testtools-dev team mailing list archive

[Robert Collins <654474@xxxxxxxxxxxxxxxxxx>]

I think its reasonable for unexpected successes to fail the test run
(though a weaker statement is ok for some folk).

Separately I hold the view that that which fails the test run should be
have its details shown to the user and be counted in any aggregates
done.

-- 
An unexpectedSuccess is like a failure not a success
https://bugs.launchpad.net/bugs/654474
You received this bug notification because you are a member of testtools
developers, which is subscribed to testtools.

Status in Bazaar Version Control System: Confirmed
Status in SubUnit: New
Status in testtools: In Progress

Bug description:
When the function passed to expectedFailure did not raise an assertion, bzr used to do:

    self.fail('Unexpected success.  Should have failed: %s' % reason)

With testtools, bazaar now treats that as a success, as jam noted in February:
<https://lists.ubuntu.com/archives/bazaar/2010q1/066898.html>

And in September he pointed out the same thing, but with subunit:
<https://lists.ubuntu.com/archives/bazaar/2010q3/070192.html>