Dear List -

Thanks for the discussion on the issue of the real strength of the Python/matplotlib/numpy/scipy combo.

I use Python for both development and teaching, but my biggest question concerned teaching.

When I teach, I need something easy and powerful, but also something that is easy to install and 'feels' like other Windows software.

Furthermore, I don't want to teach them matlab (too expensive and too restrictive) or any of its clones (cheaper, but still too restrictive).

So I settled on Python with matplotlib and am very happy with it.
In class, we always use it in interactive mode.
I use IDLE because it has a nice Windows feel to it and it comes with an editor, even though I understand that ipython is much more powerful.
It is my experience that one of the big hurdles of using the Python/matplotlib/numpy/scipy combo is installation. Many people are just not comfortable installing a whole bunch of packages to get something to work. In that respect the Enthought edition has been super (my only request to them would be to make a new version available more frequent, but I know they do this all for free so I even feel bad asking).

Regarding the "don't confuse the newbie" comment, I disagree. Many people that come with a small programming background or a matlab background don't get confused with the current documentation. I think it is pretty well done. Maybe we need separate docs for inexperienced and experienced programmers?

The changes I suggested were to get more inexperienced programmers to join the Python/matplotlib/numpy/scipy world.

I was re-reading the documentation on numpy in the bus this morning, in preperation for a workshop I got to give (to mostly matlab guys). And boy, is numpy nice. I have given the workshop several times, but always with Numeric. I got quite some converts to matplotlib (from matlab) just because they like the graphical output much better.