Lars Wirzenius writes about measuring unit test coverage with coverage.py.
I've been doing the same thing with trace.py from the Python standard library, and the Zope 3 test runner. The pretty HTML output is generated by z3c.coverage by a nightly buildbot run, and any regressions in test coverage are reported via email ("module foo.bar got 5 new untested lines of code"), also thanks to z3c.coverage.I wonder about the differences between coverage.py and trace.py sometimes, but so far I haven't found the time to investigate.
This is the point where I wish PyBlosxom had a simple-to-install, spam-proof comment plugin. Oh well, there's a mailto: link below each post.