Re: [Axiom-developer] Active arch branches listed somewhere?

[C Y]

--- Tim Daly <address@hidden> wrote:

> >Given Axiom's trend for doing things the "right way" it would be
> >asthetically satisfying if that trend could be continued in the
> >development tools, although I grant you that's probably a rather
> >silly impulse ;-).  
> 
> umm, presuming you think my way is the "right way", which not even
> *I* tend to agree with some days :-)

Well, from what I've seen you're doing very impressive work.  Your
documentation system is very much the "right way" - adding the research
and the code together with literate programming is probably something
Knuth would dream of seeing achieve fruition.  I was also thinking of
Axiom itself though - it seems to go to great lengths to do the
mathematics the "right way" rather than just a "get an answer" way. 
This isn't always the "best" way for things like practical
get-the-job-done applied math, but for mathematical research it seems
to make Axiom unique.

> but, yes, i'd like to see some global re-think of the idea of a 
> code repository/configure/make. as my programming world gets ever
> more complex i'm beginning to identify issues that needs clever
> solutions. one issue that is beginning to surface is related to 
> Doyen, which is intended to be a scientific platform built on the
> LiveCD/Quantian idea. It needs to be able to pull/patch multiple
> projects that it does not host. These issues are also there in
> Axiom but were never as clear.

Hmm.  Don't know anything clever there.  Usually people try to work
with the 3rd party programs to incorporate the patches, but I guess for
something like Axiom that may not be possible.

> In fact, if we take the "30 year horizon" view it is clear that we
> need to think about very complex applications that transparently
> include customized sub-projects. 

Sounds like CORBA :-).

> in the current instance we have the notion of an "office suite"
> (openoffice) which really is just a large number of cooperating 
> applications. a project that includes many other projects (as Doyen 
> intends) will need more than a code repository. it will need 
> a "library" system that can pull/patch whole cross-sections of code.

But Doxyen isn't a single integrated environment, correct?  I don't see
how that is possible - Axiom might reach the point where all of its
underlying assumptions are well described at all levels of the system
(the "right way" ;-) but I don't think there is any other system I am
aware of that is even close to that level of sophistication.  If there
is a more specialized program (like Macaulay2, Singular, GAP) trying to
function as a subsystem of a broader program, the broader program might
not have even defined all the parameters the specialized program would
need, except as implicit assumptions.  And then if the subprogram does
return a result, there may be parameters used in the subsystem which
SHOULD impact how subsequent calculations should be done in the calling
program, but the parent program might not even be programmed to handle.
 I can certainly understand the desire to incorporate the excellent
work done in these programs, but I'm not quite sure how you could
ensure that the mathematical "environment" in which the calculation
takes place is sufficiently uniform to be correct.  If somebody knows a
way to do this short of rewriting the programs as parts of Axiom (which
I suppose is not out of the question, except that version of Axiom
would be GPL) please please pretty please tell me I'm wrong and it is
possible.

If you don't integrate systems like this, it seems like normal
Debian/Gentoo/BSD style package management should handle obtaining the
various needed sources and patching them.  I have seen gentoo do this
in the past - in fact I think the Macaulay2 ebuild does just that.

> I have a new open source lab with two dozen machines I'm configuring.

Must... control... geek... envy...

> I'm setting up a web server on one now to try to debug this problem
> (which is clearly just a lack of understanding on my part). Once I
> get it working locally I'll turn my attention back to
> axiom-developer.org and get it working there. Perhaps I can convince 
> a student to undertake the "library" program problem (but I doubt 
> it; 'tisn't sexy, yaknow?)

[snip]

> yes, that's exactly it. i have no plans to change the savannah CVS
> site as it is globally known and easy to find. However I'd like to 
> encourage more participation (having already given out pieces of code
> like hyperdoc to various people) that didn't require me to be the
> bottleneck.

Makes sense :-).

> so a while ago I started unwinding the current code pile I maintain
> locally to split out the various interesting pieces into branches.
> that way people could hack particular branches independently. the
> arch server i set up identified about a dozen separate projects.

That reminds me - what's the status of the Axiom book?  Is it up
somewhere as a PDF?  Is anybody planning to get a new edition published
somewhere when Axiom 1.0 (or whatever the version is) is released?

CY

P.S. - Has anybody read a copy of Michael J. Wester's Computer Algebra
Systems : A Practical Guide?  It seems like it might be a good place to
look for ways to make Axiom (or any free CAS for that matter) better.
http://www.amazon.com/exec/obidos/tg/detail/-/0471983535/qid=1099596462/sr=1-1/ref=sr_1_1/104-8290387-0282321?v=glance&s=books


                
__________________________________ 
Do you Yahoo!? 
Check out the new Yahoo! Front Page. 
www.yahoo.com