It seems the "official" way to do it would be to prepare the source and then upload it as a patch to sourceforge. At least I would like the people on python-dev to comment (most of them aren't subscribed to distutils-sig).
Good point - that would make sense.
Instead of discussing what the "right" way to do testing really is, why not simply leave this decision to the package author and only ask for an entry point, e.g.
setup(... tests = [ TestSuite('mytests', run=myunittests.run, verbose=1), TestSuite('moretests', run=mymodule._test) ], ...)
The only question I see coming up is whether to run tests automatically or only at request.
Hear, hear. There are too many different testing methodologies to try to pick the "right" one for distutils to support. Within my project alone (SCons) we use two separate testing methodologies, PyUnit for per-module unit tests and a TestCmd.py module for end-to-end command testing. Anything more restrictive than a simple, generic "run the tests" interface would be a problem. --SK