#3070: floor(0/0) should not be defined -------------------------------------+------------------------------------- Reporter: carette | Owner: squadette Type: bug | Status: new Priority: lowest | Milestone: 7.12.1 Component: Prelude | Version: 6.10.1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: 9276 | Blocking: Related Tickets: #10754 | Differential Revisions: -------------------------------------+------------------------------------- Comment (by bgamari): While this doesn't really address the crux of the problem, one option here would be to introduce another pair of floating point types which ensure proper treatment of all IEEE 754 constructs at the expense of performance. I really don't like the fact that this would make you choose between correctness and speed (especially since arguably the default should be correctness), but it is (I think) an option on the table. As far as I can tell, this could be strictly a library change (with the types wrapping the unboxed `Float#` and `Double#` types; we would just need to double-check that the constant folding rules in GHC are correct with respect to the special floating-point values). -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/3070#comment:29 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler