On Tue, 2002-04-02 at 16:58, Ben Collins wrote:
> cd my-branch
> svn merge http://server/svn/trunk (syntax?)
>
> IOW, I want to magically merge in all changes to trunk, from the time
> that my-branch was created. IMO, this is a very important feature for
> branches. You could fit this into your model by noting what rev of trunk
> you copied my-branch, and continually note the new rev everytime you
> merged, but that's a hassle, prone to mistakes.
Agreed, it's not friendly to force the user to remember exactly which
revision the branch was created in. If the user *does* remember, then
obviously, the command would be
svn merge http://server/svn/trunk -r N:HEAD
The point you bring up is a good one, but it falls into the "phase 2"
category of our merge implementation. "phase 1" is just getting the
basic mechanics working. That's the interface I'm trying to finish, and
what my last mail was about. "Phase 2" is about using metadata to make
the whole branching/switching/merging experience more user-friendly.
Examples of things in phase 2:
* when I create a branch, some svn: property should remember the URL the
branch was copied-from, and in which revision this happened.
* svn merge should make use of that information.
* svn merge should also set svn: properties on files which record the
changesets that have been merged in, so as to avoid the repeated-merge
problem.
But, one phase at a time. Let me get the un-friendly mechanics working
first. :-)
Received on Wed Apr 3 19:19:36 2002