Not enough Python library development [was PEP scepticism]

[Paul Prescod]

Max Møller Rasmussen wrote:
> 
> ....
> 
> I don't have the impression that Python modules are large archives. I would
> assume that most are in th KB range some in the 100KB range and a few in the
> MB range. That should be a nice fit for Zope.

I think that the system will run faster if you let Zope manage the
metadata but not the zipfiles...especially when you get into the
thousands of modules. But it doesn't matter much.

>...
> 
> I imagined that the interdependencies would be set manual when uploading the
> modules. So far that would be the only way to do it anyway.

So then how would the client know, when it was removing a module,
whether it was breaking dependencies?

> Something like this (Very rough)::
> 
>    http://maxm.normik.dk/pypan/

You might want to add authentication. :) 

>...
> 
> Some kind of tree structure naturally.

I'm saying that one of the things you have to work out is *what* tree
structure.

>...
> > So Zope is a good start (just as Distutils is a good start).
> 
> It would be really good for making something work quickly on the server side
> instead of just talking about it endlesly ;-)

Right, that's why I said Zope is a good start. ActiveState is working on
a Zope-based seed project.

> It could simply talk to the client through webdav of xml-rpc.

That is great as a wire protocol, but what actual messages go across the
wire?

Zope is great as an implementation technology, just as Python is. But
that doesn't remove the need for design and actual coding. Saying "just
use Zope" is like saying "just use Python." You still have to write the
thing!
-- 
Take a recipe. Leave a recipe.  
Python Cookbook!  http://www.ActiveState.com/pythoncookbook