Paul Eggert wrote: > As you noted, there's no language reason to do it. It would be a change > put in only to pacify Clang, in an area where Clang is buggy.
As noted in my other mail 5 minutes ago, it's not a clang bug. It's documented clang behaviour, that the user could have avoided by reading the clang documentation and by selecting another set of -fsanitize=... options. Bruno
