I’ve done it at http://reinout.vanrees.org/weblog. The key trick is to add a preprocessor step. I’ve got my blog entries in a weblog/yyyy/mm/dd/ folder structure.
A script iterates through that folder structure, creating index.txt files in every directory, listing the sub-items. The normal Sphinx process then renders those index.txt files.
I added a custom Sphinx processor for tags. So “.. tags:: python, buildout” somewhere at the top of my weblog entry generates the tags. And the preprocessor again collects those entries and writes out a weblog/tags/TAGNAME.txt file which Sphinx again renders normally.
The preprocessor also creates the root weblog/index.txt with the latest 10 entries. And an weblog/atom.xml in (hardcoded) the output directory for the rss feed.
So: you need some custom stuff, but it is pretty much plain text, so for me it was a nice exercise. And you get to write some helper scripts to make life easy, for instance one that copies a textfile from somewhere to today’s weblog directory (including creation of missing directories and an “svn add”).