[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Python testsuite on windows

From: David James <james82_at_gmail.com>
Date: Mon, 27 Apr 2009 14:02:52 -0700

Hi Jens,

Thanks for your post. Could you try this again using an older version
of Python, say, Python 2.5? I think Python 2.6 may have made a few
changes that are breaking the test suite.

> D:\develop\svn\subversion-1.6>python
> subversion\bindings\swig\python\tests\clien
> t.py
> F............
> ======================================================================
> FAIL: Test playing with C batons
> ----------------------------------------------------------------------
> Traceback (most recent call last):
> File "subversion\bindings\swig\python\tests\client.py", line 49, in
> testBatonP
> lay
> str(self.client_ctx.auth_baton))
> AssertionError: "<libsvn.core.svn_auth_baton_t; proxy of <Swig Object of
> type 'svn_auth_baton_t *' at 0x0000000003953780> >" !=
> "<libsvn.core.svn_auth_baton_t;
> proxy of <Swig Object of type 'svn_auth_baton_t *' at 0x0000000003953750> >"

Those two strings look identical to me. In fact, I copied them into a
text editor and found that they were identical. This is likely a
peculiarity in Python 2.6. Arfrever, any idea why these two strings
would not compare identical? Note that these are string
representations of pointers.

On Mon, Apr 27, 2009 at 1:11 PM, Jens Peters <jpeters7677_at_gmx.de> wrote:
> Merge-info:
>
> d:\opt\Subversion\python\svn\core.py:47: DeprecationWarning:
> BaseException.messa
> ge has been deprecated as of Python 2.6
> self.message = message
> EEEEEE
> ======================================================================
> ERROR: test_mergeinfo_get (__main__.SubversionMergeinfoTestCase)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
> File "subversion\bindings\swig\python\tests\mergeinfo.py", line 55,
> in tearDow
> n
> repos.delete(REPOS_PATH)
> File "d:\opt\Subversion\python\libsvn\repos.py", line 93, in
> svn_repos_delete
> return _repos.svn_repos_delete(*args)
> SubversionException: ("Can't remove
> 'd:\\tmp\\tmpxyfbvp-trac-svnrepos\\db\\rep-c
> ache.db'", 720032)
>
> All 6 tests are failing with the same message.
>
> Ra:
> 16 of 17 tests are failing with "Can't remove ..". The first test
> finishes fine.
>
> WC:
> 22 of 23 tests are failing with "Can't remove ..". The first test
> finishes fine.
>
> Repository:
> I dont see a summary, but fails with "Can't remove ..". The first test
> seems to finish fine, according to the number of dots :) ).

Okay, I find this a bit confusing.
 1) client.py is having no problem deleting repositories
 2) From the message you sent, I can't tell whether mergeinfo.py is
able to delete repositories at all. Are all six exceptions identical,
or is there a slight difference in the exception? E.g., is the first
exception thrown during the second tearDown instead of the first
tearDown?
 3) repository.py, ra.py and wc.py can delete a repository, but only
during the first run.

What's the difference between the four classes? Well, it looks like
the classes that experience problems all keep references to an FS
object inside the test object (self.fs). client.py seems to not do
that and maybe that's why it's working. Maybe we should teach the
tearDown method of the classes to kill references to self.fs etc. That
might fix the problem.

Jens, could you also try updating mergeinfo.py to set self.fs = None
and self.repos = None and see if that helps? I've attached an
(untested) patch which does that.

Cheers,

David

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=1951895

Received on 2009-04-27 23:03:24 CEST

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.