P2174 Compound Literals

[wg21bot]

P2174R0 Compound Literals (Zhihao Yuan)

[jfbastien]

Discussed in today's EWG telecon. One notable action item is to sync with the C committee people on this.

Polls:

Add compound literal to C++ (assuming my favored solution to the lifetime issues are addressed)

SF	F	N	A	SA
3	2	12	3	1

C++ compound literals should only work for trivial types (so that destructors aren't an issue)

SF	F	N	S	SA
1	8	7	0	4

C++ compound literals should match C++'s temporary object lifetime semantics

SF	F	N	S	SA
2	6	4	2	1

C++ compound literals should match C's block lifetime semantics

SF	F	N	S	SA
2	3	6	3	1

[AaronBallman]

This paper was discussed at the Apr 1, 2022 SG22 teleconference and we took one poll:

Does SG22 prefer C semantics (lifetime, storage duration, value category) for C types (std::is_trivially_destructible) for compound literals in P2174R0?

Committee	For	Against	Abstain	Notes
WG14	4	0	0	consensus
WG21	5	1	0	consensus

Overall: consensus, but low attendance

[wg21bot]

P2174R1 Compound Literals (Zhihao Yuan)

[erichkeane]

EWG would be interested in seeing compound literals if they ONLY included types without destructors.

SF	F	N	A	SA
1	2	8	1	1

Result: Not Consensus

EWG would be interested in seeing compound literals with the subset of storage class specifiers in the new C syntax for N3038.

SF	F	N	A	SA
0	3	6	1	0

Result: Not Consensus

[erichkeane]

Closing, there was not consensus for further work here, will re-open if a followup paper comes with additional motivation.