Op een zonnige winterdag (Thursday 06 January 2005 17:11),schreef Tels:
> Moin,
>
> On Thursday 06 January 2005 15:57, Steve Peters wrote:
> > On Thu, Jan 06, 2005 at 07:30:00AM -0600,
>
> [EMAIL PROTECTED] wrote:
> > > Automated smoke report for 5.9.2 patch 23756
> > > mccoy.peters.homeunix.org: Intel Pentium III ("GenuineIntel"
> > > 686-class, 512KB L2 cache) (i386/1 cpu) on openbsd - 3.6
> > > using cc version 2.95.3 20010125 (prerelease, propolice)
> > > smoketime 7 hours 24 minutes (average 55 minutes 35 seconds)
> > > Failures:
> > > [stdio] -DDEBUGGING
> > > Inconsistent test results (between TEST and harness):
> > > ../lib/Math/BigInt/t/mbi_rand.t.........FAILED at test 144
> > >
> > > --
> > > Report by Test::Smoke v1.19#716 running on perl 5.8.5
> > > (Reporter v0.016 / Smoker v0.015)
> >
> > Since the integration of Math::BigInt v1.74, Math::BigRat v0.14, bignum
> > v0.16, there appears to be random (no pun intended) test failures in
> > ../lib/Math/BigInt/t/mbi_rand.t. Test failures have appeared on my
> > OpenBSD smokes. Looking through the other build reports, they have
> > also appeared on OpenVMS, HP-UX, AIX, and other systems as well.
Perhaps the TinySmoke db can help your search in the future (sorry for the
wrapped URL):
http://www.test-smoke.org/cgi/
tsdb?frtext=1&rtext=mbi_rand.t&pversion=5.9.2&lmode=List+reports
> > To
> > add to this problem, all the failures are occuring under differing
> > configurations at different test cases.
>
> Ah crap :-/
>
> Can you please run the test manually, in a loop and send me the output of
> one of the failing ones? It should include diagnostic output (and the
> random seed to reproduce the failure).
From the smoke log of my linuxbox:
# Test 482 got: "3" (../lib/Math/BigInt/t/mbi_rand.t at line 67 fail #121)
# Expected: "2"
# ../lib/Math/BigInt/t/mbi_rand.t line 67 is: print "# seed: $seed, \$ADB *
\$B / \$B = ", $ADB * $B / $B, " != $ADB (\$B=$B)\n"
lib/Math/BigInt/t/mbi_rand................FAILED at test 482
config:
http://www.test-smoke.org/cgi/tsdb?mode=report&rid=24349&top=24349
>
> I bet the div() shortcut in Calc is the culprit. The random tests are
> specifically designed to catch such things (glad they did). I have to
> revise that shortcut if it is not working on some systems.
>
> Best wishes,
>
> Tels
Good luck,
Abe
--
Either I need more coffee... or the "self-resume" code is somewhat
unlikely to be run, unless our script is called baron_munchausen.pl.
-- Jarkko Hietaniemi on p5p @ 2002-01-21
