MIME-Version: 1.0
In-Reply-To: <87ipfktkgy.fsf@gmail.com>
References: <1337953411-21000-1-git-send-email-pioto@pioto.org>
	<87ipfktkgy.fsf@gmail.com>
From: Mike Kelly <pioto@pioto.org>
Date: Fri, 25 May 2012 16:02:34 -0400
Message-ID:
 <CAFb0NsJ7Bv-Lp03WHvZQh_yaimzaj34VmkVQCjr_u5MDrfqAKw@mail.gmail.com>
Subject: Re: FreeBSD Support Patches
To: Dmitry Kurochkin <dmitry.kurochkin@gmail.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Cc: notmuch@notmuchmail.org
Precedence: list

On Fri, May 25, 2012 at 10:15 AM, Dmitry Kurochkin
<dmitry.kurochkin@gmail.com> wrote:
> Why do we need to explicitly declare Emacs dependency for tests? =C2=A0Th=
ere
> should be no need for it. =C2=A0We have "implicit" dependencies which are
> declared once (see test_declare_external_prereq calls at the end of
> test-lib.sh) and are automatically handled when a test tries to use a
> missing binary. =C2=A0Explicit dependencies are hard to maintain (e.g. yo=
ur
> patch adds explicit emacs dependency for crypto test but misses gpg).
> With rare exceptions we should not use explicit dependencies.

Because not every test actually has those implicit dependencies. For
example, some of the crypto tests depend upon emacs_deliver_message
working correctly for subsequents tests. Those emacs_deliver_message
tests are skipped, but not the ones after it that try to do something
with that injected message.

For the emacs-* test files, there are some tests that act the same
way. However, it is also a minor speed improvement to say that,
obviously, none of the emacs tests are going to work, so just don't
bother.

--=20
Mike Kelly