[libcxx-commits] [PATCH] D135781: [libc++] Assume that builtins for math.h functions are available

[Eric Fiselier via Phabricator via libcxx-commits]

EricWF added a comment.

In D135781#3895032 <https://reviews.llvm.org/D135781#3895032>, @philnik wrote:

> In D135781#3895015 <https://reviews.llvm.org/D135781#3895015>, @EricWF wrote:
>
>> ~~~What about `-Xclang -fno-math-builtins`? Clang provides an option to turn the builtins off, and libc++ currently works when it's specified. We should state why we don't care about supporting this option.~~~
>
> AFAICT this is still supported. Clang just doesn't add any attributes to the math functions anymore: https://godbolt.org/z/EfKhx896Y

Great, thanks for correcting my ignorance here.



================
Comment at: libcxx/include/math.h:432
 
-template <class _A1, class _A2>
-_LIBCPP_HIDE_FROM_ABI
-bool
-__libcpp_islessgreater(_A1 __x, _A2 __y) _NOEXCEPT
-{
-    return islessgreater(__x, __y);
+template <class _Tp, std::__enable_if_t<std::is_integral<_Tp>::value, int> = 0>
+_LIBCPP_HIDE_FROM_ABI bool isnan(_Tp) _NOEXCEPT {
----------------
Were'n


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D135781/new/

https://reviews.llvm.org/D135781