P3404 R0 std::at : Range-checked accesses to arbitrary containers

[wg21bot]

P3404R0 std::at : Range-checked accesses to arbitrary containers (Andre Kostur)

[foonathan]

SG9 did discuss it in Wroclaw and gave feedback to the author. We took the following polls.

We want the function std::at (namespace to be determined) with the semantics proposed in the paper (only forward to .at() or ADL at + C arrays). This function should also work for std::map and related types.

SF	F	N	A	SA
3	3	0	2	0

Attendance: 9

Author: SF

Consensus.

We want the function std::at (namespace to be determined) with the semantics proposed in the paper (only forward to .at() or ADL at + C arrays). This function should not work for std::map and related types.

SF	F	N	A	SA
0	0	1	3	2

Attendance: 9

Author: A

Consensus against.

If a type does not provide .at/ADL at, std::at (namespace to be determined) does not work. We would like this to somehow synthesize the operation using e.g. operator[], so we don't require all container implementations to have a .at/ADL at.

SF	F	N	A	SA
1	4	2	2	0

Attendance: 9

Author: A

Encouragement for the author to explore. Author is free to not do that.

---

Author to come back to SG9 once the paper has wording.

[rogerorr]

P3404R0 was seen in SG23 on Thurs in Wrocław

POLL: SG23 would like to see P3404R0 move forward.

Favor	Neutral	Against
11	5	0

POLL: SG23 would like to see P3404 again, if the design remains unchanged.

Favor	Neutral	Against
0	0	17