[SciPy-user] [OT] advice about publishing a scientific software
Mon Jul 16 08:20:14 CDT 2007
On Mon, Jul 16, 2007 at 03:13:39PM +0200, massimo sandal wrote:
> I haven't heard nothing but harsh criticism around "eggs",
I have never heard any criticism. I much just be hidding in a hole to
> Most Python software I've seen is
> distributed via the usual setup.py / *nix packages / Windows installers.
> All three are easy and standard.
Well, I would call Unix package easy to develop. And they are
distribution specific. Windows installers... I must admit I don't have a
clue how to develop one of these, and I can't learn, I don't have a
Windows box lying around (and I am not to interested). setup.py... well,
yes, that's pretty much what eggs are, with some sugar over it.
> What are the advantages of eggs?
Do you know apt, or yum ? Well aggs are pretty much that for Python (like
debs or rpm): an egg is versions, the "easy_install" script used to
install eggs knows where to place them automatically, it can put your
scripts in the path, deal with dependencies (eg by downloading eggs
from pypi). Of course eggs are not perfect, they are no where as nice as
debs, but they are easier to create.
> Where can I find info about them? Can a software redistributed with
> eggs play well (i.e. be repackaged) with Linux packaging or Windows
There are projects doing this (http://stdeb.python-hosting.com to makes
debs. I don't use windows, so I never tried this out, but it seems
possible to make a windows installer (using the "bdist_wininst" to the
setup.py script, that also can be used to generate the eggs).
Info can be found at http://peak.telecommunity.com/DevCenter/setuptools .
Try it out, it feels a bit strange in the beginning, as it looks like
you have written no code, so nothing can happen, but its quite handy.
More information about the SciPy-user