> I don't remember whether ECL is faster than SBCL or vice versa.
> In any event, if I recall correctly, the difference is something like
> 25 % on the Maxima test suite.
> So my guess is that there is not much to gain by trying to get SBCL to work.
>
Thanks for the information. Since I already started looking at SBCL, I'll
try to get it working (with some effort). Who knows, perhaps for our
usage of Maxima it might be different (basically "processing big lists of sets").
> Clisp is, by far, the slowest of the Lisp implementations; it is something
> like 1/3 the speed of GCL on the Maxima test suite.
> That's worth considering, if speed is a consideration.
> But all the others that I've worked with (ECL, GCL, CCL, SBCL)
> are not so different from each other.
>
I just tried to install GCL, but failed. It's rather old, 2.6.7 is from 2005,
and the build failures are reported also elsewhere, with some patches "somewhere"
(if only people would write complete e-mails --- nearly everything written
for example in the context of Linux distributions like Debian is unreadable for
an outsider), so I don't follow this up.
So well, we have invested already quite some time into ECL, so I guess
we have to stick to it.
Thanks
Oliver
P.S. http://www.gnu.org/software/gcl/ is very outdated, speaking only
about 2.6.7, while on their mailing lists they mention 2.7 and other versions.
One could try to fight with their cvs-code, but it's all so old, doesn't
enthuse me much ...