P3796 R1 Coroutine Task Issues

[wg21bot]

P3796R0 Coroutine Task Issues (Dietmar Kühl)

[wg21bot]

P3796R1 Coroutine Task Issues (Dietmar Kühl)

[inbal2l]

2025-08-26 Library Evolution Telecon

P3796R1 Coroutine Task Issues
(Seen with: "P3801R0 Concerns about the design of std::execution::task")

2025-08-26 Library Evolution Telecon Minutes

Champion: Dietmar Kühl
Chair: Inbal Levi
Minute Taker: Andreas Weis

Summary

POLL: We see the issue presented in “3.1 Iterative code can stack overflow” as an issue with the current design of std::execution::task.

SF	F	N	A	SA
4	3	5	5	1

Attendance: 22
Author's Position: SF (P3801), SA (P3796)
Outcome: No Consensus

POLL: We consider the example presented in 3.3 (potentially creating dangling reference) as require solving.

SF	F	N	A	SA
1	2	6	5	4

Attendance: 22
Author's Position: SF (P3801), A (P3796)
Outcome: Consensus against

POLL: We approve the changes proposed in P3796:

3.5.5 unhandled_stopped() Isn’t noexcept
3.5.16 Missing Rvalue Qualification

Attendance: 22
Outcome: No objection to unanimous concent

Next Steps

We discussed the following (with #2405):

• Sections 3.1, 3.3 from P3801 - LEWG did not see them as issues for task for 26.
• Section 3.2 from P3801 - authors of both papers agreed on the resolution, the wording were not available for review.
• Changes proposed in sections 3.5.5, 3.5.16 of P3796 were approved.
• Changes in sections 3.5.3, 3.5.13 were discussed but we didn’t have time to vote, we will continue next week.
  LEWG will continue to review issues in P3796 during the next telecon (and other telecons later, if needed).

[inbal2l]

2025-09-02 Library Evolution Telecon

P3796R1 Coroutine Task Issues
(Seen with: "P3801R0 Concerns about the design of std::execution::task")

2025-09-02 Library Evolution Telecon Minutes

Champion: Dietmar Kühl
Chair: Fabio Fracassi
Minute Taker: Khalil Estell

Summary

POLL: We would like to drop the scheduler argument from “affine_on”, and avoid forwarding the stop token (along the Lines of LWG4331 affine_on shape).

SF	F	N	A	SA
2	6	1	0	0

Attendance: 16
Author's Position: SF
Outcome: Consensus in favor

POLL: We encourage solving the design issues (specifying affine_on in terms of other sender algorithms) presented in:

• LWG4330 affine_on semantics
• LWG4344 affine_on has no spec for default
• LWG4329 affine_on for other senders

In the C++26 timeframe.

SF	F	N	A	SA
4	6	0	0	0

Attendance: 16
Author's Position: WF
Outcome: Consensus in favor

POLL: We would like to use the allocator provided by the receivers env instead of the one from the coroutine frame.

SF	F	N	A	SA
0	0	5	0	0

Attendance: 14
Author's Position: N
Outcome: Strictly neutral (author should explore preferred solution and come back with rational)

Next Steps

LWG issues opened (for future reference):

• LWG4330 affine_on semantics
• LWG4329 affine_on for other senders
• LWG4331 affine_on shape
• LWG4332 affine_on vs. Stop token
• LWG4344 affine_on has no spec for default

Next revision will contain solutions for the issues raised, LEWG will see them (either in this paper, as NB comments or as feedback on LWG issues).

[inbal2l]

Associated NB comments (design):

• https://github.com/cplusplus/nbballot/issues/949
• https://github.com/cplusplus/nbballot/issues/960
• https://github.com/cplusplus/nbballot/issues/961
• https://github.com/cplusplus/nbballot/issues/966
  (+ LWG-targeting comments)