jhuber6 wrote:

> > The downside here is that old HIP code gets a bunch of warnings, but 
> > nothing breaks.
> 
> But things do break. Users see the deprecation warning, but assume that it's 
> okay to continue using them because they are only _deprecated_, and it's not 
> an error. At the same time, other users who see the warning might decide to 
> use the Clang macros. Both of these simply translate to integers in the 
> preprocessor, with the same values mapped to different scopes. Sema has no 
> way to distinguish between these two situations.

This might just me being stubborn, but I don't think it's our responsibility to 
work around users ignoring compiler warnings. The deprecation warning could 
just mention like "Don't use this btw its numbers are wrong."

https://github.com/llvm/llvm-project/pull/185408
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to