https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115566
--- Comment #16 from Harald van Dijk <harald at gigawatt dot nl> --- (In reply to Joseph S. Myers from comment #13) > I think it's clear that where the question is the *syntax* of an > initializer, an actual string literal token is required up to C23 You may think it was perfectly clear one way. I thought it was perfectly clear the other way, and already thought so before that was confirmed to be the intended interpretation, so one thing we should definitely agree on at this point is that it wasn't clear enough. When wording isn't clear enough, and there is a later interpretation of that wording, IMO it does a disservice to users to not use that later interpretation consistently.
