[p4] Question about branching/integrating

David Rybolt dagg at yahoo.com
Mon Jun 14 13:08:41 PDT 2004


--- Chuck Karish <chuck.karish at gmail.com> wrote:
> Branch from main at a point where you're sure that
> you'll take
> everything.   Integrate features from main to your
> new release branch
> as they're finished.

That won't work in this case, unfortunately.  The
4.1.0 code was integrated into main _after_ some
changes were already added to main.  That means I
can't get all the 4.1.0 code integrated into my new
branch unless:

  1. I also integrate some of the new "main" stuff
(which I don't want).

  2. Or I branch from 4.1.0 .

However if I branch from 4.1.0, that'll make it hard
for me to integrate to/from main later.  I can get it 
to work with some baseless merges (-i or -I), but
there seems to be a lot of repetitive and hard
to manage conflict resolutions once I go down that
road.

> A more radical approach:  Make a permanent
> forward-development branch,
> and integrate features from //depot/dev only when
> theu're ready to be
> released.  The main line stays clean, the developers
> grumble about
> doing one more integration, the project managers
> love that release
> readiness is predictable and that they control
> release contents.

I'll likely utilize this approach at a latter
date, but I still fear issues like this:

  main 
    |
    +-- dev1
    |
    |
    +-- dev2

Both dev1 and dev2 are branched from main.  What do
I do if the dev1 folks want some of the dev2 code?
I can integrate dev2 to main, and then main
to dev1, but then my main codeline will have
some stuff that may or may not ever go to
production.  Alternatively, I can baseless
merge dev2 to dev1, but that will reintroduce
the long-term problems of baseless merging.

Dave


	
		
__________________________________
Do you Yahoo!?
Friends.  Fun.  Try the all-new Yahoo! Messenger.
http://messenger.yahoo.com/ 



More information about the perforce-user mailing list