https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93008
Jakub Jelinek <jakub at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |hubicka at gcc dot gnu.org, | |jakub at gcc dot gnu.org, | |jason at gcc dot gnu.org --- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> --- We do have DECL_DECLARED_INLINE_P flag, but that is set even with just constexpr keyword. For the cases we were discussing on IRC, functions which should be constexpr e.g. in C++20 and not in C++17 and earlier, but which should still have comdat semantics, but not be really suggested as better idea for inlining, I think we want an attribute, but as constexpr is something that is done very often, it might be best to have another generic FUNCTION_DECL bit next to DECL_DECLARED_INLINE_P, which the new attribute (something in between noinline on inline function and always_inline) would clear and we'd clear it on constexpr functions that aren't explicitly inline too?