This format makes it easier to correlate SVN revisions with Git commits
for users who depend on such behaviour.
Usual caveats around working with SVN revision numbers apply: they do
not identify patch/commit, they identify a change in the state of whole
SVN repository (i.e. single revision might span multiple paths,
including multiple "branches" or "tags" or "projects").
Do not depend on SVN revisions to uniquely identify a commit created by
an SVN user (especially for scripting) - you need a tuple <SVN path,
SVN revision> for that. It's easier to identify a commit by git hash
(this script displays shortened hash in a first column).
d4b74d9686