Strange details item reprs on skipped tests in babune

[Martin Packman]

Skipped tests appear in the babune output logs on Python 2.7 labelled not as 'ok' but instead 'skipped <repr-of-testtools-details>'.

This is understandable, though ugly, when it's the 'reason' attachment being output:

bzrlib.tests.per_branch.test_branch.TestReferenceLocation.test_sprout_copies_reference_location(RemoteBranchFormat-default) (subunit.RemotedTestCase)
bzrlib.tests.per_branch.test_branch.TestReferenceLocation.test_sprout_copies_reference_location(RemoteBranchFormat-default) ... skipped u'Text attachment: reason\n------------\nBranch cannot hold references.\n------------\n'

But some tests instead have the 'log':

bzrlib.tests.per_branch.test_locking.TestBranchLocking.test_lock_write_with_nonmatching_token(BranchReferenceFormat) (subunit.RemotedTestCase)
bzrlib.tests.per_branch.test_locking.TestBranchLocking.test_lock_write_with_nonmatching_token(BranchReferenceFormat) ... skipped u"Text attachment: log\n------------\n57.587 creating repository in file:///tmp/testbzr-QuHQFd.tmp/bzrlib.tests.per_branch.test_locking.TestBranchLocking.test_lock_write_with_nonmatching_token%28BranchReferenceFormat%29/work/b/.bzr/.\n57.591 opening working tree '/tmp/testbzr-QuHQFd.tmp'\n------------\nText attachment: reason\n------------\nUninitializable branch format\n------------\n"

This is likely down to testtools using the new skip handling in the Python 2.7 unittest, but I'm not sure if the bug needs fixing in bzr, or testtools, or subunit.

[John A Meinel (jameinel)]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 8/27/2010 6:05 PM, Martin [gz] wrote:
> Public bug reported:
>
> Skipped tests appear in the babune output logs on Python 2.7 labelled
> not as 'ok' but instead 'skipped <repr-of-testtools-details>'.
>
> This is understandable, though ugly, when it's the 'reason' attachment
> being output:
>
> bzrlib.tests.per_branch.test_branch.TestReferenceLocation.test_sprout_copies_reference_location(RemoteBranchFormat-default) (subunit.RemotedTestCase)
> bzrlib.tests.per_branch.test_branch.TestReferenceLocation.test_sprout_copies_reference_location(RemoteBranchFormat-default) ... skipped u'Text attachment: reason\n------------\nBranch cannot hold references.\n------------\n'
>
> But some tests instead have the 'log':
>
> bzrlib.tests.per_branch.test_locking.TestBranchLocking.test_lock_write_with_nonmatching_token(BranchReferenceFormat) (subunit.RemotedTestCase)
> bzrlib.tests.per_branch.test_locking.TestBranchLocking.test_lock_write_with_nonmatching_token(BranchReferenceFormat) ... skipped u"Text attachment: log\n------------\n57.587 creating repository in file:///tmp/testbzr-QuHQFd.tmp/bzrlib.tests.per_branch.test_locking.TestBranchLocking.test_lock_write_with_nonmatching_token%28BranchReferenceFormat%29/work/b/.bzr/.\n57.591 opening working tree '/tmp/testbzr-QuHQFd.tmp'\n------------\nText attachment: reason\n------------\nUninitializable branch format\n------------\n"
>
> This is likely down to testtools using the new skip handling in the
> Python 2.7 unittest, but I'm not sure if the bug needs fixing in bzr, or
> testtools, or subunit.

I mentioned this as part of my 'remove log from skips' branch, but we
might implement 'details' on bzr's addSkip method and then shrink the
reason way down.

That way, instead of formatting everything into the reason, we could do
just what we wanted (namely, the 'reason' detail).

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkx7t/oACgkQJdeBCYSNAAOEIwCdGnt6rPdDGkY4V7jRJvo90EU7
WCoAn1CB3jlVch2RoR/ZN8iV4OW5kppF
=5V7h
-----END PGP SIGNATURE-----

[Martin Packman (gz)]

This can get blamed on any of bzr, subunit, or testtools, but there's an easy fix in the last of those so I'll tackle it there.

[Martin Packman (gz)]

Fixed on babune by testtools upgrade, as can be seen since the natty builder was added:
<http://babune.ladeuil.net:24842/job/selftest-natty/7/>

The output is now much less confusing:

bzrlib.tests.blackbox.test_dpush.TestDpushStrictWithChanges.test_push_with_revision(uncommitted) ... skipped u'dpush does not handle --revision\n'