[SciPy-dev] scipy_distutils/*_info.py files

Fernando Pérez fperez at pizero.colorado.edu
Wed Feb 20 00:19:39 EST 2002


Hi Eric,

> Currently, scons is general purpose like make.  It doesn't serve the same
> purpose as distutils -- i.e. building Python extensions as automatically
> as possible. There is a lot of overlap though, and SCons would make a much
> more powerful underpinning for the second generation of distutils.  I'm
> very hopeful this will happen.  For now I think distutils is the best
> option.  It ain't always pretty, but you can usually get the job done.

Well, it seems that the 'software engineering' environment around python is
still very much taking shape. I hope it solidifies soon, so that individual
projects can spend more time solving their specific problems and less on the
base framework.

Scons and Qmtest are both from the original Software Carpentry competition,
which sounded like a great idea. But things seem to have fizzled out a bit
on that front. I really liked the rationale behind that contest: write 4
major applications useful for *any* software project, all using a powerful
but easy language (python) both for their implementation and for end-user
control and extension. That way developers waste much less time and effort
context-switching between shell/make/autoconf/... Let's hope those 4
projects can pick up some momentum...

But obviously you guys have done your homework, and at the moment it seems
that, 'duct-tapy' as the solutions may be, they work. So the priority is now
SciPy itself, I see that. It's refreshing to see though that you
(collectively) remain open to new ideas and alternatives.

I really see a very nice future for the SciPy project, as many things seem
to be falling in the right place.

Cheers,

f.




More information about the SciPy-Dev mailing list