LWG2095 and LWG3003: allocator-extended constructors for std::promise

[jwakely]

https://cplusplus.github.io/LWG/issue2095
https://cplusplus.github.io/LWG/issue3003

These two issues suggest two contradictory changes to std::promise. LWG would like LEWG to decide whether to add the missing constructors, or remove the existing allocator-extended constructors.

[inbal2l]

The issue was discussed in Library Evolution Telecon on 2023-07-25 Full Minutes.

Polls

No polls were taken

Next Steps

Jonathan will come back with a proposed resolution that aligns with Tim’s suggestion, LEWG will vote on the proposed resolution.

[jwakely]

Please see the updated proposed resolution for LWG3003, which would resolve both 3003 and 2095.

The new approach is:

• Keep support for using an allocator for shared state of a promise.
• Remove uses_allocator specialization for promise.
• Restore support for using an allocator for the shared state of packaged_task.
• Do not restore uses_allocator specialization for packaged_task.
• Update packaged_task::reset() to deal with allocators.

[inbal2l]

2024-11-18 Library Evolution Wroclaw Meeting

LWG2095 & LWG3003

2024-11-18 Library Evolution Wroclaw Meeting Minutes

Champion: Jonathan Wakely
Chair: Inbal Levi
Minute Taker: Robert Leahy

Summary

POLL: Approve the resolution proposed in “LWG3454: pointer_traits::pointer_to should be constexpr”, and send the issue back to LWG for C++26.

SF	F	N	A	SA
11	4	0	0	0

Attendance: 17 (IP) + 8 (R)
Author's Position: -
Outcome: Strong consensus in favor

POLL: Approve the resolution proposed in “LWG2991: variant copy constructor missing noexcept(see below)”, and send the issue back to LWG for C++26.

SF	F	N	A	SA
12	2	0	0	0

Attendance: 17 (IP) + 8 (R)
Author's Position: -
Outcome: Strong consensus in favor.

POLL: Close LWG532 as “Not A Defect” - without prejudice.

Attendance: 17 (IP) + 5(R)
Outcome: No objection to unanimous consent

Issues discussed (from: https://cplusplus.github.io/LWG/lwg-status.html#LEWG):

• LWG3454 (sent back to LWG)
• LWG2991 (sent back to LWG)
• LWG2985 (Require rebasing on P0551, may be obsolete, and if not will be seen again)
• LWG4042 (P3107 Reviewd, issue closed)
• LWG3114 & LWG3445 (Networking TS, sent to SG4)
• LWG2095 & LWG3003 (later one proposes more resent wording, need to be seen with a paper (“P3503R0: Make type-erased allocator use in promise and packaged_task consistent”)
• LWG532 (closed, without prejudice)

Next Steps

The following issues where resolved: LWG3454, LWG2991, LWG4042, LWG3114, LWG3445, LWG532
The following issue needs to be rebased, likely to be closed: LWG2985
The following issues need to get back to LEWG: LWG3003 (LWG2095)

[jwakely]

The polls above seem to be unrelated to LWG2095 and LWG3003, but in any case those issues were resolved by adoption of P3503R3 in Sofia. I think this can be closed.