--- Daniel Lakeland <dlakelan@street-artists.org> wrote:
> Out of curiosity, is there work being done on the manuals? I see so
> many inconsistencies (like plot2d vs plot2 and ode vs ode2 and
> similar) that I've been baffled by what exactly is in the current
> version of maxima, and what isn't. Generally I resort to trial and
> error.
At the moment, not a lot. Part of the problem is I'm going through a
major transition period (grad student to job) and it's going to be
months at best before I can look at it again. But there's another
reason for the lack of activity on docs - there are going to be too
many significant changes between now and 6.0 to warrant a lot of effort
on documentation at this stage. It would just have to be refactored
again and again until we reach a stable point. We aren't kidding when
we say these are development releases ;-). They are released in the
hope that they may be useful for some people/tasks, but we don't
pretend they are a polished, finished product by modern standards.
If you want a well documented system you might consider checking out
Axiom, which is under development as well but has its original
documentation which is probably still fairly current. They have some
very interesting ideas about documenting their system, and it is
probable they will always be much better in this regard than Maxima.
Read their list archives for the details - it's a remarkable read.
> I know you guys are working on the great downcasing, and some general
> infrastructure stuff so perhaps there's a lot of changing going on
> under the covers?
Yep, and a lot more to come. Probably not even so much under the
covers - there's a LOT of work to be done at all levels of the system.
A wide variety of mathematical bugs to be fixed, some parts perhaps
needing to be redone, removing custom pre-ANSI coding features no
longer needed, removing conditionals and cruft code, consideration of
fundamental design decisions and whether they need to be revisited,
working with external interfaces, etc, etc, etc.
After all the dust settles on these issues, it makes sense to document
the system. Believe me, we appreciate the need for good documentation,
and it is part of our plan to create it.
> finally, I noticed some rigamarole in the mailing list archive about
> dealing with CVS and the great source downcasing vs. others who are
> busy doing enhancements. Has anyone worked with the "aegis" system
> for source configuration management? I've used it quite a bit, and I
> think it would work very well for a mature source base like maxima
> where incremental changes from various sources for different purposes
> are going on and there is a strong emphasis on unit and regression
> testing and soforth.
Undoubtedly there are several better systems we could use - IIRC Axiom
uses GNU Arch which has a lot of very nice features. The thing is we
are now fairly plugged into the sourceforge system, and despite its
annoying quirks it does have a fair number of benefits. (The biggest
probably being its longevitiy - we have been there for a couple years
now and with any luck sourceforge itself will be around quite a while.)
The technology it uses is no longer cutting edge, but I suspect it
would take a lot for us to move off of it. If we run into a LOT of
issues we might someday do what Axiom seems to be doing, which is using
Arch behind the scenes for a lot of different development directions,
and then syncing them into the savannah cvs tree when they're ready.
But as yet we haven't run into enough trouble to warrant such measures.
Typically if someone is working on a major change he just uses a local
copy and then makes a bunch of diffs for people to test.
CY
_______________________________
Do you Yahoo!?
Declare Yourself - Register online to vote today!
http://vote.yahoo.com