[llvm] APFloat: Add minimumnum and maximumnum (PR #96304)

[Thorsten Schütt via llvm-commits]

================
@@ -1483,6 +1483,19 @@ inline APFloat minimum(const APFloat &A, const APFloat &B) {
   return B < A ? B : A;
 }
 
+/// Implements IEEE 754-2019 minimumNumber semantics. Returns the smaller
+/// of 2 arguments, not propagating NaNs and treating -0 as less than +0.
+LLVM_READONLY
+inline APFloat minimumnum(const APFloat &A, const APFloat &B) {
+  if (A.isNaN())
+    return B.isNaN() ? B.makeQuiet() : B;
----------------
tschuett wrote:

To follow up, is it standard-conform to pick an arbitrary value when both are nan? What happens when GCC picks A?

https://github.com/llvm/llvm-project/pull/96304