Re: What's cooking in git.git (Aug 2020, #07; Thu, 27)

[Taylor Blau <me@ttaylorr.com>]

On Thu, Aug 27, 2020 at 04:36:43PM -0700, Junio C Hamano wrote:
> Jeff King <peff@peff.net> writes:
>
> > On Thu, Aug 27, 2020 at 02:43:02PM -0700, Junio C Hamano wrote:
> >
> >> * tb/repack-clearing-midx (2020-08-26) 1 commit
> >>   (merged to 'next' on 2020-08-27 at a465875cbb)
> >>  + builtin/repack.c: invalidate MIDX only when necessary
> >>
> >>  When a packfile is removed by "git repack", multi-pack-index gets
> >>  cleared; the code was taught to do so less aggressively by first
> >>  checking if the midx actually refers to a pack that no longer
> >>  exists.
> >>
> >>  Will merge to 'master'.
> >
> > This seems to break t7700 when run with midx support:
> >
> >   $ git checkout e08f7bb093
> >   HEAD is now at e08f7bb093 builtin/repack.c: invalidate MIDX only when necessary
> >
> >   $ make && (cd t && GIT_TEST_MULTI_PACK_INDEX=1 ./t7700-repack.sh -i)
> >   [...]
> >   + git repack -a -d
> >   Enumerating objects: 10, done.
> >   Counting objects: 100% (10/10), done.
> >   Delta compression using up to 16 threads
> >   Compressing objects: 100% (5/5), done.
> >   Writing objects: 100% (10/10), done.
> >   Total 10 (delta 1), reused 10 (delta 1), pack-reused 0
> >   fatal: error preparing packfile from multi-pack-index
> >   error: last command exited with $?=128
> >   not ok 6 - packed obs in alt ODB are repacked when local repo has packs
> >   #
> >   #		rm -f .git/objects/pack/* &&
> >   #		echo new_content >>file1 &&
> >   #		git add file1 &&
> >   #		test_tick &&
> >   #		git commit -m more_content &&
> >   #		git repack &&
> >   #		git repack -a -d &&
> >   #		test_no_missing_in_packs
> >
> > I didn't look into whether it's a bug in the actual code, or just a
> > weird interaction with the way GIT_TEST_MULTI_PACK_INDEX triggers
> > git-repack to write a midx. But either way we should figure that out
> > before it graduates.
>
> Thanks for stopping us ;-)

Thanks indeed. I started looking into this tonight thinking that it'd be
an easy fix, but I think there is a deeper bug that is worth
investigating further.

Let's eject this for now. If the bug turns out to be easier than I
thought, I'll send the patch again, otherwise I'll send it with my
larger series when that's ready.


Thanks,
Taylor