On Sun, Apr 5, 2009 at 16:36, Arfrever Frehtes Taifersar Arahesis
<Arfrever.FTA_at_gmail.com> wrote:
> 2009-04-05 12:49:30 Bert Huijben napisał(a):
>>
>> > -----Original Message-----
>> > From: Arfrever Frehtes Taifersar Arahesis
>> > [mailto:arfrever.fta_at_gmail.com]
>> > Sent: Sunday, April 05, 2009 6:11 AM
>> > To: Bert Huijben; Subversion Development
>> > Subject: Re: Windows buildbot errors (A lot!)
>> >
>> > 2009-04-05 02:10:56 Bert Huijben napisał(a):
>> > > Hi,
>> > >
>> > > Somewhere in the last 24-36 hours the number of errors on the Windows
>> > > buildbot exploded from just 1 (revert_tests.py 1) to 96.
>> > >
>> > > Rerunning the tests at my pc I get the following summary (with the
>> > same 96
>> > > failures):
>>
>> <snip />
>> > > Summary of test results:
>> > > 972 tests PASSED
>> > > 43 tests SKIPPED
>> > > 32 tests XFAILED (4 WORK-IN-PROGRESS)
>> > > 96 tests FAILED
>> > >
>> > > r37006 is responsible for 73 of the 96 test failures;
>> >
>> > I fixed these 73 test failures in r37008.
>>
>> If I do a
>> svn merge -c -37008,-37006,-37005,-37004,-37003,-37002,-37001,-37000,-36999,-36998,-36997,-36996,-36995,-36987,-36986,-36985,-36984,-36983 ^/trunk .
>>
>> before running the tests the other 22 tests are succeeded, so it looks
>> like 95 of these test failures where caused by your changes for
>> introducing some python 3 compatibility in the test suite.
>
> prop_tests.py 13: some svn: properties should be converted
>
> testlog:
> ================================================
> CMD: svn.exe propset svn:executable "*" "svn-test-work\working_copies\prop_tests-13\A\B\lambda" --config-dir "M:\svn-auto-test\fsfs\subversion\tests\cmdline\svn-test-work\local_tmp\config" --password rayjandom --no-auth-cache --username jrandom <TIME = 0.578000>
> svn: 'svn-test-work' is not a working copy
> Traceback (most recent call last):
> File "C:\svn-builder\djh-xp-vse2005\build\subversion\tests\cmdline\svntest\main.py", line 1172, 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/prop_tests.py", line 869, in prop_value_conversions
> svntest.actions.set_prop('svn:executable', '*', lambda_path)
> File "C:\svn-builder\djh-xp-vse2005\build\subversion\tests\cmdline\svntest\actions.py", line 1573, in set_prop
> main.run_svn(expected_err, 'propset', name, value, path)
> File "C:\svn-builder\djh-xp-vse2005\build\subversion\tests\cmdline\svntest\main.py", line 594, in run_svn
> return run_command(svn_binary, error_expected, 0,
> File "C:\svn-builder\djh-xp-vse2005\build\subversion\tests\cmdline\svntest\main.py", line 363, in run_command
> None, *varargs)
> File "C:\svn-builder\djh-xp-vse2005\build\subversion\tests\cmdline\svntest\main.py", line 528, in run_command_stdin
> raise Failure
> Failure
> FAIL: prop_tests.py 13: some svn: properties should be converted
> ================================================
>
> Windows seems to incorrectly handle "*". Please test the attached patch (subversion-set_prop.patch).
What?! It worked JUST FINE before your changes.
Adding workarounds because the test suite is all gummed up because of
useless Py3 compatibility code is the wrong way to go. Why should we
add cruft? That cruft has a *cost*. And I'll say it again: I see zero
benefit in return for it.
-g
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=1553866
Received on 2009-04-05 21:51:39 CEST