Re: [Axiom-developer] Axiom, FriCAS, forks and teeth

[C Y]

Bill Page wrote:
> On 7/9/07, C Y wrote:
>> ... Literate programming is not a mainstream methodology (in my opinion)
>> because few developers are willing to accept the long lead times and hard
>> work of researching the necessary background to make a good literate
>> document. ...
> 
> I think you are absolutely right about that. The trouble is: What are
> we going to do about it?

Whatever we can.  I will make the best attempt I can to do what seems to
be the Right Thing, and hope to either a) prod someone with better
programming skills into doing it right and then document it (cl-web) or
b) eventually through stumbling around learning reach a point where the
tool can be built on.

In some ways, developers can "get away" without doing their homework
because they do not have to compete with programs that HAVE been written
with that attention to detail and design.  Someone has to start doing it
right, or no one will.  For decades, the focus has been on "getting
something working."  This is understandable - the market pays for it and
it is very useful practically.  But now I think we have enough of these
tools that we can take a deep breath, step back, and examine how to put
what has been learned into practice.

Axiom seems the perfect opportunity to me to start "doing it right."
Let's try to be trend setters.  We have an opportunity here to do
pioneering work - let's blaze some trails!

> Bug fixes and new programming can only be
> done by developers and the Axiom project still has far too few
> developers to beat the odds. And to make matters worse, at the risk of
> making a completely unsubstantiated but I think plausible
> generalization, those developers who have the strongest programming
> skills are least likely to have the skills required to write good
> documentation (and vice versa).

I don't know if that's true or not, but I can see how it might be - once
you DO know how to program, writing documentation feels like writing the
same thing twice in two languages (one being rather imprecise and verbose.)

> I used to think that one possible solution would be for those people
> who do have writing skills to work in a team with an experienced
> develop who has the requisite programming skill. But with the Axiom
> project at it's current size and the reluctance that many people seem
> to have to really collaborate with others, even this seems to be
> impractical.

Collaboration requires agreement on direction.  I don't think it has
been proven impractical yet.  Anyway I'll keep trying.  "Nothing worth
doing is easy."

> So I am inclined to want to proceed with the Axiom project without
> requiring literate programming (a situation no worse then at any other
> time in Axiom's history) and take advantage now of the interest of
> open source Internet community where and when it is possible to find
> it.

Isn't this one of the operating principles behind the FriCAS project?  I
support that goal - if the Axiom experiment ultimately doesn't succeed I
would want FriCAS to survive and preserve the effort already put forth -
it is still very useful.  I personally would stay with the Axiom
project, but I can see how FriCAS would be a better fit for many.

> This way Axiom can progress instead of dying a slow death while we
> all grow older.

FriCAS I think is the perfect place to pursue that (worthy) goal.

> This does not mean abandoning the goal of a well
> documented computer algebra system at the level of the program code.

If we are not to abandon it, work must continue on it.  I see no reason
that the Axiom project can't eventually study and incorporate
innovations made in FriCAS - it would be slower perhaps but FriCAS
addresses the immediate availability issue.

> If at some time in the future when/if the available literate
> programming tools can be improved and/or if it is possible to attract
> developers who have both the required skills or if it is somehow
> possible to form working groups with people who are willing to combine
> their skills, then perhaps the goal of improving Axiom in this way can
> also be pursued.

Which is why I think the current situation makes sense.  Axiom can
continue to drive in the same directions it has already driven, and
FriCAS can pursue more "practically oriented" directions.

> But as you implied, solving the problem of how to do
> literate programming in a manner that will be accepted and used by
> most developers today seems to be a very hard problem that has not
> been solved yet.

Then that becomes the first problem the Axiom project must solve.  There
is no way around the hard work of doing the background research, so the
only possible way over that hurdle is to have people motivated to climb
over it.  Beyond that, the tools should be smooth, helpful, and robust -
but they won't write themselves, so those who want them will have to get
to it ;-).

Cheers,
CY