Op een grimmige herfstdag (Monday 08 November 2004 08:43),schreef H.Merijn
Brand:
> On Mon 08 Nov 2004 02:06, Nicholas Clark <[EMAIL PROTECTED]> wrote:
> > On Mon, Nov 08, 2004 at 01:02:00AM +0000, Steve Hay wrote:
> > > O X -Duseithreads
> > > O X -Duseithreads -Dusemymalloc
> > > O O -Duseithreads -Duselargefiles
> > > O O -Duseithreads -Duselargefiles -Dusemymalloc
> > >
> > > | +--------- -DDEBUGGING
> > >
> > > +----------- no debugging
> > >
> > >
> > > Failures: (common-args) -DINST_TOP=$(INST_DRV)\Smoke\doesntexist
> > > [default] -DDEBUGGING -Duseithreads
> > > [default] -DDEBUGGING -Duseithreads -Dusemymalloc
> > > Inconsistent test results (between TEST and harness):
> > > ../lib/DBM_Filter/t/01error.t...........dubious
> >
> > Is there any to be worried about by these? Failing this test with "X"
> > seems to have been a common problem since (about) 21st October, but it
> > doesn't repeat perfectly every night. 5.9.2 isn't failing in this way, so
> > I'm somewhat confused.
>
> I have noticed that 'X's also come from "skip" failed tests.
> These are more a warning than a failure, if warning at all.
IMO skipped tests shouldn't fail. Either the test is skipped (not executed)
and thus PASS or it is a TODO test. That way everybody knows it is no
surprise that the test fails (or passes).
> IMHO skips are "not ok" OR "ok" tests that are skipped for
> a more or less valid reason.
>
> My recent HP-UX 10.20 skips all changed from "F" to "X",
> because the normal test run sees lines like
>
> not ok 43 - skip blah blah blah
>
> and harness filters them and reports them as skipped, regardless
> if they where ok or not ok
>
> Abe, would it be possible for the primary filter of T::S to skip
> these too? Meaning for /^(not\s+)?ok\s+\d+\s+[-#]\s+(?i:skip\S*[: ])/
> not to see it as "F"?
I'll try, but don't hold your breath here. Too much work at the daytime job
and my head is still with the VMS porting...
(I only want this in with some tests, both in t/skip_filter.t and my private
test-suite)
Good luck,
Abe
--
"Crashes Perl (or Used To)" is not a really useful classifying
criterion, it's about as useful as "the number of characters in
the test is divisible by 73".
-- Jarkko Hietaniemi on p5p @ 2001-10-30
