Re: [Axiom-developer] Re: Fricas

[C Y]

--- Bill Page <address@hidden> wrote:

> Right now we are lucky if we can even find
> 5 out out of the nearly 100 subscribers in the axiom-devel mailing
> list who are willing to make a public statement about any of issues
> that have been considered for vote-taking here recently.

I think people are wary of poking their heads into this kind of
argument.  I know in my case, as I am not nearly as significant a
contributor as people on BOTH sides of this argument, I was reluctant
to make any comments which would further inflame anyone.

> I get the
> feeling either everyone is watching and waiting to see what happens
> next, or else (maybe) very few people really care that much and they
> really aren't watching at all.

I have made some comments in the past.  I guess I can be a little more
definite in what my take on things is:

a)  First, a bias statement - I work on Axiom only for a hobby and do
not depend on it being in a functioning state for any professional or
academic need.  Of necessity, this makes my outlook different than that
of many people here.

b)  I am almost certainly one of the youngest, least experienced
members of the Axiom community - I do not have any formal training in
CAS and only minimal training in programming.  Thus, my opinions skirt
dangerously close to being uninformed, another reason I don't want them
to inflame things.

c)  That said, my interest and hope for Axiom is that it proposes a new
way of designing and developing a CAS - one that is intended to scale
and be comprehensible in the long term.  We discussed that earlier
(7/11/07), and both you and I concluded my goals are "disconnected from
reality," but that does not alter them - they'll stay disconnected
unless someone tries to make them into a reality.  Whether they are
representative of one "side" or the other in this debate I don't really
know - probably I am closer to Tim's goals.

d)  One area where I do have somewhat different opinions is the
changing of the current code base, which except in the strictly
technical sense is almost completely non-literate.  To me it makes
little difference whether we decode and correctness check IBM/NAG's
non-literate code or Waldek's non-literate code - in fact the latter
would probably be easier.  I see little point AT THIS STAGE in keeping
the Axiom codebase slowly changing, because I don't see any special
merit the current codebase has with respect to being mathematically
reliable.  Can it generate a proof trail of a calculation that can be
verified by COQ/ACL2/Metamath/HOL/whatever?  If not, why do we trust
the current code more than Waldek's code?  If we don't, Waldek's code
works better with modern tools and for that reason alone would be
easier to unravel.

e)  That said, there is are arguments for merging in only well
documented, compartmentalized changes.  The most compelling one
socially is that if literate files are not mandatory it will be VERY
tempting for people to ignore the requirement.  FriCAS I see as having
a legitimate purpose - it is trying a different development style than
Axiom.  It may succeed (for some definitions of success) better than
Axiom, and that's great.  But Axiom AS THE PROJECT WAS DEFINED (or at
least how I understood it to be defined) is about more than just a
working CAS.  That's why Steve and I are spending time creating the
tools we want to handle literate files - because those are foundational
tools to the direction Axiom wants to head and we need to resolve them
BEFORE other work begins.  I would much prefer to be working on Units
and Dimensions but to make that possible the foundation must be solid. 
That means the literate tools must be solid, SPAD must be solid, and
the interp environment must be solid.  Solid, by the Axiom project's
definition, means well documented literate pamphlets.  So, that's where
we must begin.  After that will probably come a systematic evaluation
of the pieces of Axiom comprising interp, and hopefully Steve's work on
a new SPAD.   Then we consider the foundations of the algebra, and I
will push at that time for a re-grounding in Category Theory.  Probably
the question of proof systems would need to be addressed by then as
well.  This of course means the final system would be YEARS away from
even basic functionality.  Per my stated bias this is OK with me, but I
understand not with others.

f)  I had hoped FriCAS would be a project for those who need a running
system now, and Axiom could continue with it's "do it Right" approach. 
In the end the two systems might wind up looking quite different, but I
think each would be useful in its own way.  To risk an analogy, it
might be that FriCAS would be the Concise  Oxford Dictionary of CAS -
high quality and widely used - and Axiom would be the full Oxford
Unabridged Dictionary - attempts to be the last word but is not widely
used outside of academic environments.

g) In the end, I think much of this comes down to project goals and how
they are set.  My understanding was that Tim from the outset had very
specific directions and goals he wanted to pursue, and he created the
Axiom project in order to pursue those specific goals.  The central
question to me seems to be whether these goals can be altered by the
Axiom community or they remain a fixed property of the project itself,
with other goals to be pursued in other projects.  That was my
understanding of the reason Tim is asking what he is asking - FriCAS
has DIFFERENT goals than Axiom and he wants to make sure Axiom as a
project is and remains defined by its foundational goals.


Whether the Axiom Project's defining goals are or should be subject to
community alteration seems to me to be the root question here.


Tim, perhaps you could upload a page to the wiki defining in a concise
manner the original foundational project goals, so we have a place to
refer to for this discussion?

Cheers,
CY

P.S.  This is a philosophical discussion that does not and is not
intended to address issues like who commits what and with or without
review.  That would be a separate and far more personal discussion, and
I'm deliberately attempting to avoid it - the more fundamental
questions I think are the critical points to resolve.


       
____________________________________________________________________________________
Take the Internet to Go: Yahoo!Go puts the Internet in your pocket: mail, news, 
photos & more. 
http://mobile.yahoo.com/go?refer=1GNXIC