Hi Joseph, On Mon, Nov 17, 2025 at 10:21:27PM +0000, Joseph Myers wrote: > On Sun, 16 Nov 2025, Alejandro Colomar wrote: > > > Allow it syntactically, but diagnose it with a new -Wmaxof-bool, enabled > > by -Wextra. > > I don't see much reason for this option, and certainly not for including > it in -Wextra.
> Warning for this feature implies someone might write code > that (a) uses this feature (probably in type-generic macros), (b) does so > on bool, and (c) does not in fact expect to get the answers 1 and 0 from > applying it to bool, or is somehow applying it to bool accidentally and > unintentionally in a context that won't actually work for bool. I agree with a, b, and c. From c, I think the second alternative is the one I'm worried about: applying it accidentally and unintentionally in a context that wouldn't work for bool. > Warnings shouldn't be in -Wextra simply because of not seeing a use for a > particular feature or combination of features, there should be some reason > to believe that the use of the feature actually indicates a defect in the > code (whether one that makes it misbehave at runtime, or simply one that > indicates that code not using that feature would in some way be cleaner). I'm okay with removing it from -Wextra. After all, I don't expect this feature to be common enough to matter that much. A dangling warning flag should be okay. Also, macros that do type-generic stuff and use _Maxof() and/or _Minof are likely to also use _Generic() for other parts of the macro. That _Generic() selector will likely reject bool if it doesn't belong there, so maybe the diagnostic is superfluous. I'll remove the diagnostic entirely. Have a lovely night! Alex -- <https://www.alejandro-colomar.es> Use port 80 (that is, <...:80/>).
signature.asc
Description: PGP signature
