Names from namespace std::pmr need to be qualified LWG 2818

[jwakely]

[contents] says:

Whenever a name x defined in the standard library is mentioned, the name x is assumed to be fully qualified as ::std::x, unless explicitly described otherwise.

Names in subclause [mem.res] are in the nested namespace std::pmr so names like get_default_resource() need to be qualified, otherwise [contents] means that it refers to ::std::get_default_resource() (which doesn't exist).

Maybe at the start of a subclause like [mem.res] we need to say that for the rest of the subclause names are fully qualified as either ::std::x or std::pmr::x as appropriate.

[time] and [filesystem] have similar issues for names in ::std::chrono and std::filesystem respectively. See also #1793.

[CaseyCarter]

I noticed this - and went to look at chrono and filesystem to see what we do there - when I was thinking about the same problem for std::ranges. Perhaps we need to change the blanket wording to more generically claim that all names are fully qualified without reference to a specific namespace?

[CaseyCarter]

...and at the same time establish an editorial convention for annotating names that are not intended to be fully qualified. They're rare enough that it would be helpful to call attention to them.

[jensmaurer]

@CaseyCarter, the whole point of this "fully qualified" stuff seems to be to disable ADL. So, we should simply say that in [contents] and stop repeating ourselves in sub-namespaces. See #1913 for the editorial meeting discussion.

[jwakely]

I'm assigned to fix [contents] via LWG 2818.

[frederick-vs-ja]

The resolution of LWG2818 seems to leave this issue nothing to do. Perhaps this can be considered fixed by #4695 (9a4fe7d).