On Tue, Aug 04, 2009 at 01:00:51PM +0200, Daniel Näslund wrote:
> > Hi Daniel,
> >
> > I've attached an updated patch this time which includes my suggested
> > changes. Please check my version of the patch to see if you are happy
> > with it.
> >
> >
> > I've slightly tested the attached patch now and produces splendid
> > diffs for multi-line property values. Thanks a lot! :)
> >
> > Before we commit this, as a final verification step, can you take the
> > time to run 'make check' on the patch and clean up any fallout in the
> > test suite? Or have you run the regression tests on this already?
> >
> > Thanks,
> > Stefan
>
> When running depth_tests.py 19 and diff_tests.py 42 I get failures.
> Theese two test uses the --depth option for diffing. The tests display
> the diffs for each target from the top of the tree and downward. That
> is:
>
> top_dir -
> | file
> |--subdir
>
> will be displayed in order top_dir-file-subdir if depth is set to
> infinity. But after applying the patch for issue 1493 the order is
> subdir-file-top_dir. How can that be? AFAIK I haven't touch anything
> that handles the depth. Using other the other depth options (empty,
> files, immediates) produces the same result. The targets are displayed
> from bottom up instead of up to down.
If I remember correctly the order which 'svn diff' uses for its
files is undefined. Julian wanted to fix this ages ago but for some
reason his patch wasn't applied.
So if the test assumes a fixed order, it's broken.
Stefan
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2379965
Received on 2009-08-04 13:31:08 CEST