Re: What do we mean by "Identical" ?

Basically, XLispStat was LispStat in XLisp. CommonLispStat was to be
LispStat in CommonLisp. CommonLispStat derived from the same base as
Luke's XLispStat, but targetted Common Lisp implementations (Kyoto,
Allegro, and a third I'm forgetting, via CLtL1 and CLtL2 -- this was
'89/'90, after all!)

Luke stopped with an initial alpha version, for various sane reasons --
my goals right now are:

1. to hone my Common Lisp skills to a decent level
2. to finish the conversion (updating to ANSI, targetting CMUCL, SBCL,
and CLISP initially)
3. to add a graphics subsystem to replace the XLisp crossplatform (i.e.
nearly identical on Windows, Macs, Unices -- who would've though about
that in '89?) system -- this is what I think will be the hardest.
Luke has notes about that, with respect to using Garnet. I'm starting
to look at that, McCLIM, and Cello as possible targets.
4. rationalize the FFI with either UFFI or CFFI. Verrazano + CFFI
looks good for what I'd like (I had a running ultra-low cost scalable
virtual reality statistical data analysis environment based on
VRJuggler and R (laptop to CAVE/C-6 configuration) that I'd like to
eventually reimplement. Current FFI is adhoc hacks into the three CL's
that it was being targetted for. Not terribly portable.
5. get it back into the game (currently taken by R, but it's always
good to have living competition).

More aggressive on the list would be to replace the Object system
(prototype based) with CLOS or reimplement prototyping on MOP. This
would make it an interactive statistical system on Lisp, rather than a
"LispStat" (as defined by Luke's book) since it would have a different
object system.

Note that R's "S4" object system _IS_ generic-function based. So CLOS
feels rather comfortable to me. But rewriting a working system is a
big job.

Also, I'm looking at a CLOS-based statistical environment. But that's
even futher in the future, since it's just a
"kill-time-on-the-tram-commute" project. On the other hand, doing
statistical design and analysis using only generics such as:

specifiy-model
design-study
create-data
analyze-model
display

seems like a nice goal.

best,
-tony

blindglobe@xxxxxxxxx
Muttenz, Switzerland.
"Commit early,commit often, and commit in a repository from which we
can easily roll-back your mistakes" (AJR, 4Jan05).

.