From: Junio C Hamano <gitster@pobox.com>
Date: Mon, 14 Dec 2015 21:43:15 GMT
Subject: Re: [PATCH v2 2/3] format-patch: add an option to suppress commit hash
Message-ID: <xmqq6100iu4s.fsf@gitster.mtv.corp.google.com>
In-Reply-To: <1450027638-788102-3-git-send-email-sandals@crustytoothpaste.net>

"brian m. carlson" <sandals@crustytoothpaste.net> writes:

> +test_expect_success 'format-patch --zero-commit' '
> +	git format-patch --zero-commit --stdout v2..v1 >patch2 &&
> +	cnt=$(egrep "^From 0{40} Mon Sep 17 00:00:00 2001" patch2 | wc -l) &&
> +	test $cnt = 3
> +'

This test is not wrong per-se, but it makes the test as a whole
somewhat brittle.  People cannot add new commits in the history
being tested, which would add to the number of patches, without
adjusting this test, even though all this test cares about is that
all the mbox "From " lines record the zeroed commit object name.

    git format-patch --zero-commit --stdout v2..v1 |
    grep "^From " | sort | uniq >actual &&
    echo "From $_z40 Mon Sep 17 00:00:00 2001" >expect &&
    test_cmp expect actual

or something like that instead?