[m-rev.] Fast string library

[Peter Schachte]

> > The big problem with this kind of string representation is that
> > the builtin comparison operation won't work properly, and we
> > don't yet allow user-defined comparison predicates. This means
> > that using `str's as keys in a map won't do what you want.
> > 
> > This module probably isn't really usable until that's fixed.
> 
> You're right, of course.  Typeclasses seem like the right thing
> for the job...  another one for v2.

I don't think type classes are quite right for this, since comparison is
available for all types.  I think we need Mercury to support a "where
comparison is" clause on type definitions.  Or perhaps this idea should be
generalized to the concept of a user-defined type class that has a default
implementation of all methods.  The default definitions can be used to make
every type an honorary member of the type class; an instance declaration
would only be needed where an alternative to the default implementation for
some methods is wanted.

-- 
Peter Schachte <schachte at cs.mu.OZ.AU>  Ultimately, there's no such thing as
http://www.cs.mu.oz.au/~schachte/      "organizational behavior"; it's all
Phone:  +61 3 8344 9166                behavior of the people in the
Fax:    +61 3 9348 1184                organization. -- Stephen Covey, et
--------------------------------------------------------------------------
mercury-reviews mailing list
post:  mercury-reviews at cs.mu.oz.au
administrative address: owner-mercury-reviews at cs.mu.oz.au
unsubscribe: Address: mercury-reviews-request at cs.mu.oz.au Message: unsubscribe
subscribe:   Address: mercury-reviews-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------