Re: Bug in `git branch --delete main` when on other orphan brancht

From: Jeff King <peff@peff.net>
Date: Nov 1, 2022, 21:45 UTC
Message-ID: <Y2GTgB0oTW3rR1HZ@coredump.intra.peff.net>
In-Reply-To: <Y2F+MA+cAsEB0Glb@nand.local>

On Tue, Nov 01, 2022 at 04:14:40PM -0400, Taylor Blau wrote:

Show 21 quoted lines

> Actually, the test doesn't need "other" here, since we aren't actually
> going to delete the target branch (for the exact same reason that you
> pointed out to Martin earlier in the thread).
> 
> So it could actually be as small as something like this:
> 
> --- >8 ---
> diff --git a/t/t3200-branch.sh b/t/t3200-branch.sh
> index 7f605f865b..464d3f610b 100755
> --- a/t/t3200-branch.sh
> +++ b/t/t3200-branch.sh
> @@ -279,6 +279,13 @@ test_expect_success 'git branch -M and -C fail on detached HEAD' '
>  	test_cmp expect err
>  '
> 
> +test_expect_success 'git branch -d on detached HEAD' '
> +	test_when_finished git checkout main &&
> +	git checkout --orphan orphan &&
> +	test_must_fail git branch -d main 2>err &&
> +	grep "not fully merged" err
> +'

I think there's a more interesting case, which is when "main" has a configured upstream to which it's fully merged. And then the deletion actually succeeds (and the bug is not just giving us a crappy message, but actually doing causing the wrong outcome).

And for that we'd probably not want to use "main", since a successful test will actually delete it. ;)

I'll try later tonight to integrate that test into the patch you've written.

-Peff