[libc-commits] [libc] [libc][math] Add C23 ldexpf128 math function and fix DyadicFloat conversions for subnormal ranges and 80-bit floating points. (PR #81780)

[via libc-commits]

================
@@ -735,8 +735,8 @@ struct FPRepImpl : public FPRepSem<fp_type, RetT> {
   // FIXME: Use an uint32_t for 'biased_exp'.
   LIBC_INLINE static constexpr RetT
   create_value(Sign sign, StorageType biased_exp, StorageType mantissa) {
-    static_assert(fp_type != FPType::X86_Binary80,
-                  "This function is not tested for X86 Extended Precision");
+    // static_assert(fp_type != FPType::X86_Binary80,
----------------
lntue wrote:

I removed this static_assert, and add comments about X86 extended precision for this function.

https://github.com/llvm/llvm-project/pull/81780