Re: Merging from a branch several times

[Mark D. Baushke]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Kyrre Kristiansen <address@hidden> writes:

> Hello, all
> 
> There's been a few discussions here where I work about
> branches and merging. The Cederquist manual states
> that:
> 
> "If you just use the cvs update -j R1fix m.c command
> again, CVS will attempt to merge again the changes
> which you have already merged, which can have
> undesirable side effects." [1]
> 
> We've been tagging all the merging we've done between
> our main development branch (MAIN) and a branch for
> bug fixing (Patch) on every merge, and have used the
> suggested method of using two -j options when
> merging.[2]
> 
> My question is: Is this really necesary? 

Sadly, yes. It looks like the CVSNT folks have
fixed this for simple merges by adding a mergpoint
extension to the RCS files. At the present time,
CVS does not have any similar annotation to find
the most recent equivalence points and so must
continue to use the greatest common ancestor
algorithms.

> If I want to merge the head of the patch branch
> with the head of the main branch, is it not
> exough to do "update -r Patch ."?

If you have a tree with lots of merges on a branch
and mainline go ahead and run the experiment in a
checked out top-of-tree. It will not be a problem
for you unless you commit the results.

I suspect you will find lots of conflict blocks
in the resulting files that need to be resolved
for which you have already done resolution once.

        Good luck,
        -- Mark

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (FreeBSD)

iD8DBQFBQcvl3x41pRYZE/gRAn9EAJ0VMglp/L2u7zRGAOOqL0grnKEmSgCg06Mn
BdXnHqQWUZqpA5g81jOA+vg=
=bUT8
-----END PGP SIGNATURE-----