On Fri, Nov 22, 2019 at 03:10:20PM -0800, Brian Norris wrote:
> Hi! I'm using git 2.24 (or, a variant of that packaged my distro -- I
> can try to build my own if this is deemed not reproducible), and I
> feel like I've been seeing a regression here:
>
> Previously, when reverting multiple commits (with the default --edit
> behavior), as soon as I'm done editing the first revert commit
> message, the second revert commit pops up an editor, and I'm on my
> way. Now, it seems to require an extra 'git revert --continue' prompt
> in between, yet it doesn't actually recommend that. This is highly
> confusing, and seemingly unnecessary.
Thanks for the report, this is indeed a regression in v2.24.0: it
bisects down to a47ba3c777 (rebase -i: check for updated todo after
squash and reword, 2019-08-19) [Cc'ing Phillip]. It's not specific to
'git revert', but with a slight variation affects 'git cherry-pick' as
well.
> An annotated transcript provided below.
I transcribed your transcript into tests that can be run in our test
framework and demonstrate this regression:
--- >8 ---
#!/bin/sh
test_description='test'
. ./test-lib.sh
test_expect_success "Brian's revert regression" '
test_create_repo revert &&
(
cd revert &&
echo 1 >file &&
git add file &&
git commit -m first &&
echo 2 >file &&
git commit -am second &&
echo 3 >file &&
git commit -am third &&
git checkout -b branch &&
git revert --edit HEAD HEAD^ &&
echo 1 >expect &&
test_cmp expect file
)
'
test_expect_success "a variant of Brian's regression for cherry-pick" '
test_create_repo cherry-pick &&
(
cd cherry-pick &&
echo 1 >file &&
git add file &&
git commit -m first &&
echo 2 >file &&
git commit -am second &&
echo 3 >file &&
git commit -am third &&
git checkout -b branch HEAD^^ &&
git cherry-pick --edit master^ master &&
echo 3 >expect &&
test_cmp expect file
)
'
test_done
--- >8 ---
They both succeed on a47ba3c777's parent, but fail on a47ba3c777 when
the 'git revert' or 'git cherry-pick' commands return with exit code 1
after reverting/cherry-picking the first commit instead of processing
the second commit:
+ git revert --edit HEAD HEAD^
[branch 88ea48c] Revert "third"
Author: A U Thor <author@xxxxxxxxxxx>
1 file changed, 1 insertion(+), 1 deletion(-)
On branch branch
Revert currently in progress.
nothing to commit, working tree clean
error: last command exited with $?=1
not ok 1 - Brian's revert regression
+ git cherry-pick --edit master^ master
[branch 2cb3f74] second
Author: A U Thor <author@xxxxxxxxxxx>
Date: Sat Nov 23 00:17:32 2019 +0000
1 file changed, 1 insertion(+), 1 deletion(-)
On branch branch
Cherry-pick currently in progress.
nothing to commit, working tree clean
The previous cherry-pick is now empty, possibly due to conflict
resolution.
If you wish to commit it anyway, use:
git commit --allow-empty
If you wish to skip this commit, use:
git reset
Then "git cherry-pick --continue" will resume cherry-picking
the remaining commits.
error: last command exited with $?=1
not ok 2 - a variant of Brian's regression for cherry-pick
These test should probably be squeezed into
't3508-cherry-pick-many-commits.sh', but Friday has just turned into
Saturday around here, so that's enough from me for now.
> Note that none of this happens with --no-edit; my reverts happen
> smoothly, with no extra need for --continue.
>
> Regards,
> Brian
>
> $ mkdir tmp
> $ cd tmp
> /tmp$ git init
> Initialized empty Git repository in [...]/tmp/.git/
> /tmp$ touch foo
> /tmp$ git add foo
> /tmp$ echo bar >> foo
> /tmp$ git commit -am baz
> [master (root-commit) a388f78d9013] baz
> 1 file changed, 1 insertion(+)
> create mode 100644 foo
> /tmp$ echo pow >> foo
> /tmp$ git commit -am whizzbang
> [master 51753222dd9a] whizzbang
> 1 file changed, 1 insertion(+)
> /tmp$ echo pop >> foo
> /tmp$ git commit -am nothing
> [master 7153607b11e0] nothing
> 1 file changed, 1 insertion(+)
> /tmp$ git revert HEAD HEAD^
> ## EDITOR pops up, as expected
> [master 586469974ec2] Revert "nothing"
> 1 file changed, 1 deletion(-)
> On branch master
> Revert currently in progress.
>
> nothing to commit, working tree clean
> ## Unexpected, confusing pause? No prompt to '--continue'
> /tmp$ git log --oneline
> 586469974ec2 (HEAD -> master) Revert "nothing"
> 7153607b11e0 nothing
> 51753222dd9a whizzbang
> a388f78d9013 baz
> /tmp$ git status
> On branch master
> Revert currently in progress.
> (run "git revert --continue" to continue)
> (use "git revert --skip" to skip this patch)
> (use "git revert --abort" to cancel the revert operation)
>
> nothing to commit, working tree clean
> /tmp$ git revert --continue
> ## EDITOR pops up, as expected
> [master b8dd23f61d07] Revert "whizzbang"
> 1 file changed, 1 deletion(-)
> On branch master
> Revert currently in progress.
>
> nothing to commit, working tree clean
> ## Unexpected state; both reverts happened, but revert is still in progress?
> /tmp$ git log --oneline
> b8dd23f61d07 (HEAD -> master) Revert "whizzbang"
> 586469974ec2 Revert "nothing"
> 7153607b11e0 nothing
> 51753222dd9a whizzbang
> a388f78d9013 baz
> /tmp$ git status
> On branch master
> Revert currently in progress.
> (run "git revert --continue" to continue)
> (use "git revert --skip" to skip this patch)
> (use "git revert --abort" to cancel the revert operation)
>
> nothing to commit, working tree clean
> ## OK...I'll run it one more time.
> /tmp$ git revert --continue
> /tmp$ git status
> On branch master
> nothing to commit, working tree clean
> ## *Now* I'm done
> /tmp$ git log --oneline
> b8dd23f61d07 (HEAD -> master) Revert "whizzbang"
> 586469974ec2 Revert "nothing"
> 7153607b11e0 nothing
> 51753222dd9a whizzbang
> a388f78d9013 baz
