On 10 September 2014 18:25, <kotkov_at_apache.org> wrote:
> Author: kotkov
> Date: Wed Sep 10 14:25:49 2014
> New Revision: 1624011
>
> URL: http://svn.apache.org/r1624011
> Log:
> Fix a broken assumption within the r1603605 changeset.
>
> When format 1 and 2 filesystems are being upgraded, the upgrade routine
> leaves the db/current contents as is. As a consequence, there is a window
> when a filesystem has a new format, but the 'current' file still contains the
> additional IDs left from an old format, i.e. it could look like "359 j5 v\n"
> (last values are bogus and won't be there after the next transaction commit).
> We should be able to parse it, otherwise any upgraded pre-1.4-compatible
> repository will block the commits.
>
> Loosen the corresponding checks.
>
[...]
>
> Modified: subversion/trunk/subversion/tests/cmdline/svnadmin_tests.py
> URL: http://svn.apache.org/viewvc/subversion/trunk/subversion/tests/cmdline/svnadmin_tests.py?rev=1624011&r1=1624010&r2=1624011&view=diff
> ==============================================================================
> --- subversion/trunk/subversion/tests/cmdline/svnadmin_tests.py (original)
> +++ subversion/trunk/subversion/tests/cmdline/svnadmin_tests.py Wed Sep 10 14:25:49 2014
> @@ -2928,6 +2928,20 @@ def freeze_same_uuid(sbox):
> sys.executable, '-c', 'True')
>
>
> +@Skip(svntest.main.is_fs_type_fsx)
I think this test should be skipped for all FS types except FSFS. I.e.
@SkipUnless(svntest.main.is_fs_type_fsfs)
Did I miss something?
--
Ivan Zhakov
Received on 2014-09-10 16:58:29 CEST