[jboss-jira] [JBoss JIRA] (JBRULES-3466) sum with no matching values results in NPE on reverse

[Chris Dolan (JIRA)]

    [ https://issues.jboss.org/browse/JBRULES-3466?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12685464#comment-12685464 ] 

Chris Dolan commented on JBRULES-3466:
--------------------------------------

It looks like the NPE is an symptom of corrupted working memory rather than a cause. When I turned on TRACE (which I should have thought to do right away...) I get this:

{noformat}
java.lang.IllegalStateException: Score corruption: the workingScore (0hard/-25200soft) is not the uncorruptedScore (0hard/-27600soft):
  The workingMemory has 0 ConstraintOccurrence(s) lacking:
    jobDelay/NEGATIVE_SOFT:[JobStart(JobSlot(s2,4),Job(j5),2400000)]=2400
  Check the score rules who created those ConstraintOccurrences. Verify that each ConstraintOccurrence's causes and weight is correct.
{noformat}

I have some work to do to understand this error, but at least now I know why my unit test passed and my real-world problem fails.
                
> sum with no matching values results in NPE on reverse
> -----------------------------------------------------
>
>                 Key: JBRULES-3466
>                 URL: https://issues.jboss.org/browse/JBRULES-3466
>             Project: Drools
>          Issue Type: Bug
>      Security Level: Public(Everyone can see) 
>          Components: drools-core  (expert)
>    Affects Versions: 5.4.0.CR1
>         Environment: drools-planner 5.4.0-SNAPSHOT, snapshot downloaded on Apr 9, 2012
>            Reporter: Chris Dolan
>            Assignee: Edson Tirelli
>
> I have a rule with this element in the when clause:
> {noformat}
>         $beforePrediction : Number() from accumulate(
>              JobSlot(job != NULL_JOB, $j : job, $service == service, $ordinal > ordinal) and $p : Prediction(job == $j, service == $service),
>              sum($p.getMillisMid())
>         )
> {noformat}
> When I execute this I have cases where there are no matching Prediction instances, so the sum is zero. Later on retract, I get this:
> {noformat}
> Caused by: java.lang.NullPointerException: null
> 	at org.drools.base.accumulators.SumAccumulateFunction.reverse(SumAccumulateFunction.java:83)
> 	at org.drools.base.accumulators.JavaAccumulatorFunctionExecutor.reverse(JavaAccumulatorFunctionExecutor.java:130)
> 	at org.drools.rule.Accumulate.reverse(Accumulate.java:208)
> {noformat}
> The relevant code is this:
> {noformat}
>     public void accumulate(Serializable context,
>                            Object value) {
>         SumData data = (SumData) context;
>         data.total += ((Number) value).doubleValue();
>     }
> {noformat}
> and this:
> {noformat}
>     public void reverse(Object workingMemoryContext,
>                         Object context,
>                         Tuple leftTuple,
>                         InternalFactHandle handle,
>                         Declaration[] declarations,
>                         Declaration[] innerDeclarations,
>                         WorkingMemory workingMemory) throws Exception {
>         final Object value = ((JavaAccumulatorFunctionContext) context).reverseSupport.remove( Integer.valueOf( handle.getId() ) );
>         this.function.reverse( ((JavaAccumulatorFunctionContext) context).context,
>                                value );
>     }
> {noformat}
> A breakpoint reveals that handle is a Prediction instance. Somehow that handle was not added to reverseSupport, apparently.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.jboss.org/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira