On Python packaging

In the perl world, I:

- use apt to manage what's installed on the system

- use CPAN to manage additional stuff installed per-user (you just specify a standard install prefix for all CPAN modules)

that way I don't mess with the system (any packages wanted by system components see the version they are expecting) but I can have (per-user) additional pkgs + updated pkgs, which is often useful.

This seems to be the best of both worlds, to me.

The thing I've always hated about eggs is that invariably something that uses an egg won't have the permissions to write to the default egg cache. It has gotten to the point now that if I know something is going to install an egg, I just do setup.py build and copy the build folder into site-packages. It's just not worth the annoyance for what is, at least as far as I can tell, no gain.

(Edit: read cache)

Short version: don't reinvent the wheel. And, going back to an analogous issue in the Ruby world mentioned by the author, I hope RubyGems team will finally accept the reality we live in and fix that little "parallel dimension" they created.

Ian Bicking's long reply: http://blog.ianbicking.org/2008/12/14/a-few-corrections-to-o...

Cached version: http://is.gd/bFtu