[clang] [llvm] [Clang] restrict use of attribute names reserved by the C++ standard (PR #106036)
Aaron Ballman via llvm-commits
llvm-commits at lists.llvm.org
Tue Jan 14 06:50:29 PST 2025
AaronBallman wrote:
> > Oh, I see, you're suggesting we remove the `getStdNamespace` check from this PR. Yeah, I think that's reasonable.
>
> Yep, that is my suggestion, sorry I was insufficiently clear.
>
> > But I'd somewhat question whether this PR and warning really has anything to do with the attribute names being "reserved" at that point -- we're not checking whether they're reserved or not, and it really doesn't matter. Warning on a `#define` that clobbers the name of a standard attribute is just generally a good thing to do, regardless of whether you're using the standard library.
>
> I agree with this 100%. The link to the 'reserved by the standard' is I think a good additional justification.
>
> I think the proposal, complaining about these as reserved, is a good idea/good patch. BUT I think getting caught up in the "well, when is it technically NOT UB" is a waste of time, given that the warning is a good idea even without that justification.
I think the warning is justified even without a standard library header being included, but I also wonder if that means putting this under `-Wreserved-identifier` is the wrong home and maybe this is a `-Wattributes` warning group instead. We could reword the diagnostic to something along the lines of "macro name conflicts with the name of a %select{vendor attribute prefix|standard attribute|attribute name}0" and we warn on all three of these cases:
```
#define msvc 12 // conflicts with [[msvc::no_unique_address]]
#define annotate 12 // conflicts with [[clang::annotate]]
#define nodiscard 12 // conflicts with [[nodiscard]]
```
WDYT?
https://github.com/llvm/llvm-project/pull/106036
More information about the llvm-commits
mailing list