[libcxx-commits] [PATCH] D136868: [libc++][math.h] Remove unnecessary uses of __promote

[Louis Dionne via Phabricator via libcxx-commits]

ldionne added a comment.

In D136868#3892502 <https://reviews.llvm.org/D136868#3892502>, @EricWF wrote:

> What happens here with __fp16 or other extended floating point types.
> Users have been known to specialize is_floating_point to make these things work.

Users are not allowed to do that because `__fp16` is not a user-defined type, so they can't specialize a type trait in namespace `std` AFAICT. And if someone did do that, TBH it would be broken in way more ways than just this.

IMO the path forward here for custom floating point types is to do http://wg21.link/p1467 and consider the problem in its entirety.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D136868/new/

https://reviews.llvm.org/D136868