[jboss-jira] [JBoss JIRA] Commented: (JBRULES-740) IncompatibleClassChangeError working with ShadowProxy

[Dirk Bergstrom (JIRA)]

    [ http://jira.jboss.com/jira/browse/JBRULES-740?page=comments#action_12357818 ] 
            
Dirk Bergstrom commented on JBRULES-740:
----------------------------------------

I have what might be an even better idea...

Seems to me that the fields of a given shadow proxy object will never
change.  If that's true, then shadows don't need to calculate their
hashcode, they can simply use a cached value.  That value can be
aquired by calling hashCode() on the original object at the time that
the rest of the shadow fields are populated.  This would have the
added benefit of making shadow.hashCode() very fast.

Unfortunately, this doesn't help as much with equals().  It does make
it easy to fail-fast -- if the hashcodes are different, the objects
are not equal.

> IncompatibleClassChangeError working with ShadowProxy
> -----------------------------------------------------
>
>                 Key: JBRULES-740
>                 URL: http://jira.jboss.com/jira/browse/JBRULES-740
>             Project: JBoss Rules
>          Issue Type: Bug
>      Security Level: Public(Everyone can see) 
>    Affects Versions: 3.1-m2
>         Environment: Trunk (revision 10170)
>            Reporter: Dirk Bergstrom
>         Assigned To: Edson Tirelli
>             Fix For: 3.1-m2
>
>         Attachments: Record.java, RLIRecord.java, stacktraces.txt
>
>
> I'm getting an exception while asserting objects into the working memory.  Depending on the timing, I get different stacktraces leading up the same exception on the same generated class.  I have 15 rules and three types of objects in the working memory.  2500 User objects, 37 Milestone objects, and 2000 RLI objects.  If I assert all the Users and Milestones first, I get the first exception below during assertion of the RLI objects, at the 257th object asserted.  If I arrange to skip that particular record, I can assert a few more before another record trips the bug.  If I put a try/catch block around the assertion and continue after hitting an exception, I can assert about 1600 of the records successfully, and the others reliably throw (ie. if I try to assert them later).  If I assert RLIs first, I get the second stacktrace during assertion of Milestones.  Looks like the problem happens during execution of one or more rules.  I'll attach the RLI class, and its parent class (which is also the parent of the other two classes).
> Exception in thread "RLIs Processor" java.lang.IncompatibleClassChangeError
> 	at net.juniper.dash.data.RLIRecordShadowProxy.hashCode(Unknown Source)
> 	at java.util.HashMap.put(HashMap.java:418)
> 	at java.util.HashSet.add(HashSet.java:194)
> 	at org.drools.reteoo.CollectNode.assertTuple(CollectNode.java:135)
> 	at org.drools.reteoo.CollectNode.assertObject(CollectNode.java:212)
> 	at
> org.drools.reteoo.SingleObjectSinkAdapter.propagateAssertObject(SingleObjectSinkAdapter.java:20)
> 	at org.drools.reteoo.AlphaNode.assertObject(AlphaNode.java:145)
> 	at
> org.drools.reteoo.CompositeObjectSinkAdapter.propagateAssertObject(CompositeObjectSinkAdapter.java:304)
> 	at org.drools.reteoo.ObjectTypeNode.assertObject(ObjectTypeNode.java:176)
> 	at org.drools.reteoo.Rete.assertObject(Rete.java:121)
> 	at org.drools.reteoo.ReteooRuleBase.assertObject(ReteooRuleBase.java:196)
> 	at
> org.drools.reteoo.ReteooWorkingMemory.doAssertObject(ReteooWorkingMemory.java:68)
> 	at
> org.drools.common.AbstractWorkingMemory.assertObject(AbstractWorkingMemory.java:729)
> 	at
> org.drools.common.AbstractWorkingMemory.assertObject(AbstractWorkingMemory.java:548)
> Exception in thread "Milestones and NPI Processor" java.lang.IncompatibleClassChangeError
> 	at net.juniper.dash.data.RLIRecordShadowProxy.hashCode(Unknown Source)
> 	at java.util.HashMap.put(HashMap.java:418)
> 	at java.util.HashSet.add(HashSet.java:194)
> 	at org.drools.reteoo.CollectNode.assertTuple(CollectNode.java:135)
> 	at org.drools.reteoo.CompositeTupleSinkAdapter.propagateAssertTuple(CompositeTupleSinkAdapter.java:30)
> 	at org.drools.reteoo.JoinNode.assertTuple(JoinNode.java:116)
> 	at org.drools.reteoo.SingleTupleSinkAdapter.createAndPropagateAssertTuple(SingleTupleSinkAdapter.java:55)
> 	at org.drools.reteoo.LeftInputAdapterNode.assertObject(LeftInputAdapterNode.java:147)
> 	at org.drools.reteoo.SingleObjectSinkAdapter.propagateAssertObject(SingleObjectSinkAdapter.java:20)
> 	at org.drools.reteoo.AlphaNode.assertObject(AlphaNode.java:145)
> 	at org.drools.reteoo.SingleObjectSinkAdapter.propagateAssertObject(SingleObjectSinkAdapter.java:20)
> 	at org.drools.reteoo.ObjectTypeNode.assertObject(ObjectTypeNode.java:176)
> 	at org.drools.reteoo.Rete.assertObject(Rete.java:121)
> 	at org.drools.reteoo.ReteooRuleBase.assertObject(ReteooRuleBase.java:196)
> 	at org.drools.reteoo.ReteooWorkingMemory.doAssertObject(ReteooWorkingMemory.java:68)
> 	at org.drools.common.AbstractWorkingMemory.assertObject(AbstractWorkingMemory.java:729)
> 	at org.drools.common.AbstractWorkingMemory.assertObject(AbstractWorkingMemory.java:548)
> 	at net.juniper.dash.data.DataSource.reconcileAssertedRecords(DataSource.java:190)
> 	at net.juniper.dash.data.DataSource.populateRecords(DataSource.java:150)
> 	at net.juniper.dash.Updater$DataSourceProcessor.work(Updater.java:181)
> 	at net.juniper.dash.Refresher.run(Refresher.java:70)

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira