Re: Build-bots always failing [was: svn trunk r37930: FAIL (win32-xp VS2005)]

Hyrum K. Wright wrote:
> On Jun 4, 2009, at 11:10 PM, Julian Foad wrote:
>
> > What's up with the Windows and Ubuntu build-bots always failing? (They
> > seem to have been failing for many weeks at least.)
>
> The windows buildbot started failing a couple of weeks ago when I
> tweaked the notification framework a bit. Since I've no Windows box
> to test on, and little experience with the "proper" way to fix the
> problem on Windows, I've left that to somebody else to fix. Bert
> asked about it in IRC today, but I'm not sure what his plan is.

I tried looking for the details...

1. The failure email says "Full details are available at:
http://crest.ics.uci.edu/buildbot/builders/win32-xp%
2520VS2005/builds/230" but
<http://crest.ics.uci.edu/buildbot/builders/win32-xp%
2520VS2005/builds/230> says "404 - No Such Resource". The correct
address is <http://crest.ics.uci.edu/buildbot/builders/win32-xp%
20VS2005/builds/230>. (The problem is that the "%" was double-escaped in
the email.)

2. The failure seems to be "FAIL: update_tests.py 32: update wc on the
root of a Windows (virtual) drive". Here's the end of that test's
output:

CMD: svn.exe up -r 1 G:/ --config-dir "M:\svn-auto-test\fsfs\subversion
\tests\cmdline\svn-test-work\local_tmp\config" --password rayjandom
--no-auth-cache --username jrandom

<TIME = 4.297000>

U A\mu

D dir1

D zeta

Updated to revision 1.

Couldn't find node 'zeta' in expected output tree

 * Node name: zeta

    Path: __SVN_ROOT_NODE\zeta

    Contents: None

    Properties: {}

    Attributes: {'status': 'D '}

    Children: None (node is probably a file)

ACTUAL OUTPUT TREE:

svntest.wc.State('G:/\', {

})

EXCEPTION: SVNTreeUnequal

Traceback (most recent call last):

  File "C:\svn-builder\djh-xp-vse2005\build\subversion\tests\cmdline\svntest\main.py", line 1138, in run

    rc = self.pred.run(sandbox)

  File "C:\svn-builder\djh-xp-vse2005\build\subversion\tests\cmdline\svntest\testcase.py", line 114, in run

    return self.func(sandbox)

  File "C:\svn-builder\djh-xp-vse2005\build\subversion/tests/cmdline/update_tests.py", line 2200, in update_wc_on_windows_drive

    '-r', '1', wc_dir)

  File "C:\svn-builder\djh-xp-vse2005\build\subversion\tests\cmdline\svntest\actions.py", line 694, in run_and_verify_update

    check_props)

  File "C:\svn-builder\djh-xp-vse2005\build\subversion\tests\cmdline\svntest\actions.py", line 605, in verify_update

    tree.compare_trees("output", actual_output, output_tree)

  File "C:\svn-builder\djh-xp-vse2005\build\subversion\tests\cmdline\svntest\tree.py", line 628, in compare_trees

    singleton_handler_a(a_child, a_baton)

  File "C:\svn-builder\djh-xp-vse2005\build\subversion\tests\cmdline\svntest\tree.py", line 525, in default_singleton_handler

    raise SVNTreeUnequal

SVNTreeUnequal

FAIL: update_tests.py 32: update wc on the root of a Windows (virtual) drive

That looks like a genuine "Subversion doesn't work properly on that
platform" bug.

> The Ubuntu buildbot is failing due to a couple of the bindings tests
> segfaulting. I've tried pinging the bindings maintainers directly,
> but to no avail, so the problems will persist until somebody cares
> enough to fix them. <insert rant about bindings being first class
> citizens here>

I can't tell whether that indicates a bug in Subversion (bindings
support) or a mis-configuration of the Ubuntu build-bot.

Thanks for the info.

- Julian

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2359670
Received on 2009-06-05 13:53:44 CEST