svn branches in svn-current
Ronald Blaschke
ron at rblasch.org
Mon May 3 17:59:40 UTC 2010
On 03.05.2010 16:48, Will Coleda wrote:
> On Mon, May 3, 2010 at 10:32 AM, Will Coleda <will at coleda.com> wrote:
>> From http://svnbook.red-bean.com/nightly/en/svn.branchmerge.basicmerging.html#svn.branchemerge.basicmerging.stayinsync
>>
>> Keeping branch in sync with trunk: svn merge ^/trunk
>>
>> Merging branch back to trunk: svn merge --reintegrate ^/branches/my-calc-branch
>> - at which point you need to delete the branch and start over if you
>> have more changes related to this feature.
>>
>> I think the problem we've been having is partially following the old
>> school of branching, with specific revisions and no --reintegrate.
I think that's good practice. Also, when merging, it's best to watch
out for the following:
* There should be no local modifications.
* The working copy should be on a single revision.
* There should be no switched directories.
* The working copy should be complete (i.e., no sparse checkout)
It's easiest to check for this with "svnversion."
$ cd Parrot/trunk
$ svnversion
46219
If you see a single revision everything is fine. Otherwise you'll get
something like:
4123:4168 mixed revision working copy
4168M modified working copy
4123S switched working copy
4123P partial working copy, from a sparse checkout
4123:4168MS mixed revision, modified, switched working copy
Relying on merge info mean that every client must participate, of
course. See Blocking Merge-Unaware Clients
http://svnbook.red-bean.com/nightly/en/svn.branchmerge.advanced.html#svn.branchmerge.advanced.pre1.5clients
>> For those of us using straight svn instead of git-svn, we should
>> probably follow these guidelines; I'm not sure if we need to do
>> anything in particular to the existing svn:mergeinfo properties. I
>> suspect we might need to remove them from trunk, but I'm not sure that
>> it's necessary.
> FYI, when committing this merge for branches/codestring, I got the
> following error:
[snip]
> .. And so on. I eventually had to nuke the checkout from orbit and
> start over. Thankfully I didn't have any uncommitted changes.
Hmm, maybe you got hit by this bug.
http://subversion.tigris.org/issues/show_bug.cgi?id=3198
Ron
More information about the parrot-dev
mailing list