RE: [Axiom-developer] Aldor and Lisp

[Page, Bill]

Martin, 

On Friday, October 14, 2005 4:58 AM you wrote:
> 
> I intend to send the following text to the lisp newsgroup
> today afternoon. Comments welcome.
> 
> ...

I think this attempt to interest people who read the lisp
news group is a good idea.

I am skeptical however about the hope to extended the SPAD
compiler to include Aldor functionality. I think that one
should presume that such an approach was considered by the
original Axiom development team and rejected for good
reasons (at least what probably seemed like good reasons
at the time). SPAD was (and still is) no doubt a rather
tangled mass of code, having evolved over a long period
of time and in spite of having been completely re-written
several times. But most programs reach a point where it
is no longer economical to continue to extend and
maintain them.

Since we have both done some programming in SPAD and in
Aldor, I am sure that you will understand what I mean
when I say that I greatly prefer Aldor over SPAD, even
though the fundamental idea of the two languages are
largely the same. Because of obscure error messages and
extraneous compiler output, using SPAD has the feeling
of working with someone's unfinished experimental code.
There seem to be many cases when things I write that I
think should work, don't work even though usually I find
a way around it only to discover that I was not quite
using SPAD in the way it was intended.

Aldor on the other hand has quite clear and informative
compiler output. It's behavior seems more predictable
and as a consequence I find it is easier and faster to
get some algorithm working first in Aldor and then
back-convert it to SPAD for use in Axiom.

In my opinion, our best option is still to find a way
around the red tape that is currently making it impossible
for us to include Aldor as part of Axiom.

That said, I do think that there is a lot of work that
needs to be done on Axiom by people with serious lisp
experience. The Interpreter in particular should be
analyzed carefully and extended where possible to better
support the Aldor extensions, e.g. dependent types
should work in the Interpreter.

Work also needs to continue on the development of the
new user interface (AxiomUI). A large part of this work
will consist in completing the implementation of the
application programmer interface - new module that was
written in lisp by Kai Kaminski as part of the Summer of
Code project.

-----

Anyway, I hope you catch some fish! :)

Regards,
Bill Page.