[PYTHON MATRIX-SIG] Where to go (was Numerical Recipes)

[Konrad Hinsen]

> Will the numerical extension become a huge package, or is it better to
> build some more packages, wich need the numerical extension.

The already existing add-ons (linear algebra, FFT etc.) point in the
second direction, and I can't see any reason not to proceed in that
way.

> Is it always better, to take some routines in languages like Fortran
> or C and build a python-extension, or should the routines build in
> python. I think the last way is not so efficent but more flexibel and

Ideally, one would start designing a Python interface, write as much
as possible in Python, and resort to C when efficiency demands it.  In
practice, I expect the use of existing C/Fortran/whatever code to be
so attractive that people will prefer to design Python wrappers around
them instead of starting from scratch. I just hope they'll do it
right.

> How is Numpy mainly used? Allways as a programing language for large
> scripts, or as tool to apply functions (or methods) to some data
> arrays interactiv and to plot the results. (For me this is one

I have done both and intend to continue. My applications of NumPy range
from throw-away scripts (reading an array and interpolating from it or
things like that) to a large (4000 lines and constantly growing) package
for molecular modelling.

> Is it possible to use OOP generaly? The longer I think about my own

I've found an OO approach to be the best solution to any problem that
exceeds the throw-away script level. And with Python it's so easy and
quick to define classes that there is no excuse for not doing so!

> How should the many evolving packages be documented? Are there any
> plans to build a help system, perhaps something like the one in the
> gist-extension? 

Good point. I try to put doc strings (at least at the module level) into
all my code, in the hope that some day someone will produce a nice
help browser using them. For large packages, some additional documentation
will be necessary. Unfortunately the current proliferation of popular
text formats (HTML, TeX, GNU-info, plain ASCII) creates a lot of chaos...

> But is this "Matlab-like"-approach the right view. I'm not sure.

If it works for you, then it is. There is no point in forcing a
single view on everyone. As long as we are all happy with NumPy,
we can afford to have different views...

Konrad.
-- 
-------------------------------------------------------------------------------
Konrad Hinsen                          | E-Mail: hinsen@ibs.ibs.fr
Laboratoire de Dynamique Moleculaire   | Tel.: +33-76.88.99.28
Institut de Biologie Structurale       | Fax:  +33-76.88.54.94
41, av. des Martyrs                    | Deutsch/Esperanto/English/
38027 Grenoble Cedex 1, France         | Nederlands/Francais
-------------------------------------------------------------------------------

=================
MATRIX-SIG  - SIG on Matrix Math for Python

send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
=================