From: Jameson Graef Rollins <jrollins@finestructure.net>
To: Mike Kelly <pioto@pioto.org>, notmuch@notmuchmail.org
Subject: Re: [PATCH v2 7/9] test/count: fix numeric comparision tests on
	FreeBSD
In-Reply-To: <1338361324-57289-8-git-send-email-pioto@pioto.org>
References: <1338361324-57289-1-git-send-email-pioto@pioto.org>
	<1338361324-57289-8-git-send-email-pioto@pioto.org>
User-Agent: Notmuch/0.13+24~g16915cd (http://notmuchmail.org) Emacs/23.4.1
	(x86_64-pc-linux-gnu)
Date: Wed, 30 May 2012 13:28:37 -0700
Message-ID: <877gvtxvka.fsf@servo.finestructure.net>
MIME-Version: 1.0
Content-Type: multipart/signed; boundary="=-=-=";
	micalg=pgp-sha256; protocol="application/pgp-signature"
Precedence: list

--=-=-=

On Wed, May 30 2012, Mike Kelly <pioto@pioto.org> wrote:
> FreeBSD's `wc -l` includes some white space in front of the number.
> Instead, we add a test_expect_equal_num() to test-lib.sh, which ensures
> we do a proper numeric comparision, and in a portable way.

Is there a way we can avoid adding another new test function here?  This
new function is almost completely identical with the old one, except
that it uses "test .. -eq" instead of "[ .. =".  If the problem is just
that the arguments occasionally contain superfluous spaces, I would
rather see the existing function just strip the spaces.  Of maybe add an
option to the test to ask for arithmetic rather than string comparison.
That would be preferable to introducing an new function that is
confusingly similar to an existing one.

> +# Like test_expect_equal, but does a numeric comparision instead of a
> +# string comparision
> +test_expect_equal_num () {
> +	exec 1>&6 2>&7		# Restore stdout and stderr
> +	inside_subtest=
> +	test "$#" = 3 && { prereq=$1; shift; } || prereq=
> +	test "$#" = 2 ||
> +	error "bug in the test script: not 2 or 3 parameters to test_expect_equal"
> +
> +	output="$1"
> +	expected="$2"
> +	if ! test_skip "$test_subtest_name"
> +	then
> +		if test "$output" -eq "$expected"
> +		then
> +			test_ok_ "$test_subtest_name"
> +		else
> +			test_failure_ "$test_subtest_name" "$output != $expected"
> +		fi
> +	fi
> +}

Also, the new function is missing the diff output, which is important
for debugging, and the syntax is different from the standard we've been
using:

  if ...; then

We try to maintain syntax consistency across the code base.  Thanks.

jamie.

--=-=-=
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)
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=dKfi
-----END PGP SIGNATURE-----
--=-=-=--