https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120775
--- Comment #40 from Jakub Jelinek <jakub at gcc dot gnu.org> --- Regarding https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120775#c38 that testcase doesn't fail on godbolt now. For expansion statements, there are patches posted and awaiting review: https://gcc.gnu.org/pipermail/gcc-patches/2025-November/700380.html https://gcc.gnu.org/pipermail/gcc-patches/2025-November/700703.html https://gcc.gnu.org/pipermail/gcc-patches/2025-November/700762.html https://gcc.gnu.org/pipermail/gcc-patches/2025-November/701644.html although the 3rd one only implements part of the https://wg21.link/cwg3131 proposed resolution. And for _, https://forge.sourceware.org/marek/gcc/pulls/110 WIP implements hopefully the right handling for [expr.ref] splicing for data members, but not yet for member functions (and likely neither for static data member templates).
