Ketil Malde
The bombing of NaN *might* be a profound compilation option, but for people who really do numerical work, this is a blessing NOT to have it.
I'll expand a bit of this, after I've checked with Wikipedia. Please correct me (and it) if I'm wrong, but: 1) Intel CPUs generate exceptions, not NaNs (unless a NaN is already involved), so NaNs are introduced by choice in the run-time system. 2) IEE754 supports both 'signaling' and 'quiet' NaNs, so it seems the standard is not blessed in this regard. And, in Haskell, I'd consider using NaNs for missing values slightly abusive of the system, this is just a poor man's way of spelling "Maybe Double". -k -- If I haven't seen further, it is by standing in the footprints of giants