[Distutils] Annoucing distribute project

Phillip J. Eby pje at telecommunity.com
Thu Sep 25 19:20:09 CEST 2008


At 06:46 PM 9/25/2008 +0200, Andreas Jung wrote:
>We can not restart with something completely different or new. Such 
>an approach would be totally unacceptable due to the huge amount of 
>existing modules having switched to setuptools. Setuptools or an 
>improved codefork must be the tools for managing distibutions in Python.

Yes and no.  Creating a new approach will not cause setuptools to 
instantly disappear.  And assuming the new approach supports only a 
subset of what setuptools/distutils does, then it should be 
relatively simple to generate a setuptools-compatible setup.py in 
distributions created by this other tool, thereby allowing the 
existing infrastructure to be used.

On the flip side, there is no reason why the new tool(s) cannot *use* 
setuptools in order to cope with existing packages, as buildout, 
pyinstaller, and virtualenv already have shown.

At that point, it would remain only to declare distutils and 
setuptools deprecated and legacy, and provide a clear migration path 
to the new tool.

The biggest bottlenecks in setuptools development are backward 
compatibility, testing, and the distutils' excessive 
flexibility...  not to mention the nature of the codebase itself.  A 
new development would allow all of those issues to be addressed at 
the same time.



More information about the Distutils-SIG mailing list