RE: [Axiom-developer] Building on Axiom... funding Axiom...

[Page, Bill]

On Friday, October 14, 2005 12:46 AM Tim Daly wrote:
> 
> Brian Kennedy wrote:
> > 
> > If you could find a commercial system that needed a CAS
> > at its core, AND was willing to fund development of a
> > free CAS rather than developing yet-another-CAS on its
> > own nickel... would the Axiom development community be
> > interested?

My answer would be an emphatic: "Yes"! To me this seems
like a win-win situation for both the developers of the
commercial system and the Axiom community as a whole. Of
course we need to be clear on licensing issues, but I do
not see this as a problem so long as the CAS itself remains
free.

> > 
> > Two big drawbacks to just building on the free Axiom
> > code:
> > 1) I'm sure there will be a huge learning curve in
> >    figuring out how to  best leverage / integrate Axiom
> >    into my system (I don't think there's a nice .NET
> >    library for Axiom);

The interface with Axiom is currently mostly at a low level
and ad hoc. However some work was done recently as part of
the AxiomUI Google-funded Summer of Code project. Kai Kaminski
the developer of AxiomUI, started by writing a small
application programmer's interface in lisp. With sufficient
time and resources this could be developed into a .NET-like
service.

The MathAction webserver interface to Axiom on the other
hand interacts with Axiom at a high level mostly through
Python regular expression parsing, filters and pipes. But
performance and reliably suffer because of the hacks that
work around problems dealing with an interface that was
designed for interaction with a human.

Andrey Grozin has devoted some effort to correcting some
syntactic deficiencies in the Axiom user interface in order
to improve the interaction with TeXmacs.

And there is some interest in MathML.

> > 
> > ... Via MathML cut n' paste, they'll be able to use  any of
> >  the "3M's" to do that.

It is my hope that MathML (and/or OpenMath) will lead to this
sort of flexibility but I fear that the world is still not
so simple and straight-forward.

> 
> There has been some discussion of MathML and Axiom. When
> Axiom was a commercial product the interface used to work.
> The library is still there but has not been re-integrated.

I think restoring the experimental OpenMath functionality
that Axiom had as a commercial product should not be a
difficult problem. But keep in mind that this was only in
the testing stage at that time. After putting it back
together (i.e. writing the lisp bindings for the OpenMath
library), there is likely still some significant work to
do to complete the implementation.

> > 
> > Would there be any interest from the Axiom developers
> > in working with a  commercial software company to develop
> > out some functionality to make Axiom usable as a core to
> > that commercial system (funded, of course)?

Yes, interest!

> > If so, would there be any capacity to do so near-term?
> > If not, when would there be capacity? what timeline
> > should I have in mind?
>

Timeline? Capacity? Near-term? These are all terms that
are largely incompatible with open source development.
On the other hand, I am confident that an open but funded
project would generate quite a lot of interest if our
recent positive experience with the Google Summer of Code
project is any guide.
 
> ...
> If you wanted to sponsor someone to work on MathML or any
> other well-defined task that could probably be arranged.
> We also have a fund for specific tasks managed by a
> committee (of which I'm not a member).

Yes, I think that is a good approach.

> 
> > 
> > More specifically, is there interest in adding the 
> following to Axiom:
> > 1) Support for units (somewhat like Maple has)?

Yes.

> > 2) Support for content MathML, both in and out?

Yes.

> > 3) .NET interfaces to allow it to serve as the core
> > of a .NET app?

Could you tell us more about how you envisage this
working? Can you give some examples?

Regards,
Bill Page.