On Nov 8, 2006, at 10:12, Gundersen, Richard wrote:
> I haven’t had anywhere near the number of replies I was expecting
> saying my approach is awful.
>
> Perhaps this is because it’s something that’s been debated to death
> before and nobody wants to get it started again. Or perhaps it’s
> because it really is better than the release-based branching
> strategy (trying to stir up some controversy here :) )
>
> Does anyone know of any real advantages the release-based approach
> has over the feature-based way?
I never imagined that these were conflicting strategies; I think they
can and in many cases should coexist harmoniously. I'm just enjoying
reading your arguments and the responses in this discussion; I don't
have too much to add to it.
In the company where I worked, we primarily used release branches;
most changes to be made were small enough that they could be done in
one or two commits, and thereby not destabilize the trunk. However,
we did allow larger development to occur on trunk, even leaving it
unstable, while releases could still be cut from the latest stable
release branch. One of our developers liked the feature-branch idea
and used it on some occasions, but the rest of us didn't bother to
learn or practice that.
I'm not at all convinced that the way we used Subversion was the best
way, so I'm certainly interested in reading the arguments in this
thread.
> Anyone know how are Subversion releases themselves done?
Looking at the branches directory of the Subversion repository, I
think it's safe to say that they make use of both strategies:
http://svn.collab.net/repos/svn/branches/
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Thu Nov 9 00:42:06 2006