On Wed, Apr 12, 2023 at 12:22:39PM +0200, Patrick Steinhardt wrote:
> When writing the multi-pack-index with geometric repacking we will add
> all packfiles to the index that are part of the geometric sequence. This
> can potentially also include packfiles borrowed from an alternate object
> directory. But given that a multi-pack-index can only ever include packs
> that are part of the main object database this does not make much sense
> whatsoever.
>
> In the edge case where all packfiles are contained in the alternate
> object database and the local repository has none itself this bug can
> cause us to invoke git-multi-pack-index(1) with only non-local packfiles
> that it ultimately cannot find. This causes it to return an error and
> thus causes the geometric repack to fail.
>
> Fix the code to skip non-local packfiles.
>
> Co-authored-by: Taylor Blau <me@xxxxxxxxxxxx>
Thanks for listing me as a co-author. Doing so probably requires my
S-o-b, which you are more than free to forge here.
> + # Assert that we wrote neither a new packfile nor a multi-pack-index.
> + # We should not have a packfile because the single packfile in the
> + # alternate object database does not invalidate the geometric sequence.
> + # And we should not have a multi-pack-index because these only index
> + # local packfiles, and there are none.
> + find member/.git/objects/pack -type f >actual &&
> + test_must_be_empty actual
This test looks good, though one small opportunity for cleanup might be
to replace this with a single:
test_dir_is_empty member/$packdir
Thanks,
Taylor
