Re: [PATCH v2 04/11] t0002: abstract away SHA-1 specific constants

From: Eric Sunshine <sunshine@sunshineco.com>
Date: Aug 19, 2018, 20:05 UTC
Message-ID: <CAPig+cQz4oM-csmniSe5LPkHixUOh04i+CLWbJB4dn-NeYS3jg@mail.gmail.com>
In-Reply-To: <20180819175351.449973-5-sandals@crustytoothpaste.net>

On Sun, Aug 19, 2018 at 1:54 PM brian m. carlson <sandals@crustytoothpaste.net> wrote:

> Adjust the test so that it computes variables for object IDs instead of
> using hard-coded hashes.

Nit: s/hashes/hash values/

Show 14 quoted lines

> Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
> ---
> diff --git a/t/t0002-gitfile.sh b/t/t0002-gitfile.sh
> @@ -92,11 +92,12 @@ test_expect_success 'enter_repo non-strict mode' '
>         ) &&
> +       head=$(git -C enter_repo rev-parse HEAD) &&
>         [...]
> +       cat >expected <<-EOF &&
> +       $head   HEAD
> +       $head   refs/heads/master
> +       $head   refs/tags/foo
>         EOF
>         test_cmp expected actual
>  '

Okay, but...

Show 9 quoted lines

> @@ -107,20 +108,20 @@ test_expect_success 'enter_repo linked checkout' '
> -       [...]
> +       cat >expected <<-EOF &&
> +       $head   HEAD
> +       $head   refs/heads/master
> +       $head   refs/tags/foo
>         EOF
>         test_cmp expected actual
>  '

This is relying upon 'head' set inside an earlier test, which seems fragile. More robust would be to compute 'head' anew within each test which needs it (including the other new test added by this patch).