[libc-commits] [libc] [libc][math] Implement a fast pass for atan2f128 with 1ULP error using DyadicFloat<128>. (PR #133150)

[via libc-commits]

================
@@ -106,7 +110,7 @@ constexpr fputil::DoubleDouble ATAN_I[65] = {
 //        + x_lo * (1 - x_hi^2 + x_hi^4)
 // Since p.lo is ~ x^3/3, the relative error from rounding is bounded by:
 //   |(atan(x) - P(x))/atan(x)| < ulp(x^2) <= 2^(-14-52) = 2^-66.
-DoubleDouble atan_eval(const DoubleDouble &x) {
+[[maybe_unused]] DoubleDouble atan_eval(const DoubleDouble &x) {
----------------
overmighty wrote:

Why add the `[[maybe_unused]]` attribute here by the way? Isn't this always used by the implementaton of `atan` and `atan2` for `double`s? Same question for the `atan_eval` overload for `Float128`. 

https://github.com/llvm/llvm-project/pull/133150