From: Dmitry Kurochkin <dmitry.kurochkin@gmail.com>
To: Austin Clements <amdragon@mit.edu>
Subject: Re: [PATCH 2/3] test: improve known broken tests support
In-Reply-To:
 <CAH-f9WuKrBpEjBaBet0d95oHXo1kWzGv6oa7zz7Zqe4JK=4N+Q@mail.gmail.com>
References: <1309744743-8556-1-git-send-email-dmitry.kurochkin@gmail.com>
	<1309744743-8556-2-git-send-email-dmitry.kurochkin@gmail.com>
	<CAH-f9WuKrBpEjBaBet0d95oHXo1kWzGv6oa7zz7Zqe4JK=4N+Q@mail.gmail.com>
User-Agent: Notmuch/0.5-321-g41686e2 (http://notmuchmail.org) Emacs/23.3.1
	(x86_64-pc-linux-gnu)
Date: Mon, 04 Jul 2011 07:46:37 +0400
Message-ID: <877h7y4siq.fsf@gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Cc: notmuch@notmuchmail.org
Precedence: list

On Sun, 3 Jul 2011 23:42:13 -0400, Austin Clements <amdragon@mit.edu> wrote:
> Great idea!
>=20
> Just a heads-up, this will conflict with the first two patches in the
> atomicity series (though git may not detect the second conflict).
> Both are trivial patches, though, so the merge should be easy.
>=20

Oh.  I hope atomicity will be there soon.

> Three minor comments below.
>=20
> On Sun, Jul 3, 2011 at 9:59 PM, Dmitry Kurochkin
> <dmitry.kurochkin@gmail.com> wrote:
> > There is existing support for broken tests. =C2=A0But it is not conveni=
ent
> > to use. =C2=A0The primary issue is that we have to maintain a set of
> > test_expect_*_failure functions which are equivalent to the normal
> > test_expect_* counterparts except for what functions are called for
> > result reporting. =C2=A0The patch adds test_subtest_known_broken functi=
on
> > which marks a subset as broken, making the normal test_expect_*
> > functions behave as test_expect_*_failure. =C2=A0All test_expect_*_fail=
ure
> > functions are removed. =C2=A0Test_known_broken_failure_ is changed to
> > format details the same way as test_failure_ does.
> >
> > Another benefit of this change is that the diff when a broken test is
> > fixed would be small and nice.
> >
> > Documentation is updated accordingly.
> > ---
> > =C2=A0test/README =C2=A0 =C2=A0 =C2=A0| =C2=A0 17 ++++++++---------
> > =C2=A0test/test-lib.sh | =C2=A0 53 +++++++++++++++---------------------=
-----------------
> > =C2=A02 files changed, 23 insertions(+), 47 deletions(-)
> >
> > diff --git a/test/README b/test/README
> > index a245bf1..f926b9f 100644
> > --- a/test/README
> > +++ b/test/README
> > @@ -132,20 +132,19 @@ library for your script to use.
> > =C2=A0 =C2=A0<script>. =C2=A0If it yields success, test is considered
> > =C2=A0 =C2=A0successful. =C2=A0<message> should state what it is testin=
g.
> >
> > - test_expect_failure <message> <script>
> > -
> > - =C2=A0 This is NOT the opposite of test_expect_success, but is used
> > - =C2=A0 to mark a test that demonstrates a known breakage. =C2=A0Unlike
> > - =C2=A0 the usual test_expect_success tests, which say "ok" on
> > - =C2=A0 success and "FAIL" on failure, this will say "FIXED" on
> > - =C2=A0 success and "still broken" on failure. =C2=A0Failures from the=
se
> > - =C2=A0 tests won't cause -i (immediate) to stop.
> > -
> > =C2=A0test_begin_subtest <message>
> >
> > =C2=A0 =C2=A0Set the test description message for a subsequent test_exp=
ect_equal
> > =C2=A0 =C2=A0invocation (see below).
> >
> > + test_subtest_known_broken
> > +
> > + =C2=A0 Mark the current test as broken. =C2=A0Such tests are expected=
 to fail.
> > + =C2=A0 Unlike the normal tests, which say "PASS" on success and "FAIL=
" on
> > + =C2=A0 failure, these will say "FIXED" on success and "BROKEN" on fai=
lure.
> > + =C2=A0 Failures from these tests won't cause -i (immediate) to stop. =
=C2=A0This
> > + =C2=A0 must be called before any test_expect_* function.
>=20
> Perhaps "A test must call this before any test_expect_* function"?
> There can be earlier test_expect_* calls from other tests.
>=20

makes sense

> > +
> > =C2=A0test_expect_equal <output> <expected>
> >
> > =C2=A0 =C2=A0This is an often-used convenience function built on top of
> > diff --git a/test/test-lib.sh b/test/test-lib.sh
> > index 22e387e..0cd4170 100755
> > --- a/test/test-lib.sh
> > +++ b/test/test-lib.sh
> > @@ -424,6 +424,7 @@ test_begin_subtest ()
> > =C2=A0 =C2=A0 =C2=A0 =C2=A0error "bug in test script: Missing test_expe=
ct_equal in ${BASH_SOURCE[1]}:${BASH_LINENO[0]}"
> > =C2=A0 =C2=A0 fi
> > =C2=A0 =C2=A0 test_subtest_name=3D"$1"
> > + =C2=A0 =C2=A0test_subtest_known_broken_=3D
> > =C2=A0 =C2=A0 # Remember stdout and stderr file descriptors and redirec=
t test
> > =C2=A0 =C2=A0 # output to the previously prepared file descriptors 3 an=
d 4 (see
> > =C2=A0 =C2=A0 # below)
> > @@ -484,29 +485,6 @@ test_expect_equal_file ()
> > =C2=A0 =C2=A0 fi
> > =C2=A0}
> >
> > -test_expect_equal_failure ()
> > -{
> > - =C2=A0 =C2=A0 =C2=A0 exec 1>&6 2>&7 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=
=A0# Restore stdout and stderr
> > - =C2=A0 =C2=A0 =C2=A0 inside_subtest=3D
> > - =C2=A0 =C2=A0 =C2=A0 test "$#" =3D 3 && { prereq=3D$1; shift; } || pr=
ereq=3D
> > - =C2=A0 =C2=A0 =C2=A0 test "$#" =3D 2 ||
> > - =C2=A0 =C2=A0 =C2=A0 error "bug in the test script: not 2 or 3 parame=
ters to test_expect_equal"
> > -
> > - =C2=A0 =C2=A0 =C2=A0 output=3D"$1"
> > - =C2=A0 =C2=A0 =C2=A0 expected=3D"$2"
> > - =C2=A0 =C2=A0 =C2=A0 if ! test_skip "$@"
> > - =C2=A0 =C2=A0 =C2=A0 then
> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 if [ "$output" =3D "=
$expected" ]; then
> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0=
 =C2=A0 test_known_broken_ok_ "$test_subtest_name"
> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 else
> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0=
 =C2=A0 test_known_broken_failure_ "$test_subtest_name"
> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0=
 =C2=A0 testname=3D$this_test.$test_count
> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0=
 =C2=A0 echo "$expected" > $testname.expected
> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0=
 =C2=A0 echo "$output" > $testname.output
> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 fi
> > - =C2=A0 =C2=A0fi
> > -}
> > -
> > =C2=A0NOTMUCH_NEW ()
> > =C2=A0{
> > =C2=A0 =C2=A0 notmuch new | grep -v -E -e '^Processed [0-9]*( total)? f=
ile|Found [0-9]* total file'
> > @@ -568,12 +546,20 @@ test_have_prereq () {
> > =C2=A0# the text_expect_* functions instead.
> >
> > =C2=A0test_ok_ () {
> > + =C2=A0 =C2=A0 =C2=A0 if [ "$test_subtest_known_broken_" =3D 1 ]; then
> > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 test_known_broken_ok=
_ "$@"
> > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 return
> > + =C2=A0 =C2=A0 =C2=A0 fi
> > =C2=A0 =C2=A0 =C2=A0 =C2=A0test_success=3D$(($test_success + 1))
> > =C2=A0 =C2=A0 =C2=A0 =C2=A0say_color pass "%-6s" "PASS"
> > =C2=A0 =C2=A0 =C2=A0 =C2=A0echo " $@"
> > =C2=A0}
> >
> > =C2=A0test_failure_ () {
> > + =C2=A0 =C2=A0 =C2=A0 if [ "$test_subtest_known_broken_" =3D 1 ]; then
> > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 test_known_broken_fa=
ilure_ "$@"
> > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 return
> > + =C2=A0 =C2=A0 =C2=A0 fi
> > =C2=A0 =C2=A0 =C2=A0 =C2=A0test_failure=3D$(($test_failure + 1))
> > =C2=A0 =C2=A0 =C2=A0 =C2=A0say_color error "%-6s" "FAIL"
> > =C2=A0 =C2=A0 =C2=A0 =C2=A0echo " $1"
> > @@ -592,7 +578,10 @@ test_known_broken_ok_ () {
> > =C2=A0test_known_broken_failure_ () {
> > =C2=A0 =C2=A0 =C2=A0 =C2=A0test_broken=3D$(($test_broken+1))
> > =C2=A0 =C2=A0 =C2=A0 =C2=A0say_color pass "%-6s" "BROKEN"
> > - =C2=A0 =C2=A0 =C2=A0 echo " $@"
> > + =C2=A0 =C2=A0 =C2=A0 echo " $1"
> > + =C2=A0 =C2=A0 =C2=A0 shift
> > + =C2=A0 =C2=A0 =C2=A0 echo "$@" | sed -e 's/^/ =C2=A0 =C2=A0 =C2=A0 =
=C2=A0/'
> > + =C2=A0 =C2=A0 =C2=A0 if test "$verbose" !=3D "t"; then cat test.outpu=
t; fi
>=20
> Would it be possible to keep just the one copy of the failure printing
> code in test_failure_?
>=20

Sure, it is possible if I am not so lazy.

> > =C2=A0}
> >
> > =C2=A0test_debug () {
> > @@ -636,20 +625,8 @@ test_skip () {
> > =C2=A0 =C2=A0 =C2=A0 =C2=A0esac
> > =C2=A0}
> >
> > -test_expect_failure () {
> > - =C2=A0 =C2=A0 =C2=A0 test "$#" =3D 3 && { prereq=3D$1; shift; } || pr=
ereq=3D
> > - =C2=A0 =C2=A0 =C2=A0 test "$#" =3D 2 ||
> > - =C2=A0 =C2=A0 =C2=A0 error "bug in the test script: not 2 or 3 parame=
ters to test-expect-failure"
> > - =C2=A0 =C2=A0 =C2=A0 if ! test_skip "$@"
> > - =C2=A0 =C2=A0 =C2=A0 then
> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 test_run_ "$2"
> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 if [ "$?" =3D 0 -a "=
$eval_ret" =3D 0 ]
> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 then
> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0=
 =C2=A0 test_known_broken_ok_ "$1"
> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 else
> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0=
 =C2=A0 test_known_broken_failure_ "$1"
> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 fi
> > - =C2=A0 =C2=A0 =C2=A0 fi
> > +test_subtest_known_broken () {
> > + =C2=A0 =C2=A0 =C2=A0 test_subtest_known_broken_=3D1
>=20
> The convention in test-lib appears to be "t" for true variables.
>=20

Thanks for the hint.

Will follow with an amend.

Regards,
  Dmitry

> > =C2=A0}
> >
> > =C2=A0test_expect_success () {
> > --
> > 1.7.5.4