On Tue, Jul 4, 2023 at 5:00 AM Jonathan Wakely <
cxx@kayari.org> wrote:
Why would C not want to use the fetch_min and fetch_max names?
C doesn't support atomic_fetch_add etc on floating-point types. Is there any suggestion that WG14 might want to change that? Because if they don't support atomic arithmetic on floating-point types, it seems unlikely they'd want to support atomic min/max on floating-point types. In which case, it doesn't matter what semantics C++ gives those names for floating-point types.
Good point. I am forwarding this question (whether the C committee has interest in whether atomic_fetch_min/atomic_fetch_max applies fminimum/fmaximum/fminimum_num/fmaximum_num/etc. semantics) to the C/C++ liaison, WG 14, and C Floating Point study group reflectors.
[Cringing at the cross-list posting, but not sure what else to try.]