Re: [PATCH] Fix revert --abort on Windows

[Jonathan Nieder <jrnieder@gmail.com>]

Johannes Sixt wrote:

> From: Johannes Sixt <j6t@kdbg.org>
>
> On Windows, it is not possible to rename or remove a directory that has
> open files. 'revert --abort' renamed .git/sequencer when it still had
> .git/sequencer/head open. Close the file as early as possible to allow
> the rename operation on Windows.

Nice catch.

Acked-by: Jonathan Nieder <jrnieder@gmail.com>

Speaking of which, it doesn't make a lot of sense for "git revert
--abort" to be leaving a .git/sequencer-old directory around.  How
about this on top?

-- >8 --
Subject: revert --abort: do not leave behind useless sequencer-old directory

The "git cherry-pick --abort" command currently renames the
.git/sequencer directory to .git/sequencer-old instead of removing it
on success due to an accident.  cherry-pick --abort is designed to
work in three steps:

 1) find which commit to roll back to
 2) call "git reset --merge <commit>" to move to that commit
 3) remove the .git/sequencer directory

But the careless author forgot step 3 entirely.  The only reason the
command worked anyway is that "git reset --merge <commit>" renames the
.git/sequencer directory as a secondary effect --- after moving to
<commit>, or so the logic goes, it is unlikely but possible that the
caller of git reset wants to continue the series of cherry-picks that
was in progress, so git renames the sequencer state to
.git/sequencer-old to be helpful while allowing the cherry-pick to be
resumed if the caller did not want to end the sequence after all.

By running "git cherry-pick --abort", the operator has clearly
indicated that she is not planning to continue cherry-picking.  Remove
the (renamed) .git/sequencer directory as intended all along.

Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
---
By the way, as the length of the second-to-last paragraph above might
have hinted, I am not convinced that allowing "git reset --hard" as an
escape route from a cherry-pick sequence was very sensible.  It
_would_ be nice to have a command to return to a known state,
discarding progress in all pending multiple-command guided workflows
(am, rebase, bisect), but git reset is not that command.

 builtin/revert.c          |    1 +
 t/t7106-reset-sequence.sh |    8 ++++++++
 2 files changed, 9 insertions(+), 0 deletions(-)

diff --git a/builtin/revert.c b/builtin/revert.c
index 5dedb51c..818b4abb 100644
--- a/builtin/revert.c
+++ b/builtin/revert.c
@@ -942,6 +942,7 @@ static int sequencer_rollback(struct replay_opts *opts)
 	}
 	if (reset_for_rollback(sha1))
 		goto fail;
+	remove_sequencer_state(1);
 	strbuf_release(&buf);
 	return 0;
 fail:
diff --git a/t/t7106-reset-sequence.sh b/t/t7106-reset-sequence.sh
index 3f86e8c5..83f7ea59 100755
--- a/t/t7106-reset-sequence.sh
+++ b/t/t7106-reset-sequence.sh
@@ -41,4 +41,12 @@ test_expect_success 'reset --hard cleans up sequencer state, providing one-level
 	test_path_is_missing .git/sequencer-old
 '
 
+test_expect_success 'cherry-pick --abort does not leave sequencer-old dir' '
+	pristine_detach initial &&
+	test_must_fail git cherry-pick base..anotherpick &&
+	git cherry-pick --abort &&
+	test_path_is_missing .git/sequencer &&
+	test_path_is_missing .git/sequencer-old
+'
+
 test_done
-- 
1.7.8.rc3