On Wed, Oct 29, 2003 at 12:03:09PM -0800, Bradd W. Szonye wrote:
> > Bradd W. Szonye wrote:
> >> A programmer cannot implement a new collection type without knowing
> >> the details of the dispatch method ....
>
> xxxxxx@freenetproject.org wrote:
> > I agree.  Thats why this is mentioned in the issues.  But thats a
> > 'solve once' problem for the implementer.
>
> It's a "must be solved" problem for everyone who implements the core of
> SRFI-44 *and* for everyone who wants to add a new collection to the
> system.

Yes, but its not that hard.  They can integrate with Tiny-CLOS if the
SRFI-44 implementation they're using isn't using a more native feature
of the Scheme system, or that system otherwise.  Adding the collection
to whatever dispatch mechanism is being used is pretty trivial compared
to actually implementing the collection.

> >> Between R5RS, SRFI-1, SRFI-13, and a little glue code, a programmer
> >> can actually do *better* than he can with SRFI-44.
>
> > Yes, but not for future collections.
>
> SRFI-44 doesn't let him do better for future collections, because it
> doesn't specify how to add new collections to the existing set.

And it can't until a dispatch mechanism is standardized and everywhere,
but its not as bad as you make it sound.

	Scott