On Thu, Jul 18, 2024 at 12:33:36PM +0200, Josef Wolf wrote: > On Thu, Jul 18, 2024 at 05:07:52AM -0500, Segher Boessenkool wrote: > > Maybe -Wstringop-overflow should not be default (as it is now)? Or > > maybe just -Wstringop-overflow=2 should not be. There are way too many > > obvious (and confusing!) problems with it. > > What good would be a broken warning if it is not active by default? Heh :-) > Even worse are the non-obvious problems. For an obvious problem, you can see > very quickly that the warning doesn't make sense. For non-obvious cases, you > would need to dive deeply into the problem before you (eventually) find that > the warning does not make any sense. The worst is that the warning message says a very definite "this or that is wrong", while that is obviously not true. Warning messages should never lie. Many (most?) warnings are heuristical (otherwise they can typically be errors, instead: it usually is not "this is wrong", but just "this looks wrong"). But the diagnostic we output should include that fact! It should never says "this *is* wrong" if we do not know for sure. Doing that is a disservice to our users. Segher
