Given the Axiom CMU port that was done, I think we can probably do this in a reasonable amount of time as long as we just patch it enough to get it working. Re-factoring it to make it into clean, sensible ANSI code is probably another story... however, we can do it incrementally once we can run on an ANSI lisp in the first place. As appealing as a from the ground up re-write is, I think a gradual re-factor is more realistic and a lot less up front work, for some significant gains.
I took a stab at SBCL once, and I have a feeling it will be the most challenging of the ANSI lisps (it has a reputation as a stickler). However, since SBCL is also being (gradually) ported to Windows it becomes a more attractive target. A couple of hard things like threads aren't there yet, IIRC, but I doubt Axiom would immediately need those anyway. The CMU work doesn't seem to quite work on current cmucl versions, so there is probably some more cleanup to do even starting from those changes as a foundation, but if we get serious about it I think it is quite doable. Should we draw the line in the sand and focus on that? Cheers, CY __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com _______________________________________________ Gcl-devel mailing list Gcl-devel@gnu.org http://lists.gnu.org/mailman/listinfo/gcl-devel
