Re: [eclipselink-dev] [Fwd: [Bug 238696] Fix use of Double Checked Locki

[Mitesh Meswani <Mitesh.Meswani@xxxxxxx>]

Tom Ware wrote:
> Hi Mitesh,
>
>   I believe James has gotten back to you about the fix.
He has reviewed the fix and replied to me on a private thread. I will 
take that as approval.

>   I am wondering if your test failures are JDK 1.6-related.  See the 
> following bug:
>
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=236185
>
>   There is a set of test results listed near the bottom that has a 
> similar number of failures for JDK 1.6 compile and run scenarios.
I am compiling and running with JDK 5. So, this does not seem related. 
Almost all the tests fail with "Missing Descriptor" error as follows.

/
Error Exception Description: Fatal query exception occurred. Internal 
Exception: Exception [EclipseLink-6007] (Eclipse Persistence Services - 
1.0 (Build SNAPSHOT - 20080723)): 
org.eclipse.persistence.exceptions.QueryException Exception Description: 
Missing descriptor for [class 
org.eclipse.persistence.testing.models.employee.domain.Employee]. Query: 
ReadAllQuery(org.eclipse.persistence.testing.models.employee.domain.Employee)| 
|
/

I am running tests from command line. Above happens when tests are run 
from top level (trunk folder) using "ant test-lrg". Just now I ran the 
tests by going under foundation/eclipselink.core.test and executing "ant 
test-lrg". I did not get any failures. But the number of tests run are 
different for both the cases. Attached are junit reports from both the 
run. From my limited knowledge of ant, I do not see a difference in why 
a different set of tests should be run in both cases. How does team in 
Ottawa run the tests?

Thanks,
Mitesh
> -Tom
>
> Mitesh Meswani wrote:
> > Please review attached patch.
> > Tests run:  LRG at top level. No regression detected due to the patch 
> > (184 test are failing under core with or without changes due to this 
> > patch)
> >
> > Thanks,
> > Mitesh
> >
> > -------- Original Message --------
> > Subject:     [Bug 238696] Fix use of Double Checked Locking pattern
> > Date:     Wed, 23 Jul 2008 19:21:31 -0400
> > From:     bugzilla-daemon@xxxxxxxxxxx
> > To:     Mitesh.Meswani@xxxxxxx
> > References:     <bug-238696-44960@xxxxxxxxxxxxxxxxxxxxxx/bugs/>
> >
> >
> >
> > https://bugs.eclipse.org/bugs/show_bug.cgi?id=238696  
> > Product/Component: EclipseLink / Foundation
> >
> > Mitesh Meswani <mitesh.meswani@xxxxxxx> changed:
> >
> >           What    |Removed                     |Added
> > ---------------------------------------------------------------------------- 
> >
> > Attachment #107673|0                           |1
> >        is obsolete|                            |
> >
> >
> >
> >
> > --- Comment #7 from Mitesh Meswani <mitesh.meswani@xxxxxxx>  
> > 2008-07-23 19:21:29 -0400 ---
> > Created an attachment (id=108282)
> > --> (https://bugs.eclipse.org/bugs/attachment.cgi?id=108282)
> > Proposed patch to fix only the proven broaken code path
> >
> > The attached patch eagerly initializes platform operators to eliminate
> > possibility of a race condition. Other code path are though 
> > theoretically
> > broken, are not considered to be a risk enough to introduce 
> > performance penalty
> > of using a volatile variable
> _______________________________________________
> eclipselink-dev mailing list
> eclipselink-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/eclipselink-dev

Attachment: junit-noframes.html.ranFromTopLevel.zip
Description: Zip compressed data