Nels Tomlinson wrote:
> Hi,
>
> I think that I would be a bit bothered by a move to a non-libre lisp.
> I'd worry that Franz might at some time have a change of corporate heart
> which would leave us out in the cold, and I'd worry that I wouldn't be
> able to freely redistribute a complete, usable Maxima.
I think that we have to be sure that what we get will work now
and for the indefinite future. That is, as far as anyone can tell,
the code we get free does not expire. (Of course changes in
OS make this harder than we would like, but what can we do there?)
On the other
> hand, I won't be able to do any of the extra work that NOT using Allegro
> might burden the maintainers with, so I really can't fuss about it.
>
> If we build anything significant on the whiz-bang features which Allegro
> has and which gcl and clisp presumably don't, then it might be a very
> painful thing to try to move back if Franz decided that helping out was
> bad for their shareholders. I completely agree with Prof. Fateman that
> we do not want to see Franz come to any harm, and we certainly couldn't
> fuss if they found that they must pull the plug some time down the road.
>
> Here's a counter-suggestion: Would it make sense to invite the Franz
> folks to contribute a few bits and pieces of Allegro, under the GPL, to
> gcl ( or clisp) and thus make it possible to run Maxima nicely on both?
I doubt there is much along these lines that we could use
Some features probably
rely on proprietary features of the Allegro implementation, and so would
not simply be pieces of common lisp to be revealed. In the case that it
is just "more lisp", they already give out pieces of code like their
lisp-language web server, which I think is on sourceforge.
> In return we could promise active cooperation toward getting and
> keeping Maxima running on their lisp and ours.
Probably this makes sense to them, but it is not much to offer. My
guess is that what we have to do is make sure Maxima is ANSI CL, and
probably that would make it run on Allegro. Convergence would mean our
making GCL or whatever, more like ANSI CL.
> This would mean having
> our lisp and theirs converge, in the areas which matter to Maxima; I
> suspect there would have to be a pretty significant subset in common.
> This would probably also entail getting away from non-ANSI, gcl
> dependent code in Maxima.
It is always possible to have conditional code, and may be advantageous
when there is an ANSI CL way of doing things and a super-clever
efficient way in a particular dialect.
>
> My idea is that we would be getting a kick-start toward having a GPL'd
> ANSI CL and some nicely integrated graphics, and they would be getting a
> really nice computer algebra system that they could use as a selling
> point for Allegro. Of course, this depends on our (probably with help
> from Franz) being able to identify a set of things which Franz could
> share without harm to their business, but which would make a real
> difference to us.
Since Franz Inc already rejected the idea of buying Macsyma Inc
outright, it's clear they don't see this as a great selling point in
their current market; if there is a new market for algebra systems in
lisp, then this project, with their cooperation, will make that market
rather unproductive: all the Maxima users will get a free lisp!
As I said before, I don't really understand the financial motivation, so
I am assuming it has to do with some kind of good will towards maxima
users.
After this semester, I will perhaps have some time and will just compile
the maxima source to see what happens. It used to work (older Allegro,
older Maxima). I don't think we have to do much.
RJF