AaronBallman wrote:

> Okay, I made sure it says `argument unused during compilation: 
> '/Zc:__STDC__'` and never defines **STDC** in C++ mode, but I have couple of 
> questions:
> 
>     * Where should I put the test to make sure `-fno-ms-define-stdc` is not 
> accepted?

I think that would make sense in `clang/test/Frontend` to ensure you can't do 
`-Xclang -fno-ms-define-stdc`, probably in a new test file. (I don't see an 
existing test that would make sense to augment.)

>     * !IsCXX feels kind of wrong - is this a justified case for creating 
> `isC` in `clang/lib/Driver/Types.cpp`?

I know what you mean about it feeling kind of wrong, but it mirrors how the 
compiler's language options work. Not C++ mode means C mode, so I would 
probably not add an `isC()` interface. However, if either @MaskRay and 
@jansvoboda11 feel differently as code owners of that area, I wouldn't be 
opposed.

>       * now that I think about it, what should the behavior be with 
> Objective-C? I imagine the number of people compiling Objective-C with 
> `-fms-compatibility` is very small

Objective-C is a pure extension over C, so whatever is allowed in C should 
generally be allowed in Objective-C. Similarly for Objective-C++ and C++.

https://github.com/llvm/llvm-project/pull/68690
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to