CWG2565 Invalid types in the parameter-declaration-clause of a requires-expression

[jensmaurer]

These two examples are very similar, but the second one is ill-formed per the wording despite implementations accepting it due to user demand:

  template <typename T>
  concept C = requires (typename T::type x) {
    x + 1;
  };

  static_assert(!C<int>);

vs.

  template <typename T>
  constexpr bool b = requires (T::type x) { x + 1; };

  static_assert(!b<int>);  // de-jure ill-formed, but widely accepted by implementations

In order to make the second case well-formed, the user needs to write a nested requires that checks for the presence of T::type first.

See CWG2565 for details.

[erichkeane]

This was discussed during the Morning AM session in Tokyo, the following poll was taken:

CWG2565 EWG agrees that this is an issue worth considering, but would like to see a paper to propose a change.

SF	F	N	A	SA
1	17	7	1	0

Result: Consensus

Paper author not identified.