Thanks to preliminary testing by Paul, Mark and Philip, the symmetric merge idea is starting to show its main point, but it is far from complete.
What needs to be done? The current implementation is crude, and suffices only to get the development started by calling one or other of the current 'sync' or 'reintegrate' code paths. The two big things are: * Subtree mergeinfo. I spent part of Friday talking this through with Philip Martin and we agreed on how it should behave. I'll write this up in a separate post. * Skipping source-branch revisions that have been cherry-picked to the target branch. The present source code only does this when it invokes the 'sync-like' (base on source branch) code path; the 'reintegrate-like' (base on target branch) code path doesn't attempt to do this. We need to generalize the merge subroutines to handle both cases with the same code. These are not independent tasks; rather they are two aspects of the behaviour that the merge code will need to have. New People I'm pleased to be able to say that as of this weekend we can welcome three more people -- Stefan Fuhrmann and Cristian Camerin and Oleksiy Suschuk -- who are all now getting involved with this work. They are all working for Assembla and have just been asked to help us, and so are reading up on the merge stuff and (apart from our Stefan2) on contributing to the Subversion community. Please join me in welcoming them on board. - Julian
