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

Re: svn commit: r895653 - /subversion/trunk/subversion/bindings/swig/python/tests/mergeinfo.py

From: Paul Burba <ptburba_at_gmail.com>
Date: Mon, 4 Jan 2010 11:49:27 -0500

On Mon, Jan 4, 2010 at 10:00 AM, <hwright_at_apache.org> wrote:
> Author: hwright
> Date: Mon Jan  4 15:00:14 2010
> New Revision: 895653
>
> URL: http://svn.apache.org/viewvc?rev=895653&view=rev
> Log:
> Update a test expectation, as a result of recent mergeinfo sanitation.
>
> * subversion/bindings/swig/python/tests/mergeinfo.py
>  (test_mergeinfo_get): Expect to receive valid mergeinfo.
>
> Modified:
>    subversion/trunk/subversion/bindings/swig/python/tests/mergeinfo.py
>
> Modified: subversion/trunk/subversion/bindings/swig/python/tests/mergeinfo.py
> URL: http://svn.apache.org/viewvc/subversion/trunk/subversion/bindings/swig/python/tests/mergeinfo.py?rev=895653&r1=895652&r2=895653&view=diff
> ==============================================================================
> --- subversion/trunk/subversion/bindings/swig/python/tests/mergeinfo.py (original)
> +++ subversion/trunk/subversion/bindings/swig/python/tests/mergeinfo.py Mon Jan  4 15:00:14 2010
> @@ -129,10 +129,10 @@
>                                        False, None, None)
>     expected_mergeinfo = \
>       { '/trunk' :
> -          { 'branches/a' : [RevRange(2, 11)],
> -            'branches/b' : [RevRange(9, 13)],
> -            'branches/c' : [RevRange(2, 16)],
> -            'trunk'      : [RevRange(1, 9)],  },
> +          { '/branches/a' : [RevRange(2, 11)],
> +            '/branches/b' : [RevRange(9, 13)],
> +            '/branches/c' : [RevRange(2, 16)],
> +            '/trunk'      : [RevRange(1, 9)],  },
>       }
>     self.compare_mergeinfo_catalogs(mergeinfo, expected_mergeinfo)
>
>
>
>

Just making a record of something that puzzled me about the test
failure fixed by this commit...

The "mergeinfo sanitation" referred to in the log message is r888979
"Fix issue #3547 - 'svnadmin load --parent-dir PATH' can destroy
mergeinfo", rather than r889840 "When producing svn_mergeinfo_t from
string representations or vice-versa, tolerate relative merge source
paths, but convert such paths to absolute paths in the resulting
string/mergeinfo."

When I saw mention of this failing test on 1.6.x
(http://svn.haxx.se/dev/archive-2009-12/0508.shtml) I thought for sure
it was related to r889840, but the test still failed prior to that
change being backported on 1.6.x.

The test actually started failing with r888979 because the test repos
was loaded with an empty PARENT_DIR argument to svn_repos_load_fs2:

  def setUp(self):
    """Load the mergeinfo-full Subversion repository. This dumpfile is
       created by dumping the repository generated for command line log
       tests 16. If it needs to be updated (mergeinfo format changes, for
       example), we can go there to get a new version."""
    dumpfile = open(os.path.join(os.path.split(__file__)[0],
                                 'data', 'mergeinfo.dump'))
    # Remove any existing repository to ensure a fresh start
    self.tearDown()
    self.repos = repos.svn_repos_create(REPOS_PATH, '', '', None, None)
    repos.svn_repos_load_fs2(self.repos, dumpfile, StringIO(),
                             repos.svn_repos_load_uuid_ignore, '',
                                                               ^^^

Thus issue #3547 manifested itself and the loaded repos had relative
merge source paths. For some reason the test was incorrectly
expecting this corrupted mergeinfo, and prior to r888979 that is
exactly what it got.

Paul
Received on 2010-01-04 17:50:05 CET

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