harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Vladimir Beliaev" <vladimir.k.beli...@gmail.com>
Subject Re: [general][M3] status for milestone candidate r580985
Date Wed, 03 Oct 2007 09:52:36 GMT
I've double checked EUT3.2/winx86/jdtcoremodel suite (new 1 error and 68
failures).

They are not reproducible locally. So this is not a regression.

Based on failures stack traces it seems to be new intermittent host
configuration issue:

    java.lang.NullPointerException
         at org.eclipse.jdt.internal.compiler.batch.FileSystem.<init>(
FileSystem.java:66)
         at
org.eclipse.jdt.core.tests.compiler.regression.AbstractRegressionTest.getClassLibs
(AbstractRegressionTest.java:208)

This should not stop the release.

Still, Stepan, could you please rerun EUT3.2/winx86 to get a clear results?

Thanks
Vladimir Beliaev

2007/10/3, Mark Hindess <mark.hindess@googlemail.com>:
>
>
> On 3 October 2007 at 15:08, "Stepan Mishura" <stepan.mishura@gmail.com>
> wrote:
> > Hi all,
> >
> > Status for milestone candidate (r580985):
> >
> > * Class library unit tests:
> >     Linux: 1 crash (HARMONY-4873), 1 failure (incorrect test -
> HARMONY-4872)
> > * DRL VM tests
> >     Windows: known intermittent failure (HARMONY-4830)
> > * Eclipse Unit Tests 3.2.
> >     Windows: the exclude list was updated (1 know issue + 3
> > intermittent were added) but there are new 68 failures. Regressions
> > from the previous milestone candidate?
> >     Linux - I have to re-run the suite
> > * Eclipse Unit Tests 3.3.
> >     Linux - as for 3.2
> > * Eclipse Geronimo Application (EGA) scenario x 48h -  still running
> > on both platforms
> > * Functional Test Suite -  still running on both platforms
> >   There fixes for several tests, also an update to the exclude list
> > (HARMONY-4889)
> > * Geronimo Unit Tests
> >    2 regressions on Windows/Linux caused by applying patch for
> HARMONY-617
> > * JDKTools tests:
> >     Windows: 1 failure - reproducible only if the whole suite is run
> > (HARMONY-4868)
> > * Reliability Test Suite:
> >      Linux/Windows: 13 failures - of them are intermittent
> > * Stress Test Suite
> > Results was improved compared to the previous one: Windows - 2
> > failure, Linux - 4 failures!. (There is HARMONY-4881 with a patch to
> > all failures to exclude list.)
> >
> > As I can see we need to:
> > - investigate what happened with Eclipse Unit Tests 3.2 on Windows
> > - wait for result for EGAx48 (would be nice to see it passed on both
> > platforms), EUTs on Linux, functional suite (I don't expect big
> > changes here -  EUTs on Linux will probably  timeout again, functional
> > results should be a bit better),
>
> When are these due/likely to finish?
>
> > - make decision for HARMONY-4873 (classlib test crash on Linux)
>
> This is a regression from M2 so it would be nice to fix it.  (Looks like
> a JIT problem since it passes for me on r580985 with -Xint.)  Is someone
> actively looking at this?
>
> Depending on the complexity of the fix it is likely we'd want to restart
> testing so perhaps we might have to release with this as a known issue.
> Depends how long we are prepared to delay the milestone release.
>
> > - make decision for Geronimo unit tests - quick fix, revert patch for
> > HARMONY-617 or exclude
>
> +1 for exclude
>
> > More issues?
> >
> > Also we should keep in mind HARMONY-4790 (mentioned by Gregory)
>
> I'd be inclined to delay fixing this until M4.
>
> -Mark.

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message