LWG3545 std::pointer_traits should be SFINAE-friendly

[jwakely]

Issue: https://cplusplus.github.io/LWG/issue3545

During the LWG prioritization poll it was suggested that this should be seen by LEWG to answer some open questions.

[cor3ntin]

Mailing list discussions failed to gather a crowd. @jwakely kindly offered to write a paper with a more comprehensive fix matching his implementation, along the following lines

if pointer_traits<P>::element_type cannot be deduced, then pointer_traits<P> is an empty class.
So you could use the existence of pointer_traits<P>::element_type
(or pointer_traits<P>::pointer or pointer_traits<P>::difference_type) 
to check whether something is pointer-like

My assumption is that this fix would be applied as a fix to C++20, so it need not be seen in the context of the C++23 schedule, but I am sure implementers are eager for a resolution as this bugs is affecting users.

[billy-baker]

2022-02-22 Library Evolution Telecon

LWG3545: pointer_traits should be SFINAE-friendly

2022-02-22 Library Evolution Telecon Minutes

Chair: Billy Baker

Minute Taker: Ben Craig

POLL: pointer_traits should be SFINAE friendly (LWG3545).

No objections to unanimous consent.

We want pointer_traits to be SFINAE-friendly.