Re: What's cooking in git.git (Nov 2024, #03; Mon, 4)

From: Jonathan Tan <jonathantanmy@google.com>
Date: Nov 5, 2024, 21:28 UTC
Message-ID: <20241105212849.3759572-1-jonathantanmy@google.com>
In-Reply-To: <xmqqwmhirrvl.fsf@gitster.g>

Junio C Hamano <gitster@pobox.com> writes:

Show 16 quoted lines

> * jt/repack-local-promisor (2024-11-03) 5 commits
>  - fixup! index-pack: repack local links into promisor packs
>  - index-pack: repack local links into promisor packs
>  - t5300: move --window clamp test next to unclamped
>  - t0410: use from-scratch server
>  - t0410: make test description clearer
> 
>  "git gc" discards any objects that are outside promisor packs that
>  are referred to by an object in a promisor pack, and we do not
>  refetch them from the promisor at runtime, resulting an unusable
>  repository.  Work it around by including these objects in the
>  referring promisor pack at the receiving end of the fetch.
> 
>  Needs review.
>  Breaks CI (with a known fix).
>  source: <cover.1730491845.git.jonathantanmy@google.com>

The fixup (in your tree) looks good.

As for review, for what it's worth, Josh Steadmon has written some comments [1], which I've addressed. Regarding the big-picture issue as to whether this solution is the best one for the problem we have, both Han Young [2] and I have verified (with private repos) that this solution does not have noticeable performance issues (which was a concern with other solutions) and I think that this solution does well to preserve the concept that we know which objects are re-fetchable from the remote (even if we have created them locally) and which are not.

[1] https://lore.kernel.org/git/radxsrv6sjemdzl2mw5zzkieyim6xfikrevwggjmzi774g2sob@4nx7fwcjfk32/ [2] https://lore.kernel.org/git/CAG1j3zGiNMbri8rZNaF0w+yP+6OdMz0T8+8_Wgd1R_p1HzVasg@mail.gmail.com/

Show 11 quoted lines

> * jt/commit-graph-missing (2024-11-04) 3 commits
>  - SQUASH???
>  - fetch-pack: warn if in commit graph but not obj db
>  - Revert "fetch-pack: add a deref_without_lazy_fetch_extended()"
> 
>  A regression where commit objects missing from a commit-graph can
>  cause an infinite loop when doing a fetch in a partial clone has
>  been fixed.
> 
>  Waiting an ack for CI breakage fix and possibly a reroll.
>  source: <cover.1730409376.git.jonathantanmy@google.com>

Through reading this I've realized that the "warn if" should be "die if", so I've made a reroll [3].

[3] https://lore.kernel.org/git/cover.1730833754.git.jonathantanmy@google.com/