Re: [PATCH] t2018: move checkout case-insensitive test from t0050

From: Elijah Newren <newren@gmail.com>
Date: Feb 23, 2026, 17:04 UTC
Message-ID: <CABPp-BENP2tUt1MufKeGxXwgr+iHEDnXjOLsaFk7OPkbOsLAMw@mail.gmail.com>
In-Reply-To: <xmqqjyw3o2w2.fsf@gitster.g>

On Mon, Feb 23, 2026 at 8:12 AM Junio C Hamano <gitster@pobox.com> wrote:

Show 30 quoted lines

>
> "Md Ferdous Alam via GitGitGadget" <gitgitgadget@gmail.com> writes:
>
> Adding CC the authors of 69f272b9 (dir: special case check for the
> possibility that pathspec is NULL, 2019-10-01) and 06d53148 (t[01]*:
> adjust the references to the default branch name "main", 2020-11-18)
> the block of lines removed by this patch are blamed for.
>
> One thing that I noticed that needs checking but I didn't do so
> myself is that the original is prepared not to break after Git 3.0
> by using 'main' in t0050 (which forces the initial branch name to be
> 'main').  Whereever the corresponding new code goes, there needs a
> similar provision to prevent the test from getting broken with the
> default change.
>
> I think specifying the initial branch name explicitly when the test
> creates "repo-case" test repository and use that to go back to that
> branch would be better for the purpose of this single test, rather
> than using the blanket "GIT_TEST_DEFAULT_INITIAL_BRANCH_NAME" for
> the entire script.
>
>
> > From: mdferdousalam <mdferdousalam1989@yahoo.com>
> >
> > The test 'checkout with no pathspec and a case insensitive fs' in
> > t0050 does not really belong there as it tests branch checkout
> > behavior, not filesystem properties.  It also had an unnecessary
> > CASE_INSENSITIVE_FS prereq since the sequence of commands should
> > succeed on any filesystem, and it did not verify the resulting
> > worktree contents.

Actually, the CASE_INSENSTIVE_FS was very much intentional and critical to triggering the original bug before it was fixed, and in fact was overlooked by both Denton (the original reporter) and I, which made us unable to figure out how to reproduce the segfault he had triggered. (The testcase does use both "Gitweb" and "gitweb" in it as paths). Luckily, Gábor came along and noticed the case differences and created a testcase for us, and then I dug further for other surrounding issues with that initial guide. See the threads around https://lore.kernel.org/git/20190925215530.GA9013@generichostname/ and https://lore.kernel.org/git/20190927021746.GL2637@szeder.dev/

> > Move it to t2018-checkout-branch.sh where it belongs, drop the
> > prereq,

I think it belongs where it already is, and the prereq should be kept.

> > and add a check that the expected file is present after
> > the checkout.

Not sure if that's worth changing, but if others feel strongly then it doesn't hurt anything.

> >
> > Signed-off-by: mdferdousalam <mdferdousalam1989@yahoo.com>

From Documentation/SubmittingPatches:

""" Please use a known identity in the `Signed-off-by` trailer, since we cannot accept anonymous contributions. It is common, but not required, to use some form of your real name. We realize that some contributors are not comfortable doing so or prefer to contribute under a pseudonym or preferred name and we can accept your patch either way, as long as the name and email you use are distinctive, identifying, and not misleading.

The goal of this policy is to allow us to have sufficient information to contact you if questions arise about your contribution. """

Since your patch was sent by "Md Ferdous Alam via GitGitGadget" I suspect that your Signoff should have been "Signed-off-by: Md Ferdous Alam <mdferdousalam1989@yahoo.com>" (and the From line updated to match). If I'm wrong about that, it might be helpful for you to include an explanation of the name differences with or before your next patch submission.