[libcxx-commits] [libcxx] [libcxx] makes `tanh(complex<float>)` work for large values (PR #122194)

[via libcxx-commits]

lntue wrote:

> > > > I think an approach similar to #99677 would be better. That way we avoid a bunch of code and most likely get a better implementation.
> > > 
> > > 
> > > When will #99677 land? I'd be happy to reduce this PR to just changes to the tests, if it does.
> > 
> > 
> > I don't know when it'll land, but I'm also not touching `tanh` in there right now. I'm only talking about the approach (that being just calling the libc versions instead of implementing them on our own), not the patch itself.
> 
> Ah, you're editing `atanh`, which I misread as `tanh`. In that case, yes, I'll try that out (@lntue I think this answers your question about using the C99 complex functions).

So I assume we can do the same for `std::sqrt(std::complex<T>) --> csqrt*()`?

https://github.com/llvm/llvm-project/pull/122194