https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117249
--- Comment #8 from rguenther at suse dot de <rguenther at suse dot de> --- On Tue, 22 Oct 2024, jakub at gcc dot gnu.org wrote: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117249 > > --- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> --- > Then we could use [[assume (EXPR)]]; if the attribute is supported, that would > never evaluate the side-effects but still optimize the non-side-effect cases > and some easy side-effect cases too if inlining + VRP could prove something. That sounds good I guess.
