P1021R4
Mike Spertus, Symantec
mike_spertus@symantec.com
Timur Doumler
papers@timur.audio
Richard Smith
richardsmith@google.com
20190617
Audience: Core Working Group
Filling holes in Class Template Argument Deduction
This paper proposes filling several gaps in Class Template Argument Deduction.
Document revision history
R0, 20180507: Initial version
R1, 20181007: Refocused paper on filling holes in CTAD
R2, 20181126: Following EWG guidance, removed proposal for CTAD from partial template argument lists
R3, 20190121: Add wording and change target to Core Working Group
R4, 20190617: Updating wording in response to CWG comments
Rationale
As one of us (Timur) has noted when giving public presentations on using
class template argument deduction, there are a significant number of
cases where it cannot be used. This always deflates the positive
feelings from the rest of the talk because it is accurately regarded as
artificially inconsistent. In particular, listeners are invariably
surprised that it does not work with aggregate templates, type aliases,
and inherited constructors.
We will show in this paper
that these limitations can be safely removed. Note that some of these
items were intentionally deferred from
C++17 with the intent of adding them in C++20.
Class Template Argument Deduction for aggregates
We propose that Class Template Argument Deduction works for aggregate initialization as one
shouldn't have to choose between having aggregates and deduction. This is well illustrated
by the following example:
C++17  Proposed 
template < class T>
struct Point { T x; T y; };
Point< double > p{3.0, 4.0};
Point< double > p2{.x = 3.0, .y = 4.0};


template < class T>
struct Point { T x; T y; };
Point p{3.0, 4.0};
Point p2{.x = 3.0, .y = 4.0};


For the code on the right hand side to work in C++17, the user would
have to write an explicit deduction guide. We believe that this is
unnecessary and errorprone boilerplate and that the necessary deduction
guide should be implicitly synthesized by the compiler from the
arguments of a braced or designated initializer during aggregate
initialization.
Algorithm
In the current Working Draft, an aggregate class is defined as a
class with no userdeclared or inherited constructors, no private or
protected nonstatic data members, no virtual functions, and no virtual,
private or protected base classes. While we would like to generate an
aggregate deduction guide for class templates
that comply with these rules, we first need to consider the case
where there is a dependent base
class that may have virtual functions, which would violate the rules
for aggregates. Fortunately,
that case does not cause a problem because any deduction guides that
require one or more arguments
will result in a compiletime error after instantiation, and
nonaggregate classes without
userdeclared or inherited constructors generate a zeroargument
deduction guide anyway.
Based on the above, we can safely generate an aggregate deduction
guide for class templates
that comply with aggregate rules.
When P0960R0
was discussed in Rapperswil, it was voted that in order to allow
aggregate initialization from a parenthesized list of arguments,
aggregate initialization should proceed as if there was a synthesized
constructor. We can use the same approach to also synthesize the
required additional deduction guide during class template argument
deduction as follows:
 Given a primary class template C, determine whether it satisfies all the conditions for an aggregate class ([dcl.init.aggr]/1.1  1.4).
 If yes, let T_1, ...., T_n denote the types of the N elements ([dcl.init.aggr]/2) of the aggregate (which may or may not depend on its template arguments).
 Form a hypothetical constructor C(T_1, ..., T_N).
 For every constructor argument of type T_i, if all types T_i ... T_n are defaultconstructible, add a default argument value zeroinitializing the argument as if by T_i = {}.
 For the set of functions and function templates formed for
[over.match.class.deduct], add an additional function template derived
from this hypothetical constructor as described in
[over.match.class.deduct]/1.
There is a slight complication resulting from subaggregates, and the
fact that nested braces can be omitted when instantiating them:
struct Foo { int x, y; };
struct Bar { Foo f; int z; };
Bar bar{1, 2};

In this case, we have two initializers, but they do not map to the two elements of the aggregate type Bar, instead initializing the subelements of the first subaggregate element of type Foo.
For complicated nested aggregates, there are potentially many
different combinations of valid mappings of initializers to subaggregate
elements. It would be unpractical to create hypothetical constructors
for all of those combinations. Additionally, whether or not an aggregate
type has subaggregate elements may depend on the template arguments:
template < typename T>
struct Bar { T f; int z; };

This information is not available during class template argument deduction, because for this we first need to deduce T.
We therefore propose simply to avoid all of these problems by
prohibiting the omission of nested braces when performing class template
argument deduction.
Class Template Argument Deduction for alias templates
While Class Template Argument Deduction makes type inferencing easier
when constructing classes,
it doesn't work for type aliases, penalizing the use of type aliases and
creating unnecessary inconsistency.We propose allowing Class Template
Argument Deduction for type aliases as in the following example.
vector  pmr::vector (C++17)  pmr::vector (proposed) 
vector v = {1, 2, 3};
vector v2(cont.begin(), cont.end());


pmr::vector< int > v = {1, 2, 3};
pmr::vector<decltype(cont)::value_type> v2(cont.begin(), cont.end());


pmr::vector v = {1, 2, 3};
pmr::vector v2(cont.begin(), cont.end());


pmr::vector also serves to illustrate another interesting case. While one might
be tempted to write
pmr::vector pv({1, 2, 3}, mem_res);

this example is illformed by the normal rules of template argument deduction because
pmr::memory_resource fails to
deduce pmr::polymorphic_allocator<int> in the second argument.
While this is to be expected, one suspects that had class template argument deduction
for alias templates been around when pmr::vector was being designed,
the committee would have considered allowing such an inference as safe and useful in this context.
If that was desired, it could easily have been achieved
by indicating that the pmr::allocator
template parameter should be considered nondeducible:
namespace pmr {
template < typename T>
using vector = std::vector<T, type_identity_t<pmr::polymorphic_allocator<T>>>;
}
pmr::vector pv({1, 2, 3}, mem_res);

Finally, in the spirit of alias templates being simply an alias for the type, we do not propose
allowing the programmer to write explicit deduction guides specifically for an alias
template.
Algorithm
For deriving deduction guides for the alias templates from guides in the class, we use the following approach (for which we are very grateful for the invaluable assistance of Richard Smith):
 Deduce template parameters for the deduction guide by deducing the right hand side of
the deduction guide from the alias template. We do not require that this deduces all the template
parameters as nondeducible contexts may of course occur in general
 Substitute any deductions made back into the deduction guides. Since the previous step may not
have deduced all template parameters of the deduction guide, the new guide may have template
parameters from both the type alias and the original deduction guide.
 Derive the corresponding deduction guide for the alias template by
deducing the alias from the result type. Note that a constraint may be necessary
as whether and how to deduce the alias from the result type may depend on the
actual argument types.
 The guide generated from the copy deduction guide should be given
the precedence associated with copy deduction guides during overload resolution
Let us illustrate this process with an example. Consider the following example:
template < class T> using P = pair< int , T>;

Naively using the deduction guides from pair is not ideal because they
cannot necessarily deduce objects of type P even
from arguments that should obviously work, like P({}, 0). However,
let us apply the above procedure. The relevant deduction guide is
template < class A, class B> pair(A, B) > pair<A, B>

Deducing (A, B)
from (int, T)
yield int
for A
and T
for B
. Now substitute back into the deduction guide to get
a new deduction guide
template < class T> pair( int , T) > pair< int , T>;

Deducing the template arguments for the alias template from this gives us the following deduction guide for the alias template
template < class T> P( int , T) > P<T>;

A repository of additional expository materials and worked out examples used in the refinement of this algorithm
is maintained online.
Deducing from inherited constructors
In C++17, deduction guides (implicit and explicit) are not inherited when constructors are inherited.
According to the C++ Core Guidelines C.52,
you should “use inheriting constructors to import constructors into a
derived class that does not need further explicit initialization”. As
the creator of such a thin wrapper has not asked in any way for the
derived class to behave differently under construction, our experience
is that users are
surprised that construction behavior changes:
template < typename T> struct CallObserver requires Invocable<T> {
CallObserver(T &&) : t(std::forward<T>(t)) {}
virtual void observeCall() { t(); }
T t;
};
template < typename T> struct CallLogger : public CallObserver<T> {
using CallObserver<T>::CallObserver;
virtual void observeCall() override { cout << "calling" ; t(); }
};


C++17  Proposed 
CallObserver observer([]() { });
CallLogger< > logger([]() { });


CallObserver observer([]() { });
CallLogger logger([]() { });


Note that inheriting the constructors of a base class must include
inheriting all the deduction guides, not just the implicit ones. As a
number of standard library
writers use explicit guides to behave “asif” their classes were defined
as in the standard, such internal implementation details
details would become visible if only the internal guides were inherited.
We of course use the same algorithm
for determining deduction guides for the base class template as
described above for alias templates.
Wording
In over.match.class.deduct, append to paragraph 1 as follows:
 […]
 For each deductionguide, […]
 The template parameters, if any, […]
 The return type […]
In addition, if C satisfies the conditions for an aggregate class with the assumption that any dependent base class has no virtual functions and no virtual base classes, let e_{1}, ... e_{n} be the elements of C ([dcl.init.aggr]).
If there is at least one such element, and the initializer is a nonempty bracedinitlist or parenthesized expressionlist, the set contains an additional function template, called the aggregate deduction candidate, defined as follows. Let x_{1}, ..., x_{m}
be the elements of the initializerlist, designatedinitializerlist, or expressionlist. The aggregate deduction candidate is derived as above from a hypothetical constructor C(T_{1}, ..., T_{m}), where T_{i} is the declared type of the element e_{i} of C which would be explicitly initialized by x_{i} as described in [dcl.init.aggr].
In addition, if C inherits constructors (namespace.udecl 9.8)
from a base
class denoted by a simpletemplateid B,
the set contains the functions
and function templates formed from an
alias template (over.match.alias.deduct) whose template parameters are those of C
and whose simpletemplateid is B.
Add a new subsubsection after over.match.class.deduct titled “over.match.alias.deduct”
When resolving a placeholder for a deduced class type (dcl.spec.auto 9.1.7.5) where the
templatename names an alias template A
whose definingtypeid is a simpletemplateid
naming a specialization of a class template C with
template argument list L, a
set of functions and function templates is formed as follows. For each function or function
template f formed for C by the process in
(over.match.class.deduct 11.3.1.8), form a function or function
template f' according to the following procedure:
 Follow the procedure below for deducing the arguments of the result type of f
from C<L> with the exception that deduction does
not fail if not all template arguments are deduced.
 Form f' as follows:
 The parameters and return type are constructed by substituting
into
the parameters (including default arguments) and return type of f
the deductions
of all template parameters that were successfully deduced in the deduction of
the result type of f.
 The template parameters consist of those template parameters of f or A
that appear in the parameters and return type constructed in the previous step.
 The associated constraints (temp.constr.decl 12.4.2) are the conjunction of the associated constraints
of f and a constraint that is
satisfied if and only if the arguments of A
are deducible (see below) from the return type.
 If f is a copy deduction candidate (over.match.class.deduct 11.3.1.8), then f'
is considered to be so as well.

If f was generated from a deductionguide (over.match.class.deduct 11.3.1.8), then f'
is considered to be so as well.
 If f was generated from a constructor or deductionguide
with an explicitspecifier, then
f' is considered to be so as well.
The arguments of a template A are said to be deducible from a type T if, given a class template
template <typename> class AA;
with a single partial specialization whose template parameter list is that of A and whose template argument list is a specialization of A with the template argument list of A (temp.deduct.type 12.9.2.5), AA<T> matches the partial specialization.
Initialization and overload resolution are performed as described in
(dcl.init 9.3) and (over.match.ctor 11.3.1.3),
(over.match.copy 11.3.1.4), or (over.match.list 11.3.1.7) (as
appropriate for the type of initialization performed) for an object of a hypothetical class type, where the
selected functions and function templates are considered to be the constructors of that class type for the
purpose of forming an overload set, and the initializer is provided by the context in which class template
argument deduction was performed. As an exception, the first phase in 11.3.1.7 (considering initializerlist
constructors) is omitted if the initializer list consists of a single expression of type cv U,
where U is a specialization of C or a class derived from a specialization of C.
If the function or
function template was generated from a constructor or deductionguide
that had an explicitspecifier, each such notional constructor is considered to
have that same explicitspecifier. All such notional constructors are considered
to be public members of the hypothetical class type. [Note: The constraint ensures that
a specialization of A can be deduced from the return type. — end note]
[Example:
template <class T, class U> struct C {
C(T, U);
};
template<class T, class U>
C(T, U) > C<T, std::type_identity_t<U>>;
template<class V>
using A = C<V *, V *>;
int i{};
double d{};
A a1(&i, &i); // Deduces A<int>
A a2(i, i); // Error: cannot deduce V * from i
A a3(&i, &d); // Error: cannot deduce alias template from C<int *, double *>
Possible exposition only implementation of the above procedure:
//The following concept ensures a specialization of A is deduced
template <class> class AA;
template <class V> class AA<A<V>> { };
template <class T> concept deduces_A = requires { sizeof(AA<T>); };
// f is formed (over.match.class.deduct 11.3.1.8) from the deductionguide of C
template<class T, class U> auto f(T, U) > C<T, std::type_identity_t<U>>;
// Deducing arguments for C<T, std::type_identity_t<U>> from C<V *, V*> deduces T as V *
// f' is obtained by transforming f as described by the above procedure
template<class V, class U>
auto f_prime(V *, U) > C<V *, std::type_identity_t<U>>
requires deduces_A<C<V *, std::type_identity_t<U>>>;
— end example]
Insert a bullet after over.match.best/1.9 as follows
 F1 is generated from class template argument deduction (over.match.class.deduct 11.3.1.8) for
a class D, F2 is generated from inheriting constructors from
a base class of D, and for all arguments the corresponding
parameters of F1 and F2 have the same type, or if not that,