Re: [PATCH 3/4] repack: retain cruft packs in MIDXs after incremental repacks

From: Jeff King <peff@peff.net>
Date: Sep 30, 2026, 20:45 UTC
Message-ID: <20260930204529.GB747209@coredump.intra.peff.net>
In-Reply-To: <1774fed77be11b37ce9eb4b7806f5f14539503fb.1790731662.git.me@ttaylorr.com>

On Tue, Sep 29, 2026 at 08:28:53PM -0500, Taylor Blau wrote:

Show 13 quoted lines

> When the 'repack.midxMustContainCruft' configuration is set to "false",
> writing the first MIDX after such a repack may omit that cruft pack. The
> new pack bypasses the `!names.nr` fallback, and there are no previous
> MIDX packs for `midx_has_unknown_packs()` to check. Selecting the new
> commit for bitmap coverage then fails because its reachable objects are
> not all in the MIDX.
> 
> The omission dates all the way back to 5ee86c273bf (repack: exclude
> cruft pack(s) from the MIDX where possible, 2025-06-23). It relies on
> geometric repacking to copy once-cruft objects with
> '--stdin-packs=follow'. However, an ordinary incremental repack makes no
> such guarantee. Require the MIDX to include cruft packs in that case,
> even when a new pack was written.

OK. So this is a problem with just incremental repacks, but _not_ geometric repacks? And only when those incremental repacks write a midx?

If so, that makes sense to me (and the fix seems reasonable).

BTW, write_midx_incremental() does not check midx_must_contain_cruft. So I think you'd have the same problem with --write-midx=incremental. Adding that to the tests causes them to fail. I thought it might also fail with GIT_TEST_MULTI_PACK_INDEX_WRITE_INCREMENTAL=1, but doesn't seem to.

That's not a new problem, but just a spot where the fix doesn't extend. Not sure how important it is to do now, or if it can wait for future work.

-Peff