Elijah Newren <new...@gmail.com> writes:
> Interesting; tbdiff looks cool. Junio hasn't queued this series yet,
> but it's easy enough to reconstruct the old one. It does weigh in
> pretty heavy, and I'm slighly worried about gmail mangling all the
> lines, but I'm going to give it a shot anyway. If it's too mangled,
> I'll try to repost using git-send-email. It does weigh in at over
> 1600 lines, so it's not small.
It seems that you have installed tbdiff correctly. The below seems
to match what I saw when I queued this round, relative to the
previous one.
What I often do when I see a new round of patches is:
$ git checkout en/rename-directory
$ git checkout master... ;# detach at the base of the old
$ git am -s mbox ;# take the new
$ git tbdiff ..@{-1} @{-1}..
to compare the old and new. Often (but not with this topic) earlier
parts of the topic are identical between the old and the new, so I
may rebase the new to preserve the commit timestamp of the old one
when it happens after the above sequence of commands. For example,
if I see these in tbdiff
1: 7893bf1720 = 1: f17207893b commit #1
2: c291293b2e = 2: 93b2ec2912 commit #2
3: a7d3c870a3 ! 3: 87b5e236a1 commit #3
@@ ... @@@
then we know up to commit #2 are the same as before, so I'd do
$ git rebase --onto c29129eb2e 93b2ec2912
by using the two commit object names on the last "=" line in the
output. Then, running the same tbdiff again:
$ git tbdiff ..@{-1} @{-1}..
would now show the output starting from "commit #3".
