[m-dev.] for review: resolve semantic problems with IEEE float

[Mark Anthony BROWN]

> 
> Hi,
> 
> Anyone care to review this one?
> (If I don't get any comments within a few days, then I'll just
> go ahead and commit it.)
> 
> library/float.m:
> 	Document the intended handling of IEEE NaNs and signed zeros.
> 	The idea is that the semantic inconsistencies between IEEE
> 	floating point and Mercury's declarative semantics should be
> 	resolved by saying for any operation where the two specify
> 	inconsistent results, executing that operation should result
> 	in a run-time error (or exception).
> 

That looks fine.

Cheers,
Mark
-- 
Mark Brown  (dougl at cs.mu.oz.au)     )O+     |
MEngSc student,                             |  John Howard is the Anti-Gough
Dept of Computer Science                    |
University of Melbourne                     |