Re: [Haskell-cafe] instance Enum Double considered not entirely great?

On Wed, Sep 21, 2011 at 12:47 AM, Richard O'Keefe wrote:

In any mathematical partial order, we expect        x `le` x to be a law.  But in IEEE arithmetic, if x is a NaN, x `le` x is false.  I don't see how to reconcile these.

I agree that a standard partial order class would be nice.

logfloat package's PartialOrd solution is to have le :: PartialOrd a => a -> a -> Maybe Bool In IEEE arithmetic, if x is a NaN and y is anything else, x `le` y == y `le` x == Nothing. Cheers, -- Felipe.