https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121936

--- Comment #27 from Iain Sandoe <iains at gcc dot gnu.org> ---
(In reply to Jan Hubicka from comment #26)

> So while we can patch visibility to have AVAIL_AVAILABLE_IF_CLONED value
> and disable IPA propagation across those, we are giving up on quite
> considerable amount of current and future optimization oppurtunities.

I see, I suppose the thing that I don't fully understand is how we prevent this
kind of thing happening even for code within GCC (I suppose that we can be
confident that the std:: space will not see this between GCC and clang -
because the libc++ version is behind an inline namespace) - presumably we are
relying on similar conditions being applied to the library build and the
consumer?

Reply via email to