This page is a snapshot from the LWG issues list, see the Library Active Issues List for more information and the meaning of New status.
Section: 25.6.4 [istreambuf.iterator] Status: New Submitter: Jonathan Wakely Opened: 2019-02-21 Last modified: 2019-10-30
Priority: 3
View other active issues in [istreambuf.iterator].
View all other issues in [istreambuf.iterator].
View all issues with New status.
Discussion:
The current working paper requires iterator_traits<Iter>::pointer to be void if Iter doesn't define operator-> (25.3.2.3 [iterator.traits]). We recently removed operator-> from istreambuf_iterator via LWG 2790, therefore either its pointer member should be void, or there should be a partial specialization of iterator_traits. Do we want to change unspecified to void in the class synopsis in [istreambuf.iterator]?
[2019-03-03, Daniel provides concrete wording]
[2019-03-05 Priority set to 3 after reflector discussion]
[2019-03-05, Daniel comments]
With the acceptance of P1252R2 the committee decided to deprecate operator-> of move_iterator, interestingly without mentioning what should happen with its current pointer typedef (which is equal to the template parameter Iterator and thus never void). Effectively this is a very similar situation as for the here discussed istreambuf_iterator case and it seems attractive to me to solve both cases similarly.
[2019-10-30, Jonathan comments]
Also, reverse_iterator::operator->() is now constrained and so not always defined, but reverse_iterator::pointer is defined unconditionally.
Proposed resolution:
This wording is relative to N4800.
Change class template istreambuf_iterator synopsis, 25.6.4 [istreambuf.iterator], as indicated:
template<class charT, class traits = char_traits<charT>>
class istreambuf_iterator {
public:
using iterator_category = input_iterator_tag;
using value_type = charT;
using difference_type = typename traits::off_type;
using pointer = voidunspecified;
using reference = charT;
[…]
};