[Axiom-developer] Re: current state and CCL

[root]

Bill,

I did manage to build a running CCL many moons ago but never
managed to get it to run axiom. the build steps were never 
clear to me and after 2 months i gave up and switched to gcl.

CCL is still in the queue of things to do and is likely to be
the third lisp port. somebody on this list managed to get 
axiom running on CMUCL, i believe, and i need to integrate
that piece of work. CLISP is likely to be the fourth lisp
porting effort.

axiom used to run on several lisps including vmlisp, maclisp,
franz lisp, symbolics common lisp, golden common lisp, lucid
common lisp, spice lisp (aka cmucl) and akcl (aka gcl) 
and i even had a scheme cover version for a brief
moment in time but that was entirely interpreted. the big issue
is the ability to save an image with compiled code inside. the
other issue is that most of the axiom build-time steps have
system-dependent commands (e.g. GCL's save-system command).
a great deal of effort went into optimizing axiom to run on 
AKCL (aka GCL) and SPICE (aka CMUCL). in particular, CMUCL has
a really sweet disassemble output which really helps to optimize
lisp code and GCL has a profiler. axiom also has a trace profiler
(src/interp/monitor.lisp.pamphlet) which can handle the algebra.

unfortunately the various special purpose branches of this code
never made it off my desk at ibm. they were not part of the NAG
original distribution as they were very "experimental", meaning
they only ran on my local group of machines. some of the code is
still there as you can see if you grep for the "#+" and "#-" sexprs.

we could easily start a CCL arch branch and restart that effort.
(assuming i get my ass in gear and restart arch).

t