P0849 auto(x): DECAY_COPY in the language

[wg21bot]

P0849R0 auto(x): DECAY_COPY in the language (Zhihao Yuan)

[jensmaurer]

Marked as "backlog" on San Diego EWG-I wiki.

[jfbastien]

Discussed in EWG-I in Kona, want to see again with updates.

[wg21bot]

P0849R1 auto(x): decay-copy in the language (Zhihao Yuan)

[jfbastien]

EWGI in Cologne:
Continue spending committee time on this versus other proposals, given that time is limited?
SF F N A SA
2 6 5 0 0
Forward to EWG with the additional discussions requested.
SF F N A SA
1 4 5 0 0

[wg21bot]

P0849R2 auto(x): decay-copy in the language (Zhihao Yuan)

[jfbastien]

EWG Prague Thursday morning:

auto(expr) is a problem worth solving.

SF	F	N	A	SA
4	6	5	1	1

decltype(auto) is a problem worth solving.

SF	F	N	A	SA
0	3	7	6	1

We like the solution for auto(expr) as presented, have LWG review the updated paper to confirm what they would use, then send it back to us.

SF	F	N	A	SA
4	6	1	0	2

[wg21bot]

P0849R3 auto(x): decay-copy in the language (Zhihao Yuan)

[wg21bot]

P0849R4 auto(x): decay-copy in the language (Zhihao Yuan)

[jwakely]

LWG reviewed Part 2 of the wording in our telecon. It was observed that some of the changes have semantic changes for the better (e.g. auto(t.begin()) is a no-op if t.begin() returns a prvalue, where decay-copy(t.begin()) forces temporary materialization and a copy). Some of the changes are not necessarily improvements (for some range adaptors we would need to say that auto(E) is materialized, because it's not implementable otherwise). In other cases it's semantically equivalent but the proposed new wording seems slightly clearer about those semantics.

LWG has no objection to EWG proceeding with this, but it would need another LWG review.

[wg21bot]

P0849R5 auto(x): decay-copy in the language (Zhihao Yuan)

[JeffGarland]

@jfbastien have LWG review the updated paper to confirm what they would use, then send it back to us.

I believe this belongs back on EWG plate (also still target in paper) for formal approval after LWG confirms desire to use. Switching labels.

[jfbastien]

The paper was discussed in an EWG telecon on December 3rd 2020.

Polls:

Instead of spelling it auto, spell it more explicitly for example prvalue_cast.

SF	F	N	A	SA
0	4	2	8	6

We should hold P0849 for a while to consider whether instead of or in addition to P0849, C++ needs the capability to forward prvalue through functions in C++ (for example, as suggested in P0927 or P1221).

SF	F	N	A	SA
1	2	6	10	1

We’d like to see the feedback items brought up today before forwarding P0849

SF	F	N	A	SA
1	2	12	3	1

Forward P0849 to LEWG / LWG for feedback on the library parts. If they have no language feedback, forward the language part to CWG.

SF	F	N	A	SA
4	10	3	0	1

We'll need to do electronic polling of this paper, and hope that the R6 update will come this month.

[wg21bot]

P0849R6 auto(x): decay-copy in the language (Zhihao Yuan)