Dotted revno "algebra"

[John Yates]

I would like to see a discussion of this (Ben Finney's) perspective added
to Matthew Fuller's Numbering Revisions page:

http://wiki.bazaar.canonical.com/MatthewFuller/SpotDocs/RevNumbering

/john
________________________________________

For my part, I find the branch-point-based revno useful because it is in
terms of what I'm interested in: what is the distance between this merge
and the previous common ancestor along the branch.

It's a measure of conceptual size of the divergence that is now being
resolved. The measure is in two dimensions that I care about: how far
back on this branch, and how far back on the other branch. That's been
helpful many times to me in forming a conceptual model of the potential
impact of what is merged. It's in *addition* to the size and contents of
the diff, and makes the conceptual model richer.