Re: [PATCH] combine-diff: handle --find-object in multitree code path

From: Jeff King <peff@peff.net>
Date: Sep 30, 2020, 22:54 UTC
Message-ID: <20200930225441.GD1908000@coredump.intra.peff.net>
In-Reply-To: <xmqqmu16gbqk.fsf@gitster.c.googlers.com>

On Wed, Sep 30, 2020 at 03:07:15PM -0700, Junio C Hamano wrote:

Show 10 quoted lines

> > +test_expect_success 'do not detect merge that does not touch blob' '
> > +	git checkout -B merge interesting &&
> > +	git merge -m "untouched blob" base &&
> > +	git diff-tree --format=%s --find-object=$blob -c --name-status HEAD >actual &&
> 
> You learn new things every day ;-)
> 
> I've always thought that for --find-object to do a good job, you'd
> need "--full-history" and perhaps "-m".  Especially, I didn't expect
> "-c" or "--cc" to make a difference.

You don't need --full-history, since there's no history simplification going on here. And "-m" isn't very helpful with --find-object. It's going to find every merge that crosses a boundary where the object was introduced, since it will be introduced in the diff against the _other_ parent. E.g., in:

  A -- B
   \    \
    C -- M

If B introduces object X, then:

  git log --find-object=X -m M

is going to see the diff of C to M as introducing X. But M didn't do anything interesting there; it just picked it up from the branch with B.

My concrete use case here, btw, is reporting to users which commit introduced a blob (and at which path). It mostly works if you ignore merges, but misses out on any evil merges which introduce the object. Adding "-c" fixes that (but before this patch is slow and misses the case where the merge removes the object).

-Peff