[Bug libstdc++/125624] C++23: support of std::float16_t for std::uniform_real_distribution

[gccbugs@andreas-borchert.de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125624

--- Comment #5 from Andreas F. Borchert <gccbugs@andreas-borchert.de> ---
I am not assuming anything here and this is not about whether C++23 requires
these concepts or associated requires-clauses to work or to be well-defined. I
just tested if this could possibly work with libstdc++. I understand that a
conforming implementation is free to do anything where the effect is undefined
as in the case of instantiating and using
std::uniform_real_distribution<std::float16_t>.

Nonetheless, it is my opinion that it would be more friendly (and also
conforming) to use options that follow the principle of least surprise like
static assertions or concepts or requires-clause-based variants.

Please feel free to close this.