Yes.. but you can include these two lines before this, to check.
Operator op = BeforeEvaluatorDefinition.BEFORE;
Operator op = AfterEvaluatorDefinition.AFTER;
The problem in that, for example, for the before,
https://github.com/droolsjbpm/drools/blob/5.5.x/drools-core/src/main/java...
If you don-t mention not use this class before trying to unmarshall it, the
operator is not added to the operators registry, and thus NPE is thrown.
public static final Operator BEFORE =
Operator.addOperatorToRegistry( "before",
false );
In the ticket I included a patch to fix it, I have one thing pending to add
to the pull request in order to be ready to be merged. But the workaround
of just "mentioning" the evaluator definitions so they are added to the
operators list, before doing the unmarshall, has worked for me.
Thanks,
Demian
Demian
On Fri, Oct 11, 2013 at 12:09 PM, Alexander Wolf <mail(a)alexander-wolf.net>wrote:
Hey Demian,
thank you! This might actually be related. But how do I get the reference
to these Operators to (re)set them?
This is my unmarshalling code (pretty simple and basic..):
ByteArrayInputStream bais = new ByteArrayInputStream(
serializedKnowledgeBase);
ObjectInputStream ois = new ObjectInputStream(bais);
kbase = (KnowledgeBase)ois.readObject();
Marshaller marshaller = MarshallerFactory.newMarshaller(kbase);
ByteArrayInputStream bais2 = new ByteArrayInputStream(
serializedKnowledgeSession);
knowledgeSession = marshaller.unmarshall(bais2);
Also: Is this going to be fixed in 5.6.0 ? (@Mark Proctor?)
- Alex
On 11.10.2013, at 16:09, Demian Calcaprina <calcacuervo(a)gmail.com> wrote:
Hey Alexander,
Would this be related to this?
https://issues.jboss.org/browse/DROOLS-243
If it is this one, you can do something like this as a workaround before
constructing the KBase:
Operator op = BeforeEvaluatorDefinition.BEFORE;
Operator op = AfterEvaluatorDefinition.AFTER;
This will cause the evaluation definition to be put in memory.
Hope this helps!
Demian
On Fri, Oct 11, 2013 at 9:21 AM, Alexander Wolf <mail(a)alexander-wolf.net>wrote:
> [Drools 5.5.0 Final]
>
> I wrote a webserver that keeps StatefulKnowledgeSessions running in
> Stream mode. When the server is stopped, the sessions are serialized and
> stored in a database. When the server is restarted, the existing sessions
> are restored / deserialized.
>
> Some time after recreating the session and inserting some events I get
> the following exception:
>
> java.lang.NullPointerException: null
> at
>
org.drools.base.evaluators.AfterEvaluatorDefinition$AfterEvaluator.evaluateCachedLeft(AfterEvaluatorDefinition.java:351)
> ...
>
> Apparently the exception seems to be related to a rule that uses temporal
> reasoning.
> After getting it once, the session seems to be corrupt -> from now on
> every time I call ksession.insert(event) the same exception occurs.
> When I serialize the session and attempt to deserialize it, I get a long
> stack trace --> I can not deserialize it anymore. (as apparently it is
> corrupted).
> When I run the server with the same events/sessions without stopping
> (serialization/deserialization) the exception does not occur)
>
> Can I do anything to prevent this behaviour? I thought maybe the "after"
> timer/trigger is somehow still active when I serialize the ksession.
> I tried ksession.halt() before serialization, but with no effect.
>
> Any ideas?
>
> _______________________________________________
> rules-users mailing list
> rules-users(a)lists.jboss.org
>
https://lists.jboss.org/mailman/listinfo/rules-users
>
_______________________________________________
rules-users mailing list
rules-users(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users
_______________________________________________
rules-users mailing list
rules-users(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users