[PATCH] D69281: [FPEnv] Constrained FCmp intrinsics

[Ulrich Weigand via Phabricator via llvm-commits]

uweigand added a comment.

How can we get to a decision on this?  At some point we need to move ahead one way or the other.

I'd still argue that using separate ISD nodes for signaling vs. quiet compares is preferable, both from a complexity of implementation perspective and also conceptually: I feel that a signaling and a quiet compare implement the same **predicates** (and also return the same values in the absence of traps), it's just that operation is different in which exceptions are signaled ...


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D69281/new/

https://reviews.llvm.org/D69281