PQM: test failure

[John Arbash Meinel]

Andrew Bennetts wrote:
> John Arbash Meinel wrote:
>> Olexandr Byelchenko wrote:
>>> FAIL: test_known_failure_ok_run (bzrlib.tests.test_selftest.TestRunner)
>> ...
>>> I think it's a very bad style to make test depends on time intervals.
>>>
>>> [µ]
>>>
>>>
>> I don't think it was meant to be dependent on time interval. So we could
>> instead use:
>>         self.assertContainsRe(stream.getvalue(),
>>             '\n'
>>             '-*\n'
>>             'Ran 1 test in 0.0\\d\\ds\n'
>>             '\n'
>>             'OK \\(known_failures=1\\)\n')
>>
>> It works for me, at least.
>> (the \\ are a little ugly, but necessary since we need \n)
> 
> You version is still dependent on the time interval, even though it's more
> permissive than the current version.
> 
> Why not: 'Ran 1 test in .*s\n'?  That way it will match precisely
> what matters, and no more.  You could make it check for "\\d+.\\d{3}", but
> that's not the point of this test.  Time reporting is (or at least should be)
> tested elsewhere.  This test is to verify that a known failure is reported in
> the final output.
> 
> Keep the test simple, and keep it focussed one doing just one thing (and doing
> it well).
> 
> -Andrew.

As I didn't write the original test, I was being conservative. But I
have no problem with:

Ran 1 test in .*\n

Though what is the test actually testing for, then. I think it is just
the 'known_failures=1' line. Which means we should realy change it to

self.assertContainsRe(stream.getvalue(), r'\(known_failures=1\)')

And then we don't have to worry about anything else.

John
=:->