On Fri, Feb 20, 2009 at 11:20 AM, Stefan Sperling <stsp_at_elego.de> wrote:
> On Fri, Feb 20, 2009 at 08:57:28AM -0700, Nathan Nobbe wrote:
>> we are migrating to svn1.5 at the office. ive been running some tests
>> using the svn1.5.5 merge tracking and ive run a simple bidirectional
>> merge test, and i dont see it leveraging the merge tracking. let me
>> illustrate the flow, and open up the floor to folks who can clarify
>> things for me.
>> imagine there is a file in a branch, $branches/test/a.c, then a branch
>> is created from there,
>> svn cp $branches/test $branches/test-b
>> ok, now, we checkout both branchs, and make a change to a.c, in
>> test-b, and commit it.
>> cd test-b
>> vi a.c # add a line
>> svn commit a.c
>> then we merge back to upstream,
>> cd ..
>> svn up test
>> svn merge --reintegrate $branches/test-b test
>> svn commit -m 'merge --reintegrate test-b' test
>> now the change from test-b is in test. so now, we go into test-b, and
> Don't go on using $branches/test-b after you have reintegrated it.
> You want to svn rm $branches/test-b at this point.
> If you need it again, re-branch.
> See http://blogs.open.collab.net/svn/2008/07/subversion-merg.html
The alternative is that you can use the svn merge --record-only option
to record the merge on test-b so that subsequent merges do not attempt
to merge back that revision.
Personally, I think removing the branch and recreating when/if there
is more to do is cleaner.
To unsubscribe from this discussion, e-mail: [users-unsubscribe_at_subversion.tigris.org].
Received on 2009-02-20 17:25:04 CET