On Sep 3, 2004, at 12:33 PM, Ben Collins-Sussman wrote:
> On Fri, 2004-09-03 at 11:24, Scott Palmer wrote:
>
>> Is it really that hard? Isn't this a fairly simple extension of what
>> is already in place? always remember the version at which the code
>> branched or was merged. When a merge is done update a "merge-point"
>> record that says 'changes prior to this point are already merged'.
>> Subsequent merges would only merge changes from the last
>> "merge-point".
>> Is it really 2 years to do that feature? Or am I way
>> over-simplifying
>> like I usually do :) ?
>
> You've described a simple solution for the case where you repeatedly
> merge contigous ranges of revisions from one branch to another. It
> solves the problem of the user asking, "hm, where did that last merge
> end? Let me pick up from there."
>
> But in your own case, you've done 'cherry-picking'. You've selectively
> ported a bunch of random, non-contiguous changesets. Now you want to
> merge the "whole branch" to another, and not repeat changesets you've
> already got. Much more complicated.
>
> Take a look at the Arch and SVK homepages. There are volumes of
> philosophy dedicated to solving this problem.
Point taken, I guess the "simple case" is all I've ever needed.
Scott
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Fri Sep 3 18:57:01 2004