P2997 R1 Removing the common reference requirement from the indirectly invocable concepts

[wg21bot]

P2997R0 Removing the common reference requirement from the indirectly invocable concepts (Barry Revzin, Tim Song)

[dhollman]

I'm planning to schedule this for Kona.

[dhollman]

SG9 Kona 2023

P2997R0: Removing the common reference requirement from the indirectly invocable concepts

2023-11-06 SG9 Kona Minutes

Champion: Barry Revzin

Chair: Daisy Hollman

Polls

We took the following polls:

"SG9 recommends forwarding P2997R0 to LEWG for consideration in C++26."

SF	F	N	A	SA
7	1	0	0	0

"SG9 recommends forwarding P2997R0 to LEWG for consideration as a defect report to C++20. (With the undersatnding that someone who has constrained their code on indirectly_unary_invocable (and friends) and then use invocable_with<iter_common_reference_t<t>> constraint would now have underconstrained code.)"

SF	F	N	A	SA
3	4	1	0	0

Summary

This was very uncontroversial and passed with very little discussion (and no opposition once the room understood the paper). It was noted that the standard essentially admits (in the very next constraint, common_reference_with<...>) that the invocable_with<iter_common_reference_t<T>> constraint is unused (in, e.g., indirectly_unary_invocable), since the return type of the invocation is never constrained to be related in any way to the other invocation constraints:

template<class F, class I>
  concept indirectly_unary_invocable =
    indirectly_readable<I> &&
    copy_constructible<F> &&
    invocable<F&, indirect-value-t<I>> &&
    invocable<F&, iter_reference_t<I>> &&
    invocable<F&, iter_common_reference_t<I>> &&
    common_reference_with<
      invoke_result_t<F&, indirect-value-t<I>>,
      invoke_result_t<F&, iter_reference_t<I>>>;

Outcome

Forwarded to LEWG for C++26