On 7/11/06, Reedick, Andrew <Andrew.Reedick@bellsouth.com> wrote:
> Thanks for the fast response.
>
> So what is going on here? Is it a case of 'evil twins', aka the
> 'deleted and added instead of copied' issue? Along the lines of:
>
> foo.java(1) and foo.java(2) have the same name but are different
> objects with different history, ancestry, etc. on different branches.
> Evil twins are normally created when you do an 'svn add' to add foo.java
> to two different branches (creating two unique objects,) instead of an
> 'svn add' on branch one, and then an 'svn copy' to branch two (which
> creates two references to a single unique object.)
>
> So two merges are done. One to merge the objects (hence the A
> or U reported by 'svn merge',) and the 2nd to diff/merge the two
> objects' files' contents. So it's really (A or U) plus C?
>
>
> More importantly, I have a vague, queasy feeling that since the ancestry
> is different for each foo.java object, future merges may have subtle
> problems?
Actually, sorry, my bad. I was thinking of a different problem. The
issue we fixed was where status reported something as added, when in
fact it was conflicted. Could be something similar going on with
merge I suppose though. You might want to try out the 1.4.x branch
(release candidate coming real soon now!) and see if it fixes the
issue.
It sounds like the problems are related though, and you're right, the
issue I fixed was one where there was an add from a merge and then an
edit to the same file that resulted in a conflict. I'm not sure how
that would happen in a single merge though...
If you can find a simple reproduction recipie that would be great...
-garrett
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Tue Jul 11 22:15:12 2006