P0177 Cleaning up allocator_traits

[jensmaurer]

P0177R1 Cleaning up allocator_traits (Alisdair Meredith)

https://issues.isocpp.org/show_bug.cgi?id=151

[jensmaurer]

P0177R2 Cleaning up allocator_traits (Alisdair Meredith)

[mclow]

Alisdair says that this was written for against C++14/17, and the underlying text has changed significantly. This paper should be revised, and then the updated paper should be back in LEWG.

[tituswinters]

Updated version in the mailing: http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2016/p0177r2.html

[jensmaurer]

@tituswinters, this "updated version" is from 2016 (also linked earlier in this issue), and substantially predates Marshall's February comment.

[tituswinters]

OI, indeed. (Was confused because the first comment here linked r1 and wg21.link found me an r2).

I'm ... gonna close for now.

[jwakely]

Alisdair says that this was written for against C++14/17, and the underlying text has changed significantly.

The Allocator requirements are no longer presented as tables, so that part of the paper needs restructuring, but the changes should all still be relevant once presented in the modern style.

The paragraph starting "User specializations of allocator_traits are permitted only if ..." should be removed, as we no longer permit any specializations (P2652R2).

The changes to vector<bool> should be removed, as they were done by LWG3778.

The changes to match_results are mostly still relevant, except adding the two allocator-extended constructors which was done by LWG2195.

This paper should be revised, and then the updated paper should be back in LEWG.

Does it need to go back to LEWG if it's just rebasing the wording? The changes seem editorial.