Paul Eggert wrote: > > - Should glibc define SA_RESETHAND as ((int)0x80000000) ? > > Then SA_RESETHAND could not be used in preprocessor directives any more. > > POSIX would allow that, as it doesn't require SA_RESETHAND to be usable > in preprocessor directives. However, too much software uses it that way > anyway (e.g., squid/src/tools.cc has "#if SA_RESETHAND == 0 && > !_SQUID_WINDOWS_"). So I have my doubts whether this change would be > adopted. > > > - Should clang be silent about this case of implicit conversion? > > That would solve the problem, although the people who want lots of > warnings might want one here too. > > > - Should we discourage users from using > > -fsanitize=implicit-integer-sign-change? > > For me that flag tends to cause more problem than it cures. So we could > tell people that Gnulib won't worry about that warning.
Thanks for your evaluation. I think this should best be fixed by clang, and therefore have registered a clang bug: https://bugs.llvm.org/show_bug.cgi?id=46025 Bruno
