Re: [Axiom-mail] Re: Axiom and OpenMath/MathMl

[C Y]

--- root <address@hidden> wrote:
> 
> We're looking into using a Magnus-type of front end which has a
> near-zero learning curve. It offers a completely different way of
> thinking about the system than any other interface I've ever seen.
> The casual user should find it easy to learn. I expect that the
> more sophisticated user will want command-line or TeXmacs.

Yes, we actually also have a similar division of users - there are the
advanced users like Dr. Fateman who prefer the power of the programming
language, and then there are wimps like me who prefer GUIs.  That's why
we've never really considered trying to build an interface that fits
all purposes - it really doesn't work.  Emacs is usually our interface
of choice for more powerful usage - you can do command line or use a
mode for putting maxima sessions into latex documents.  That's how our
Maxima Book is being written. 
 
> > That said, our game plan is to work on fixing the code and
> > documentation until we can produce a release which fixes all known
> 
> .. perhaps we can use the same documenting technology. Certainly
> if the Maxima people document their algorithms sufficiently well
> we could stea...ummm, share them :-)

There are some very powerful algorighms, no question, but you'll
probably faint if you try to read our code.  Apparently when it was
written computers were so limited that commenting the code was
considered a poor use of space, and also encouraged users to change
(break) things.  Obviously things have changed considerably now, and it
is on our long term plan to document the code, but it is likely to be a
work of years and Dr. Fateman has said that for many parts of the code
it might be simpler to just rewrite it than figure out the old stuff. 
So if you can figure out the algorithms, we would be interested :-) 
 
One possibly useful tool I've been looking at is xref, which trys to
generate a map of what calles what within a lisp file.  It generates
alarmingly huge outputs, but I don't have a problem believing the code
is that complex.  I don't know if it would work for pamphlet files, or
even be useful, but in case you want to check it out...

> Since we both use Lisp as a base (indeed, GCL is a shared common lisp
> base) we could, in principle, define the semantics of operations by
> sending lambda expressions as well as syntax embedded in the
> stream. Even in this case it is not clear what Maxima would do with
> Axiom's type information. 

Yes, I'm not sure quite how that would work.  Probably in the long run
it makes more sense for each system to learn from the other and
impliment whatever concepts are relevant in the context of their own
system - the different philosophy of the two programs is just too big a
complication.

> One of the key reasons I've been pushing "command line" as the first
> interface is that it works everywhere. I agree that it is primitive.

Yes, command line needs to come first.
 
> GCL has a tcl connection. We're not exploiting it but I see the code
> there. I'm not in a great hurry to invent something specific. We did
> that once with hypertex (we had hyperlinking before browsers were
> big) and it was great at the time. Now it just looks primitive.

Well, I can't say I'm too worried about looking primitive (within
reason) as long as the system is easy to use.  Garnet allows custom
look and feel designs, so if someone really doesn't like an interface
implimented there they can redefine it. Old interfaces in the
literature still impliment innovative ideas, and I would be glad to use
them if they still worked/were available :-).  To me thats the more
major issue, whether the code can survive.  Lisp code seems to do
extraordinarily well in that department.  (Witness Maxima and Axiom. 
Also I downloaded Xref from the CMU archive, and as far as I know it
hasn't been maintained since the early ninties.  One trivial change and
I was able to run it on Clisp and CMUCL.  GCL has some problems I
haven't figured out yet.  That's a pretty good average.)

> If TeXmacs can grow to support pamphlet and booklets, can be driven
> from Axiom as a "display control", can embed graphics, etc and works
> cross-platform I think that it will be sufficient for our needs.
> Hopefully the TeXmacs crowd will give us a good API from automation
> as well as a good editor. Having done graphics all the way from
> setting registers on hardware display chips to Java 3D I'd much
> rather someone else figure it all out. 

Can't blame you there.  I'm sure TeXmacs will be fine once it runs on
Windows  - it's a truly amazing piece of work.
 
CY

__________________________________________________
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com