Fork VPython?
kirby...@gmail.com
Arthur is asking for advice regarding forking VPython (vpython.org),
even with the new beta taking a promising direction and so on (e.g.
adding textures, other OpenGL features).
Based on my limited understanding of this picture (described below,
warning: may contain errors of fact as well as misguided speculations)
I'd say "go right ahead" -- but with caveats (while I'll also
describe).
My understanding is that Arthur wants to directly harness the power of
NumPy, its linear algebra in particular, and merge this into Visual's
real time capabilities, making it one engine.
That sounds like a really promising move, speaking with my math teacher
hat on, as VPython is already a prime training ground for learning
about vector arithmetic.
A more capable matrix handling capability, other multi-dimensional
array powers (ala J and APL), and without reinventing the wheel, would
seem just one of the advantages that better integration with NumPy (aka
Numeric) might provide.
My caveat is I think any forking should involve a renaming /
rebranding, such that the new NumPy-aware engine might be clearly
differentiated in its public profile, keeping students aware of the
fork and the resulting tree of options (which I'm not trying to predict
at this point).
It wouldn't be a hostile takeover, but a trip down some alternative
fork on the river that the VPython team just doesn't have time to
explore. Life is short and Lewis & Clark knew they weren't going to
see everything on just that one trip. One purpose of making source
open is precisely this: to help different teams "fan out" and explore
more territory.
These are off the cuff recommendations and non-reflective of any banter
with principals (e.g. Bruce Sherwood and others).
I've been using VPython somewhat extensively in the current Pythonic
Math cycle, and am more persuaded than ever of its power and continuing
relevance.
More on that here:
http://mail.geneseo.edu/pipermail/math-thinking-l/2006-November/001073.html
Kirby
PS: some reading over my shoulder will object that Arthur doesn't
personally have all the C++ savvy one'd need to tackle the sourcecode
tarball with any chance of success, to which I'd reply (a) Arthur is a
smart cookie well on his way to building his kudos in C++ world and (b)
I always assume we're talking about teams of coders backing the
principals, so in saying "why not go for it" I'm really talking to
"Arthur's team" to whatever extent he assembles it.