4vtomat wrote:

> RISCVUsage says we don't have any experimental intrinsics, but we should 
> probably have a view on whether these are defined for them or not - 
> understanding that experimental intrinsics can and have changed.

Yeah I think it's reasonable to let user know if intrinsic is experimental, 
some ideas come up to my mind:
1. append "experimental" prefix to defined macro, e.g. #define 
__riscv_v_intrinsic_experimental_xxxx
2. emit an error with message if experimental feature is not presented for 
those intrinsics that are needed

I would prefer 2 because users may not know if the intrinsic is ratified so 
they won't be able to know whether to append "experimental" prefix or not. It's 
also much clear to tell user what's exactly happening 

https://github.com/llvm/llvm-project/pull/187197
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to