https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121267
--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> --- (In reply to Jan Schultke from comment #3) > I need to ask CWG about this; it could just be that Clang is wrong, and I've > had some false assumptions. Note EDG also accepts it ... So 3 out of 4 accept while clang rejects it.