[Answer]-Django+Selenium tests give different results with and without coverage

I figured it out. I was using this as a reference, and the problem was…

1. When using Selenium find_element_by_blah(), if the element is loaded via Ajax, you need to wait until it’s loaded before looking for it. Hence the wait_for_css() method in the link. (I was indeed using that method).
2. The difference between the manage.py test runner and coverage test running is probably due to time – if I added a time.sleep(2) before checking for the element, the test would pass in both cases.
3. But why doesn’t the wait_for_css method prevent this problem???

4. Because I naively created the NoSuchElementException, like

   class NoSuchElementException(Exception)
       pass
   

   instead of importing it: from selenium.common.exceptions import NoSuchElementException (and thus find_css method from the link was raising a different exception than what Selenium was expecting, and failing and exiting instead of waiting).

Oy.