[cvsnt] Re: Merging branches with mergepoints

Tony Hoyle tony.hoyle at march-hare.com
Sat Dec 3 04:06:51 GMT 2005


Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to sales@march-hare.com.


Detlef Marschall wrote:

> 
> There's a problem. The user must know that this is the first merge to B2
> for the specific source. Then he must use "-jTAG -jBRANCH" for the
> merge, otherwise he can merge with "-jBRANCH". For only a few sources is
> this OK. But there are 2816 sources, without includes, and nobody can
> say how many he must change for a problem. Only when all sources was
> merged from B1 to B2 then we can use mergepoints with no problem.

The problem is created by the fact that B2 has the wrong base.

If you're promoting HEAD->B1->B2 then B2 should be based on B1 (or 
possibly the same revision of HEAD).  By creating B2 out of a different 
revision there is no possible way of retionalising it without an 
existing mergepoint.

You've got the situation:

1.3 -> B1 + changes (eg. 1.3.2.5)
1.8 -> B2 + changes (eg 1.8.2.21)

The 'common ancestor' of both branches is 1.3, so the only possible (and 
logical) merge of B1->B2 is the entireity of B1 onto B2.  Once a 
mergepoint is there it fixes this as it knows where the first merge was.

The only revisions being considered during such a merge are 1.3->1.3.2.5 
(common ancestor to branch head) being patched onto revision 1.8.2.21 
(current sandbox revision).  There are no mergepoints within that frame 
of reference.  Once a mergepoint is registered at 1.3.2.5, any future 
merges will use that as their starting point.

Tony



More information about the cvsnt mailing list
Download the latest CVSNT, TortosieCVS, WinCVS etc. for Windows 8 etc.
@CVSNT on Twitter   CVSNT on Facebook