Also, are you going to modify hg next --rebase to use this new restack functionality? Presently it just calls _restackonce() (which itself is not ideal for that case since it rebases all of the children at each level of the stack rather than just the one on the path to the destination).

Does this mean that the entire restack will continue after a merge conflict with hg rebase --continue? (Since the entire thing is just a single rebase call now.) If so, that's really great.

I'm seeing a unit test failure I don't understand in test-check-config-hg.t. Any idea what's going on?

I'm going to accept this since this is blocking our internal release, but there are some portability concerns here since POSIX advisory locking isn't available on Windows, for example. That should probably be addressed in a separate patch. It might be worth making note of this as a comment in the code?