https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119011
--- Comment #3 from Alejandro Colomar <alx at kernel dot org> --- (In reply to Richard Biener from comment #2) > Your point that unsigned long == -1 vs. unsigned long == -1u is not the same > is a very good point to continue diagnosing this. But the bug is actually in ==-1u. So we're diagnosing good code, and not diagnosing bogus code. I one wanted UINT_MAX, one would use UINT_MAX. What we want is to be able to express -1 converted (not casted) to whatever time_t is. unsigned long == -1 does The Right Thing(tm). The -1 is first sign-extended, and afterwards converted, which is what programmers want. I don't know if anyone would want to keep that diagnostic, but I can conceive it could happen, which is why I initially proposed the separation of levels =1 and =2, so that people could continue with the current diagnostics by using =2, and I could drop that diagnostic using =1.
