[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
RE: [wtp-dev] Still waiting on smoke test results for WTP 2.0 andfailing JUnits?

Bug https://bugs.eclipse.org/bugs/show_bug.cgi?id=164216 has been reported for the thread issue.  It includes a procedure to demonstrate in a debug session both the NullPointerException a tester here at SAS encountered as well as a ConcurrentModificationException.  The later doesn't have exactly the same stacktrace, but I believe the same basic cause could be the culprit.  I have attached to the bug some thread dumps from my test runs in case others have trouble duplicating it.
 
Cheers,
Larry


From: wtp-dev-bounces@xxxxxxxxxxx [mailto:wtp-dev-bounces@xxxxxxxxxxx] On Behalf Of David M Williams
Sent: Friday, November 10, 2006 11:32 PM
To: General discussion of project-wide or architectural issues.
Subject: [wtp-dev] Still waiting on smoke test results for WTP 2.0 andfailing JUnits?


Larry, a test case would be great,  but I know these threading issues aren't easy to make test cases for.

Nitin, I'm pretty sure this same test fails about 1 out of every 8 or so runs ... so presumably it too is some threading issue?

Our JUnit's are supposed to give us, and others, a warm fuzzy feeling that all is well and we haven't broken anything.
For these tests that fail occasionally due to threading problems, they do not serve this purpose, and maybe have the negative
side effect of "teaching" us to ignore any failures.

So, I suggest when we have a test that fails occasionally, but can't be reproduced, and is presumably a threading problem,
that we open a 'major' bug against that JUnit, and remove the JUnit from the test bucket until the problem is understood and fixed.
Leaving it there and having it usually pass and occasionally fail doesn't do us much good, other than to learn to ignore failing JUnits!  :(

I want us to get so we never declare a build with failing JUnit's, and when one fails, we know something is badly broken and needs immediate attention.
(I did want to declare this one, until we get this settled).

So .. Nitin, I assume you can handle the JSP one ... Chris or Kathy, can you handle the WS failing one?

BTW, to emphasize it, I'm not saying these tests are not important ... just the opposite ... they are all the more important since they fail occasionally, but until
someone invests in fixing it, it makes no sense to leave it there to fail occasionally. One alternative, if it makes sense for the specific case, is to add more
diagnostics to the code or to the test so the source of the problem can be tracked down when it does occur.






Nitin Dahyabhai/Raleigh/IBM@IBMUS
Sent by: wtp-dev-bounces@xxxxxxxxxxx

11/10/2006 01:34 PM

Please respond to
"General discussion of project-wide or architectural issues." <wtp-dev@xxxxxxxxxxx>

To
wtp-dev@xxxxxxxxxxx
cc
Subject
RE: [wtp-dev] Still waiting on smoke test results for WTP 2.0 and        failing Junits?





The JSP failing test has yet to be reproduced locally.  We're not sure if
it's just a quirk of the one build.

---
Nitin Dahyabhai




"Larry Isaacs" <Larry.Isaacs@xxxxxxx>
Sent by: wtp-dev-bounces@xxxxxxxxxxx
11/10/2006 12:50 PM
Please respond to
"General discussion of project-wide or architectural issues."
<wtp-dev@xxxxxxxxxxx>


To
"General discussion of project-wide or architectural issues."
<wtp-dev@xxxxxxxxxxx>
cc

Subject
RE: [wtp-dev] Still waiting on smoke test results for WTP 2.0   andfailing
Junits?






With respect to the org.eclipse.jst.ws.tests, I have recently found a
thread safety issue with respect to ModuleStructuralModel via its
dependence on org.eclipse.emf.ecore.resource.impl.ResourceSetImpl, which
in turn depends on org.eclipse.emf.common.util.BasicEList.  I believe the
exception that appears in the failure fits one of the scenarios that can
occur with concurrent access to ModuleStructuralModel.  If so, this would
be a hit or miss occurrence.

I started filling a bug report, but got stuck trying to come up with a
reproducible test case to demonstrate the problem that doesn't involve our
SAS plug-ins.  I'll go ahead file report today with explanations and
hopefully add a test case to demonstrate it in a development environment
later.

Cheers,
Larry


From: wtp-dev-bounces@xxxxxxxxxxx [mailto:wtp-dev-bounces@xxxxxxxxxxx] On
Behalf Of John Lanuti
Sent: Friday, November 10, 2006 12:18 PM
To: wtp-dev@xxxxxxxxxxx
Subject: [wtp-dev] Still waiting on smoke test results for WTP 2.0
andfailing Junits?


Hey guys,  (sorry for hitting the general list, but I know a lot of
component leads are out and people may not be following the releng list)

So far, only server, SSE, WSDL, and XSD have reported any smoke test
results for WTP 2.0.  As a reminder, this is something that MUST be done
before noon on Friday's so we can predictably deliver integration drivers.
What's going on with the rest of the teams?

Also, we have 2 failing JUnits from web services and SSE.  Care to
comment?  Should we not declare an integration build this week?

org.eclipse.jst.jsp.core.tests_

org.eclipse.jst.ws.tests_

Thanks,

John Lanuti
IBM Web Tools Platform Technical Lead, IBM Rational
IBM Software Lab - Research Triangle Park, NC
jlanuti@xxxxxxxxxx
t/l 441-7861_______________________________________________
wtp-dev mailing list
wtp-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/wtp-dev


_______________________________________________
wtp-dev mailing list
wtp-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/wtp-dev