On Thu, May 30, 2013 at 1:37 AM, Johan Corveleyn <jcorvel_at_gmail.com> wrote:
> On Thu, May 23, 2013 at 12:50 AM, Stefan Sperling <stsp_at_elego.de> wrote:
>> The 1.7.10 release artifacts are now available for testing/signing.
>> Please get the tarballs from
>> https://dist.apache.org/repos/dist/dev/subversion
>> and add your signatures there. I plan to try and release on May
>> 30th so please try and get your votes/signatures in place by May 28th.
>
> I had one failure with serf: update_tests.py#45 (tree conflicts 2.1:
> leaf edit, tree del on update). But I can't reproduce it during
> subsequent runs (I made a copy of the svn-test-work directory after
> the failure).
>
> See the dav-fails.log file in attachment.
>
> Some relevant portion:
> [[[
> =============================================================
> Expected 'beta' and actual 'beta' in UNQUIET STATUS tree are different!
> =============================================================
> EXPECTED NODE TO BE:
> =============================================================
> * Node name: beta
> Path:
> svn-test-work\working_copies\update_tests-45\local_leaf_edit_incoming_tree_del\DF\D1\beta
> Contents: None
> Properties: {}
> Attributes: {'status': 'MM', 'copied': '+', 'wc_rev': '-'}
> Children: None (node is probably a file)
> =============================================================
> ACTUAL NODE FOUND:
> =============================================================
> * Node name: beta
> Path:
> svn-test-work\working_copies\update_tests-45\local_leaf_edit_incoming_tree_del\DF\D1\beta
> Contents: None
> Properties: {}
> Attributes: {'status': ' M', 'copied': '+', 'wc_rev': '-'}
> Children: None (node is probably a file)
> ]]]
>
> I'm suspecting a timestamp / sleep issue, which gave me a "spurious
> test failure". But I'm not sure (how can I verify this?). Is this
> failure log consistent with such a problem?
>
> If this proves to be a correct hypothesis, I can sign the release (I
> reran the test 3 times without problems). If not, I'll have to dig a
> bit deeper.
>
> Could this mean some sleep_for_timestamp is missing in this test?
I don't get it. When I look at the working copy from the failed test
now, I can't see the problem:
[[[
(with svn.exe being the 1.7.10 I built and ran the tests with, running
the command exactly as logged by the testsuite)
R:\test\subversion\tests\cmdline>svn.exe status -v -u
svn-test-work\working_copies\update_tests-45\local_leaf_edit_incoming_tree_del
--config-dir R:\test\subversion\tests\cmdline\svn-test-work\local_tmp\config
--password rayjandom --no-auth-cache --username jrandom
...
MM + - 2 jrandom
svn-test-work\working_copies\update_tests-45\local_leaf_edit_incoming_tree_del\DF\D1\beta
...
]]]
Perhaps the (timestamp?) issue is gone now, because I copied the
svn-test-work to C: (as a backup, before rerunning the test), and now
copied it back (R: is a ramdrive).
I'm confused ...
I suppose I can either continue to stare at this (and not deliver a
signature), or just let it go (and sign, and consider this an
unexplained non-reproducible local error that's probably not relevant
...). Anyway, it seems Mark just delivered the third Windows sig, so
it doesn't matter too much.
Lesson for the future: first try to investigate the failed stuff
closely before moving files around.
--
Johan
Received on 2013-05-30 21:30:02 CEST