On Oct 27, 2009, at 10:21 AM, Tarek Ziadé wrote:
Maybe something like a TEST_README documenting the tests that prove
the that
the closed ticket's bugs have a test that proves that the former bad
behaviour was fixed.
Maybe a simpler way would be to use Bitbucket issue tracker features ?
Are you suggesting maybe getting at that through the (undocumented)
API, or something else?
For the last two bugs I've fixed, they have a corresponding test in
the commit.
And if you go to the issue, you have a link to the commit with a diff,
so you can see it.
That's done with the special "fixes #N" bit in the commit messages,
where N is the issue number. Bitbuckets links the tyicket with the
commits when its pushed there. So we could maybe add the link to
the issue in CHANGES ?
I'm not sure what mechanisms we have for this, do you know?
Although some parts are cleary undertested yet, so even if it's
unpleasant to add tests in an undertested code base, I am +1 in
making the tests mandatory when the code is not yet covered for now on
That was what we discussed at the sprint and I'm doing what I feel is
my job as QA PITA (pain in the ass) by insisting that A> we have tests
for anything fixed and especially not covered and B> that we document
the test related to the fix so that whomever reported the bug can
verify that the test does indeed cover the use-case they reported.
What is the next step?
S
_______________________________________________
Distutils-SIG maillist - [email protected]
http://mail.python.org/mailman/listinfo/distutils-sig
