[libcxx-commits] [libcxx] [libc++][math] Generalize workaround for UCRT to `signbit` (PR #212398)

[Lucas Mellone via libcxx-commits]

================
@@ -65,18 +59,15 @@ struct TestInt {
   template <class T>
   TEST_CONSTEXPR_CXX23 void operator()() {
     test<T>();
-#if TEST_STD_VER >= 23
-    static_assert(test<T>());
-#endif
   }
 };
 
 template <typename T>
 struct ConvertibleTo {
-  operator T() const { return T(); }
+  TEST_CONSTEXPR_CXX23 operator T() const { return T(); }
----------------
lknknm wrote:

If we are touching this test here I'd remove this entirely and just have the `TEST_CONSTEXPR_CXX23 void operator()() const` in the same place as the `test()` following the same "pattern" I've applied in  https://github.com/llvm/llvm-project/pull/210075

Same for `using lim = std::numeric_limits<T>;`.

If you prefer I can open up a follow-up for the conformance of these tests.


https://github.com/llvm/llvm-project/pull/212398