[libc-commits] [libc] [libc][math] Implement double precision sin correctly rounded to all rounding modes. (PR #95736)

[via libc-commits]

lntue wrote:

> I still have a question. Since the full set of hard-to-round cases for sin is not known, in CORE-MATH we issue an error message when we are not able to guarantee correct rounding. How do you deal with this issue?

@zimmermann6 : right now we will probably rely on fuzz testing for it, comparing against MPFR outputs.

@nickdesaulniers : shall we add accuracy assertion under `#ifndef NDEBUG` guard?

https://github.com/llvm/llvm-project/pull/95736