Look into code coverage more
Stem has quite a bit of tests but I haven't personally run a code coverage tool to figure out what we're missing. Sean had a suggestion on this...
http://nedbatchelder.com/code/coverage/
http://pypi.python.org/pypi/coverage
I've used this on three projects (not counting Stem) and it helps
check that written tests are reaching all the dark crevices of one's
code. If you have questions, I'm happy to help. Once installed, do:
1) coverage run --parallel-mode --branch --omit="test*" ./run_tests.py
-u -i -t RUN_NONE (or some variation)
2) coverage combine
3) coverage html
4) read the nice reports in the htmlcov/ directory