Re: Integration of undo-tree in Emacs

[Stefan Monnier]

> Yes, makes sense. So the code wouldn't trim every undo/redo pair but
> just those that don't lose information about the other branches that
> exist. So if the user goes up and down a branch twice, we only trim
> out one round trip.

Right.  That is imposed by the current buffer-undo-list representation.

> You lost me at "the most recent state as the root".

"most recent state" = "current buffer state".

> If you mean most recently visited, that is the user's current node and
> is not necessarily the root either.

We're discussing *choosing* a node as being the root.  So, it doesn't
have to be the root, but it can (always) be chosen as the root.
Technically, any one of the nodes can be chosen as the root of "a"
tree.  Obviously, some choices are more meaningful to the user than others.


        Stefan