cor3ntin wrote:

> As a side note, I hate this. It really shouldn't be that hard to decide 
> whether producing a nan is UB or not. I don't think following GCC is _that_ 
> important in this case. But I'll let someone else decide.

I think we really should go to WG14/WG21/SC22 before even considering anything 
in this space. It has a lot of ramifications:
  - NTTPs and mangling
  - Signaling NaNs
  - Trial evaluation
  - How and when NaNs propagate
  - Impact on standard libraries
  - Whether it make sense to allow in constant expressions something that might 
otherwise be caught by some sanitizer

https://github.com/llvm/llvm-project/pull/167892
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to