RE: [Axiom-developer] New blood in Axiom (was: Community)

[Bill Page]

On May 20, 2007 1:44 AM Alasdair McAndrew wrote:

> Well, I can tell you why I have hesitated about joining the
> Axiom effort:
>
> 1)  Axiom gives the impression of being more difficult and
> abstruse than other systems;

The words abstruse and recondite often used to refer to the
calculations performed by mathematicians and has a connotation
of being somehow at the same time "deep", "unfathomable" and
maybe profound. So I suppose as a first impression of Axiom
this is not so bad.

> the moment you get started you are inundated with information
> about categories, domains and operations.  This has enormous
> strength, but nonetheless makes Axiom appear bewilderingly
> complicated. 

Axiom was developed as an advanced research project by people
working at was then (circa 1985) the leading edge of the
application of computer systems to mathematics (as opposed to
the numerical applications of computers which until then had
been the main focus). This legacy still shows in spite of the
fact that a somewhat more advanced version of Axiom was briefly
marketed as a commercial product by Numerical Algorithms Group.
The version of Axiom that became available as open source is
more closely related to the original IBM research project.

Maybe these rough edges that have the potential to bewilder new
users of Axiom can be reduced over time if the Axiom project
ever attracts some developers with a strong interest and ability
in user interface design. It would be nice to recover at least
what Axiom once had in it's commercial incarnation.

> 2)  The online documentation is very poor.  HyperDoc (which is
> itself incomplete) under unix, and some confusing system commands
> such as ")d op differentiate", ")wh th int" which produce output
> meaningless to the beginning user.

Perhaps these commands a little less confusing if you spell them
out in full? ;)

  )display operation differentiate

But your point about the readability of the output for new users
is well taken. It obviously assumes some level of understanding
of how the interpreter disambiguates the often highly overloaded
operation names in Axiom. Of course understanding this is a
prerequisite for using Axiom in the first place so I am not so
sure that greatly simplifying the output of these commands would
be such a good thing for the average Axiom user.

Still, you are right. A better way to get started in Axiom is
required.

> There are the books, but they are no substitute for good online
> help. 

I would like to hear more about your ideas for good online help
for Axiom. There is supposed to be a )help command in Axiom but
at the moment it provides nothing very helpful:

(1) -> )help
   If the system command or synonym  exists, help information is not
      available for it. Issue )what commands or )what synonyms to
      determine is  is a valid name.

Perhaps this could give at least some kind of basic introduction
for Axiom console users. If so, what would you suggest?

The last commercial version of Axiom included a very nice tutorial
that could be run from with the techexplorer browser interface.
Several years ago I spent some time to convert this tutorial to a
TeXmacs document. I think that is still around somewhere. If you
are interested I could help you resurrect it.

> 3)  There is not much in the way of a decent user interface.
>  I think TeXmacs and console is about it?  And HyperDoc isn't
> obtainable from TeXmacs.

The Axiom wiki includes some information at

http://wiki.axiom-developer.org/address@hidden
eloper.org

from Andrey Grozin about how to start hyperdoc from TeXmacs.
Of course to do this you need to have the source code for
TeXmacs or more specifically the code for 'tm_axiom.c'. It
isn't necessary to recompile all of TeXmacs from source, but
just this interface program. If you have trouble locating it
please let me know.

> 4)  It doesn't seem to be particularly cross-platform. 

The new 'build-improvements' branch of the Axiom source is
considerably more cross-platform than the older versions.

> 5)  There seems to be some lack of cohesion amongst the Axiom
> developers as to the direction in which Axiom should go.

Basically I think there are just too few developers for any
pressure for consensus to exist. This comment from you as an
"outsider" is certainly a push in the right direction.

> I must admit that my main interest is in mathematics education,
> in which I think a GPL CAS has enormous potential.  But at the
> moment I think Maxima is ahead of Axiom on all the points I
> have mentioned. 

I would probably choose Sage rather than Maxima since it includes
Maxima plus several other open source packages (including Axiom),
but I agree with you completely. Of course the right choice of CAS
for mathematics education depends greatly on what you mean by
"mathematics education". If you are talking about high school and
the first few years of university education then I think this
comment is right on track.

But if you are talking about graduate education and research in
the use of computers for abstract or advanced applied mathematics
then I think you have come to the right place. Axiom was really
designed for this purpose. I would say that this also applies
specifically to research into new computer languages specifically
for use in mathematics and it might also apply also to the advanced
use of abstract mathematics in physics.

Regards,
Bill Page.




-Alasdair