[TIP] Test discovery for unittest

Douglas Philips dgou at mac.com
Mon Apr 6 08:07:39 PDT 2009

On or about 2009 Apr 6, at 7:27 AM, Robert Collins indited:
> With the current TextTestRunner yes you do; what I mean to say is that
> TextTestRunner currently does things that are IMO properly the
> responsibility of the Result object.

Yes, I too have found unittest to be overly tangled. Its been a while  
since I've looked closely at that code, since I did our own variation  
on it quite a while back and haven't had to make any changes to it  
recently... Will be soon though, as we're refactoring out knowledge  
about which tests are 'expected' to fail (i.e. hardware has known  
issues) and moving to just reporting the test status. That is  
currently blocked on the replacement downstream piece. Hopefully when  
Vicki reveals the sekrit scrolls this week that'll be a good spring  
board for working out what the "wire" protocol is for decoupling the  
running and the higher level test-results-analysis pieces.


More information about the testing-in-python mailing list