On 10.03.2025 15:06, Martin Storsjö wrote:
On Wed, 5 Mar 2025, Jacek Caban wrote:
I left FE_*_ENV handling unchanged, but we could consider making them
proper pointers to structs filled with the appropriate values instead.
Regarding this, what does MSVC do?
I don't think MSVC has anything similar; this seems to be a mingw-w64
invention. I'd argue that users would be better off just using
_fpreset(), so I wouldn't mind removing them entirely, though that might
be more controversial.
If we do that, that's also a bit of an ABI break, so if we do want to
do it, we should probably do it quite soon afterwards.
Sure.
Additionally, it would probably be good to even more explicitly spell
out in the commit message, that this is an ABI break - and what cases
it affects. E.g. it's probably uncommon to expose fenv_t as such
directly in public DLL interfaces, so breakage on that level is
probably not going to crop up. But e.g. if you have object files (or
static libraries!) that use fenv.h, that have been compiled before,
and those are linked with an updated mingw-w64-crt without rebuilding
those object files or static libraries, then anything regarding fenv.h
will be absolutely broken.
OK, I will be more explicit.
Thanks,
Jacek
_______________________________________________
Mingw-w64-public mailing list
Mingw-w64-public@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
