https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106579
Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Ever confirmed|0 |1
Last reconfirmed| |2022-08-11
--- Comment #5 from Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> ---
Previous discussion on the list:
- when I first implemented it in 2013:
https://gcc.gnu.org/legacy-ml/gcc/2013-11/msg00109.html
- more recent: https://www.mail-archive.com/gcc@gcc.gnu.org/msg97340.html
I think Joseph pointed (either in another bugzilla issue, or on the list) to a
first tentative implementation of __builtin_issignaling in the middle-end by
Sandra Loosemore, which had to be reversed because it caused trouble on some
targets. The difficulty that was cited (and why I didn't get to it myself) is
the large number of floating-point formats to be handled.
That said, having __builtin_issignaling() would be great and we could emit a
lot more of the IEEE functions in the front-end, if it were available.
