of operator== "Returns: lhs.compare(rhs) == 0" would require call to compare() and no longer consider the report valid.
Proposed resolution:
This wording is relative to N4618.
Preferred: NAD
Alternative:
Modify 21.3.3.2 [string.operator==] p1 as shown:
template<class charT, class traits, class Allocator> bool operator==(const basic_string<charT, traits, Allocator>& lhs, const basic_string<charT, traits, Allocator>& rhs) noexcept;-1- Returns: lhs.size() == rhs.size() && lhs.compare(rhs) == 0.
Modify 21.4.3 [string.view.comparison] as shown:
[Example: A sample conforming implementation for operator== would be:
template<class T> using __identity = decay_t<T>; template<class charT, class traits> constexpr bool operator==(basic_string_view<charT, traits> lhs, basic_string_view<charT, traits> rhs) noexcept { return lhs.size() == rhs.size() && lhs.compare(rhs) == 0; } template<class charT, class traits> constexpr bool operator==(basic_string_view<charT, traits> lhs, __identity<basic_string_view<charT, traits>> rhs) noexcept { return lhs.size() == rhs.size() && lhs.compare(rhs) == 0; } template<class charT, class traits> constexpr bool operator==(__identity<basic_string_view<charT, traits>> lhs, basic_string_view<charT, traits> rhs) noexcept { return lhs.size() == rhs.size() && lhs.compare(rhs) == 0; }— end example]
template<class charT, class traits> constexpr bool operator==(basic_string_view<charT, traits> lhs, basic_string_view<charT, traits> rhs) noexcept;-2- Returns: lhs.size() == rhs.size() && lhs.compare(rhs) == 0.
template<class charT, class traits> constexpr bool operator!=(basic_string_view<charT, traits> lhs, basic_string_view<charT, traits> rhs) noexcept;-3- Returns: lhs.size() != rhs.size() || lhs.compare(rhs) != 0.
Section: 30.6.9 [futures.async] Status: New Submitter: Andrey Davydov Opened: 2017-01-20 Last modified: 2017-03-13
Priority: 2
View other active issues in [futures.async].
View all other issues in [futures.async].
View all issues with New status.
Discussion:
Because the destructor of the std::future returned from the std::async blocks until the asynchronous operation completes, discarding the std::async return value leads to the synchronous code execution, which is pointless. For example, in the following code
void task1();
void task2();
void foo()
{
std::async(std::launch::async, task1),
std::async(std::launch::async, task2);
}
void bar()
{
std::async(std::launch::async, task1);
std::async(std::launch::async, task2);
}
task1 and task2 will be concurrently executed in the function 'foo', but sequentially in the function 'bar'.
[2017-01-27 Telecon]
Priority 2; Nico to provide wording.
[2017-03-04, Kona]
This should be handled by Nico's paper P0600.
Proposed resolution:
Section: 24.5.1 [reverse.iterators] Status: New Submitter: Hubert Tong Opened: 2017-01-28 Last modified: 2017-03-13
Priority: 4
View all other issues in [reverse.iterators].
View all issues with New status.
Discussion:
Further to LWG 2472, the case of reverse_iterator comparisons is a regression introduced by LWG 280.
Consider the following program:
#include <utility>
#include <iterator>
using namespace std::rel_ops;
bool f(std::reverse_iterator<int *> it) { return it != it; }
Under C++03, the operator!= in lib.reverse.iterator is more specialized than the operator!= in std::rel_ops.
Following LWG 280, neither operator!= candidate is more specialized than the other. The program is observed to fail with libc++. Online compiler example, see here. Suggested resolution: Reintroduce the homogeneous comparison operators from C++03 alongside the new ones.[2017-03-04, Kona]
Set priority to 4. STL to write a paper deprecating relops Alisdair to provide an example for Annex C.
Proposed resolution:
This wording is relative to N4618.
Modify 24.5.1 [reverse.iterators], class template reverse_iterator synopsis, as indicated:
template <class Iterator1, class Iterator2>
constexpr bool operator==(
const reverse_iterator<Iterator1>& x,
const reverse_iterator<Iterator2>& y);
template <class Iterator1, class Iterator2>
constexpr bool operator<(
const reverse_iterator<Iterator1>& x,
const reverse_iterator<Iterator2>& y);
template <class Iterator1, class Iterator2>
constexpr bool operator!=(
const reverse_iterator<Iterator1>& x,
const reverse_iterator<Iterator2>& y);
template <class Iterator1, class Iterator2>
constexpr bool operator>(
const reverse_iterator<Iterator1>& x,
const reverse_iterator<Iterator2>& y);
template <class Iterator1, class Iterator2>
constexpr bool operator>=(
const reverse_iterator<Iterator1>& x,
const reverse_iterator<Iterator2>& y);
template <class Iterator1, class Iterator2>
constexpr bool operator<=(
const reverse_iterator<Iterator1>& x,
const reverse_iterator<Iterator2>& y);
template <class Iterator>
constexpr bool operator==(
const reverse_iterator<Iterator>& x,
const reverse_iterator<Iterator>& y);
template <class Iterator>
constexpr bool operator<(
const reverse_iterator<Iterator>& x,
const reverse_iterator<Iterator>& y);
template <class Iterator>
constexpr bool operator!=(
const reverse_iterator<Iterator>& x,
const reverse_iterator<Iterator>& y);
template <class Iterator>
constexpr bool operator>(
const reverse_iterator<Iterator>& x,
const reverse_iterator<Iterator>& y);
template <class Iterator>
constexpr bool operator>=(
const reverse_iterator<Iterator>& x,
const reverse_iterator<Iterator>& y);
template <class Iterator>
constexpr bool operator<=(
const reverse_iterator<Iterator>& x,
const reverse_iterator<Iterator>& y);
Modify 24.5.1.3.13 [reverse.iter.op==] as indicated:
template <class Iterator1, class Iterator2> constexpr bool operator==( const reverse_iterator<Iterator1>& x, const reverse_iterator<Iterator2>& y); template <class Iterator> constexpr bool operator==( const reverse_iterator<Iterator>& x, const reverse_iterator<Iterator>& y);-1- Returns: x.current == y.current.
Modify 24.5.1.3.14 [reverse.iter.op<] as indicated:
template <class Iterator1, class Iterator2> constexpr bool operator<( const reverse_iterator<Iterator1>& x, const reverse_iterator<Iterator2>& y); template <class Iterator> constexpr bool operator<( const reverse_iterator<Iterator>& x, const reverse_iterator<Iterator>& y);-1- Returns: x.current > y.current.
Modify 24.5.1.3.15 [reverse.iter.op!=] as indicated:
template <class Iterator1, class Iterator2> constexpr bool operator!=( const reverse_iterator<Iterator1>& x, const reverse_iterator<Iterator2>& y); template <class Iterator> constexpr bool operator!=( const reverse_iterator<Iterator>& x, const reverse_iterator<Iterator>& y);-1- Returns: x.current != y.current.
Modify 24.5.1.3.16 [reverse.iter.op>] as indicated:
template <class Iterator1, class Iterator2> constexpr bool operator>( const reverse_iterator<Iterator1>& x, const reverse_iterator<Iterator2>& y); template <class Iterator> constexpr bool operator>( const reverse_iterator<Iterator>& x, const reverse_iterator<Iterator>& y);-1- Returns: x.current < y.current.
Modify 24.5.1.3.17 [reverse.iter.op>=] as indicated:
template <class Iterator1, class Iterator2> constexpr bool operator>=( const reverse_iterator<Iterator1>& x, const reverse_iterator<Iterator2>& y); template <class Iterator> constexpr bool operator>=( const reverse_iterator<Iterator>& x, const reverse_iterator<Iterator>& y);-1- Returns: x.current <= y.current.
Modify 24.5.1.3.18 [reverse.iter.op<=] as indicated:
template <class Iterator1, class Iterator2> constexpr bool operator<=( const reverse_iterator<Iterator1>& x, const reverse_iterator<Iterator2>& y); template <class Iterator> constexpr bool operator<=( const reverse_iterator<Iterator>& x, const reverse_iterator<Iterator>& y);-1- Returns: x.current >= y.current.
Section: 18.6.4 [ptr.launder] Status: Core Submitter: Hubert Tong Opened: 2017-01-31 Last modified: 2017-05-31
Priority: 2
View other active issues in [ptr.launder].
View all other issues in [ptr.launder].
View all issues with Core status.
Discussion:
Given:
struct A { int x, y; };
A a[100];
The bytes which compose a[3] can be reached from &a[2].x: reinterpret_cast<A *>(&a[2].x) + 1 points to a[3], however, the definition of "reachable" in [ptr.launder] does not encompass this case.
[2017-03-04, Kona]
Set priority to 2. Assign this (and 2860) to Core.
Proposed resolution:
This wording is relative to N4618.
Modify 18.6.4 [ptr.launder] as indicated:
template <class T> constexpr T* launder(T* p) noexcept;[…]
-3- Remarks: An invocation of this function may be used in a core constant expression whenever the value of its argument may be used in a core constant expression. A byte of storage b is reachable through a pointer value that points to an object Y if there is an object Z, pointer-interconvertible with Y, such that bitis within the storage occupied by ZY,an object that is pointer-interconvertible with Y,or the immediately-enclosing array object if ZYis an array element. The program is ill-formed if T is a function type or (possibly cv-qualified) void.
Section: 18.6.4 [ptr.launder] Status: Core Submitter: Hubert Tong Opened: 2017-01-31 Last modified: 2017-05-31
Priority: 2
View other active issues in [ptr.launder].
View all other issues in [ptr.launder].
View all issues with Core status.
Discussion:
There is an apparent oversight in the wording for launder that allows it to return base class subobjects which differ in their polymorphic behaviour between calls to launder.
This can be fixed by restricting launder from returning pointers to base class subobjects:
always, or
only for polymorphic class types.
[2017-03-04, Kona]
Set priority to 2. This was discussed in EWG (via a paper). Assign this (and 2859) to Core.
Proposed resolution:
This wording is relative to N4618.
Option 1:
Modify 18.6.4 [ptr.launder] as indicated:
template <class T> constexpr T* launder(T* p) noexcept;-1- Requires: p represents the address A of a byte in memory. An object X that is within its lifetime (3.8 [basic.life]) and whose type is similar (4.5 [conv.qual]) to T is located at the address A. X shall either be a most derived object, or pointer-interconvertible with a most derived object that is within its lifetime. All bytes of storage that would be reachable through the result are reachable through p (see below).
Option 2:
Modify 18.6.4 [ptr.launder] as indicated:
template <class T> constexpr T* launder(T* p) noexcept;-1- Requires: p represents the address A of a byte in memory. An object X that is within its lifetime (3.8 [basic.life]) and whose type is similar (4.5 [conv.qual]) to T is located at the address A. If T is a polymorphic class type, then X shall be a most derived object. All bytes of storage that would be reachable through the result are reachable through p (see below).
Section: 26.5.7 [complex.value.ops] Status: New Submitter: Japan Opened: 2017-02-03 Last modified: 2017-02-03
Priority: Not Prioritized
View all other issues in [complex.value.ops].
View all issues with New status.
Discussion:
Addresses JP 25
Parameter theta of polar has the type of the template parameter. Therefore, it needs to change the default initial value to T(). The change of the declaration of this function in 26.5.1 [complex.syn] is accompanied by this change.
Proposed change:
template<class T> complex<T> polar(const T& rho, const T& theta =0T());
[2017-02 pre-Kona]
(twice)
Proposed resolution:
This wording is relative to N4618.
Modify 26.5.1 [complex.syn], header <complex> synopsis, as indicated:
template<class T> complex<T> polar(const T&, const T& =0T());
Modify 26.5.7 [complex.value.ops] as indicated:
template<class T> complex<T> polar(const T& rho, const T& theta =0T());
Section: 20.15.2 [meta.type.synop] Status: New Submitter: United States Opened: 2017-02-03 Last modified: 2017-02-05
Priority: Not Prioritized
View other active issues in [meta.type.synop].
View all other issues in [meta.type.synop].
View all issues with New status.
Discussion:
Addresses US 10
A user specialization of any type trait should produce an ill-formed program, not merely one whose behavior is unspecified. See also the related comment re 20.19.3 [execpol.type].
Proposed change:
Reword the paragraph as follows:
Unless otherwise specified, a program that adds specializations for any of the templates defined in this subclause is ill-formed; no diagnostic required.
See also 2909
Proposed resolution:
This wording is relative to N4618.
Modify 20.15.2 [meta.type.synop] as indicated:
-1-
The behavior of a program that adds specializations for any of the templates defined in this subclause is undefined unless otherwise specifiedUnless otherwise specified, a program that adds specializations for any of the templates defined in this subclause is ill-formed; no diagnostic required.
Section: 25.4.3 [algorithms.parallel.exec] Status: New Submitter: Switzerland Opened: 2017-02-03 Last modified: 2017-02-03
Priority: Not Prioritized
View other active issues in [algorithms.parallel.exec].
View all other issues in [algorithms.parallel.exec].
View all issues with New status.
Discussion:
Addresses CH 10Parallel implementations of algorithms may be faster if not restricted to the complexity specifications of serial implementations.
Proposed change: Add a relaxation of complexity specifications for non-sequenced policies.
Proposed resolution:
Section: 20.7.3 [variant.variant] Status: New Submitter: Switzerland Opened: 2017-02-03 Last modified: 2017-02-05
Priority: Not Prioritized
View other active issues in [variant.variant].
View all other issues in [variant.variant].
View all issues with New status.
Discussion:
Addresses CH 7Consider making the variant statically !valueless_by_exception() for cases where is_nothrow_move_constructible_v<T_i> for all alternative types Ti
Proposed change: Adopt section III of P0308R0.
Proposed resolution:
Section: 21.4 [string.view] Status: New Submitter: Switzerland Opened: 2017-02-03 Last modified: 2017-02-05
Priority: Not Prioritized
View other active issues in [string.view].
View all other issues in [string.view].
View all issues with New status.
Discussion:
Addresses CH 9The standard library should provide string_view parameters instead or in addition for functions defined with char const * or string const & as parameter types. Most notably in cases where both such overloads exist or where an internal copy is expected anyway.
It might be doubted that the non-null termination of string_view could be an issue with functions that pass the char * down to OS functions, such as fstream_buf::open() etc. and those shouldn't provide it and favour generating a std::string temporary instead in that case. However, std::path demonstrates it is usable to have string_view overloads and there might be many places where it can be handy, or even better.
Proposed change: Provide the overloads for std::regex, the exception classes, std::bitset, std::locale and more.
Proposed resolution:
Section: 23 [containers], 20 [utilities] Status: New Submitter: Finland Opened: 2017-02-03 Last modified: 2017-02-05
Priority: Not Prioritized
View other active issues in [containers].
View all other issues in [containers].
View all issues with New status.
Discussion:
Addresses FI 16Relational operators for containers should sfinae; if the underlying type is not comparable, neither should the container be. Same applies to tuple and pair.
Proposed change: Make the relational operators of containers and utility components reflect the validity of the underlying element types.
Proposed resolution:
Section: 23 [containers], 20 [utilities] Status: New Submitter: Finland Opened: 2017-02-03 Last modified: 2017-02-05
Priority: Not Prioritized
View other active issues in [containers].
View all other issues in [containers].
View all issues with New status.
Discussion:
Addresses FI 17The relational operators of optional and variant completely reflect the semantics of the element types; this is inconsistent with other types in the library, like pair, tuple and containers. If we believe it's important that we don't synthesize relational operators for wrapper types, we should believe it's important for other types as well. Otherwise comparing containers of floating-point types and tuples/pairs etc. of floating point types will give incorrect answers.
Proposed change: Make the relational operators of containers and utility components reflect the semantics of the operators for the underlying element types.
Proposed resolution:
Section: 20.8.3.4 [any.observers] Status: New Submitter: Finland Opened: 2017-02-03 Last modified: 2017-05-31
Priority: Not Prioritized
View all issues with New status.
Discussion:
Addresses FI 7The proposal p0032 has multiple problems: 1) it turns member function .empty() into .has_value(), negating the logic. Refactoring e.g. existing uses of std::experimental::any to use std::any thus involve non-trivial refactorings that are error- prone and can't be done via simple search-and-replace if there are containers in the same source files for which .empty() is used (based on the implementation experience of making the change in libstdc++ and refactoring the testsuite). Whilst any is not a container, the library is failing to go towards a direction where there would be a generic way to query for emptiness. 2) The use of function references for tag types makes the interface hard to use. The tag types do not have value semantics like every other tag type has, the tag types are hard to construct, and present surprises for certain kinds of overload sets. Furthermore, any attempts to decay the tag types produces a really surprising effect — as opposed to what the other tag types do, which is that the result of decaying them is the tag type itself, decaying these new tag types results in a pointer to function.
Proposed change: Keep the .empty() functions (and introduce them to all the types that are supposed to have a homogeneous interface), and make the tag types be regular tag types that are not references to functions.
[2017-02-03, Marshall notes]
P0504R0 (adopted in Issaquah) removed use of function references for tag types
Proposed resolution:
Section: 17.5.5 [conforming] Status: New Submitter: Great Britain Opened: 2017-02-03 Last modified: 2017-02-05
Priority: Not Prioritized
View all other issues in [conforming].
View all issues with New status.
Discussion:
Addresses GB 35Most implementations have poor testing and support for instantiating standard library templates with volatile-qualified types. We should grant a library-freedom to conforming implementations so that support for volatile (and const volatile) qualified types in standard library templates is not required unless explicitly specified — and mandate such support for all templates in the <type_traits> header. Additional support is already specified in most places we would be interested (e.g., tuple API). We may want to additionally guarantee support through forwarding references.
Proposed change: add a new 17.6.5.x Volatile Qualified Types [res.on.volatile.type] describing the intended level of support for volatile qualifiers.
Proposed resolution:
Section: 17.5.5.6 [constexpr.functions] Status: New Submitter: Great Britain Opened: 2017-02-03 Last modified: 2017-03-05
Priority: Not Prioritized
View all other issues in [constexpr.functions].
View all issues with New status.
Discussion:
Addresses GB 38Relax the prohibition on libraries adding constexpr; this was a constraint requested by library implementers when constexpr was new, and those same implementers now feel unduly constrained.
Proposed change: Rewrite the whole sub-clause to support libraries adding constexpr in a compatible manner, much like the freedom to add a noexcept specification.
[2017-02-20, Marshall adds wording]
The simplest change would be to strike the sentence "An implementation shall not […] explicitly required". However, people seem to want a definite permission here, so I have provided one.
Proposed resolution:
This wording is relative to N4640.
Modify 17.5.5.6 [constexpr.functions] as indicated:
This International Standard explicitly requires that certain standard library functions are constexpr (7.1.5). An implementation may declare additional standard library function signatures as constexpr
An implementation shall not declare any standard library function signature as constexpr except for those where it is explicitly required. Within any header that provides any non-defining declarations of constexpr functions or constructors an implementation shall provide corresponding definitions.
Section: 20 [utilities] Status: New Submitter: Great Britain Opened: 2017-02-03 Last modified: 2017-03-05
Priority: Not Prioritized
View all other issues in [utilities].
View all issues with New status.
Discussion:
Addresses GB 45If P0067R4 is applied consider how to parse hexadecimally:
to_chars(beg, end, 42, 16); 16 for hex
to_chars(beg, end, 4.2, true); true means hex
to_chars(beg, end, 4.2, chars_format::hex);
to_chars(beg, end, 4.2, chars_format::hex, 2);
That is: We have 3 different formats to specify hex depending on value types and whether to use precision. Which application programmer should remember this?
May be even worse (I am not sure):
to_chars(beg, end, 4.2, 16);
would silently convert 4.2 to 4 and
to_chars(beg, end, 4, chars_format::hex);
would silently convert 4 to 4.000000.
Proposed change: The various options should be harmonized, possibly by use of an extended enum approach, having the values: dec, hex, scientific, fixed, general with dec (new!) as default for integral values and general for floats
[2017-02-23, Jens Maurer comments]
This issue is NAD; the facility was redesigned with these considerations in mind before it went into the working draft.
Proposed resolution:
Section: 20.14.4 [func.invoke] Status: Open Submitter: Great Britain Opened: 2017-02-03 Last modified: 2017-03-05
Priority: Not Prioritized
View other active issues in [func.invoke].
View all other issues in [func.invoke].
View all issues with Open status.
Discussion:
Addresses GB 51The function template std::apply() in 20.5.3.5 [tuple.apply] is required to be constexpr, but std::invoke() in 20.14.4 [func.invoke] isn't. The most sensible implementation of apply_impl() is exactly equivalent to std::invoke(), so this requires implementations to have a constexpr version of invoke() for internal use, and the public API std::invoke, which must not be constexpr even though it is probably implemented in terms of the internal version.
Proposed change: Add constexpr to std::invoke.
[2017-02-20, Marshall adds wording]
[Kona 2017-03-01]
We think this needs CWG 1581 to work; accepted as Immediate to resolve NB comment.
Friday: CWG 1581 was not moved in Kona. Status back to Open.
Proposed resolution:
This wording is relative to N4640.
Modify 20.14.4 [func.invoke] as indicated:
template <class F, class... Args>
constexpr result_of_t<F&&(Args&&...)> invoke(F&& f, Args&&... args);
Section: 23.3.7.1 [array.overview] Status: New Submitter: Russia Opened: 2017-02-03 Last modified: 2017-05-31
Priority: 2
View other active issues in [array.overview].
View all other issues in [array.overview].
View all issues with New status.
Discussion:
Addresses RU 3Force the literal type requirement for the iterator and const_iterator in the std::array so that iterators of std::array could be used in constexpr functions.
Proposed change: Add to the end of the [array.overview] section: iterator and const_iterator shall be literal types.
[2017-02-20 Marshall adds wording]
I used the formulation "are literal types", rather than "shall be", since that denotes a requirement on the user.
Previous resolution [SUPERSEDED]:
This wording is relative to N4640.
Add a new paragraph at the end of 23.3.7.1 [array.overview]:
-?- iterator and const_iterator are literal types.
[2017-03-01, Kona]
Antony Polukhin provides revised wording.
[2017-03-02, Kona]
Wording tweaks suggested by LWG applied.
[2017-03-02, Tim Song comments]
I don't believe the blanket "all operations" wording is quite correct.
First, T t; (required by DefaultConstructible) isn't usable in a constant expression if the iterator is a pointer, since it would leave it uninitialized. Second, an explicit destructor call u.~T() (required by Destructible) isn't usable if the iterator is a class type because it explicitly invokes a non-constexpr function (the destructor); see [expr.const]/2.2.[2017-03-04, Kona]
Set priority to 2. Lisa and Alisdair to work with Antony to come up with better wording. The same wording can be used for 2938.
Previous resolution [SUPERSEDED]:
This wording is relative to N4640.
Add a new paragraph at the end of 23.3.7.1 [array.overview]:
-3- An array satisfies all of the requirements of a container […]
-?- All operations on iterator and const_iterator that are required to satisfy the random access iterator requirements are usable in constant expressions.
[2017-04-22, Antony Polukhin provides improved wording]
Proposed resolution:
This wording is relative to N4659.
Add a new paragraph at the end of 24.2.1 [iterator.requirements.general] as indicated:
-12- An invalid iterator is an iterator that may be singular.(footnote: […])
Iterators are called constexpr iterators if all defined iterator category operations, except a pseudo-destructor call (5.2.4 [expr.pseudo]) or the construction of an iterator with a singular value, are usable as constant expressions. [Note: For example, the types "pointer to int" and reverse_iterator<int*> are constexpr random access iterators. — end note] -13- In the following sections, a and b denote […]
Add a new paragraph at the end of 23.3.7.1 [array.overview]:
-3- An array satisfies all of the requirements of a container […]
-?- iterator and const_iterator satisfy the constexpr iterator requirements (24.2.1 [iterator.requirements.general]).
Section: 20.5 [tuple], 20.6 [optional], 20.11.1.2.1 [unique.ptr.single.ctor] Status: New Submitter: United States Opened: 2017-02-03 Last modified: 2017-03-05
Priority: Not Prioritized
View all other issues in [tuple].
View all issues with New status.
Discussion:
Addresses US 110The move constructors for tuple, optional, and unique_ptr should return false for is_(nothrow_)move_constructible_v<TYPE> when their corresponding Requires clauses are not satisfied, as there are now several library clauses that are defined in terms of these traits. The same concern applies to the move-assignment operator. Note that pair and variant already satisfy this constraint.
[2017-02-26, Scott Schurr provides wording]
Proposed resolution:
This wording is relative to N4640.
Modify 20.5.3 [tuple.tuple] as indicated:
// 20.5.3.1, tuple construction EXPLICIT constexpr tuple(); EXPLICIT constexpr tuple(const Types&...); // only if sizeof...(Types) >= 1 template <class... UTypes> EXPLICIT constexpr tuple(UTypes&&...) noexcept(see below); // only if sizeof...(Types) >= 1 tuple(const tuple&) = default; tuple(tuple&&) = default; template <class... UTypes> EXPLICIT constexpr tuple(const tuple<UTypes...>&); template <class... UTypes> EXPLICIT constexpr tuple(tuple<UTypes...>&&) noexcept(see below); template <class U1, class U2> EXPLICIT constexpr tuple(const pair<U1, U2>&); // only if sizeof...(Types) == 2 template <class U1, class U2> EXPLICIT constexpr tuple(pair<U1, U2>&&) noexcept(see below); // only if sizeof...(Types) == 2 […] // 20.5.3.2, tuple assignment tuple& operator=(const tuple&); tuple& operator=(tuple&&) noexcept(see below); template <class... UTypes> tuple& operator=(const tuple<UTypes...>&); template <class... UTypes> tuple& operator=(tuple<UTypes...>&&) noexcept(see below); template <class U1, class U2> tuple& operator=(const pair<U1, U2>&); // only if sizeof...(Types) == 2 template <class U1, class U2> tuple& operator=(pair<U1, U2>&&) noexcept(see below); // only if sizeof...(Types) == 2
Modify 20.5.3.1 [tuple.cnstr] as indicated:
template <class... UTypes> EXPLICIT constexpr tuple(UTypes&&... u) noexcept(see below);-8- Effects: Initializes the elements in the tuple with the corresponding value in std::forward<UTypes>(u).
-9- Remarks: This constructor shall not participate in overload resolution unless sizeof...(Types) == sizeof...(UTypes) and sizeof...(Types) >= 1 and is_constructible_v<Ti, Ui&&> is true for all i. The constructor is explicit if and only if is_convertible_v<Ui&&, Ti> is false for at least one i. The expression inside noexcept is equivalent to the logical AND of the following expressions:is_nothrow_constructible_v<Ti, Ui&&>where Ti is the ith type in Types, and Ui is the ith type in UTypes.
[…]
template <class... UTypes> EXPLICIT constexpr tuple(tuple<UTypes...>&& u) noexcept(see below);-16- Effects: For all i, initializes the ith element of *this with std::forward<Ui>(get<i>(u)).
-17- Remarks: This constructor shall not participate in overload resolution unless […] The constructor is explicit if and only if is_convertible_v<Ui&&, Ti> is false for at least one i. The expression inside noexcept is equivalent to the logical AND of the following expressions:is_nothrow_constructible_v<Ti, Ui&&>where Ti is the ith type in Types, and Ui is the ith type in UTypes.
[…]
template <class U1, class U2> EXPLICIT constexpr tuple(pair<U1, U2>&& u) noexcept(see below);-21- Effects: Initializes the first element with std::forward<U1>(u.first) and the second element with std::forward<U2>(u.second).
-22- Remarks: This constructor shall not participate in overload resolution unless sizeof...(Types) == 2, is_constructible_v<T0, U1&&> is true and is_constructible_v<T1, U2&&> is true. -23- The constructor is explicit if and only if is_convertible_v<U1&&, T0> is false or is_convertible_v<U2&&, T1> is false. The expression inside noexcept is equivalent to:is_nothrow_constructible_v<T0, U1&&> && is_nothrow_constructible_v<T1, U2&&>
Modify 20.5.3.2 [tuple.assign] as indicated:
template <class... UTypes> tuple& operator=(tuple<UTypes...>&& u) noexcept(see below);-12- Effects: For all i, assigns std::forward<Ui>(get<i>(u)) to get<i>(*this).
-13- Remarks: This operator shall not participate in overload resolution unless is_assignable_v<Ti&, Ui&&> == true for all i and sizeof...(Types) == sizeof...(UTypes). The expression inside noexcept is equivalent to the logical AND of the following expressions:is_nothrow_assignable_v<Ti&, Ui&&>where Ti is the ith type in Types, and Ui is the ith type in UTypes.
-14- Returns: *this.
[…]
template <class U1, class U2> tuple& operator=(pair<U1, U2>&& u) noexcept(see below);-18- Effects: Assigns std::forward<U1>(u.first) to the first element of *this and std::forward<U2>(u.second) to the second element of *this.
-19- Remarks: This operator shall not participate in overload resolution unless sizeof...(Types) == 2 and is_assignable_v<T0&, U1&&> is true for the first type T0 in Types and is_assignable_v<T1&, U2&&> is true for the second type T1 in Types. The expression inside noexcept is equivalent to:is_nothrow_assignable_v<T0&, U1&&> && is_nothrow_assignable_v<T1&, U2&&>-20- Returns: *this.
Section: 20.7.3 [variant.variant] Status: New Submitter: United States Opened: 2017-02-03 Last modified: 2017-03-05
Priority: Not Prioritized
View other active issues in [variant.variant].
View all other issues in [variant.variant].
View all issues with New status.
Discussion:
Addresses US 113Variants cannot properly support allocators, as any assignment of a subsequent value throws away the allocator used at construction. This is not an easy problem to solve, so variant would be better served dropping the illusion of allocator support for now, leaving open the possibility to provide proper support once the problems are fully understood.
Proposed change: Strike the 8 allocator aware constructor overloads from the class definition, and strike 20.7.2.1 [variant.ctor] p34/35. Strike clause 20.7.12 [variant.traits]. Strike the specialization of uses_allocator for variant in the <variant> header synopsis, 20.7.1 [variant.general].
[2017-02-28, Kona, Casey provides wording]
Proposed resolution:
This wording is relative to N4640.
Change 20.7.2 [variant.syn], header <variant> synopsis, as follows:
[…]// 20.7.13, allocator-related traits template <class T, class Alloc> struct uses_allocator; template <class... Types, class Alloc> struct uses_allocator<variant<Types...>, Alloc>;
Change 20.7.3 [variant.variant], class template variant synopsis, as follows:
[…]// allocator-extended constructors template <class Alloc> variant(allocator_arg_t, const Alloc&); template <class Alloc> variant(allocator_arg_t, const Alloc&, const variant&); template <class Alloc> variant(allocator_arg_t, const Alloc&, variant&&); template <class Alloc, class T> variant(allocator_arg_t, const Alloc&, T&&); template <class Alloc, class T, class... Args> variant(allocator_arg_t, const Alloc&, in_place_type_t<T>, Args&&...); template <class Alloc, class T, class U, class... Args> variant(allocator_arg_t, const Alloc&, in_place_type_t<T>, initializer_list<U>, Args&&...); template <class Alloc, size_t I, class... Args> variant(allocator_arg_t, const Alloc&, in_place_index_t<I>, Args&&...); template <class Alloc, size_t I, class U, class... Args> variant(allocator_arg_t, const Alloc&, in_place_index_t<I>, initializer_list<U>, Args&&...);
Modify 20.7.3.1 [variant.ctor] as indicated:
template <class Alloc> variant(allocator_arg_t, const Alloc& a); template <class Alloc> variant(allocator_arg_t, const Alloc& a, const variant& v); template <class Alloc> variant(allocator_arg_t, const Alloc& a, variant&& v); template <class Alloc, class T> variant(allocator_arg_t, const Alloc& a, T&& t); template <class Alloc, class T, class... Args> variant(allocator_arg_t, const Alloc& a, in_place_type_t<T>, Args&&... args); template <class Alloc, class T, class U, class... Args> variant(allocator_arg_t, const Alloc& a, in_place_type_t<T>, initializer_list<U> il, Args&&... args); template <class Alloc, size_t I, class... Args> variant(allocator_arg_t, const Alloc& a, in_place_index_t<I>, Args&&... args); template <class Alloc, size_t I, class U, class... Args> variant(allocator_arg_t, const Alloc& a, in_place_index_t<I>, initializer_list<U> il, Args&&... args);
-34- Requires: Alloc shall meet the requirements for an Allocator (17.5.3.5).-35- Effects: Equivalent to the preceding constructors except that the contained value is constructed with uses-allocator construction (20.10.7.2).
Modify 20.7.13 [variant.traits] as indicated:
template <class... Types, class Alloc> struct uses_allocator<variant<Types...>, Alloc> : true_type { };
-1- Requires: Alloc shall be an Allocator (17.5.3.5).-2- [Note: Specialization of this trait informs other library components that variant can be constructed with an allocator, even though it does not have a nested allocator_type. — end note]
Section: 20.7.3 [variant.variant] Status: New Submitter: United States Opened: 2017-02-03 Last modified: 2017-02-13
Priority: Not Prioritized
View other active issues in [variant.variant].
View all other issues in [variant.variant].
View all issues with New status.
Discussion:
Addresses US 114variant needs to know the size of an object in order to compute the size of its internal buffer, so require that any cv-qualified object type in Types... be a complete type.
Proposed change: Add 'complete' in p2
Proposed resolution:
This wording is relative to N4618.
Edit 20.7.3 [variant.variant]/2 as indicated:
All types in Types... shall be (possibly cv-qualified) complete object types that are not arrays.
Section: 20.11.2.2.1 [util.smartptr.shared.const] Status: New Submitter: United States Opened: 2017-02-03 Last modified: 2017-02-05
Priority: Not Prioritized
View all other issues in [util.smartptr.shared.const].
View all issues with New status.
Discussion:
Addresses US 130There is no ability to supply an allocator for the control block when constructing a shared_ptr from a unique_ptr. Note that no further shared_ptr constructors need an allocator, as they all have pre-existing control blocks that are shared, or already have the allocator overload.
Proposed change: Add an additional shared_ptr constructor, template <class Y, class D, class A> shared_ptr(unique_ptr<Y, D>&& r, A alloc), with the same semantics as the existing constructor taking a unique_ptr, but using the alloc argument to supply memory as required.
Proposed resolution:
Section: 20.11.2.2.2 [util.smartptr.shared.dest] Status: New Submitter: United States Opened: 2017-02-03 Last modified: 2017-02-05
Priority: Not Prioritized
View other active issues in [util.smartptr.shared.dest].
View all other issues in [util.smartptr.shared.dest].
View all issues with New status.
Discussion:
Addresses US 134The semantics for destroying the deleter and the control-block are unclear. In particular, it is not clear that we guarantee a lack of race conditions destroying the control-block and deleter. Possible race-free implementations might destroy the deleter after running d(p), and before giving up the weak reference held by this shared_ptr; running the destructor for d only when the last weak_ptr is destroyed, potentially at a much later date, but ensuring that d(p) completes before the shared_ptr gives up its weak reference; making a copy of d in the destructor before manipulating the weak count, and then using this copy to run d(p), even while the control-block could be concurrently reclaimed with an expiring weak_ptr in another thread. Note that this may be related to LWG 2751. (Also, see the note in 20.11.2.2.10 p1 [util.smartptr.getdeleter])
Proposed change: Clarify that the shared_ptr weak ownership of the control block is released at the end of the destructor, and not as the destructor begins. Otherwise, the deleter might be destroyed even before the destructor gets to move a copy to call safely.
Proposed resolution:
Section: 20.19.3 [execpol.type] Status: New Submitter: United States Opened: 2017-02-03 Last modified: 2017-06-01
Priority: Not Prioritized
View all issues with New status.
Discussion:
Addresses US 14A user specialization of any is_execution_policy should produce an ill-formed program, not merely one whose behavior is unspecified. See also the related comment re [meta.type.synop] (20.15.2). (US 10)
Proposed change: Reword the paragraph as follows:
See also LWG 2871.Unless otherwise specified, a program that adds specializations for is_execution_policy is ill-formed; no diagnostic required.
[2017-03-03, Kona Friday AM]
Jonathan: This breaks all known implementations of parallel algorithms, since they use custom execution policies.
Alisdair to check with SG1 to see if that's what they meant.
Proposed resolution:
Section: 20.15.2 [meta.type.synop] Status: New Submitter: United States Opened: 2017-02-03 Last modified: 2017-03-05
Priority: Not Prioritized
View other active issues in [meta.type.synop].
View all other issues in [meta.type.synop].
View all issues with New status.
Discussion:
Addresses US 142An alias template using the new template template auto deduction would make integral_constant slightly easier to use.
Proposed resolution:
This wording is relative to N4618.
Add to the synopsis of <type_traits>:
template <auto N> using integer_constant = integral_constant<decltype(N), N>;
Section: 24.5.2 [insert.iterators] Status: New Submitter: United States Opened: 2017-02-03 Last modified: 2017-02-05
Priority: Not Prioritized
View all other issues in [insert.iterators].
View all issues with New status.
Discussion:
Addresses US 151The three insert iterators should each have an instantiation guide to initialize from a container.
Proposed change:
Add to the <iterator> header synopsis:
template <class Container> back_insert_iterator(Container&) -> back_insert_iterator<Container>; template <class Container> front_insert_iterator(Container&) -> back_insert_iterator<Container>; template <class Container> insert_iterator(Container&, typename Container::iterator) -> insert_iterator<Container>;
Proposed resolution:
This wording is relative to N4618.
Add to the synopsis of <iterator>:
template <class Container> back_insert_iterator(Container&) -> back_insert_iterator<Container>; template <class Container> front_insert_iterator(Container&) -> back_insert_iterator<Container>; template <class Container> insert_iterator(Container&, typename Container::iterator) -> insert_iterator<Container>;
Section: 20.15.2 [meta.type.synop] Status: New Submitter: United States Opened: 2017-02-03 Last modified: 2017-03-05
Priority: Not Prioritized
View other active issues in [meta.type.synop].
View all other issues in [meta.type.synop].
View all issues with New status.
Discussion:
Addresses US 171The *_constant<> templates (including the proposed addition, bool_constant<>) do not make use of the new template<auto> feature.
Proposed change: Add a constant<> (subject to bikeshedding) template which uses template<auto>.
Define integral_constant<> as using integral_constant<T, V> = constant<T(V)> or integral_constant<T, V> = constant<V>.
Either remove bool_constant, define it as using bool_constant = constant<bool(B)> or using bool_constant = constant<B>.
[2017-03-03, Kona, LEWG]
Straw polls:
| constant | 3 |
| numeric_constant | 8 |
| static_constant | 1 |
| scalar_constant | 7 |
| integer_constant (Over LWG's dead body) | 1 |
| auto_constant | 4 |
| integral_c | 7 |
| int_ | 0 |
| | |
| scalar_constant | 6 |
| numeric_constant | 3 |
| integral_c | 5 |
Accept P0377 with "scalar_constant" for C++17 to address LWG 2922 and US 171:
SF | F | N | A | SA 0 | 1 | 3 | 7 | 5Proposed resolution:
Section: 26.9.1 [cmath.syn] Status: New Submitter: United States Opened: 2017-02-03 Last modified: 2017-02-05
Priority: Not Prioritized
View other active issues in [cmath.syn].
View all other issues in [cmath.syn].
View all issues with New status.
Discussion:
Addresses US 172noexcept is inconsistently applied across headers which import components of the C standard library into the C++ library; some functions (std::abort(), std::_Exit(), etc) are defined as noexcept in some places, but not in others. Some functions which seem like they should be noexcept (std::abs(), std::div(), etc) are not defined as noexcept.
Proposed change: Make the majority of the C library functions (with exceptions such as std::qsort() and std::bsearch(), which can call user code) noexcept.
Proposed resolution:
Section: 21.3.2.6.3 [string.assign] Status: New Submitter: Jonathan Wakely Opened: 2017-02-03 Last modified: 2017-05-31
Priority: 3
View all other issues in [string.assign].
View all issues with New status.
Discussion:
basic_string::assign(size_type n, charT c); says:
Effects: Equivalent to assign(basic_string(n, c)).
This requires that a new basic_string is constructed, using a default-constructed allocator, potentially allocating memory, and then that new string is copy-assigned to *this, potentially propagating the allocator. This must be done even if this->capacity() > n, because memory allocation and allocator propagation are observable side effects. If the allocator doesn't propagate and isn't equal to this->get_allocator() then a second allocation may be required. This can't be right; it won't even compile if the allocator isn't default constructible.
basic_string::assign(InputIterator first, InputIterator last) has a similar problem, even if the iterators are random access and this->capacity() > distance(first, last). basic_string::assign(std::initializer_list<charT> doesn't say "Equivalent to" so maybe it's OK to not allocate anything if the list fits in the existing capacity. basic_string::append(size_type, charT) and basic_string::append(InputIterator, InputIterator) have the same problem, although they don't propagate the allocator, but still require at least one, maybe two allocations. A partial fix would be to ensure all the temporaries are constructed with get_allocator() so that they don't require default constructible allocators, and so propagation won't alter allocators. The problem of observable side effects is still present (the temporary might need to allocate memory, even if this->capacity() is large) but arguably it's unspecified when construction allocates, to allow for small-string optimisations.[2017-03-04, Kona]
Set priority to 3. Thomas to argue on reflector that this is NAD.
[2017-03-07, LWG reflector discussion]
Thomas and Jonathan remark that LWG 2788 fixed most cases except the allocator respectance and provide wording for this:
Proposed resolution:
This wording is relative to N4659.
Change 21.3.2.6.3 [string.assign] as indicated:
basic_string& assign(size_type n, charT c);-22- Effects: Equivalent to assign(basic_string(n, c, get_allocator())).
Section: 24.4.3 [iterator.operations] Status: Open Submitter: Morwenn Opened: 2017-02-04 Last modified: 2017-03-13
Priority: 3
View all other issues in [iterator.operations].
View all issues with Open status.
Discussion:
It seems that std::next is missing an optimization opportunity when taking a single parameter. The standard mandates that std::next shall call std::advance on the passed iterator and return it. For random-access iterators, it means that operator+= will be called on the iterator. However, if a single-argument overload was added to std::next, it could call ++it directly instead of std::advance(it, 1), which means that operator++ would be called instead of operator+=. This might make a small performance difference for complicated iterators such as std::deque's ones, where operator++ has a simpler logic than operator+=.
An equivalent optimization could be allowed by adding a single-argument overload to std::prev too.[2017-03-04, Kona]
Set priority to 3. Alisdair to provide wording.
Proposed resolution:
Section: 25.4.3 [algorithms.parallel.exec] Status: Ready Submitter: Dietmar Kühl Opened: 2017-02-05 Last modified: 2017-03-13
Priority: Not Prioritized
View other active issues in [algorithms.parallel.exec].
View all other issues in [algorithms.parallel.exec].
Discussion:
Section 25.4.3 [algorithms.parallel.exec] specifies constraints a user of the parallel algorithms has to obey. Notably, it specifies in paragraph 3 that executions of element access functions are indeterminately sequenced with respect to each other. Correspondingly, it is the user's obligation to ensure that these calls do not introduce data races (this is also clarified in a note on this section).
Unfortunately, there is no constraint that, at least, mutating element access functions like operator++() on an iterator are called on different objects. An odd implementation of a parallel algorithm could increment a shared iterator from two threads without synchronisation of its own and the user would be obliged to make sure there is no data race! For example:
template <typename FwdIt>
FwdIt adjacent_find(std::execution::parallel_policy, FwdIt it, FwdIt end)
{
if (2 <= std::distance(it, end)) {
FwdIt tmp(it);
auto f1 = std::async([&tmp](){ ++tmp; });
auto f2 = std::async([&tmp](){ ++tmp; });
f1.get();
f2.get();
}
return std::adjancent_find(it, end);
}
This code is, obviously, a contrived example but with the current specification a legal implementation of adjacent_find(). The problem is that, e.g., for pointers there is a data race when incrementing tmp, i.e., the function can't be used on pointers. I don't think any of the containers makes a guarantee that their iterators can be incremented without synchronisation, i.e., the standard library doesn't have any iterators which could be used with this algorithm!
Of course, no sane implementation would do anything like that. However, they are allowed to be implemented as above, making it unnecessarily hard and probably inefficient to use the algorithms: in portable code any user of the parallel algorithms needs to deal with the possibility that mutating operations on the same object are called from different threads. There is a constraint missing for the parallel algorithm implementations which limits calls to, at least, some element access functions to be applied only to different objects if there is synchronisation done by the algorithm. At least, obviously mutating operations like the iterator increment/decrement operators need to be correspondingly constrained. On the other hand, it seems reasonable that a shared data structure stores, e.g., a predicate used concurrently by all threads. This use is quite reasonable and there is nothing wrong. That is, demanding that all element access functions are called on different objects between different threads would possibly adversely over-constraining the algorithms. Only the element access functions deliberately changing the object need to be constrained. Based on checking all algorithms in the standard library taking an ExecutionPolicy template parameter there are broadly four groups of template parameters:Parameters with a known set of possible arguments: ExecutionPolicy (execution policies listed in 20.19 [execpol]).
Parameters specifying types of objects which are expected not to change: BinaryOperation, BinaryPredicate, Compare, Function, Predicate, UnaryFunction, UnaryOperation, and T (all but the last one are function objects although I couldn't locate concepts for some of them — that may be a separate issue).
Parameters of mutable types which are also meant to be mutated: InputIterator, OutputIterator, ForwardIterator, BidirectionalIterator, RandomAccessIterator, and Size (the concept for Size also seems to be unspecified).
Some algorithms use Generator which seems to be a mutable function object. However, I couldn't locate a concept for this parameter.
The problematic group is 3 and possibly 4: mutations on the objects are expected. It seem the general approach of disallowing calling non-const functions without synchronisation applies. Note, however, that prohibiting calling of any non-const function from the algorithms would put undue burden on the implementation of algorithms: any of the accessor functions may be non-const although the concept assume that the function would be const. The constraint should, thus, only apply to functions which may mutate the object according to their respective concept.
Suggested Resolution:
Add a statement prohibiting unsequenced calls to element access functions on the same object which are not applicable to const objects according to the corresponding concept. I'm not sure how to best specify the constraint in general, though.
Since the current algorithms use relatively few concepts there are fairly few operations actually affected. It may be reasonable at least for the initial version (and until we could refer to constructs in concepts in the language) to explicitly list the affected operations. I haven't done a full audit but iterator ++, --, @= (for @ being any of the operators which can be combined with an assignment), and assignments on all objects may be the set of affected element access functions whose use needs to be constrained. Here is a concrete proposal for the change: In 25.4.2 [algorithms.parallel.user] add a paragraph: Parallel algorithms are constrained when calling mutating element access functions without synchronisation: if any mutating element access function is called on an object there shall be no other unsynchronised accesses to this object. The mutating element access functions are those which are specified as mutating object in the concept, notably assignment on any object, operators ++, --, +=, and -= on any of the iterator or Size parameters, and any @= operators on the Size parameters.Previous resolution [SUPERSEDED]:
This wording is relative to N4640.
Modify 25.4.2 [algorithms.parallel.user] as indicated:
-1- Function objects passed into parallel algorithms as objects of type Predicate, BinaryPredicate, Compare, and BinaryOperation shall not directly or indirectly modify objects via their arguments.
-?- Parallel algorithms are constrained when calling mutating element access functions without synchronisation: If any mutating element access function is called on an object there shall be no other unsynchronised accesses to this object. The mutating element access functions are those which are specified as mutating object in the concept, notably assignment on any object, operators ++, --, +=, and -= on any of the iterator or Size parameters, and any @= operators on the Size parameters.
[2017-03-03, Kona]
Dietmar provides improved wording. Issues with the PR before the change:
The part before the colon is redundant: we don't need to state that.
Replace "notably" with "specifically"
swap() needs to be in the list.
Not sure what "called on an object means"
The assignment side is overconstrained: the right hand side is allowed.
Previous resolution [SUPERSEDED]:
This wording is relative to N4640.
Modify 25.4.2 [algorithms.parallel.user] as indicated:
-1- Function objects passed into parallel algorithms as objects of type Predicate, BinaryPredicate, Compare, and BinaryOperation shall not directly or indirectly modify objects via their arguments.
-?- If an object is mutated by an element access function the algorithm will perform no other unsynchronized accesses to that object. The mutating element access functions are those which are specified as mutating the object in the relevant concept, such as swap(), ++, --, @=, and assignments. For the assignment and @= operators only the left argument is mutated.
[2017-03-03, Kona]
Dietmar finetunes wording after review by SG1.
[2017-03-03, Kona]
Move to Ready
Proposed resolution:
This wording is relative to N4640.
Add a new paragraph following 25.4.3 [algorithms.parallel.exec] p1 as indicated:
-1- Parallel algorithms have template parameters named ExecutionPolicy (20.19) which describe the manner in which the execution of these algorithms may be parallelized and the manner in which they apply the element access functions.
-?- If an object is modified by an element access function the algorithm will perform no other unsynchronized accesses to that object. The modifying element access functions are those which are specified as modifying the object in the relevant concept [Note: For example, swap(), ++, --, @=, and assignments modify the object. For the assignment and @= operators only the left argument is modified. — end note]. -2- […]
Section: 20.5.3.4 [tuple.creation] Status: New Submitter: Eric Fiselier Opened: 2017-02-06 Last modified: 2017-06-01
Priority: 3
View other active issues in [tuple.creation].
View all other issues in [tuple.creation].
View all issues with New status.
Discussion:
The current PR for LWG 2773 changes std::ignore to be a constexpr variable. However it says nothing about whether using std::ignore in std::tie is a constant expression. I think the intent was clearly to allow this. Therefore I suggest we update the resolution to explicitly call this out in a note. (I don't think new normative wording is needed).
I propose we update the current PR as follows:Keep the current changes proposed by the PR.
Add a note after [tuple.creation]/p7 (std::tie):
[Note: The constructors and assignment operators provided by ignore shall be constexpr]
Perhaps LWG feels the existing wording is clear enough, but if not I think the above changes sufficiently clarify it.
The ability to constexpr assign to std::ignore can be important: Here is an extremely contrived example:
constexpr bool foo() {
auto res = std::tie(std::ignore);
std::get<0>(res) =42;
return true;
}
static_assert(foo());
[2017-03-04, Kona]
Set priority to 3. P/R is incorrect; it should NOT be a note. Marshall to work with Eric to get better wording. STL says "use an exposition-only class".
Proposed resolution:
This wording is relative to N4640.
Modify 20.5.3.4 [tuple.creation] as indicated:
template<class... TTypes> constexpr tuple<TTypes&...> tie(TTypes&... t) noexcept;-7- Returns: […]
-?- [Note: The constructors and assignment operators provided by ignore shall be constexpr. — end note] -8- [Example: […] — end example]
Section: 27.10.15.6 [fs.op.create_directories], 27.10.15.7 [fs.op.create_directory] Status: New Submitter: Billy Robert O'Neal III Opened: 2017-02-15 Last modified: 2017-03-05
Priority: Not Prioritized
View all issues with New status.
Discussion:
The create_directory and create_directories functions have a postcondition that says is_directory(p), but it is unclear how they are supposed to provide this. Both of their effects say that they create a directory and return whether it was actually created. It is possible to interpret this as "if creation fails due to the path already existing, issue another system call to see if the path is a directory, and change the result if so" — but it seems unfortunate to require both Windows and POSIX to issue more system calls in this case.
In email discussion Davis Herring and Billy O'Neal discussed this issue and agreed that this was probably unintentional. Special thanks for Johnathan Wakely's suggested change to create_directories' Returns clause.Proposed resolution:
This wording is relative to N4640.
Make the following edits to 27.10.15.6 [fs.op.create_directories]:
bool create_directories(const path& p); bool create_directories(const path& p, error_code& ec) noexcept;
-1- Effects: Establishes the postcondition by calling Calls create_directory() for anyeach element of p that does not exist.
-5- Complexity: 𝒪(n) where n is the number of elements of p that do not exist.
Make the following edits to 27.10.15.7 [fs.op.create_directory]:
bool create_directory(const path& p); bool create_directory(const path& p, error_code& ec) noexcept;
-1- Effects: Establishes the postcondition by attempting to createCreates the directory
p resolves to, as if by POSIX mkdir() with a second argument of static_cast<int>(perms::all).
Creation failure because p resolves to an existing directory shall not be treated asalready exists
is not an error.
Section: 99 [fs.path.compare] Status: New Submitter: Billy Robert O'Neal III Opened: 2017-02-21 Last modified: 2017-06-01
Priority: Not Prioritized
View all issues with New status.
Discussion:
Currently, path comparison is defined elementwise, which implies a conversion from the native format (implied by native() returning const string&). However, the conversion from the native format to the generic format might not be information preserving. This would allow two paths a and b to say a.compare(b) == 0, but a.native().compare(b.native()) != 0 as a result of this missing information, which is undesirable. We only want that condition to happen if there are redundant directory separators. We also don't want to change the path comparison to be in terms of the native format, due to Peter Dimov's example where we want path("a/b") to sort earlier than path("a.b"), and we want path("a/b") == path("a//////b").
Citing a Windows example, conversion to the generic format is going to have to drop alternate data streams. This might give path("a/b:ads") == path("a/b"). I think I should consider the alternate data streams as part of the path element though, so this case might be fine, so long as I make path("b:ads").native() be "b:ads". This might not work for our z/OS friends though, or for folks where the native format looks nothing like the generic format. Additionally, this treats root-directory specially. For example, the current spec wants path("c:/a/b") == path("c:/a////b"), but path("c:/a/b") != path("c:///a/b"), because native() for the root-directory path element will literally be the slashes or preferred separators. This addresses similar issues to those raised in US 57 — it won't make absolute paths sort at the beginning or end but it will make paths of the same kind sort together.[2017-03-04, Kona Saturday morning]
We decided that this had seen so much churn that we would postpone looking at this until Toronto
Proposed resolution:
This wording is relative to N4640.
Make the following edits to 99 [fs.path.compare]:
int compare(const path& p) const noexcept;-1- Returns:
— Let rootNameComparison be the result of this->root_name().native().compare(p.root_name().native()). If rootNameComparison is not 0, rootNameComparison; otherwise,
— If this->has_root_directory() and !p.has_root_directory(), a value less than 0; otherwise,
— If !this->has_root_directory() and p.has_root_directory(), a value greater than 0; otherwise,
— A value greater than, less than, or equal to 0, ordering the paths in a depth-first traversal order.
-?- [Note: For POSIX and Windows platforms, this is accomplished by lexicographically ordering the half-open ranges [begin(), end()) of this->relative_path() and p.relative_path() as follows:
— A value less than 0, if native() for the elements of
*this->relative_path() are lexicographically less than native() for the elements of p.relative_path(); otherwise,— a value greater than 0, if native() for the elements of
*this->relative_path() are lexicographically greater than native() for the elements of p.relative_path(); otherwise,— 0.
— end note]
-2- Remarks: The elements are determined as if by iteration over the half-open range [begin(), end()) for *this and p.int compare(const string_type& s) const int compare(basic_string_view<value_type> s) const;
-3- Returns: compare(path(s))[Editor's note: Delete paragraph 3 entirely and merge the value_type overload with those above.]
int compare(const value_type* s) const-4-
ReturnsEffects: Equivalent to return compare(path(s));.
Section: 27.10.15.12 [fs.op.equivalent] Status: Tentatively Ready Submitter: Billy Robert O'Neal III Opened: 2017-02-27 Last modified: 2017-03-13
Priority: 0
View all other issues in [fs.op.equivalent].
View all issues with Tentatively Ready status.
Discussion:
See discussion on the LWG mailing list with subject "Is equivalent("existing_thing", "not_existing_thing") an error?", abreviated below:
Billy O'Neal:The existing "an error is reported" effects say that an error is reported for !exists(p1) && !exists(p2), but I'm not sure that treating equivalent("existing_thing", "not_existing_thing") as "false, no error" makes any more sense than for equivalent("not_existing_thing", "not_existing_thing").
It's also unfortunate that the current spec requires reporting an error for is_other(p1) && is_other(p2) — there's no reason that you can't give a sane answer for paths to NT pipes. (Do POSIX FIFOs give garbage answers here?)
Davis Herring:
I'm fine with an error if either path does not exist. See also Late 29: I would much prefer
file_identity identity(const path&, bool resolve = true);which would of course produce an error if the path did not exist (or, with the default resolve, was a broken symlink).
See Late 30 and 32 (31 has been resolved). FIFOs pose no trouble: you can even fstat(2) on the naked file descriptors produced by pipe(2). (That said, I observe the strange inconsistency that Linux but not macOS gives both ends of a pipe the same st_ino.) POSIX has no reason that I know of to treat any file type specially for equivalent().
Billy O'Neal:
I think such a file_identity feature would be useful but we can always add it in addition to equivalent post-C++17.
Beman Dawes:
Looks good to me. Maybe submit this as an issue right away in the hopes it can go in C++17?
[2017-03-04, Kona]
Set priority to 0; Tentatively Ready
Proposed resolution:
This wording is relative to N4640.
Make the following edits to 27.10.15.12 [fs.op.equivalent]:
bool equivalent(const path& p1, const path& p2); bool equivalent(const path& p1, const path& p2, error_code& ec) noexcept;
-1- Let s1 and s2 be file_statuss determined as if by status(p1) and status(p2), respectively.-2- Effects: Determines s1 and s2. If (!exists(s1) && !exists(s2)) || (is_other(s1) && is_other(s2)) an error is reported (27.10.7).-3- Returns: true, ifs1 == s2 andp1 and p2 resolve to the same file system entity, else false. The signature with argument ec returns false if an error occurs. -4- Two paths are considered to resolve to the same file system entity if two candidate entities reside on the same device at the same location. [Note: On POSIX platforms, tThis is determined as if by the values of the POSIX stat structure, obtained as if by stat() for the two paths, having equal st_dev values and equal st_ino values. — end note] -?- Remarks: !exists(p1) || !exists(p2) is an error. -5- Throws: As specified in 27.10.7.
Section: 21.4.2 [string.view.template] Status: New Submitter: Antony Polukhin Opened: 2017-03-01 Last modified: 2017-03-13
Priority: 2
View all issues with New status.
Discussion:
Force the literal type requirement for the const_iterator in the std::basic_string_view so that iterators of std::basic_string_view could be used in constexpr functions.
[2017-03-02, Kona]
Wording tweaks suggested by LWG applied.
[2017-03-02, Tim Song comments]
I don't believe the blanket "all operations" wording is quite correct.
First, T t; (required by DefaultConstructible) isn't usable in a constant expression if the iterator is a pointer, since it would leave it uninitialized. Second, an explicit destructor call u.~T() (required by Destructible) isn't usable if the iterator is a class type because it explicitly invokes a non-constexpr function (the destructor); see [expr.const]/2.2.[2017-03-04, Kona]
Set priority to 2. Lisa and Alisdair to work with Antony to come up with better wording. The same wording can be used for 2897.
Proposed resolution:
This wording is relative to N4640.
Add to the end of the 21.4.2 [string.view.template] section:
-1- In every specialization basic_string_view<charT, traits>, the type traits shall satisfy the character traits requirements (21.2), and the type traits::char_type shall name the same type as charT.
-?- All operations on iterator and const_iterator that are required to satisfy the random access iterator requirements are usable in constant expressions.
Section: 20.15.2 [meta.type.synop] Status: New Submitter: Daniel Krügler Opened: 2017-03-02 Last modified: 2017-03-13
Priority: 2
View other active issues in [meta.type.synop].
View all other issues in [meta.type.synop].
View all issues with New status.
Discussion:
LWG 2797 (RU 2) suggests that certain type-traits should be required to diagnose violations of their pre-conditions. The basic idea is founded and I see no problems for requiring this for the mentioned traits alignment_of or is_base_of, for example. But if we want to require this diagnostics for some other traits, such as is_convertible, is_constructible (and friends), or is_callable (and possibly some others), we really should be sure that our current requirements are OK.
Unfortunately, there exists some cases, where we currently overspecify imposing complete type requirements where they are not actually required. For example, for the following situation the answer of the trait could be given without ever needing the complete type of X:struct X; // Never defined static_assert(std::is_convertible_v<X, const X&>);
Unfortunately we cannot always allow incomplete types, because most type constructions or conversions indeed require a complete type, so generally relaxing the current restrictions is also not an option.
The core language has a solution for this "small" gap of situations, where the response of the compiler might depend on type completeness: Undefined behaviour. So, I believe we need a somewhat more detailled form to express the intend here. Informally, I would suggest that the program should only be ill-formed in the situation described by LWG 2797, if there exists the possibility that the compiler would require complete types for the considered operation. The example shown above, std::is_convertible_v<X, const X&>, would never require the need to complete X, so here no violation should exist. The presented example might seem a tiny one, but the Standard Library type traits are extreme fundamental tools and we should try to not give the impression that an approximate rule of the current type constraints breaks reasonable code. It is correct, that above example has currently undefined behaviour due to the breakage of pre-conditions, therefore this issue suggests to fix the current situation before enforcing a diagnostic for such valid situations.[2017-03-04, Kona]
Set priority to 2. Is related to 2797, but really needs an audit of the type traits.
Proposed resolution:
Section: D.11 [depr.meta.types] Status: Tentatively Ready Submitter: Daniel Krügler Opened: 2017-03-04 Last modified: 2017-06-15
Priority: 3
View all other issues in [depr.meta.types].
View all issues with Tentatively Ready status.
Discussion:
P0604R0 missed a similar adjustment that was performed for the deprecated type trait templates is_literal_type and is_literal_type_v via LWG 2838:
Moving the result_of to Annex D means that the general prohibition against specializing type traits in [meta.type.synop]/1 does no longer exist, so should be explicitly spelled out. Wording will be provided after publication of the successor of N4640.[2017-03-04, Kona]
Set priority to 3
[2017-04-24, Daniel provides wording]
Instead of enumerating all the templates where adding a specialization is valid, a general exclusion rule is provided. Albeit currently not needed for the templates handled by this sub-clause, a potential exception case is constructed to ensure correctness for the first deprecated trait template that would permit specializations by the user.
[2017-06-10, Moved to Tentatively Ready after 6 positive votes on c++std-lib]
Proposed resolution:
This wording is relative to N4659.
Change D.11 [depr.meta.types] as indicated:
-4- The behavior of a program that adds specializations for any of the templates defined in this subclause
is_literal_type or is_literal_type_vis undefined, unless explicitly permitted by the specification of the corresponding template.
Section: 30.2.4 [thread.req.timing] Status: New Submitter: Jonathan Mcdougall Opened: 2017-03-07 Last modified: 2017-03-12
Priority: Not Prioritized
View all other issues in [thread.req.timing].
View all issues with New status.
Discussion:
In 30.2.4 [thread.req.timing], both /3 and /4 talk about "member functions whose names end in _for" and "_until", but these clauses also apply to this_thread::sleep_for() and this_thread::sleep_until(), which are namespace-level functions (30.3.2).
Proposed resolution:
This wording is relative to N4640.
Modify 30.2.4 [thread.req.timing] as indicated::
[…]
-3- Thememberfunctions whose names end in _for take an argument that specifies a duration. […] -4- Thememberfunctions whose names end in _until take an argument that specifies a time point. […]
Section: 20.11.2.3.5 [util.smartptr.weak.obs] Status: Tentatively Ready Submitter: Tim Song Opened: 2017-03-09 Last modified: 2017-06-15
Priority: Not Prioritized
View all other issues in [util.smartptr.weak.obs].
View all issues with Tentatively Ready status.
Discussion:
The NB comment asked for noexcept on shared_ptr::owner_before, weak_ptr::owner_before, and owner_less::operator(), but the PR of LWG 2873 only added it to the first and the third one.
[2017-06-03, Moved to Tentatively Ready after 6 positive votes on c++std-lib]
Proposed resolution:
This wording is relative to N4659.
Edit 20.11.2.3 [util.smartptr.weak], class template weak_ptr synopsis, as indicated:
[…] // 20.11.2.3.5 [util.smartptr.weak.obs], observers […] template<class U> bool owner_before(const shared_ptr<U>& b) const noexcept; template<class U> bool owner_before(const weak_ptr<U>& b) const noexcept;
Edit 20.11.2.3.5 [util.smartptr.weak.obs] immediately before p4 as indicated:
template<class U> bool owner_before(const shared_ptr<U>& b) const noexcept; template<class U> bool owner_before(const weak_ptr<U>& b) const noexcept;-4- Returns: An unspecified value such that […]
Section: 27.9.2.3 [filebuf.members] Status: New Submitter: Tim Song Opened: 2017-03-09 Last modified: 2017-03-19
Priority: Not Prioritized
View all other issues in [filebuf.members].
View all issues with New status.
Discussion:
LWG 2676 specified basic_filebuf::open(const std::filesystem::path::value_type* s, ios_base::openmode mode) by simply reusing the specification for the const char* overload, but that specification is incorrect for the wide overload: it says that s is an NTBS — a null-terminated byte string — which it isn't. Moreover, it specifies that the file is opened as if by calling fopen(s, modstr), but that call is ill-formed if s isn't a const char*.
Proposed resolution:
Section: 20.11.1.2.1 [unique.ptr.single.ctor] Status: New Submitter: Tim Song Opened: 2017-03-11 Last modified: 2017-05-31
Priority: Not Prioritized
View all other issues in [unique.ptr.single.ctor].
View all issues with New status.
Discussion:
The wording simplification in LWG 2905 accidentally deleted the requirement that construction of the deleter doesn't throw an exception. While this isn't the end of the world since any exception will just run into the noexcept on the constructor and result in a call to std::terminate(), the other unique_ptr constructors still have a similar no-exception Requires: clause, leaving us in the odd situation where throwing an exception results in undefined behavior for some constructors and terminate() for others. If guaranteeing std::terminate() on exception is desirable, that should be done across the board.
The proposed wording below simply restores the nothrow requirement along with the Copy/MoveConstructible requirement. Wording for the other alternative (guaranteed std::terminate()) can be produced if desired.[2017-03-16, Daniel comments]
The publication of the new working draft is awaited, before proposed wording against that new working draft is formally possible.
[2017-05-03, Tim comments]
The suggested wording has been moved to the PR section now that the new working draft is available.
Proposed resolution:
This wording is relative to N4659.
Insert a paragraph after 20.11.1.2.1 [unique.ptr.single.ctor] p11:
unique_ptr(pointer p, see below d1) noexcept; unique_ptr(pointer p, see below d2) noexcept;-9- The signature of these constructors depends upon whether D is a reference type. If D is a non-reference type A, then the signatures are:
[…] -10- If D is an lvalue reference type A&, then the signatures are: […] -11- If D is an lvalue reference type const A&, then the signatures are: […] -??- Requires: For the first constructor, if D is not a reference type, D shall satisfy the requirements of CopyConstructible and such construction shall not exit via an exception. For the second constructor, if D is not a reference type, D shall satisfy the requirements of MoveConstructible and such construction shall not exit via an exception.
Section: 20.6.8 [optional.comp_with_t] Status: New Submitter: Jonathan Wakely Opened: 2017-03-13 Last modified: 2017-03-19
Priority: Not Prioritized
View all other issues in [optional.comp_with_t].
View all issues with New status.
Discussion:
LWG 2934 added an additional template parameter to the comparison operators for std::optional, but the ones that compare U with optional<T> have the parameters backwards compared to the function parameters:
template <class T, class U> constexpr bool operator==(const U&, const optional<T>&);
Ville confirmed there's no particular reason for this, it's just how he wrote the proposed resolution, but as this has normative effect we should consider if we really want the template parameters and function parameters to be in different orders or not.
Proposed resolution:
Section: 21.3.2.2 [string.cons], 21.3.2.6.2 [string.append], 21.3.2.6.3 [string.assign], 21.3.2.7 [string.ops] Status: New Submitter: Daniel Krügler Opened: 2017-03-17 Last modified: 2017-03-19
Priority: Not Prioritized
View all other issues in [string.cons].
View all issues with New status.
Discussion:
LWG 2758 corrected newly introduced ambiguities of std::string::assign and other functions that got new overloads taking a basic_string_view as argument, but the assignment operator of basic_string and other functions taking a parameter of type basic_string_view<charT, traits> were not corrected. Similar to the previous issue the following operations lead now to an ambiguity as well:
#include <string>
int main()
{
std::string s({"abc", 1});
s = {"abc", 1};
s += {"abc", 1};
s.append({"abc", 1});
s.assign({"abc", 1});
s.insert(0, {"abc", 1});
s.replace(0, 1, {"abc", 1});
s.replace(s.cbegin(), s.cbegin(), {"abc", 1});
s.find({"abc", 1});
s.rfind({"abc", 1});
s.find_first_of({"abc", 1});
s.find_last_of({"abc", 1});
s.find_first_not_of({"abc", 1});
s.find_last_not_of({"abc", 1});
s.compare({"abc", 1});
s.compare(0, 1, {"abc", 1});
}
The right fix is to convert all member functions taken a basic_string_view<charT, traits> parameter into constrained function templates.
When doing so, it turns out that there occurs an additional problem: The functions that had been massaged by LWG 2758 are all functions that are not specified to be noexcept, but the wider range of "string operation" functions taking a basic_string_view<charT, traits> parameter are mostly noexcept because they had a wide contract. Now with the approach of LWG 2758, there are all types allowed that are convertible to basic_string_view<charT, traits>, but the conversion occurs now in the function body, not outside of it. So, if these conversion would be potentially exception-throwing, this would lead to a call to std::terminate, which is a semantic change compared to the previous specification. There are several options to handle this situation:Ignore that and let std::terminate come into action. This is a different way of saying that we impose the requirement of a nothrowing operation.
Remove noexcept from all the affected functions.
Make these functions conditionally noexcept.
The submitter has a personal preference for option (3), except that this would complicate the wording a bit, because unfortunately there exists yet no trait std::is_nothrow_convertible (See LWG 2040). A seemingly low-hanging fruit would be the attempt to use std::is_nothrow_constructible instead, but this trait describes a potentially different initialization context and is therefore inappropriate. Option (1) would conserve the existing noexcept guarantee for all non-throwing conversions, but now these functions become narrow-contract functions and at least according to the current noexcept guidelines such functions should not be marked as noexcept. But there are exceptions possible for that rule, and the initially suggested proposed wording below argues that this exception is reasonable here, because the required wording fixes just an unintended side-effects of transforming the functions into functions templates, but it doesn't intend to change the actual functionality.
Some of the below suggested overload exclusion constraints technically don't require the additional is_convertible_v<const T&, const charT*> == false requirement, but the submitter of this issue suggests a more advanced approach that should be applied in a synchronous wording adjustment combined with the existing LWG 2758 wording: It would presumably life easier for implementations (which are allowed to provide additional member function overloads as conforming extensions), when we would define a mini requirement set for template parameter type T below:is_convertible_v<const T&, basic_string_view<charT, traits>> is true.
is_convertible_v<const T&, const charT*> is false.
The implicit conversion to basic_string_view<charT, traits> shall not throw an exception.
But the corresponding slightly revised wording taking advantage of this "concept-like" requirements set will not be suggested before the upcoming working draft has been published to allow a simpler coordinated adjustment together with the LWG 2758 wording.
It should also be noted that these changes have impact on deduction behaviour and therefore may require further adjustments of the deduction rules.Proposed resolution:
This wording is relative to N4640.
Edit 21.3.2 [basic.string], class template basic_string synopsis, as indicated:
[…] // 21.3.2.2, construct/copy/destroy […] template<class T> explicit basic_string(basic_string_view<charT, traits> svconst T& t, const Allocator& a = Allocator()); […] template<class T> basic_string& operator=(basic_string_view<charT, traits> svconst T& t); basic_string& operator=(const charT* s); […] // 21.3.2.6, modifiers […] template<class T> basic_string& operator+=(basic_string_view<charT, traits> svconst T& t); […] template<class T> basic_string& append(basic_string_view<charT, traits> svconst T& t); […] template<class T> basic_string& assign(basic_string_view<charT, traits> svconst T& t); […] template<class T> basic_string& insert(size_type pos,basic_string_view<charT, traits> svconst T& t); […] template<class T> basic_string& replace(size_type pos1, size_type n1,basic_string_view<charT, traits> svconst T& t); […] template<class T> basic_string& replace(const_iterator i1, const_iterator i2,basic_string_view<charT, traits> svconst T& t); […] // 21.3.2.7, string operations […] template<class T> size_type find (basic_string_view<charT, traits> svconst T& t, size_type pos = 0) const noexcept; […] template<class T> size_type rfind(basic_string_view<charT, traits> svconst T& t, size_type pos = npos) const noexcept; […] template<class T> size_type find_first_of(basic_string_view<charT, traits> svconst T& t, size_type pos = 0) const noexcept; […] template<class T> size_type find_last_of (basic_string_view<charT, traits> svconst T& t, size_type pos = npos) const noexcept; […] template<class T> size_type find_first_not_of(basic_string_view<charT, traits> svconst T& t, size_type pos = 0) const noexcept; […] template<class T> size_type find_last_not_of (basic_string_view<charT, traits> svconst T& t, size_type pos = npos) const noexcept; […] template<class T> int compare(basic_string_view<charT, traits> svconst T& t) const noexcept; […] template<class T> int compare(size_type pos1, size_type n1,basic_string_view<charT, traits> svconst T& t) const; […]
Edit 21.3.2.2 [string.cons] as indicated:
template<class T> explicit basic_string(basic_string_view<charT, traits> svconst T& t, const Allocator& a = Allocator());-9- Effects:
-?- Remarks: This constructor shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.Same as basic_string(sv.data(), sv.size(), a).Creates a variable, sv, as if by basic_string_view<charT, traits> sv = t; and then behaves the same as basic_string(sv.data(), sv.size(), a).[…]
template<class T> basic_string& operator=(basic_string_view<charT, traits> svconst T& t);-25- Effects: Equivalent to:
{ basic_string_view<charT, traits> sv = t; return assign(sv); }-?- Remarks: This function shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.
Edit 21.3.2.6.1 [string.op+=] as indicated:
template<class T> basic_string& operator+=(basic_string_view<charT, traits> svconst T& t);-3- Effects: Creates a variable, sv, as if by basic_string_view<charT, traits> sv = t; and then c
-4- Returns: *this. -?- Remarks: This function shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.Calls append(sv).
Edit 21.3.2.6.2 [string.append] as indicated:
template<class T> basic_string& append(basic_string_view<charT, traits> svconst T& t);-6- Effects: Equivalent to:
{ basic_string_view<charT, traits> sv = t; return append(sv.data(), sv.size()); }-?- Remarks: This function shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.
Edit 21.3.2.6.3 [string.assign] as indicated:
template<class T> basic_string& assign(basic_string_view<charT, traits> svconst T& t);-8- Effects: Equivalent to:
{ basic_string_view<charT, traits> sv = t; return assign(sv.data(), sv.size()); }-?- Remarks: This function shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.
Edit 21.3.2.6.4 [string.insert] as indicated:
template<class T> basic_string& insert(size_type pos,basic_string_view<charT, traits> svconst T& t);-5- Effects: Equivalent to:
{ basic_string_view<charT, traits> sv = t; return insert(pos, sv.data(), sv.size()); }-?- Remarks: This function shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.
Edit 21.3.2.6.6 [string.replace] as indicated:
template<class T> basic_string& replace(size_type pos1, size_type n1,basic_string_view<charT, traits> svconst T& t);-5- Effects: Equivalent to:
{ basic_string_view<charT, traits> sv = t; return replace(pos1, n1, sv.data(), sv.size()); }-?- Remarks: This function shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.
[…]
template<class T> basic_string& replace(const_iterator i1, const_iterator i2,basic_string_view<charT, traits> svconst T& t);-21- Requires: [begin(), i1) and [i1, i2) are valid ranges.
-22- Effects: Creates a variable, sv, as if by basic_string_view<charT, traits> sv = t; and then cCalls replace(i1 - begin(), i2 - i1, sv). -23- Returns: *this.-?- Remarks: This function shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.
Edit 21.3.2.7.2 [string.find] as indicated:
template<class T> size_type find(basic_string_view<charT, traits> svconst T& t, size_type pos = 0) const noexcept;-?- Requires: The initialization of sv, as specified below, shall not throw an exception.
-1- Effects: Creates a variable, sv, as if by basic_string_view<charT, traits> sv = t; and then dDetermines the lowest position xpos, if possible, such that both of the following conditions hold: […] -2- Returns: xpos if the function can determine such a value for xpos. Otherwise, returns npos.-?- Remarks: This function shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.
Edit 21.3.2.7.3 [string.rfind] as indicated:
template<class T> size_type rfind(basic_string_view<charT, traits> svconst T& t, size_type pos = npos) const noexcept;-?- Requires: The initialization of sv, as specified below, shall not throw an exception.
-1- Effects: Creates a variable, sv, as if by basic_string_view<charT, traits> sv = t; and then dDetermines the highest position xpos, if possible, such that both of the following conditions hold: […] -2- Returns: xpos if the function can determine such a value for xpos. Otherwise, returns npos.-?- Remarks: This function shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.
Edit 21.3.2.7.4 [string.find.first.of] as indicated:
template<class T> size_type find_first_of(basic_string_view<charT, traits> svconst T& t, size_type pos = 0) const noexcept;-?- Requires: The initialization of sv, as specified below, shall not throw an exception.
-1- Effects: Creates a variable, sv, as if by basic_string_view<charT, traits> sv = t; and then dDetermines the lowest position xpos, if possible, such that both of the following conditions hold: […] -2- Returns: xpos if the function can determine such a value for xpos. Otherwise, returns npos.-?- Remarks: This function shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.
Edit 21.3.2.7.5 [string.find.last.of] as indicated:
template<class T> size_type find_last_of(basic_string_view<charT, traits> svconst T& t, size_type pos = npos) const noexcept;-?- Requires: The initialization of sv, as specified below, shall not throw an exception.
-1- Effects: Creates a variable, sv, as if by basic_string_view<charT, traits> sv = t; and then dDetermines the highest position xpos, if possible, such that both of the following conditions hold: […] -2- Returns: xpos if the function can determine such a value for xpos. Otherwise, returns npos.-?- Remarks: This function shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.
Edit 21.3.2.7.6 [string.find.first.not.of] as indicated:
template<class T> size_type find_first_not_of(basic_string_view<charT, traits> svconst T& t, size_type pos = 0) const noexcept;-?- Requires: The initialization of sv, as specified below, shall not throw an exception.
-1- Effects: Creates a variable, sv, as if by basic_string_view<charT, traits> sv = t; and then dDetermines the lowest position xpos, if possible, such that both of the following conditions hold: […] -2- Returns: xpos if the function can determine such a value for xpos. Otherwise, returns npos.-?- Remarks: This function shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.
Edit 21.3.2.7.7 [string.find.last.not.of] as indicated:
template<class T> size_type find_last_not_of(basic_string_view<charT, traits> svconst T& t, size_type pos = npos) const noexcept;-?- Requires: The initialization of sv, as specified below, shall not throw an exception.
-1- Effects: Creates a variable, sv, as if by basic_string_view<charT, traits> sv = t; and then dDetermines the highest position xpos, if possible, such that both of the following conditions hold: […] -2- Returns: xpos if the function can determine such a value for xpos. Otherwise, returns npos.-?- Remarks: This function shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.
Edit 21.3.2.7.9 [string.compare] as indicated:
template<class T> int compare(basic_string_view<charT, traits> svconst T& t) const noexcept;-?- Requires: The initialization of sv, as specified below, shall not throw an exception.
-1- Effects: Creates a variable, sv, as if by basic_string_view<charT, traits> sv = t; and then dDetermines the effective length rlen of the strings to compare as the smaller of size() and sv.size(). The function then compares the two strings by calling traits::compare(data(), sv.data(), rlen). -2- Returns: The nonzero result if the result of the comparison is nonzero. Otherwise, returns a value as indicated in Table 63. […]-?- Remarks: This function shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.
template<class T> int compare(size_type pos1, size_type n1,basic_string_view<charT, traits> svconst T& t) const;-3- Effects: Equivalent to:
{ basic_string_view<charT, traits> sv = t; return basic_string_view<charT, traits>(data(), size()).substr(pos1, n1).compare(sv); }-?- Remarks: This function shall not participate in overload resolution unless is_convertible_v<const T&, basic_string_view<charT, traits>> is true and is_convertible_v<const T&, const charT*> is false.
Section: 99 [fs.enum.path.format], 99 [fs.class.directory_iterator], 27.10.15.3 [fs.op.copy] Status: New Submitter: Thomas Köppe Opened: 2017-03-14 Last modified: 2017-05-31
Priority: Not Prioritized
View all issues with New status.
Discussion:
During the application of P0430R2, we came across several terms that seem insufficiently clear and lacking proper definitions.
We would like clarification on what those terms mean, and we would welcome wording suggestions, or alternatively a verbose explanation and dispensation to change the presentation editorially. The items in question are:What does it mean to 'always interpret [the character sequence] in the same way'? ([fs.enum.path.format], which is [path.format] in the paper)
What is a "directory-like" file type? ([fs.class.directory_iterator])
What is an "implementation-defined file type"? (e.g. [fs.op.copy])
Proposed resolution:
Section: 20.11.1 [unique.ptr] Status: New Submitter: Peter Dimov Opened: 2017-03-19 Last modified: 2017-05-31
Priority: Not Prioritized
View all other issues in [unique.ptr].
View all issues with New status.
Discussion:
shared_ptr does define operator<<, and unique_ptr should too, for consistency and usability reasons.
Proposed resolution:
This wording is relative to N4659.
Change 20.10.2 [memory.syn], header <memory> synopsis, as indicated:
namespace std {
[…]
// 20.11.1 [unique.ptr], class template unique_ptr
[…]
template <class T, class D>
bool operator>=(nullptr_t, const unique_ptr<T, D>& y);
template<class E, class T, class Y, class D>
basic_ostream<E, T>& operator<< (basic_ostream<E, T>& os, const unique_ptr<Y, D>& p);
[…]
}
Change 20.11.1 [unique.ptr], class template unique_ptr synopsis, as indicated:
namespace std {
[…]
template <class T, class D>
bool operator>=(nullptr_t, const unique_ptr<T, D>& y);
template<class E, class T, class Y, class D>
basic_ostream<E, T>& operator<< (basic_ostream<E, T>& os, const unique_ptr<Y, D>& p);
}
Add a new subclause following subclause 20.11.1.5 [unique.ptr.special] as indicated:
23.11.1.?? unique_ptr I/O [unique.ptr.io]
template<class E, class T, class Y, class D> basic_ostream<E, T>& operator<< (basic_ostream<E, T>& os, const unique_ptr<Y, D>& p);-?- Effects: Equivalent to os << p.get();
-?- Returns: os. -?- Remarks: This function shall not participate in overload resolution unless os << p.get() is a valid expression.
Section: 17 [library] Status: New Submitter: Jens Maurer Opened: 2017-03-20 Last modified: 2017-05-31
Priority: Not Prioritized
View all other issues in [library].
View all issues with New status.
Discussion:
This is from editorial issue #1088:
It is not always made explicit whether the requirement is referring to time or space complexity, or both."Linear time." vs. "Linear."
"Constant time." vs. "Constant."
17.4.1.4 [structure.specifications] says that the Complexity element specifies "the time and/or space complexity of the function", so being explicit about this would be good.
Examples:[sequence.reqmts] a.clear()
[deque.cons]
[vector.cons]
[map.cons]
[multimap.cons]
[set.cons]
[multiset.cons]
[unord.map.cnstr]
[unord.multimap.cnstr]
etc.
Proposed resolution:
Section: 99 [support.types.byteops] Status: New Submitter: Thomas Köppe Opened: 2017-03-24 Last modified: 2017-05-31
Priority: Not Prioritized
View all issues with New status.
Discussion:
The operations for std::byte (99 [support.types.byteops]) are currently specified to have undefined behaviour in general cases, since the type of the expression expr that appears in return byte(expr) is obtained by the arithmetic conversion rules and has higher conversion rank than unsigned char. Therefore, the value of the expression may be outside the range of the enum (for example, consider ~0), and by 5.2.9 [expr.static.cast] p10 the conversion results in undefined behaviour.
I believe the original intent of the specification could be expressed correctly with the following, more verbose sequence of casts. I will only give one representative example:byte operator<<(byte b, IntType shift)Equivalent to: return byte(static_cast<unsigned char>(static_cast<unsigned char>(b) << shift));
Proposed resolution:
Section: 24.4.1 [iterator.traits] Status: New Submitter: Billy Robert O'Neal III Opened: 2017-03-24 Last modified: 2017-05-31
Priority: Not Prioritized
View other active issues in [iterator.traits].
View all other issues in [iterator.traits].
View all issues with New status.
Discussion:
A customer recently triggered an unexpected SFINAE condition with class path. We constrain the constructor that takes const Source& by asking if iterator_traits<Source>::value_type is one that's OK. This forms iterator_traits<void*>, which is a hard error, as it tries to form void&.
Pointers-to-non-object-type can never act as iterators, as they don't support pointer arithmetic (as we recently discovered in atomic<void*> / atomic<int(*)(int)>).Proposed resolution:
This wording is relative to N4659.
Change 24.4.1 [iterator.traits] as indicated:
-3- It is specialized for pointers as
namespace std { template<class T> struct iterator_traits<T*> { using difference_type = ptrdiff_t; using value_type = T; using pointer = T*; using reference = T&; using iterator_category = random_access_iterator_tag; }; }and for pointers to const as
namespace std { template<class T> struct iterator_traits<const T*> { using difference_type = ptrdiff_t; using value_type = T; using pointer = const T*; using reference = const T&; using iterator_category = random_access_iterator_tag; }; }-?- These partial specializations for pointers apply only when T is an object type.
Section: 24.4.1 [iterator.traits] Status: New Submitter: Billy Robert O'Neal III Opened: 2017-03-27 Last modified: 2017-05-31
Priority: Not Prioritized
View other active issues in [iterator.traits].
View all other issues in [iterator.traits].
View all issues with New status.
Discussion:
iterator_traits accepts pointer to volatile T*, but then says that the value_type is volatile T, instead of T, which is inconsistent for what it does for pointer to const T. We should either reject volatile outright or give the right answer.
[2017-03-30, David Krauss comments]
volatile pointers may not be well-behaved random-access iterators. When simple access incurs side effects, the multiple-pass guarantee depends on underlying (hardware) semantics.
Proposed resolution:
This wording is relative to N4659.
Change 24.3 [iterator.synopsis] as indicated:
// 24.4 [iterator.primitives], primitives template<class Iterator> struct iterator_traits; template<class T> struct iterator_traits<T*>;template<class T> struct iterator_traits<const T*>;
Change 24.4.1 [iterator.traits] as indicated:
-3- It is specialized for pointers as
namespace std { template<class T> struct iterator_traits<T*> { using difference_type = ptrdiff_t; using value_type = remove_cv_t<T>; using pointer = T*; using reference = T&; using iterator_category = random_access_iterator_tag; }; }
and for pointers to const asnamespace std { template<class T> struct iterator_traits<const T*> { using difference_type = ptrdiff_t; using value_type = T; using pointer = const T*; using reference = const T&; using iterator_category = random_access_iterator_tag; }; }
Section: 23.3.8.4 [deque.modifiers] Status: New Submitter: Tim Song Opened: 2017-03-30 Last modified: 2017-05-31
Priority: Not Prioritized
View all other issues in [deque.modifiers].
View all issues with New status.
Discussion:
Most of the discussion of LWG 2853 applies, mutatis mutandis, to deque::erase. The relevant requirements table requires neither Copy/MoveInsertable nor Copy/MoveConstructible for the erase operations, so there's no way a copy/move constructor can safely be called.
And "assignment operator or move assignment operator" is just "assignment operator", since "move assignment operator" is just a species of "assignment operator".Proposed resolution:
This wording is relative to N4659.
Change 23.3.8.4 [deque.modifiers] as indicated:
iterator erase(const_iterator position); iterator erase(const_iterator first, const_iterator last); void pop_front(); void pop_back();-4- Effects: […]
-5- Complexity: The number of calls to the destructor of T is the same as the number of elements erased, but the number of calls to the assignment operator of T is no more than the lesser of the number of elements before the erased elements and the number of elements after the erased elements. -6- Throws: Nothing unless an exception is thrown by thecopy constructor, move constructor,assignment operator, or move assignment operatorof T.
Section: 17.5.4 [constraints] Status: Tentatively Ready Submitter: Tim Song Opened: 2017-03-31 Last modified: 2017-06-15
Priority: Not Prioritized
View all issues with Tentatively Ready status.
Discussion:
There's currently no rule against specializing the various _v convenience variable templates outside of <type_traits>.
There should be; foo_v<T> should be always equal to foo<T>::value. The correct way to influence, say, is_placeholder_v<T> should be to specialize is_placeholder, not is_placeholder_v. Otherwise, the editorial changes to use the _v form to the specification would no longer be editorial but have normative impact. Adding a global prohibition in 17.5.4.2.1 [namespace.std] seems preferable to adding individual prohibitions to each affected template; the PR below carves out an exception for variable templates that are intended to be specialized by users. As far as I know there are no such templates in the current WP, but the Ranges TS does use them.[2017-06-14, Moved to Tentatively Ready after 6 positive votes on c++std-lib]
Proposed resolution:
This wording is relative to N4659.
Add a paragraph to 17.5.4.2.1 [namespace.std], before p2:
-1- The behavior of a C++ program is undefined if it adds declarations or definitions to namespace std or to a namespace within namespace std unless otherwise specified. A program may add a template specialization for any standard library template to namespace std only if the declaration depends on a user-defined type and the specialization meets the standard library requirements for the original template and is not explicitly prohibited.(footnote: […])
-?- The behavior of a C++ program is undefined if it declares an explicit or partial specialization of any standard library variable template, except where explicitly permitted by the specification of that variable template. -2- The behavior of a C++ program is undefined if it declares […]
Section: 20.2 [utility] Status: New Submitter: Jens Maurer Opened: 2017-04-11 Last modified: 2017-05-31
Priority: Not Prioritized
View other active issues in [utility].
View all other issues in [utility].
View all issues with New status.
Discussion:
We added from_chars_result and to_chars_result into <utility>, but those contain error_code instances, which come from <system_error>. This creates a circular dependency issue which is somewhat difficult to resolve, as error_code depends on error_category which has virtuals that return string — it would effectively force putting all of <string> into <utility>.
Since (I believe) nobody has implemented this yet, would there be support for changing the member of from_chars_result / to_chars_result to an errc instead of an error_code (as a fast-track DR against 17)?Proposed resolution:
Section: 27.10.15.2 [fs.op.canonical] Status: New Submitter: Sergey Zubkov Opened: 2017-04-21 Last modified: 2017-05-31
Priority: Not Prioritized
View all issues with New status.
Discussion:
This is from editorial issue #1620:
Since the resolution of US-78 was applied (as part of P0492R2), std::filesystem::absolute(path, base) no longer exists. However, std::filesystem::canonical is still defined in terms of absolute(p, base).Proposed resolution:
Section: 20.14.11.3 [func.bind.bind] Status: New Submitter: Tim Song Opened: 2017-05-04 Last modified: 2017-05-31
Priority: Not Prioritized
View all other issues in [func.bind.bind].
View all issues with New status.
Discussion:
According to 20.14.11.3 [func.bind.bind]/1.2,
fd is an lvalue of type FD constructed from std::forward<F>(f),
and then uses fd throughout the specification, seemingly without regard to the cv-qualification of the call wrapper g. But this definition means that fd is always cv-unqualified, rather than having its cv-qualification change with that of g as intended.
LWG 2545 accidentally exacerbated the problem by removing any hint that fd's cv-qualifier followed that of the call wrapper. A similar issue affects the type of tdi for nested binds.Proposed resolution:
This wording is relative to N4659.
Edit 20.14.11.3 [func.bind.bind] as indicated:
template<class F, class... BoundArgs> unspecified bind(F&& f, BoundArgs&&... bound_args);[…]
-3- Returns: A forwarding call wrapper g (20.14.3 [func.require]). The effect of g(u1, u2, ..., uM) shall beINVOKE(static_cast<FD cv &>(fd), std::forward<V1>(v1), std::forward<V2>(v2), ..., std::forward<VN>(vN))where cv represents the cv-qualifiers of g and the values and types of the bound arguments v1, v2, . . . , vN are determined as specified below. The copy constructor and move constructor of the forwarding call wrapper shall throw an exception if and only if the corresponding constructor of FD or of any of the types TDi throws an exception.
[…]template<class R, class F, class... BoundArgs> unspecified bind(F&& f, BoundArgs&&... bound_args);[…]
-7- Returns: A forwarding call wrapper g (20.14.3 [func.require]). The effect of g(u1, u2, ..., uM) shall beINVOKE<R>(static_cast<FD cv &>(fd), std::forward<V1>(v1), std::forward<V2>(v2), ..., std::forward<VN>(vN))where cv represents the cv-qualifiers of g and the values and types of the bound arguments v1, v2, . . . , vN are determined as specified below. The copy constructor and move constructor of the forwarding call wrapper shall throw an exception if and only if the corresponding constructor of FD or of any of the types TDi throws an exception.
[…] -10- The values of the bound arguments v1, v2, ... , vN and their corresponding types V1, V2, ... , VN depend on the types TDi derived from the call to bind and the cv-qualifiers cv of the call wrapper g as follows:
(10.1) — […]
(10.2) — if the value of is_bind_expression_v<TDi> is true, the argument is static_cast<TDi cv &>(tdi)(std::forward<Uj>(uj)...) and its type Vi is invoke_result_t<TDi cv &, Uj...>&&;
(10.3) — […]
Section: 20.4.2 [pairs.pair], 20.5.3.2 [tuple.assign] Status: New Submitter: Casey Carter Opened: 2017-05-05 Last modified: 2017-06-15
Priority: Not Prioritized
View other active issues in [pairs.pair].
View all other issues in [pairs.pair].
View all issues with New status.
Discussion:
LWG 2729 constrained many pair & tuple assignment operators to "not participate in overload resolution" and "be defined as deleted if." As discussed when LWG reviewed 2756, it is undesirable to require that a move constructor or assignment operator be "defined as deleted," since it is unclear whether that operation should be implicitly deleted, and therefore not participate in overload resolution, or explicitly deleted and therefore impede overload resolution for usages that would otherwise find copies.
Proposed resolution:
This wording is relative to N4659.
Edit 20.4.2 [pairs.pair] as indicated:
pair& operator=(pair&& p) noexcept(see below);-21- Effects: Assigns to first with std::forward<first_type>(p.first) and to second with std::forward<second_type>(p.second).
-22- Remarks: This operator shallbe defined as deletednot participate in overload resolution unless is_move_assignable_v<first_type> is true and is_move_assignable_v<second_type> is true. […]
Edit 20.5.3.2 [tuple.assign] as indicated:
tuple& operator=(tuple&& u) noexcept(see below);-5- Effects: For all i, assigns std::forward<Ti>(get<i>(u)) to get<i>(*this).
-6- Remarks: This operator shallbe defined as deletednot participate in overload resolution unless is_move_assignable_v<Ti> is true for all i. […]
Section: 21.2.3.2 [char.traits.specializations.char16_t] Status: New Submitter: Jonathan Wakely Opened: 2017-05-05 Last modified: 2017-05-31
Priority: Not Prioritized
View all other issues in [char.traits.specializations.char16_t].
View all issues with New status.
Discussion:
The standard requires that char_traits<char16_t>::int_type is uint_least16_t, so when that has the same representation as char16_t there are no bits left to represent the eof value.
21.2.3.2 [char.traits.specializations.char16_t] says:— The member eof() shall return an implementation-defined constant that cannot appear as a valid UTF-16 code unit.
Existing practice is to use the "noncharacter" u'\uffff' for this value, but the Unicode spec is clear that U+FFFF and other noncharacters are valid, and their appearance in a UTF-16 string does not make it ill-formed. See here and here:
The fact that they are called "noncharacters" and are not intended for open interchange does not mean that they are somehow illegal or invalid code points which make strings containing them invalid.
In practice this means there's no way to tell if basic_streambuf<char16_t>::sputc(u'\uffff') succeeded or not. If it can insert the character it returns to_int_type(u'\uffff') and otherwise it returns eof(), which is the same value.
I believe that char_traits<char16_t>::to_int_type(char_type c) can be defined to transform U+FFFF into U+FFFD, so that the invariant eq_int_type(eof(), to_int_type(c)) == false holds for any c (and the return value of sputc will be distinct from eof). I don't think any implementation currently meets that invariant. I think at the very least we need to correct the statement "The member eof() shall return an implementation-defined constant that cannot appear as a valid UTF-16 code unit", because there are no such constants if sizeof(uint_least16_t) == sizeof(char16_t). This issue is closely related to LWG 1200, but there it's a slightly different statement of the problem, and neither the submitter's recommendation nor the proposed resolution solves this issue here. It seems that was closed as NAD before the Unicode corrigendum existed, so at the time our standard just gave "surprising results" but wasn't strictly wrong. Now it makes a normative statement that conflicts with Unicode.Proposed resolution:
Section: 99 [fund.ts.v2::meta] Status: New Submitter: Tim Song Opened: 2017-05-08 Last modified: 2017-05-31
Priority: Not Prioritized
View all issues with New status.
Discussion:
Addresses: fund.ts.v2
The definition of std::experimental::nonesuch (with a deleted default constructor, destructor, copy constructor, and copy assignment operator) means that it is an aggregate, which means that it can be initialized from {} in contexts where the availability of the destructor is not considered (e.g., overload resolution or a new-expression).
The deleted default constructor also has this effect standing alone, because it doesn't affect the formation of implicit conversion sequences (and hence overload resolution). The net result is ambiguities in situations like:
struct such {};
void f(const such&);
void f(const nonesuch&);
f({});
For a real-life example of such ambiguity, see GCC bug 79141, involving libstdc++'s internal __nonesuch type defined just like the one in the fundamentals TS.
I believe that nonesuch would be substantially more useful if the ICS from {} is gone. nonesuch should have no default constructor (rather than a deleted one), and it shouldn't be an aggregate.Proposed resolution:
This wording is relative to N4617.
Edit 20.15.2 [meta.type.synop] as indicated, moving the definition of nonesuch to [meta.detect]:
// [meta.detect], Detection idiom […] struct nonesuch{ nonesuch() = delete; ~nonesuch() = delete; nonesuch(nonesuch const&) = delete; void operator=(nonesuch const&) = delete; }; […]
Insert at the beginning of [meta.detect] the following paragraphs:
[Drafting note: The seemingly redundant statement about default and initializer-list constructors is intended to negate the usual leeway for implementations to declare additional member function signatures granted in 17.5.5.5 [member.functions]. — end drafting note]
struct nonesuch { ~nonesuch() = delete; nonesuch(nonesuch const&) = delete; void operator=(nonesuch const&) = delete; };-?- nonesuch has no default constructor (C++14 §[class.ctor]) or initializer-list constructor (C++14 §[dcl.init.list]), and is not an aggregate (C++14 §[dcl.init.aggr]).
Section: 20.12.4 [mem.res.global] Status: Tentatively Ready Submitter: Casey Carter Opened: 2017-05-09 Last modified: 2017-06-15
Priority: Not Prioritized
View all issues with Tentatively Ready status.
Discussion:
The specification of set_default_resource in N4659 20.12.4 [mem.res.global] reads:
memory_resource* set_default_resource(memory_resource* r) noexcept;-4- Effects: If r is non-null, sets the value of the default memory resource pointer to r, otherwise sets the default memory resource pointer to new_delete_resource().
-5- Postconditions: get_default_resource() == r. -6- Returns: The previous value of the default memory resource pointer. -7- Remarks: […]
It is apparent that the effects specified in para 4 cannot — and indeed should not — achieve the postcondition specified in para 5 when r is null.
[2017-05-13, Tim comments]
This is the same issue as LWG 2522, which just missed the Fundamentals TS working draft used for the merge. A similar resolution (simply striking the Postconditions: paragraph) might be better.
Previous resolution [SUPERSEDED]:
This wording is relative to N4659.
Modify 20.12.4 [mem.res.global] as follows:
memory_resource* set_default_resource(memory_resource* r) noexcept;-4- Effects: If r is non-null, sets the value of the default memory resource pointer to r, otherwise sets the default memory resource pointer to new_delete_resource().
-5- Postconditions: If r is non-null, get_default_resource() == r. Otherwise, get_default_resource() == new_delete_resource(). -6- Returns: The previous value of the default memory resource pointer. -7- Remarks: […]
[2017-06-13, Casey Carter revises proposed wording]
I suggest to strike the Postconditions paragraph ([mem.res.global]/5) completely, as in LWG 2522
[2017-06-15, Moved to Tentatively Ready after 6 positive votes on c++std-lib]
Proposed resolution:
This wording is relative to N4659.
Modify 20.12.4 [mem.res.global] as follows:
memory_resource* set_default_resource(memory_resource* r) noexcept;-4- Effects: If r is non-null, sets the value of the default memory resource pointer to r, otherwise sets the default memory resource pointer to new_delete_resource().
-5- Postconditions: get_default_resource() == r.-6- Returns: The previous value of the default memory resource pointer. -7- Remarks: […]
Section: 24.2.3 [input.iterators] Status: New Submitter: Gašper Ažman Opened: 2017-05-10 Last modified: 2017-05-31
Priority: Not Prioritized
View other active issues in [input.iterators].
View all other issues in [input.iterators].
View all issues with New status.
Discussion:
In Table 95 in 24.2.3 [input.iterators], it is specified that the expression *a returns reference, which must be convertible to value_type. This is not true for move-only types, which incidentally means that std::vector<std::unique_ptr<int>> does not possess even a lowly InputIterator, which is, of course, absurd.
With the advent of concepts as first-class citizens in the language, getting this right as soon as possible is a priority. This issue seems to be similar to both LWG 448 and LWG 484, but not the same. The proposed resolution stems from two considerations outlined below:Convertibility is too strong for all algorithms
No algorithm in the standard library requires convertibility to value_type. If algorithms require things that smell of that, they specify the assignment or constructibility flavor they need directly. I checked this by going through the specification of each and every one of them in <algorithm> and <numeric>, which highlighted several issues unrelated to this one. These issues are presented in Algorithms with underspecified iterator requirements (LWG 2963). reference needs to be related to value_type Algorithms need this for the following reasons:
lifetime-extension: served as adequately by T const& as by T. Also works for iterators that return by value. T&& also correctly binds to T const&.
passing to predicates: again, served adequately by T const&
writing to *result: not provided by the requirement anyway.
capture-by-copy: currently implicitly guaranteed, but unused in the standard library (always specified separately). A separate specification can always be made for algorithms that need to capture-by-copy.
We must give due consideration to code that so far required its inputs to be CopyConstructible implicitly by requiring convertibility to T. This is done in the issue LWG 2963, which presents the results of a comb-through of <algorithm> and <numeric> to find algorithms that have this requirement, but where it is not specified. While related issues have been identified, no algorithms seems to require more than T const& convertibility without separately requiring convertibility to T.
Since such code is already compiling today, relaxing this requirement does not break code. The only code this could possibly break is if, in a concept checking library, the InputIterator concept requirement on reference being convertible to value_type gets relaxed. Such a library, if it offered overloading based on most-specific modeled concept, could now, once fixed, resolve the call to a different algorithm, which could break user code that uses a hypothetical algorithm with a move-only container and was relying to select some other overload for move-only types based on the implicit CopyConstructible assertion provided by the iterator. In our internal concepts-checking library, we have had this issue "fixed" since the very beginning — move-only types were too important for our internal algorithms library, and also no algorithm in it seems to require something like Iterator::value_type x = *first without also requiring copy-constructibility anyway.
Proposed resolution:
This wording is relative to N4659.
Change Table 95 — "Input iterator requirements", 24.2.3 [input.iterators] as indicated:
Table 107 — Input iterator requirements (in addition to Iterator) Expression Return type Operational
semanticsAssertion/note pre-/post-condition … *a reference,
convertible to T
that binds to const T&[…] … *r++ convertible to T
that binds to const T&{ Tauto&& tmp = *r;
++r;
return tmp; }
Section: 25 [algorithms], 26 [numerics] Status: New Submitter: Gašper Ažman Opened: 2017-05-10 Last modified: 2017-05-31
Priority: Not Prioritized
View other active issues in [algorithms].
View all other issues in [algorithms].
View all issues with New status.
Discussion:
While researching whether the proposed resolution of Iterators of Containers of move-only types do not model InputIterator (LWG 2962), I came across several algorithms that underspecify their requirements, mostly with regard to some associated type of the iterator type they operate on. A list can be found below.
The list of algorithms with underspecified requirements from <algorithm> and <numeric> follows. With the advent of concepts, these algorithms will need better specifications if we are ever hoping to be allowed to overload based on them. I want this issue to bring the standard algorithms closer to having their concept requirements directly transcribable to library annotations. Suggested resolution:copy, copy_if, copy_n, copy_backward
Add to description: *result shall be assignable from *first.
move, move_backward
Add to description: *result shall be move-assignable from *first.
transform
Add to description: The result of the expression op(*first) or binary_op(*first1, *first2) shall be writable to result.
rotate_copy
Add to description: *first shall be writable to result.
merge
Add to description: *first1 and *first2 shall be writable to result..
set_union, set_intersection, set_difference, set_symmetric_difference
Add to description: *first1 and *first2 shall be writable to result.
partial_sum
acc is not defined.
Change description: acc, a variable of InputIterator's value type, shall be constructibleadjacent_difference
acc is not defined.
Change description: acc, a variable of InputIterator's value type, shall be MoveAssignable and shall be constructible from the type of *first.iota
iota is mis-specified. Since the expression we need to support is *first = value: *first is required to be of type InputIterator::reference, and value is an lvalue of type T. The current specification allows calling iota with a const output iterator!
Proposed resolution:
Section: 20.11.2.2.9 [util.smartptr.shared.cast] Status: New Submitter: Tim Song Opened: 2017-05-11 Last modified: 2017-05-31
Priority: Not Prioritized
View all other issues in [util.smartptr.shared.cast].
View all issues with New status.
Discussion:
Currently 20.11.2.2.9 [util.smartptr.shared.cast]/4 says:
Requires: The expression dynamic_cast<T*>((U*)nullptr) shall be well formed and shall have well defined behavior.
A dynamic_cast of a null pointer, if well-formed, always has well-defined behavior: it returns a null pointer. The second part is therefore redundant as currently worded. The C++14 version, on the other hand, requires dynamic_cast<T*>(r.get()) to have well-defined behavior, which actually adds something: it requires the user to not trigger the undefined case in [class.cdtor]/5, for instance.
Proposed resolution:
This wording is relative to N4659.
Edit 20.11.2.2.9 [util.smartptr.shared.cast] as indicated:
shared_ptr<T> dynamic_pointer_cast(const shared_ptr<U>& r) noexcept;-4- Requires: The expression dynamic_cast<T*>((U*)0) shall be well formed. The expression dynamic_cast<typename shared_ptr<T>::element_type*>(r.get()) shall be well formed and shall have well defined behavior.
[…]
Section: 27.10.9.1 [filesystem_error.members] Status: New Submitter: Daniel Krügler Opened: 2017-05-22 Last modified: 2017-05-31
Priority: Not Prioritized
View all issues with New status.
Discussion:
As pointed out by Jonathan Wakely and Bo Persson, 27.10.9.1 [filesystem_error.members]/7 refers to a non-existing function path::native_string:
Returns: A string containing runtime_error::what(). The exact format is unspecified. Implementations are encouraged but not required to include path1.native_string() if not empty, path2.native_string() if not empty, and system_error::what() strings in the returned string.
Existing implementations differ, as Jonathan also determined:
Boost.Filesystem uses path::string().
Libstdc++ uses path::string().
MSVC++/Dinkumware uses path::u8string().
It seems that libc++ doesn't include the paths in what().
We've had native_string() in the spec since N3239 (where it already didn't match any existing path function at that time).
Before that it was file_string() in N1975 (within that specification path was a template that was parametrized in the character type). Since it can't be path::native() because that might be the wrong type, one of path::string() or path::u8string() seems appropriate. Albeit the wording is just a non-binding encouragement to implementations, the decision on this matter should not be considered editorially due to the existing implementation variance. Any official resolution of the current state could cause a reconsideration of existing implementations, and therefore it should be documented.Previous resolution [SUPERSEDED]:
This wording is relative to N4659.
Edit 27.10.9.1 [filesystem_error.members] as indicated:
const char* what() const noexcept override;-7- Returns: A string containing runtime_error::what(). The exact format is unspecified. Implementations are encouraged but not required to include path1.
native_string() if not empty, path2.native_string() if not empty, and system_error::what() strings in the returned string.
[2017-05-25, Jonathan comments and suggests an alternative resolution]
The revised wording changes leave it up to the implementation which of the native format observers to use. The "if not empty" seems redundant, because if the path is empty then there's nothing to include anyway, but the proposed resolution preserves it.
Proposed resolution:
This wording is relative to N4659.
Edit 27.10.9.1 [filesystem_error.members] as indicated:
const char* what() const noexcept override;-7- Returns: A string containing runtime_error::what(). The exact format is unspecified. Implementations are encouraged but not required to include
path1.native_string() if not empty, path2.native_string() if not empty, and system_error::what() stringsthe system_error::what() string and the pathnames of path1 and path2 in the native format in the returned string.
Section: 99 [fs.path.native.obs] Status: Tentatively Ready Submitter: Jonathan Wakely Opened: 2017-05-25 Last modified: 2017-06-15
Priority: Not Prioritized
View all issues with Tentatively Ready status.
Discussion:
P0492R1 US-74 failed to replace one occurrence of pathstring, in [fs.path.native.obs] p8. It should be changed to native(), as with the other native format observers.
[2017-06-12, Moved to Tentatively Ready after 6 positive votes on c++std-lib]
Proposed resolution:
This wording is relative to N4659.
Edit 99 [fs.path.native.obs] as indicated:
std::string string() const; std::wstring wstring() const; std::string u8string() const; std::u16string u16string() const; std::u32string u32string() const;-8- Returns:
-9- Remarks: Conversion, if any, is performed as specified by [fs.path.cvt]. The encoding of the string returned by u8string() is always UTF-8.pathstringnative().
Section: 25.5.11 [alg.equal] Status: New Submitter: Gennaro Prota Opened: 2017-05-26 Last modified: 2017-06-15
Priority: Not Prioritized
View all issues with New status.
Discussion:
The description of the std::equal() algorithm in the standard doesn't make clear what the result of it is on empty ranges:
std::equal(first, first, second) ; // what does this return?
It should IMHO return true (two empty ranges are always equal).
Proposed resolution:
This wording is relative to N4659.
Edit 25.5.11 [alg.equal] as indicated:
[Drafting note: The current wording presented below uses two times the unusual phrase "[…] return […]" instead of "[…] returns […]". The project editor is kindly asked to consider to replace these unusual wording forms by the usual one. — end drafting note]
template<class InputIterator1, class InputIterator2> bool equal(InputIterator1 first1, InputIterator1 last1, InputIterator2 first2); […]-1- Remarks: If last2 was not given in the argument list, it denotes first2 + (last1 - first1) below.
-2- Returns: If [first1, last1) and [first2, last2) are both empty, returns true. If last1 - first1 != last2 - first2, return false. Otherwise return true if for every iterator i in the range [first1, last1) the following corresponding conditions hold: *i == *(first2 + (i - first1)), pred(*i, *(first2 + (i - first1))) != false. Otherwise, returns false. […]
Section: 21.3.2.4 [string.capacity] Status: New Submitter: Andrew Luo Opened: 2017-05-30 Last modified: 2017-06-15
Priority: Not Prioritized
View all other issues in [string.capacity].
View all issues with New status.
Discussion:
According to 21.3.2.4 [string.capacity] paragraph 11:
-11- Effects: After reserve(), capacity() is greater or equal to the argument of reserve. [Note: Calling reserve() with a res_arg argument less than capacity() is in effect a non-binding shrink request. A call with res_arg <= size() is in effect a non-binding shrink-to-fit request. — end note]
A call to basic_string's reserve function with res_arg <= size() is taken as a non-binding request to shrink the capacity, whereas for vector (and similarly for unordered_map and unordered_set) according to 23.3.11.3 [vector.capacity] p3:
-3- Effects: A directive that informs a vector of a planned change in size, so that it can manage the storage allocation accordingly. After reserve(), capacity() is greater or equal to the argument of reserve if reallocation happens; and equal to the previous value of capacity() otherwise. Reallocation happens at this point if and only if the current capacity is less than the argument of reserve(). If an exception is thrown other than by the move constructor of a non-CopyInsertable type, there are no effects.
The problem here is that the different behavior makes it that writing template code where the template argument type is a container type (for example std::string or std::vector<char>) calls to reserve can have different meaning depending on which container type the template is instantiated with. It might be a minor issue but it would be nice to fix the inconsistency. I ran into an issue around this when I was porting code from MSVC++ to G++ (For basic_string, MSVC++'s STL implementation, based on Dinkumware, ignores the call if res_arg < capacity() whereas GCC's STL implementation, libstdc++ will actually shrink the string. For the code I wrote this caused a huge performance issue since we were reallocating the entire string with every call to reserve. Of course we could have worked around it by doing the res_arg < capacity() check ourselves, but I think this inconsistency in the standard isn't desirable).
My proposal is to change 21.3.2.4 [string.capacity] paragraph 11 to read:-11- Effects: After reserve(), capacity() is greater or equal to the argument of reserve if reallocation happens; and equal to the previous value of capacity() otherwise. Reallocation happens at this point if and only if the current capacity is less than the argument of reserve().
I realize that this causes the basic_string::reserve to no longer have the secondary property of shrinking, but this is what shrink_to_fit is for.
Proposed resolution:
This wording is relative to N4659.
Edit 21.3.2.4 [string.capacity] as indicated:
void reserve(size_type res_arg=0);-10- The member function reserve() is a directive that informs a basic_string object of a planned change in size, so that it can manage the storage allocation accordingly.
-11- Effects: After reserve(), capacity() is greater or equal to the argument of reserve, if reallocation happens; and equal to the previous value of capacity() otherwise. Reallocation happens at this point if and only if the current capacity is less than the argument of reserve().[Note: Calling reserve() with a res_arg argument less than capacity() is in effect a non-binding shrink request. A call with res_arg <= size() is in effect a non-binding shrink-to-fit request. — end note]-12- Throws: length_error if res_arg > max_size().
Section: 20.12.3.2 [mem.poly.allocator.mem] Status: New Submitter: Pablo Halpern Opened: 2017-05-30 Last modified: 2017-06-15
Priority: Not Prioritized
View other active issues in [mem.poly.allocator.mem].
View all other issues in [mem.poly.allocator.mem].
View all issues with New status.
Discussion:
Section 20.12.3.2 [mem.poly.allocator.mem] defines the effect of polymorphic_allocator<T>::construct as:
Effects: Construct a T object in the storage whose address is represented by p by uses-allocator construction with allocator resource() and constructor arguments std::forward<Args>(args)....
The use of resource() is a hold-over from the LFTS, which contains a modified definition of uses-allocator construction. This revised definition was not carried over into the C++17 WP when allocator_resource and polymorphic_allocator were moved over.
Previous resolution [SUPERSEDED]:
This wording is relative to N4659.
Edit 20.12.3.2 [mem.poly.allocator.mem] as indicated:
template <class T, class... Args> void construct(T* p, Args&&... args);-5- Requires: Uses-allocator construction of T with allocator
-6- Effects: Construct a T object in the storage whose address is represented by p by uses-allocator construction with allocatorresource()*this (see 20.10.7.2 [allocator.uses.construction]) and constructor arguments std::forward<Args>(args)... is well-formed. [Note: Uses-allocator construction is always well formed for types that do not use allocators. — end note]resource()*this and constructor arguments std::forward<Args>(args).... -7- Throws: Nothing unless the constructor for T throws.template <class T1, class T2, class... Args1, class... Args2> void construct(pair<T1,T2>* p, piecewise_construct_t, tuple<Args1...> x, tuple<Args2...> y);-8- [Note: This method and the construct methods that follow are overloads for piecewise construction of pairs (20.4.2 [pairs.pair]). — end note]
-9- Effects: Let xprime be a tuple constructed from x according to the appropriate rule from the following list. [Note: The following description can be summarized as constructing a pair<T1, T2> object in the storage whose address is represented by p, as if by separate uses-allocator construction with allocatorresource()*this (20.10.7.2 [allocator.uses.construction]) of p->first using the elements of x and p->second using the elements of y. — end note] […]
[2017-06-12, Pablo comments]
The current description is correct and does not depend on changes to uses-allocator construction. It relies on the fact that memory_resource* is convertible to polymorphic_allocator.
[2017-06-13, Tim Song reopens]
While it is true that memory_resource* is convertible to polymorphic_allocator, uses-allocator construction still requires allocators, and a memory_resource* isn't an allocator.
To take a concrete example from the current WP, a pmr::vector<std::promise<int>>, as specified, will be attempting to uses-allocator construct a promise<int> with a memory_resource*, but std::promise's allocator-taking constructor expects something that satisfies the allocator requirements, rather than a memory_resource*.[2017-06-13, Daniel and Tim restore and improve the previously proposed wording]
Proposed resolution:
This wording is relative to N4659.
Edit 20.12.3.2 [mem.poly.allocator.mem] as indicated:
template <class T, class... Args> void construct(T* p, Args&&... args);-5- Requires: Uses-allocator construction of T with allocator
-6- Effects: Construct a T object in the storage whose address is represented by p by uses-allocator construction with allocatorresource()*this (see 20.10.7.2 [allocator.uses.construction]) and constructor arguments std::forward<Args>(args)... is well-formed. [Note: Uses-allocator construction is always well formed for types that do not use allocators. — end note]resource()*this and constructor arguments std::forward<Args>(args).... -7- Throws: Nothing unless the constructor for T throws.template <class T1, class T2, class... Args1, class... Args2> void construct(pair<T1,T2>* p, piecewise_construct_t, tuple<Args1...> x, tuple<Args2...> y);-8- [Note: This method and the construct methods that follow are overloads for piecewise construction of pairs (20.4.2 [pairs.pair]). — end note]
-9- Effects: Let xprime be a tuple constructed from x according to the appropriate rule from the following list. [Note: The following description can be summarized as constructing a pair<T1, T2> object in the storage whose address is represented by p, as if by separate uses-allocator construction with allocatorresource()*this (20.10.7.2 [allocator.uses.construction]) of p->first using the elements of x and p->second using the elements of y. — end note]
(9.1) — If uses_allocator_v<T1,
memory_resource*polymorphic_allocator> is false and is_constructible_v<T1,Args1...> is true, then xprime is x.(9.2) — Otherwise, if uses_allocator_v<T1,
memory_resource*polymorphic_allocator> is true and is_constructible_v<T1,allocator_arg_t,memory_resource*polymorphic_allocator,Args1...> is true, then xprime is tuple_cat(make_tuple(allocator_arg,resource()*this), std::move(x)).(9.3) — Otherwise, if uses_allocator_v<T1,
memory_resource*polymorphic_allocator> is true and is_constructible_v<T1,Args1...,memory_resource*polymorphic_allocator> is true, then xprime is tuple_cat(std::move(x), make_tuple(resource()*this)).(9.4) — Otherwise the program is ill formed.
Let yprime be a tuple constructed from y according to the appropriate rule from the following list:
(9.5) — If uses_allocator_v<T2,
memory_resource*polymorphic_allocator> is false and is_constructible_v<T2,Args2...> is true, then yprime is y.(9.6) — Otherwise, if uses_allocator_v<T2,
memory_resource*polymorphic_allocator> is true and is_constructible_v<T2,allocator_arg_t,memory_resource*polymorphic_allocator,Args2...> is true, then yprime is tuple_cat(make_tuple(allocator_arg,resource()*this), std::move(y)).(9.7) — Otherwise, if uses_allocator_v<T2,
memory_resource*polymorphic_allocator> is true and is_constructible_v<T2,Args2...,memory_resource*polymorphic_allocator> is true, then yprime is tuple_cat(std::move(y), make_tuple(resource()*this)).(9.8) — Otherwise the program is ill formed.
Section: 20.7.7 [variant.visit] Status: New Submitter: Tim Song Opened: 2017-05-31 Last modified: 2017-06-15
Priority: Not Prioritized
View all issues with New status.
Discussion:
[variant.visit]/1 correctly uses "type and value category", but then p3 describes the return type of visit to be "the common type of all possible INVOKE expressions of the Effects: element." The type of an expression is never a reference type, due to [expr]/5 removing the referenceness "prior to any further analysis", so this wording as written says that visit always returns a non-reference type, which is presumably not the intent.
Proposed resolution:
Section: 20.7.3 [variant.variant] Status: New Submitter: Agustín K-ballo Bergé Opened: 2017-05-31 Last modified: 2017-06-15
Priority: Not Prioritized
View other active issues in [variant.variant].
View all other issues in [variant.variant].
View all issues with New status.
Discussion:
The state of a variant is not specified if an exception is thrown during the destruction of the active alternative. Exceptions should not be allowed to escape during destruction of an object used as a variant alternative. Otherwise, it should be specified that the variant object might not hold a value.
Proposed resolution:
This wording is relative to N4659.
Edit 20.7.3 [variant.variant] as indicated:
-2- All types in Types... shall be (possibly cv-qualified) object types that are not arrays, and shall satisfy the requirements of Destructible (Table 27).
Section: 20.15.4.3 [meta.unary.prop] Status: New Submitter: Richard Smith Opened: 2017-06-01 Last modified: 2017-06-15
Priority: Not Prioritized
View other active issues in [meta.unary.prop].
View all other issues in [meta.unary.prop].
View all issues with New status.
Discussion:
The spec for is_trivially_destructible says the value is true if "is_destructible_v<T> is true and the indicated destructor is known to be trivial."
For a case like is_trivially_destructible_v<int>, there is no indicated destructor, so it's unclear what value the trait would have but the most plausible reading of these words is that it should be false. However, I'm confident the intent is that this trait should yield true in that situation, and that's what all the implementations I can find actually do.[2017-06-14, Daniel and Jonathan provide wording]
Proposed resolution:
This wording is relative to N4659.Change 20.15.4.3 [meta.unary.prop], Table 42 — "Type property predicates", as indicated:
Table 22 — Type property predicates Template Condition Preconditions … template <class T>
struct is_trivially_destructible;is_destructible_v<T> is true and the indicated destructor is known to be trivialremove_all_extents_t<T> is either a non-class type or a class type with a trivial destructor.T shall be a complete type, cv void, or an array of unknown bound. …
Section: 25.7.5 [alg.merge] Status: New Submitter: Billy Robert O'Neal III Opened: 2017-06-08 Last modified: 2017-06-15
Priority: Not Prioritized
View all other issues in [alg.merge].
View all issues with New status.
Discussion:
At the moment, inplace_merge requires exactly N - 1 comparisons, if enough additional memory is available (and in practice enough additional memory is always available). However, this prohibits implementing the merge operation using forms of binary search, as in Timsort's 'Galloping Mode', a useful optimization for non-uniform input data. It's not really useful to prohibit standard libraries from trying a few extra speculative compares like this, given that users must be prepared for the fallback "not enough memory" 𝒪(N lg N) algorithm.
Proposed resolution:
This wording is relative to N4659.
Edit 25.7.5 [alg.merge] as indicated:
template<class BidirectionalIterator> void inplace_merge(BidirectionalIterator first, BidirectionalIterator middle, BidirectionalIterator last); template<class ExecutionPolicy, class BidirectionalIterator> void inplace_merge(ExecutionPolicy&& exec, BidirectionalIterator first, BidirectionalIterator middle, BidirectionalIterator last); template<class BidirectionalIterator, class Compare> void inplace_merge(BidirectionalIterator first, BidirectionalIterator middle, BidirectionalIterator last, Compare comp); template<class ExecutionPolicy, class BidirectionalIterator, class Compare> void inplace_merge(ExecutionPolicy&& exec, BidirectionalIterator first, BidirectionalIterator middle, BidirectionalIterator last, Compare comp);[…]
-8- Complexity: Let N = last - first:
(8.1) — For the overloads with no ExecutionPolicy, if enough additional memory is available,
exactlyN - 1 comparisons on average, 𝒪(N) comparisons in the worst case.(8.2) — For the overloads with no ExecutionPolicy if no additional memory is available, 𝒪(N log N) comparisons.
(8.3) — For the overloads with an ExecutionPolicy, 𝒪(N log N) comparisons.
-9- Remarks: Stable (17.5.5.7 [algorithm.stable]).
Section: 20.4.4 [pair.astuple], 20.7.4 [variant.helper] Status: Tentatively Ready Submitter: Agustín K-ballo Bergé Opened: 2017-06-10 Last modified: 2017-06-15
Priority: Not Prioritized
View all other issues in [pair.astuple].
View all issues with Tentatively Ready status.
Discussion:
Instantiating tuple_element with an out-of-bounds index requires a diagnostic for tuple and array, but not for pair. The specification requires an out-of-bounds index for pair to go to the base template instead, which is undefined.
Similarly, instantiating variant_alternative with an out-of-bounds index violates a requirement, but it is not required to be diagnosed.[2017-06-12, Moved to Tentatively Ready after 5 positive votes on c++std-lib]
Proposed resolution:
This wording is relative to N4659.
Modify 20.2.1 [utility.syn], header <utility> synopsis, as indicated:
[…] // 20.4.4 [pair.astuple], tuple-like access to pair template <class T> class tuple_size; template <size_t I, class T> class tuple_element; template <class T1, class T2> struct tuple_size<pair<T1, T2>>; template <size_t I, class T1, class T2> struct tuple_element<I, pair<T1, T2>>;template <class T1, class T2> struct tuple_element<0, pair<T1, T2>>; template <class T1, class T2> struct tuple_element<1, pair<T1, T2>>;[…]
Modify 20.4.4 [pair.astuple] as indicated:
tuple_element<0, pair<T1, T2>>::type
-1- Value: The type T1.tuple_element<1, pair<T1, T2>>::type
-2- Value: The type T2.tuple_element<I, pair<T1, T2>>::type-?- Requires: I < 2. The program is ill-formed if I is out of bounds.
-?- Value: The type T1 if I == 0, otherwise the type T2.
Modify 20.7.4 [variant.helper] as indicated:
variant_alternative<I, variant<Types...>>::type-4- Requires: I < sizeof...(Types). The program is ill-formed if I is out of bounds.
-5- Value: The type TI.
Section: 20.12.3.2 [mem.poly.allocator.mem], 20.13.4 [allocator.adaptor.members] Status: New Submitter: Casey Carter Opened: 2017-06-13 Last modified: 2017-06-15
Priority: Not Prioritized
View other active issues in [mem.poly.allocator.mem].
View all other issues in [mem.poly.allocator.mem].
View all issues with New status.
Discussion:
scoped_allocator_adaptor ([allocator.adaptor.syn]) and polymorphic_allocator ([mem.poly.allocator.class]) have identical families of members named construct:
template <class T, class... Args>
void construct(T* p, Args&&... args);
template <class T1, class T2, class... Args1, class... Args2>
void construct(pair<T1,T2>* p, piecewise_construct_t,
tuple<Args1...> x, tuple<Args2...> y);
template <class T1, class T2>
void construct(pair<T1,T2>* p);
template <class T1, class T2, class U, class V>
void construct(pair<T1,T2>* p, U&& x, V&& y);
template <class T1, class T2, class U, class V>
void construct(pair<T1,T2>* p, const pair<U, V>& pr);
template <class T1, class T2, class U, class V>
void construct(pair<T1,T2>* p, pair<U, V>&& pr);
Both allocators perform uses_allocator construction, and therefore need special handling for pair constructions since pair doesn't specialize uses_allocator (tuple gets all of that magic and pair is left out in the cold). Presumably, the intent is that the construct overloads whose first argument is a pointer to pair capture all pair constructions. This is not the case: invoking construct with a pair pointer and a non-constant lvalue pair resolves to the first overload when it is viable: it's a better match than the pair-pointer-and-const-lvalue-pair overload. The first overload notably does not properly perform piecewise uses_allocator construction for pairs as intended.
Proposed resolution:
This wording is relative to N4659.
Modify 20.12.3.2 [mem.poly.allocator.mem] as indicated:
template <class T, class... Args> void construct(T* p, Args&&... args);-5- Requires: Uses-allocator construction of T with allocator resource() (see 20.10.7.2 [allocator.uses.construction]) and constructor arguments std::forward<Args>(args)... is well-formed. [Note: Uses-allocator construction is always well formed for types that do not use allocators. — end note]
-6- Effects: Construct a T object in the storage whose address is represented by p by uses-allocator construction with allocator resource() and constructor arguments std::forward<Args>(args).... -7- Throws: Nothing unless the constructor for T throws. -?- Remarks: This function shall not participate in overload resolution unless T is not a specialization of pair.
Modify 20.13.4 [allocator.adaptor.members] as indicated:
template <class T, class... Args> void construct(T* p, Args&&... args);-9- Effects: […]
-?- Remarks: This function shall not participate in overload resolution unless T is not a specialization of pair.
Section: 30.6.10 [futures.task], 30.6.2 [future.syn], 30.6.10.2 [futures.task.nonmembers] Status: New Submitter: Tim Song Opened: 2017-06-13 Last modified: 2017-06-15
Priority: Not Prioritized
View all other issues in [futures.task].
View all issues with New status.
Discussion:
When LWG 2921 removed allocator support from packaged_task, it forgot to remove the uses_allocator partial specialization.
Proposed resolution:
This wording is relative to N4659.
Modify 30.6.2 [future.syn], header <future> synopsis, and 30.6.10 [futures.task], class template packaged_task synopsis, as indicated:
template <class R, class Alloc> struct uses_allocator<packaged_task<R>, Alloc>;
Modify 30.6.10.2 [futures.task.nonmembers] as indicated:
template <class R, class Alloc> struct uses_allocator<packaged_task<R>, Alloc> : true_type { };
-2- Requires: Alloc shall be an Allocator (20.5.3.5).
Section: 23.2.7 [unord.req] Status: New Submitter: Tim Song Opened: 2017-06-14 Last modified: 2017-06-15
Priority: Not Prioritized
View other active issues in [unord.req].
View all other issues in [unord.req].
View all issues with New status.
Discussion:
As pointed out in this StackOverflow question, unordered_{map,multimap,set,multiset}::merge() may need to rehash to maintain its max_load_factor invariant, which may require allocation, which may throw.
Proposed resolution:
This wording is relative to N4659.
In 23.2.7 [unord.req], edit Table 91 "Unordered associative container requirements" as indicated:
Table 91 — Unordered associative container requirements (in addition to container) Expression Return type Assertion/note
pre-/post-conditionComplexity … a.merge(a2) void Requires: a.get_allocator() == a2.get_allocator().
Attempts to extract each element in a2 and insert it into a using the hash function and key equality predicate of a. In containers with unique keys, if there is an element in a with key equivalent to the key of an element from a2, then that element is not extracted from a2.
Postconditions: Pointers and references to the transferred elements of a2 refer to those same elements but as members of a. Iterators referring to the transferred elements and all iterators referring to a will be invalidated, but iterators to elements remaining in a2 will remain valid.
Throws: Nothing unless the hash function or key equality predicate throws.Average case 𝒪(N), where N is a2.size().
Worst case 𝒪(N*a.size()+N).…
Section: 21.3.5 [basic.string.hash], 21.3.1 [string.syn] Status: New Submitter: Tim Song Opened: 2017-06-14 Last modified: 2017-06-15
Priority: Not Prioritized
View all issues with New status.
Discussion:
In most cases, std::pmr::meow is a drop-in replacement for std::meow. The exception is std::pmr::{,w,u16,u32}string, because unlike their std:: counterparts, they don't come with enabled std::hash specializations.
The P/R below simply adds std::hash specializations for those four typedefs. An alternative approach, for which wording can be produced if desired, is to make the hash specializations for basic_string allocator-agnostic, similar to the partial specialization of hash for vector<bool>.
Proposed resolution:
This wording is relative to N4659.
Edit 21.3.1 [string.syn], header <string> synopsis, as indicated:
namespace pmr {
template <class charT, class traits = char_traits<charT>>
using basic_string = std::basic_string<charT, traits, polymorphic_allocator<charT>>;
using string = basic_string<char>;
using u16string = basic_string<char16_t>;
using u32string = basic_string<char32_t>;
using wstring = basic_string<wchar_t>;
}
// 21.3.5 [basic.string.hash], hash support
template<class T> struct hash;
template<> struct hash<string>;
template<> struct hash<u16string>;
template<> struct hash<u32string>;
template<> struct hash<wstring>;
template<> struct hash<pmr::string>;
template<> struct hash<pmr::u16string>;
template<> struct hash<pmr::u32string>;
template<> struct hash<pmr::wstring>;
namespace pmr {
template <class charT, class traits = char_traits<charT>>
using basic_string = std::basic_string<charT, traits, polymorphic_allocator<charT>>;
using string = basic_string<char>;
using u16string = basic_string<char16_t>;
using u32string = basic_string<char32_t>;
using wstring = basic_string<wchar_t>;
}
Edit 21.3.5 [basic.string.hash] as indicated:
template<> struct hash<string>; template<> struct hash<u16string>; template<> struct hash<u32string>; template<> struct hash<wstring>; template<> struct hash<pmr::string>; template<> struct hash<pmr::u16string>; template<> struct hash<pmr::u32string>; template<> struct hash<pmr::wstring>;-1- If S is one of these string types, SV is the corresponding string view type, and s is an object of type S, then hash<S>()(s) == hash<SV>()(SV(s)).
Section: 20.15.7.6 [meta.trans.other] Status: New Submitter: Tim Song Opened: 2017-06-14 Last modified: 2017-06-15
Priority: Not Prioritized
View all other issues in [meta.trans.other].
View all issues with New status.
Discussion:
aligned_union's description doesn't, but should, require the types provided to be complete object types.
Proposed resolution:
This wording is relative to N4659.
In 20.15.7.6 [meta.trans.other], edit Table 50 "Other transformations" as indicated:
Table 50 — Other transformations Template Comments […] template <size_t Len, class... Types>
struct aligned_union;The member typedef type shall be a POD type suitable for use as uninitialized storage for any object whose type is listed in Types; its size shall be at least Len. The static member alignment_value shall be an integral constant of type size_t whose value is the strictest alignment of all types listed in Types.
Requires: At least one type is provided. Each type in the parameter pack Types shall be a complete object type.[…]
Section: 20.11.2.6 [util.smartptr.shared.atomic] Status: New Submitter: Alisdair Meredith Opened: 2017-06-15 Last modified: 2017-06-15
Priority: Not Prioritized
View all other issues in [util.smartptr.shared.atomic].
View all issues with New status.
Discussion:
[util.smartptr.shared.atomic] p35 states that two shared pointers are equivalent if they store the same pointer value, and share ownership. As empty shared pointers never share ownership, it is not possible to replace an empty shared pointer using the atomic compare_exchange API.
Note that through aliasing, empty shared pointers may still point to different objects, and any resolution must allow for that case too.Proposed resolution:
This wording is relative to N4659.
Edit 20.11.2.6 [util.smartptr.shared.atomic] as indicated:
template<class T> bool atomic_compare_exchange_weak_explicit( shared_ptr<T>* p, shared_ptr<T>* v, shared_ptr<T> w, memory_order success, memory_order failure); template<class T> bool atomic_compare_exchange_strong_explicit( shared_ptr<T>* p, shared_ptr<T>* v, shared_ptr<T> w, memory_order success, memory_order failure);[…]
-35- Remarks: Two shared_ptr objects are equivalent if they store the same pointer value and share ownership, or if they store the same pointer value and both are empty. The weak form may fail spuriously. See 32.6.1.
Section: 20.14.5 [refwrap], 30.4.4.1 [thread.lock.guard], 99 [thread.lock.scoped], 30.4.4.2 [thread.lock.unique], 30.4.4.3 [thread.lock.shared] Status: New Submitter: Mike Spertus Opened: 2017-06-15 Last modified: 2017-06-18
Priority: Not Prioritized
View all other issues in [refwrap].
View all issues with New status.
Discussion:
There are several deduction guides added to the standard library by P0433R2 that have no effect probably because LWG had not considered late changes to core wording that automatically adds a "copy deduction candidate" (13.3.1.8 [over.match.class.deduct]) that renders these explicit guides moot.
Proposed resolution:
This wording is relative to N4659.
Edit 20.14.5 [refwrap], end of class template reference_wrapper synopsis, as indicated:
template<class T> reference_wrapper(reference_wrapper<T>) -> reference_wrapper<T>;
Edit 30.4.4.1 [thread.lock.guard], end of class template lock_guard synopsis, as indicated:
template<class Mutex> lock_guard(lock_guard<Mutex>) -> lock_guard<Mutex>;
Edit 99 [thread.lock.scoped], end of class template scoped_lock synopsis, as indicated:
template<class... MutexTypes> scoped_lock(scoped_lock<MutexTypes...>) -> scoped_lock<MutexTypes...>;
Edit 30.4.4.2 [thread.lock.unique], end of class template unique_lock synopsis, as indicated:
template<class Mutex> unique_lock(unique_lock<Mutex>) -> unique_lock<Mutex>;
Edit 30.4.4.3 [thread.lock.shared], end of class template shared_lock synopsis, as indicated:
template<class Mutex> shared_lock(shared_lock<Mutex>) -> shared_lock<Mutex>;
Section: 23.5.6.1 [unord.set.overview], 23.5.7.1 [unord.multiset.overview] Status: New Submitter: Mike Spertus Opened: 2017-06-16 Last modified: 2017-06-18
Priority: Not Prioritized
View all issues with New status.
Discussion:
Due to an incompletely implemented change in Kona, some of the size_type deduction guides say something like:
A size_type parameter type in an unordered_set deduction guide refers to the size_type member type of the primary unordered_set template
while others say
A size_type parameter type in an unordered_map deduction guide refers to the size_type member type of the type deduced by the deduction guide.
Clearly they should both be the same. My recollection is that the intent of the committee was to change them all to be the latter. Note, however, that this issue may be mooted if the suggestions in the upcoming P0433R3 paper are adopted as a DR in Toronto.
Proposed resolution:
This wording is relative to N4659.
Edit 23.5.6.1 [unord.set.overview] as indicated:
-4- A size_type parameter type in an unordered_set deduction guide refers to the size_type member type of the
primary unordered_set templatetype deduced by the deduction guide.
Edit 23.5.7.1 [unord.multiset.overview] as indicated:
-4- A size_type parameter type in an unordered_multiset deduction guide refers to the size_type member type of the
primary unordered_multiset templatetype deduced by the deduction guide.