Re: [PATCH 1/2] kunit: tool: Do not error on tests without test plans

[Brendan Higgins]

On Fri, Oct 22, 2021 at 5:25 PM David Gow <davidgow@xxxxxxxxxx> wrote:
>
> On Sat, Oct 23, 2021 at 6:42 AM Daniel Latypov <dlatypov@xxxxxxxxxx> wrote:
> >
> > On Thu, Oct 21, 2021 at 11:10 PM David Gow <davidgow@xxxxxxxxxx> wrote:
> > >
> > > On Fri, Oct 22, 2021 at 9:29 AM Daniel Latypov <dlatypov@xxxxxxxxxx> wrote:
> > > >
> > > > On Wed, Oct 20, 2021 at 11:28 PM David Gow <davidgow@xxxxxxxxxx> wrote:
> > > > >
> > > > > The (K)TAP spec encourages test output to begin with a 'test plan': a
> > > > > count of the number of tests being run of the form:
> > > > > 1..n
> > > > >
> > > > > However, some test suites might not know the number of subtests in
> > > > > advance (for example, KUnit's parameterised tests use a generator
> > > > > function). In this case, it's not possible to print the test plan in
> > > > > advance.
> > > > >
> > > > > kunit_tool already parses test output which doesn't contain a plan, but
> > > > > reports an error. Since we want to use nested subtests with KUnit
> > > > > paramterised tests, remove this error.
> > > > >
> > > > > Signed-off-by: David Gow <davidgow@xxxxxxxxxx>
> > > > > ---
> > > > >  tools/testing/kunit/kunit_parser.py    | 5 ++---
> > > > >  tools/testing/kunit/kunit_tool_test.py | 5 ++++-
> > > > >  2 files changed, 6 insertions(+), 4 deletions(-)
> > > > >
> > > > > diff --git a/tools/testing/kunit/kunit_parser.py b/tools/testing/kunit/kunit_parser.py
> > > > > index 3355196d0515..50ded55c168c 100644
> > > > > --- a/tools/testing/kunit/kunit_parser.py
> > > > > +++ b/tools/testing/kunit/kunit_parser.py
> > > > > @@ -340,8 +340,8 @@ def parse_test_plan(lines: LineStream, test: Test) -> bool:
> > > > >         """
> > > > >         Parses test plan line and stores the expected number of subtests in
> > > > >         test object. Reports an error if expected count is 0.
> > > > > -       Returns False and reports missing test plan error if fails to parse
> > > > > -       test plan.
> > > > > +       Returns False and sets expected_count to None if there is no valid test
> > > > > +       plan.
> > > > >
> > > > >         Accepted format:
> > > > >         - '1..[number of subtests]'
> > > > > @@ -356,7 +356,6 @@ def parse_test_plan(lines: LineStream, test: Test) -> bool:
> > > > >         match = TEST_PLAN.match(lines.peek())
> > > > >         if not match:
> > > > >                 test.expected_count = None
> > > > > -               test.add_error('missing plan line!')
> > > >
> > > > This works well, but there's an edge case.
> > > >
> > > > This patch means we no longer print an error when there are no test
> > > > cases in a subtest.
> > > > We relied on a check just a bit lower in this function.
> > > >
> > > > Consider
> > > >
> > > > $ ./tools/testing/kunit/kunit.py parse <<EOF
> > > > TAP version 14
> > > > 1..1
> > > >   # Subtest: suite
> > > >   1..1
> > > >     # Subtest: case
> > > >   ok 1 - case
> > > > ok 1 - suite
> > > > EOF
> > > >
> > > > This produces the following output (timestamps removed)
> > > >
> > > > ============================================================
> > > > ==================== suite (1 subtest) =====================
> > > > =========================== case ===========================
> > > > ====================== [PASSED] case =======================
> > > > ====================== [PASSED] suite ======================
> > > > ============================================================
> > > >
> > > > Should we surface some sort of error here?
> > >
> > > I thought about this a bit (and started prototyping it), and think the
> > > answer is probably "no" (or, perhaps, "optionally"). Largely because I
> > > think it'd be technically valid to have, e.g., a parameterised test
> > > whose generator function can legitimately provide zero subtests. And
> >
> > That's the question. Should we report PASSED in that case as we do now?
> >
> > Let's consider parameterised tests, our only current example in KUnit.
> >
> > I think in most cases, it's a bug that if we got 0 cases and we should
> > let the user know somehow.
> > Should it be an error/warning? Maybe not, but wouldn't it be better to
> > report SKIPPED?
> > (This would require a change in KUnit on the kernel side, I'm not
> > suggesting we do this in the parser)
> >
>
> Yeah: there are two sorf-of separable decisions here:
> 1) What result should a test with no subtests return?
> 2) Do we want to trigger any other errors/warnings.
>
> I think the answer to 1 is that kunit_tool should report the result
> printed in the KTAP output. I agree that, for parameterised tests,
> though, that SKIPPED makes more sense than PASSED. (kunit_tool has a
> separate NO_TESTS result, which we could maybe try to generate and
> handle explicitly. I think we might as well leave that for the "no
> tests run at all" case for now.)
>
> For 2, I feel that this definitely should count as a "warning", but
> all we have at the moment are "errors", which I feel is probably a bit
> too strong a term for this. Given errors don't actually halt parsing,
> I'm okay with generating them in kunit_tool in this case, but I'd
> probably slightly prefer to leave it with SKIPPED, and maybe add a
> warning later.

I am OK marking it as SKIPPED, but I like the idea of promoting it to
a warning in a future change.

Completely ignoring an empty test suite seems wrong, especially when
we *do* complain when there *is* a test plan, and not all test cases
are accounted for.

My 2c.

> > > while that's probably worth warning about if it's the only test
> > > running, if you're trying to run all tests, and one random subtest of
> > > a test of a suite has no subtests, that seems like it'd be more
> > > annoying to error on than anything else.
> > >
> > > That being said, I'm not opposed to implementing it as an option, or
> > > at least having the test status set to NO_ERROR. The implementation
> > > I've experimented with basically moves the check to "parse_test", and
> > > errors if the number of subtests is 0 after parsing, if parent_test is
> > > true (or main, but my rough plan was to make main imply parent_test,
> > > and adjust the various conditions to match). I haven't looked into
> > > exactly how this is bubbled up yet, but I'd be okay with having an
> > > error if there are no tests run at all.
> > >
> > > I'll keep playing with this anyway: it's definitely a bit more of a
> > > minefield than I'd originally thought. :-)
> > >
> > > -- David
> > >
> > > >
> > > >
> > > > >                 return False
> > > > >         test.log.append(lines.pop())
> > > > >         expected_count = int(match.group(1))
> > > > > diff --git a/tools/testing/kunit/kunit_tool_test.py b/tools/testing/kunit/kunit_tool_test.py
> > > > > index 9c4126731457..bc8793145713 100755
> > > > > --- a/tools/testing/kunit/kunit_tool_test.py
> > > > > +++ b/tools/testing/kunit/kunit_tool_test.py
> > > > > @@ -191,7 +191,10 @@ class KUnitParserTest(unittest.TestCase):
> > > > >                         result = kunit_parser.parse_run_tests(
> > > > >                                 kunit_parser.extract_tap_lines(
> > > > >                                 file.readlines()))
> > > > > -               self.assertEqual(2, result.test.counts.errors)
> > > > > +               # A missing test plan is not an error.
> > > > > +               self.assertEqual(0, result.test.counts.errors)
> > > > > +               # All tests should be accounted for.
> > > > > +               self.assertEqual(10, result.test.counts.total())
> > > > >                 self.assertEqual(
> > > > >                         kunit_parser.TestStatus.SUCCESS,
> > > > >                         result.status)
> > > > > --
> > > > > 2.33.0.1079.g6e70778dc9-goog
> > > > >