Scientific Libraries in Python

Horatio Davis horatio at qpsf.edu.au
Wed Nov 14 00:19:06 CET 2001


On Tue, 13 Nov 2001, Chris Barker wrote:

> But Scientific Python already is exactly what you are advocating! It
> isn't the least bit complete, but as far as I can tell, it's goals are
> exactly what you want.

Agreed. Um, this and the next sentence read together confuse me slightly:
are you saying this of SciPy by Enthought, or Scientific Python by Konrad
Hinsen? Not that it matters, the same sentiment can be applied to both
projects.

And probably a few others, which is the problem. Another poster on this
thread has already observed that users would rather be busy solving their
problem than hunting down and installing libraries.

> The only question is why other folks are still doing their own little
> projects, rather than contibuting to SciPy.

This was, in fact, the question I was getting at. Although possibly not
phrased with such certainty that SciPy is the One True Way. (:

> One anser, with Scientific Pyhton at least, is that it waws around
> long before the SciPy project got started.

Also agreed.

A minor aesthetic point: Konrad is sitting dead on top of the most
Pythonic part of the namespace for this domain. When you type

from Scientific import Foo

you know what you're getting and where it's coming from, for much the same
reasons that Numeric is named, well, Numeric instead of numpy. Two of the
minor side benefits from merging Scientific Python into SciPy (note the
order) would be that SciPy could finally use a decent top-level package
name <0.3875 wink> and it could stop smooshing together the two words in
its name. Doesn't Scientific.Compiler sound much better? (:

It's a pity the issue isn't just aesthetics (which I could cheerfully
ignore), but useability. The merge is worth doing because having a single
Scientific Python library will give more and richer functionality, with
less hassle, to scientists and engineers using Python. With the current
state of affairs, I can't really make the case to my colleagues here (or
anywhere) to use Python instead of Matlab or Mathematica.

Such a merge would imply some serious manual labour on someone's part.
While I'm willing, I'm not able for a few months, but I thought I'd get
the idea up in front of the relevant community now.

having-exhausted-his-quota-of-posting-for-2001-ly yours,

AHD






More information about the Python-list mailing list