Re: [take 2] revision parsing made incremental

Pierre Habouzit [off-list ref] writes:

Following Dscho's remarks, I reworked the series to avoid changing
setup_revisions semantics for now, and only exposed the part that groks
options (and keep pseudo revision arguments out).

It indeed makes the series smaller, even if the first patch is quite
long to read, and is just enough for simplifying git-blame in a very
satisfying way.

All looks good and the result looks much more readable.
Thanks.