why python for the academe?

[Siegfried Gonzi]

huck wrote:
> 
> hello all,
> 
> i've been lurking for a while and am amused at how
> many people in the list here use python for scientific
> computing. why not matlab (or octave or scilab or whatever)?

First of all there is a rip-off concerning the European prices. This is
one reason (personally for me here in Austria) not to use
IDL/Matlab/Mathematica. Not only do they protect a sale outside of
America (you won't get a license in America for Europe; you always must
buy here at your local dealer) they even charge more than twice of the
American prices (this cannot be explained by taxes/shipping/handling).

For me this is unacceptable (I would have really no problem when a
Third-World-Country would charge overgenerous high prices; but not
America-Europe). I only know of PV-Wave (which is just similar to IDL;
they bought the IDL code base ten years ago) which has got similar
prices for Europe and America (or Common Lisp from LispWorks; they also
charge  the same prices here in Europe as they do for America).

But the above is maybe just only valid for Europe, there must be more
reasons. 

Python with the right combination (numerical libraries and plotting
libraries) is much more flexible than lets say Matlab or IDL. What you
buy with Matlab is a huge repository of pre-compiled code. Scientist
normally do not get paid for programming (curiosity: but they do
programming  all the time) and a great many are not very well educated
how to glue together libraries. Personally, I switched to Windows (even
at home I just threw away my Macintosh; but this is another story),
because it bothered me to waste my time on Unix with  ./configure files
and make cycles where at the end the compiler reports that the program
does not compile. On Windows I get at least a pre-compiled program. 

And for me there is also the fact, that when you as a scientist write
code in Python it is more likely that a colleague in a developing
country can profit from your code. The Matlab or IDL prices for them are
unreachable.

This is also a reason why I just abandoned Fortran 90 (there are only
some commercially deliverd compilers available; yes, yes F-language,
but...), and in the meantime I think a scientist is better off when he
is using C++; especially if he writes code which others can profit from
(lets say a new radiation transfer code for example).

But if were in a company I would rethink my stance and maybe I would
rely on commercial products. It is a little bit strange to sell results
based on free software. How can I convince my customer that he should
buy my study and not expect to get it for free. Free software is not all
that good as a great many of people want to make believe us.

S. Gonzi