Re: [PATCHv2 4/4] t7500: add tests of commit --squash

From: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Date: Sep 22, 2010, 18:12 UTC
Message-ID: <AANLkTi=MjE3OBqAd8NqFCMdSb3qfNW_1QKvrL6-AK+Er@mail.gmail.com>
In-Reply-To: <AANLkTinajr6DvYeyiK79xESYqg0kegUN1s8LxEqGoUc+@mail.gmail.com>

On Wed, Sep 22, 2010 at 17:59, Pat Notz <patnotz@gmail.com> wrote:

Show 33 quoted lines

> On Tue, Sep 21, 2010 at 2:36 PM, Ævar Arnfjörð Bjarmason
> <avarab@gmail.com> wrote:
>>
>> On Tue, Sep 21, 2010 at 20:25, Pat Notz <patnotz@gmail.com> wrote:
>>
>> > +cat >editor <<\EOF
>> > +#!/bin/sh
>> > +sed -e "s/intermediate/edited/g" <"$1" >"$1-"
>> > +mv "$1-" "$1"
>> > +EOF
>> > +chmod 755 editor
>> > +
>> > +test_expect_success 'commit --squash works with -c' '
>> > +       commit_for_rebase_autosquash_setup &&
>> > +       EDITOR=./editor git commit --squash HEAD~1 -c HEAD &&
>> > +       commit_msg_is "squash! target message subject lineedited commit"
>> > +'
>>
>> Why not put the editor in t/t7500/ and use test_set_editor() like the
>> other tests?
>
> The real reason is that I'm new enough that I wasn't aware of this
> pattern.  I saw what was done in t7501-commit.sh and followed along.
> I missed the use of test_set_editor() right there in t7500-commit.sh.
> Doh!
>
> I can certainly do that if it's preferred.  I must say, though, that I
> find it odd to put test inputs in a separate file in a separate
> directory from where the test transforms those into expected outputs.
> To see what the test is doing you have to load both files and trace
> through it.
>
> Still, I'd be happy to change do this if that's the preferred way.

It's a bit odd, but it's best to following existing style within a test. Then maybe submit fixup patches to fix the whole thing later.