Re: Garnet, CMUCL & Debian

Hi,

Sorry for taking so long to reply. I think the one thing that KR lacks
compared to CLOS that nothing else can make up for is multimethods
(multiple dispatch, generic functions, or whatever you want to call
them) and method combination. This is related to the fact that KR
doesn't (and probably won't for most implementations) fit into the
Common Lisp type system. I actually don't even care that much for the
rest of CLOS, but generic functions really are the best thing since
sliced bread. The other thing about CLOS is that you can take
advantage of multimethods and the Meta-Object Protocol and mix in
stuff like object persistence (or even constraints :)) into what you
already have. It looks to me like there would be no way to do this in
KR without either changing the whole object system or doing it on an
object-by-object basis, even though slot constraints would probably be
really helpful for implementing stuff like persistence.

You can see some rather interesting looking screenshots of Cello at
Kenny Tilton's webpage (I think also that he made an early preview
release available but I don't have the link handy).
http://www.tilton-technology.com/cellophane.html

On another, non-Garnet bashing note, with any luck I will probably
have Garnet running on top of ECL sometime soon after I am done
flunking my finals (there are some ECL and CLX pixmap issues that need
fixing).

Vladimir