Mon Dec 5 17:30:49 EST 2011

Haskell type classes and optimization

I was worrying about optimization in the face of all this type class
abstraction once I'm moving to STU arrays.  But.. All numeric
operations are type classes and they are optimised properly in tests
I've seen onlin.

It seems that once the types are specific enough (which would be the
case for STU arrays), there is no reason why the "type class
arguments" can't be eliminated.