[m-dev.] discussion about the implementation of compact type representations

[Michael Day]

Hi Zoltan,

> Such code is significantly harder to get right than code that does not involve
> committed choice. This is one reason why user defined unify and compare preds
> are relatively rare. We have in the past tried out specifying unification and comparison
> predicates for some of the set types in the standard library, but (for reasons that
> I can't remember right now) we never committed them.
> 
> I therefore don't think there is any urgency in trying to optimize the representation
> of such types.

Makes sense!

Michael

-- 
Prince: Print with CSS!
http://www.princexml.com