Re: [Gcl-devel] gcl/acl2

[Camm Maguire]

Greetings!

C Y <address@hidden> writes:

> --- Camm Maguire <address@hidden> wrote:
> > Greetings!  After a few days of consolidation, it appears as if gcl
> > is now successfully building acl2 on all 11 Debian architectures(!). 
> > The same gcl is likewise successfully building and testing maxima on 
> > all 11.  To this we can probably add sparc solaris, Windows, and the
> > *BSD's with possibly a little work.  
> 
> Sweet!  Just for fun, has anyone tried on Mac OS X?
> 

So far, we have no one volunteering to support a Mac OS X native port,
which would of course require access to such a box.  But, to my
understanding, darwin has something called 'Fink' which is basically a
port of Debian ppc to Mac OS X.  It would appear that this avenue
would be straightforward.

> > I think that this fact now justifiably earns GCL a unique distinction
> > among the free lisp systems -- it is the only, to my knowledge, free
> > lisp vehicle that carries all known open-source lisp applications (I
> > am only aware of these two, alas) to practically all current
> > hardware, and at good performance.  Congratulations to all.
> 
> Well... there is Garnet ;-p (I know - not an app.  Plus it needs
> updating itself before it's likely to run on gcl)  Congrats!  Well
> done.  And when axiom gets released, it will be using GCL as well?
> 

The axiom maintainer tells me this is his intention.  But axiom has
evolved to require certain 'literate' common lisp extensions, to my
understanding, so the release could take some time.  I'm definitely
hoping that axiom will be the third major common lisp app in the
category under consideration.

> Has anyone ever tried running matlisp on gcl?
> 

Have not yet, but I do happen to have fairly extensive experience with
blas.  There has been some discussion of trying to use lisp macros to
automatically convert large floating point loops to blas calls.  This
is on the back burner, though I find it interesting.  Matlisp can
probably work, but it would in fact be easier given GCL's C roots just
to write a wrapper in the C code itself.  

What I don't understand is how foreign libraries are to be supported
in lisp *distributions*.  Does one ship an image with every available
interface compiled in, and thereby requiring the requisite libraries
in order to run?  This seems like a quite odd model, given the large
size of lisp images and the perspective of the system as a compiler.
What if gcc required libblas to run?  Or does one have a large variety
of images, gcl/blas, gcl/blas/lapack, gcl/mpi, gcl/blas/mpi, each ~ 10
Meg?

> > So GCL is shaping up to be a very portable, quick, and *practical*
> > lisp, albeit rather non-ansi compliant at present.  Once the
> > compliance is achieved, we will hopefully be in good shape to propose
> > including gcl as the standard common lisp compiler in the gcc
> > compiler suite.
> 
> That will rock!  Good stuff Camm!
> 

Thanks for your support and encouragement.  

BTW, do I undertand correctly that you wrote the new maxima web page?
Would you have any time and or interest in doing something similar for
us? Not to worry if the answer is negative :-)

BTW, again, I happen to be an alumn of the smu business school.

Take care,

> __________________________________________________
> Do you Yahoo!?
> HotJobs - Search new jobs daily now
> http://hotjobs.yahoo.com/
> 
> 

-- 
Camm Maguire                                            address@hidden
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah