Re: git branches (was: Re: Reply to David Holland's notes and comments)

["Eric S. Raymond" <esr%thyrsus.com@localhost>]

Dennis Ferguson <dennis.c.ferguson%gmail.com@localhost>:
> I guess I would argue that every one of those is a "join", in that they
> are all merging one branch's development into another. 

Well, at the content level you're certainly right.  

The argument for the other point of view is that the merge operations are not
reflected in the DAG.  No node has more than one parent.  Thus a CVS merge
is analogous not to a git merge but to a git cherry pick.  It doesn't create
any ambiguity about the graph coloring.

Which brings us back to a practical question.  Are we going to try to identify 
these sync points and turn them into real merges in git?

reposurgeon can edit a node to create a join there, but it can't identify where
this should be done.  For that to happen, some human will have to audit all 
branch tips, identify which ones are merged where, and provide a (tip, target)
list of pairs. Commands for gluing these together will go in the lift script.
-- 
		<a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>