There are some Python packages that I prefer to get from the developer instead of from my Linux distribution (Debian). Usually it is because I intend to upgrade the package before, or more often than, my distribution releases a new version. (A likely scenario if you use Debian - zing!)
So I download the source.tar.gz from the developer. I extract it into ~/src, change to the directory and run "python setup.py build". (Here you might recursively detour to install a dependency.) Then I change to the build/lib directory and in there is all of the modules. In build/scripts will be any included scripts. I copy the contents of build/lib to ~/python/lib and the scripts to ~/python/scripts. (Alternatively I could make symbolic links.) My PYTHONPATH environment variable includes ~/python/lib. (Alternatively you can modify sitecustomize.py to insert that directory into the sys.path but this could have system-wide implications.) You can also add ~/python/scripts to your command path. (I tend to manually type the path whenever I want to run something in there.) Now the package is available for import from any python program that I run. I can reproduce this on a system that I do not have root for but only a user shell account. It is easy to transfer the whole library to another computer just by copying the relevant directories. And if I want to upgrade I can download the new source tarball, build it, and change the symlinks or overwrite the old directory contents. I find this is a reasonably convenient way to have up-to-date packages from the Python world on my stable-but-not-entirely-state-of-the-art Debian system. I'm sure as time goes by I'll refine this system. Or maybe Python Eggs will solve most of the problems... -- http://mail.python.org/mailman/listinfo/python-list
