On Wed, 2003-02-19 at 09:00, Raymond Toy wrote:
> >>>>> "James" == James Amundson <amundson@fnal.gov> writes:
>
> James> On Tue, 2003-02-18 at 10:01, Raymond Toy wrote:
> >> While trying to compile maxima with ACL 6.2, I ran into a few
> >> problems. The first of which is that SI is already package in ACL.
> >>
> >> To get around this, I changed the name from SI to CL-INFO. I'd like
> >> to make this change permanent. I've already tested this with clisp,
> >> cmucl, and gcl. They all find the info docs just fine.
>
> James> Very good. I'm happier with the name "CL-INFO" than I was with "SI",
> James> anyway.
>
> Yeah, SI was the name of gcl's package that had the info stuff. Can I
> assume gcl has defpackage? I've only been using the CVS versions of
> gcl for testing.
Since GCL 2.5.0 is due out soon I think we can assume that future
versions of Maxima will require GCL 2.5.0. Don't assume that GCL will be
compiled with --ansi, however.
> >> However, one nice thing is that you can take the resulting fasl files
> >> and load them up in the trial acl 6.2 and run maxima. (Trial acl 6.2
> >> can't build it and can't load the dxl file, however.)
>
> James> Ah. Interesting. I once tried to build Maxima with the trial ACL, but I
> James> ran out of heap space. If the trial version is sufficient to run, but
> James> not compile, maxima, that's still worth something.
>
> To make it easier, we can also concatenate all of the fasl files (in
> the right order) into one giant fasl, which will make it easier to
> use.
>
> Other than some simple testing, I do not know how much heap is left
> for the trial ACL version. ACL causes my 384 MB machine to swap like
> crazy for 30 sec or so when compiling a couple of the numerical
> routines. Don't know why; CMUCL doesn't even come close to that.
Hmmm.
> I also don't know if it's legal to distribute the fasl files.
That clearly needs to be determined. I would like to see ACL support in
any case.
--
James Amundson <amundson@fnal.gov>