[PATCH v3] sequencer: avoid dropping fixup commit that targets self via commit-ish

Commit 68d5d03bc4 (rebase: teach --autosquash to match on sha1 in addition to message, 2010-11-04) taught autosquash to recognize subjects like "fixup! 7a235b" where 7a235b is an OID-prefix. It actually did more than advertised: 7a235b can be an arbitrary commit-ish (as long as it's not trailed by spaces).

Accidental(?) use of this secret feature revealed a bug where we would silently drop a fixup commit. The bug can also be triggered when using an OID-prefix but that's unlikely in practice.

Let the commit with subject "fixup! main" be the tip of the "main" branch. When computing the fixup target for this commit, we find the commit itself. This is wrong because, by definition, a fixup target must be an earlier commit in the todo list. We wrongly find the current commit because we added it to the todo list prematurely. Avoid these fixup-cycles by only adding the current commit to the todo list after we have finished looking for the fixup target.

Reported-by: Erik Cervin Edin <erik@cervined.in>
Signed-off-by: Johannes Altmanninger <aclopte@gmail.com>
---
 sequencer.c                  |  4 ++--
 t/t3415-rebase-autosquash.sh | 13 +++++++++++++
 2 files changed, 15 insertions(+), 2 deletions(-)

Changes since v2:
- minor commit message rewording and clarification
- dropped a test that was based on a wrong understanding of --autosquash.

  I saw b425461cb5 (SQUASH??? resurrect previous version of the tests,
  2022-09-21) today. I'm leaning towards not adding them back because
  those tests were only added to reflect the current behavior.  but that
  behavior is not something I care about.
  We might even want to change this behavior, by restricting OID fixup
  targets to SHAs only. So I don't think these tests are desirable,
  unless they help others understand the current state of affairs?

1:  37e00c51bd ! 1:  a25c886a78 sequencer: avoid dropping fixup commit that targets self via commit-ish
    @@ Commit message
         would silently drop a fixup commit. The bug can also be triggered
         when using an OID-prefix but that's unlikely in practice.
     
    -    Given a commit with subject "fixup! main" that is the tip of the
    -    branch "main". When computing the fixup target for this commit, we
    -    find the commit itself. This is wrong because, by definition, a fixup
    +    Let the commit with subject "fixup! main" be the tip of the "main"
    +    branch. When computing the fixup target for this commit, we find
    +    the commit itself. This is wrong because, by definition, a fixup
         target must be an earlier commit in the todo list. We wrongly find
         the current commit because we added it to the todo list prematurely.
    -    Avoid these fixup-cycles by only adding the current commit after we
    -    have finished finding its target.
    +    Avoid these fixup-cycles by only adding the current commit to the
    +    todo list after we have finished looking for the fixup target.
     
         Reported-by: Erik Cervin Edin <erik@cervined.in>
     
      ## sequencer.c ##
     @@ sequencer.c: int todo_list_rearrange_squash(struct todo_list *todo_list)
    @@ t/t3415-rebase-autosquash.sh: test_expect_success 'auto squash that matches long
      	test_line_count = 1 actual
      '
      
    -+test_expect_success 'auto squash that matches regex' '
    -+	git reset --hard base &&
    -+	git commit --allow-empty -m "hay needle hay" &&
    -+	git commit --allow-empty -m "fixup! :/needle" &&
    -+	GIT_SEQUENCE_EDITOR="cat >tmp" git rebase --autosquash -i HEAD^^ &&
    -+	sed -ne "/^[^#]/{s/[0-9a-f]\{7,\}/HASH/g;p;}" tmp >actual &&
    -+	cat <<-EOF >expect &&
    -+	pick HASH hay needle hay # empty
    -+	fixup HASH fixup! :/needle # empty
    -+	EOF
    -+	test_cmp expect actual
    -+'
    -+
     +test_expect_success 'auto squash of fixup commit that matches branch name which points back to fixup commit' '
     +	git reset --hard base &&
     +	git commit --allow-empty -m "fixup! self-cycle" &&