[rules-users] Drools Planner: entity was never added to this ScoreDirector?

Sat May 5 13:55:49 EDT 2012

It turned out to be a hidden hashCode problem on the superclass of one of the non-planning variable fields in the planning entity.  I was calling HashCodeBuilder.append(…).hashCode() instead of HashCodeBuilder.append(…).toHashCode().  It was strange that the hashCode output values in my PE's toString didn't change, but the issue appears to be resolved after that fix nonetheless.

Thanks for your help.

Best,
- C

--
Build Smarter Software.

Spantree Technology Group, LLC
813 W Randolph St, Suite 301
email: cedric at spantree.net (mailto:cedric at spantree.net) | phone: 773.359.3865 (tel:773.359.3865)
http://www.spantree.net (http://www.spantree.net/)

On Saturday, May 5, 2012 at 4:29 AM, Geoffrey De Smet wrote:

> Cedric is referring to this issue:
>   https://issues.jboss.org/browse/JBRULES-3461
> Currently, the following implementation should be respected for entities:
> - solutionHashCode() must include the variables, and hashCode() must not include the variables
> (same for solutionEquals() and equals() of course)
> 1) Check if your code follows that?
> 2) Does your code follow the equals() and hashCode() contract of Java?
>  
> I agree with you guys that a changing hashcode might have caused this.
> - The entity X is inserted into a Set or Map with a hashcode A.
> - While it's in that Set or Map, the entity X changes which also changes it's hashcode value to B.
> - This effectively corrupts the Set or Map, because X is still in bucket A%n, while it should be in bucket B%n,
> so the Set or Map will report that it doesn't contain X.
>  
> That fact that uncommented equals() and hashcode() together doesn't fix it, means that:
> a) either the superclass has an equals/hashcode problem
> b) either it's not a hashcode related problem
> c) the 2 Meeting entities are different instances, that represent the same entity. This is only possible if you create or remove entities in moves. => In this case you cannot rely on object identity and you must implement an equals/hashcode implementation based on at least some sort of id (for example a database primary key or uuid).
>  
> > It seems to happen randomly, with different entities each time If environmentMode == REPRODICBLE or DEBUG or TRACE (so not PRODUCTION)
> and you're not using simulated annealing (~ timegradient based), then 2 runs should be exactly the same always (more info in manual on environmentMode).
>  
> So, if they are still not in that case, the only thing I can think of that differs between runs is the memory addresses of the entity instances which are used in object identity.
> Object.hashCode() is implemented to return the memory address. Object.equals() returns true if it's the same memory address. The memory address doesn't change during 1 run, but it differs from run to run.
> So how can that explain that "randomly" behavior? Because in a Map of capacity 10, the hashCode() gets % 10, so about 1 out of 10 instances with different hashCode()s do end up in the same bucket (and this differs from run to run).
>  
> 3) Can you turn on environmentMode TRACE, just to verify that it doesn't fail on anything else sooner?
>  
> > I couldn't find an easy way to add a knowledge session logger in planner, but that might also be useful. I don't think there is an easy way at this time. Feel free to create a jira issue that DroolsScoreDirector or DroolsScoreDirectorFactory should support that.
>  
> Op 05-05-12 04:35, Cedric Hurst schreef:  
> > yeah, i have a solutionEquals() and solutionHashCode() as well. I tried dumping my hashCode() and equals() methods on the PlanningEntity but that didn't seem to have much effect.  My hashCode() and solutionHashCode() values now both change when I assign a different supplier, but planner still can't seem to find the entity in working memory.
> >  
> > --  
> > Build Smarter Software.
> >  
> > Spantree Technology Group, LLC  
> > 813 W Randolph St, Suite 301
> > email: [hidden email] (/user/SendEmail.jtp?type=node&node=3963892&i=0) | phone: <a href="tel:773.359.3865" value="+17733593865" target="_blank">773.359.3865
> > http://www.spantree.net (http://www.spantree.net/)  
> >  
> >  
> > On Friday, May 4, 2012 at 8:51 PM, Christopher Dolan [via Drools] wrote:
> >  
> > > Try just commenting out your .equals() and .hashCode() methods. My experience was that Planner is unexpectedly sensitive to hashcode algorithms, and identity equality just worked best. I ended up implementing a solutionEquals like some of the example code.
> > > Chris
> > >  
> > > From: [hidden email] (/user/SendEmail.jtp?type=node&node=3963850&i=0) [[hidden email] (/user/SendEmail.jtp?type=node&node=3963850&i=1)] on behalf of Cedric Hurst [[hidden email] (/user/SendEmail.jtp?type=node&node=3963850&i=2)]
> > > Sent: Friday, May 04, 2012 5:09 PM
> > > To: [hidden email] (/user/SendEmail.jtp?type=node&node=3963850&i=3)
> > > Subject: [rules-users] Drools Planner: entity was never added to this ScoreDirector?
> > >  
> > > Hi all,  
> > > I'm working on a Drools Planner 5.4.0.CR1 implementation for a scheduling app.
> > > First, some background:
> > > The solution class is 'Schedule' and the PlanningEntity is 'Meeting'. Meetings consist of timeslots (or 'slots'), and two participants in complimentary roles 'planners' and 'suppliers'. Since there are more planners than suppliers, I've structured the Schedule model so that its meeting collection is basically the cross-product of slots and planners. Suppliers are configured as the @PlanningVariable in a meeting. The meeting list in the schedule is also annotated as a @PlanningEntityCollectionProperty. I've structured three MoveFactories that add, subtract or swap suppliers from meeting to meeting.
> > > That part works fairly well, however, when I try to try to run my solution through a Solver, I am inevitably plagued with a dreaded message like this:
> > > java.lang.IllegalArgumentException: The entity (Meeting[slot: 26, planner: 29C2E0FF-00EE-409F-B05B-B175B30A6B45, supplier: BB25B297-54FA-45BD-923A-C209848E74B4, hashCode: 1219749520, solutionHashCode: -1086095971, slotHashCode: 436282962, plannerHashCode: -2037841459]) was never added to this ScoreDirector. at org.drools.planner.core.score.director.drools.DroolsScoreDirector.afterAllVariablesChanged(DroolsScoreDirector.java:110) at org.drools.planner.core.score.director.drools.DroolsScoreDirector.afterVariableChanged(DroolsScoreDirector.java:120) at com.somecompany.scheduler.moves.AddSupplierToMeetingMove.doMove(AddSupplierToMeetingMove.groovy:36) at org.drools.planner.core.localsearch.decider.DefaultDecider.doMove(DefaultDecider.java:128) at org.drools.planner.core.localsearch.decider.DefaultDecider.decideNextStep(DefaultDecider.java:103) at org.drools.planner.core.localsearch.DefaultLocalSearchSolverPhase.solve(DefaultLocalSearchSolverPhase.java:57) at org.drools.planner.core.solver.DefaultSolver.runSolverPhases(DefaultSolver.java:183) at org.drools.planner.core.solver.DefaultSolver.solve(DefaultSolver.java:151) at org.spockframework.util.GroovyRuntimeUtil.invokeMethod(GroovyRuntimeUtil.java:100) at com.somecompany.scheduler.solver.SolverSpec.should solve schedule(SolverSpec.groovy:59)  
> > > It seems to happen randomly, with different entities each time, but I've not yet had a successful run of a full solve cycle without hitting it a least once. Unfortunately, it kills my whole process.
> > > I've step debugged through the planner internals and verified that all the meetings are getting added to the working memory, including the offending one:
> > > 2012-05-04 16:30:17,379 INFO com.somecompany.scheduler.solver.SolverSpec - Meeting[slot: 26, planner: 29C2E0FF-00EE-409F-B05B-B175B30A6B45, supplier: FEA76C13-C9E2-42EC-977C-BCBB0BD1FFA0, hashCode: 1219749520, solutionHashCode: -574491332, slotHashCode: 436282962, plannerHashCode: -2037841459]  
> > > I was pretty certain my issue has to do with the meeting's hashCode value, but as you'll see they're exactly the same in working memory as they are when . I'm using commons HashCodeBuilder and EqualsBuilder to check only the fixed slot and planner values. The solutionHashCode changes based on the slot, planner and supplier-- supplier being the @PlanningVariable. I've pasted the full log, with trace enabled, here for review:
> > > https://gist.github.com/raw/93752853a0996ae1c016/d0ca35f5c531a2fa25e6a106c90257b015500931/solver-error.log  
> > > The error is listed at 2012-05-04 16:30:37,168.
> > > The original meeting in working memory is listed at 2012-05-04 16:30:17,379 (toward the middle of entries with that timestamp).
> > > I couldn't find an easy way to add a knowledge session logger in planner, but that might also be useful.
> > >  
> > > View this message in context: Drools Planner: entity was never added to this ScoreDirector? (http://drools.46999.n3.nabble.com/Drools-Planner-entity-was-never-added-to-this-ScoreDirector-tp3963528.html)
> > > Sent from the Drools: User forum mailing list archive (http://drools.46999.n3.nabble.com/Drools-User-forum-f47000.html) at Nabble.com (http://Nabble.com).
> > >  
> > > _______________________________________________  
> > > rules-users mailing list  
> > > [hidden email] (/user/SendEmail.jtp?type=node&node=3963850&i=4)  
> > > https://lists.jboss.org/mailman/listinfo/rules-users
> > >  
> > >  
> > > If you reply to this email, your message will be added to the discussion below: http://drools.46999.n3.nabble.com/Drools-Planner-entity-was-never-added-to-this-ScoreDirector-tp3963528p3963850.html  
> > > To unsubscribe from Drools Planner: entity was never added to this ScoreDirector?, click here.
> > > NAML (http://drools.46999.n3.nabble.com/template/NamlServlet.jtp?macro=macro_viewer&id=instant_html%21nabble%3Aemail.naml&base=nabble.naml.namespaces.BasicNamespace-nabble.view.web.template.NabbleNamespace-nabble.view.web.template.NodeNamespace&breadcrumbs=notify_subscribers%21nabble%3Aemail.naml-instant_emails%21nabble%3Aemail.naml-send_instant_email%21nabble%3Aemail.naml)  
> >  
> >  
> > View this message in context: Re: [rules-users] Drools Planner: entity was never added to this ScoreDirector? (http://drools.46999.n3.nabble.com/Re-rules-users-Drools-Planner-entity-was-never-added-to-this-ScoreDirector-tp3963892.html)
> > Sent from the Drools: User forum mailing list archive (http://drools.46999.n3.nabble.com/Drools-User-forum-f47000.html) at Nabble.com (http://Nabble.com).
> > _______________________________________________ rules-users mailing list rules-users at lists.jboss.org (mailto:rules-users at lists.jboss.org) https://lists.jboss.org/mailman/listinfo/rules-users  
> -- With kind regards, Geoffrey De Smet  
> _______________________________________________
> rules-users mailing list
> rules-users at lists.jboss.org (mailto:rules-users at lists.jboss.org)
> https://lists.jboss.org/mailman/listinfo/rules-users
>  
>  

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/rules-users/attachments/20120505/bbb8b611/attachment-0001.html 