Patrick Steinhardt [off-list ref] writes:

Hi,

this patch series fixes some small issues I've discovered while working
on some other patch series. I've decided to split it out of these
because I'm hitting the same issues in multiple series, and I don't want
those to become dependent on one another.

The patch series is built on top of f0ef5b6d9b with
ps/object-source-management at ac65c70663 (odb: handle recreation of
quarantine directories, 2025-11-19) merged into it.

Changes in v3:
  - Use `test_commit ()` so that we the same object at multiple tips.
  - Slightly reword the commit message.
  - Link to v2: https://lore.kernel.org/r/20251211-odb-related-fixes-v2-0-bdf875ce51fc@pks.im (local)

Changes in v2:
  - Drop the first commit that regards geometric repacking with promisor
    remotes. As it turns out my assertion was wrong: geometric repacks
    do and have to consider promisors, but they will fail to handle
    them. This is a bigger topic to fix though, so I'll rather want to
    move this into a separate patch series.
  - Tighten tests a bit for the commit-graph generation.
  - Stop referring to a "subsequent" commit that doesn't exist.
  - Link to v1: https://lore.kernel.org/r/20251205-odb-related-fixes-v1-0-ef4250abb584@pks.im (local)

Thanks!

Patrick

---
Patrick Steinhardt (2):
      builtin/gc: fix condition for whether to write commit graphs
      odb: properly close sources before freeing them

 builtin/gc.c           |  8 +++++---
 odb.c                  |  2 +-
 t/t7900-maintenance.sh | 25 +++++++++++++++++++++++++
 3 files changed, 31 insertions(+), 4 deletions(-)

Range-diff versus v2:

1:  564b26fa6b ! 1:  3ef6ea3560 builtin/gc: fix condition for whether to write commit graphs
    @@ Commit message
         establish the proper border at which the auto-condition would be
         fulfilled. As it turns out, there's another bug: if an object is at the
         tip of any reference we don't mark it as seen. Consequently, if it is
    -    reachable via any other reference, we'd count that object twice.
    +    the tip of or reachable via another ref, we'd count that object multiple
    +    times.

         Fix both of these bugs so that we properly count objects without leaking
         any memory.
    @@ t/t7900-maintenance.sh: test_expect_success 'commit-graph auto condition' '
     +	(
     +		cd repo &&
     +		git config set maintenance.auto false &&
    -+		git commit --allow-empty -m initial &&
    ++		test_commit initial &&
     +		git switch --create feature &&
    -+		git commit --allow-empty -m feature-1 &&
    -+		git commit --allow-empty -m feature-2 &&
    ++		test_commit feature-1 &&
    ++		test_commit feature-2 &&
     +		git switch - &&
    -+		git commit --allow-empty -m main-1 &&
    -+		git commit --allow-empty -m main-2 &&
    ++		test_commit main-1 &&
    ++		test_commit main-2 &&
     +		git merge feature &&
     +
    -+		# We have 6 commit, none of which are covered by a commit
    ++		# We have 6 commits, none of which are covered by a commit
     +		# graph. So this must be the boundary at which we start to
     +		# perform maintenance.
     +		test_must_fail git -c maintenance.commit-graph.auto=7 \
2:  20bb4741eb = 2:  55cad3ea0f odb: properly close sources before freeing them

---
base-commit: 2797238193944b52d12624a04a962f40b9bcad69
change-id: 20251205-odb-related-fixes-5f48a0993ef7

The changes in this version looks good to me! :)