[Gcl-devel] Re: [Maxima] maxima, defsystem and CMUCL 2005-01

[Camm Maguire]

Greetings!

James Amundson <address@hidden> writes:

> On Wed, 2005-01-12 at 14:00, Vadim V. Zhytnikov wrote:
> 
> > Unfortunately this is not so simple - there are other places which are
> > not compatible with GCL 2.6.5.
> 
> Crud. Even though I would rather not maintain our own fork of defsystem,
> it looks like we may have to do so for the time being.
> 
> However, there is one last thing to look at. I think Ray just replaced
> defsystem.lisp with the current version from clocc. We had a few local
> gcl modifications to defsystem which have therefore been lost. Would
> restoring those patches solve the remaining problems?
> 
> >   Obviously defsystem follows
> > main GCL development branch. 
> 
> As far as I can tell, defsystem doesn't follow GCL at all. There is a
> real problem if we want to use pre-existing Lisp libraries: There are
> two sets of people: people who write/maintain Lisp libraries and people
> who use GCL. There seems to be *very* little overlap between the two
> groups. Maybe GCL will seem more appealing to hard-core Lisp programmers
> when it gets closer to the ANSI standard.

I agree in general here.  GCL's main emphasis is on supporting large
existing end-user applications, many of them written in the LISP glory
days long gone when much of the heavy lifting in the lisp world was
done.  To my understanding, experienced users from this era don't
really want to futz with new language features, but just want to get a
job done with a tool they know and can rely on.  There are actually
quite a lot of these people, and they universally tend to be too busy
for email list discussions.  I think we've done a decent job of making
this work primarily authored in the past available to modern users on
a wide variety of machines.

This said, of course, the primary goal for 2.7.0 is essentially
complete ansi support.  We've already gone most of the way here.  When
released in its two modes (CLtL1 and ANSI), this should be an ideal
vehicle for bringing the lisp language enthusiasts together with the
experienced veterans.  At least that is my hope.  

I'm also in the middle of some serious type propagation work in the
compiler, which I feel will improve the quality of GCL generated code
significantly.  Realistically, I don't foresee a 2.7 release before
the summer, about a year after our last release.  I think that is a
decent schedule.

In the meantime, this heavy 2.7 work is now only possible because we
have a substantively functioning 2.6.5 beachhead holding up the
applications while we break 2.7 before fixing it.  So please, stay
clear of 2.7 on production work.  It is very much a moving target, but
should be quite nice when done.

Take care,

> 
> In the meantime, we have two possibilities: maintain our own defsystem
> or patch defsystem for GCL and try to get them accepted into the main
> defsystem branch. The latter might be the right thing to do, but I don't
> have the energy to do it myself. A volunteer would be welcome.
> 
> >  It seems that we have two
> > options
> > 
> > 1. keep our private copy of defsystem
> 
> Looks likely.
> 
> > 2. swith to GCL 2.7.X
> > 
> > We should ask Camm about 2. 
> 
> I *really* don't want to switch to GCL 2.7.x.
> 
> --Jim
> 
> 
> 
> 

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