3:28 p.m.
I am using Drools version 5.4.0.Beta1. My rules are defined in a drl file.
I cannot get a type with @role(event) to run correctly. In every case I've
tried, I get the following error at runtime:
---------------------
java.lang.NullPointerException
at
org.drools.reteoo.ReteooFactHandleFactory.newFactHandle(ReteooFactHandleFactory.java:56)
at
org.drools.common.AbstractFactHandleFactory.newFactHandle(AbstractFactHandleFactory.java:80)
at
org.drools.common.AbstractFactHandleFactory.newFactHandle(AbstractFactHandleFactory.java:65)
at org.drools.common.NamedEntryPoint.createHandle(NamedEntryPoint.java:740)
at org.drools.common.NamedEntryPoint.insert(NamedEntryPoint.java:286)
at org.drools.common.NamedEntryPoint.insert(NamedEntryPoint.java:122)
at org.drools.common.NamedEntryPoint.insert(NamedEntryPoint.java:52)
at test.DroolsTest.main(DroolsTest.java:62)
--------------------
Looking at the source:
ReteooFactHandleFactory.java:56:
if ( type.getTimestampExtractor() != null ) {
"type" is null, hence the NPE. Tracing through in Eclipse, I find that in
ReteooFactHandleFactory.java, on line 54, is the following:
TypeDeclaration type = conf.getTypeDeclaration();
getTypeDeclaration() returns null. Stepped in, goes to
AbstractRuleBase.java:745:
public TypeDeclaration getTypeDeclaration( Class< ? > clazz ) {
TypeDeclaration typeDeclaration = this.classTypeDeclaration.get(
clazz );
if ( typeDeclaration == null ) {
// check super classes and keep a score of how up in the
hierarchy is there a declaration
TypeDeclarationCandidate candidate = checkSuperClasses( clazz );
// now check interfaces
candidate = checkInterfaces( clazz,
candidate,
1 );
if ( candidate != null ) {
typeDeclaration = candidate.candidate;
}
}
return typeDeclaration;
}
On line 746,
TypeDeclaration typeDeclaration = this.classTypeDeclaration.get( clazz );
this.classTypeDeclaration is type Map<Class< ? >, TypeDeclaration>. Map
is empty (size = 0), so the get() returns null. checkSuperClasses() and
checkInterfaces() also attempt to retrieve a TypeDeclaration from this same
(empty) map. So, a null typeDeclaration is returned.
Checking classTypeDeclaration references, the put() occurs in
AbstractRuleBase.addPackages() (AbstractRuleBase.java:476), and in
AbstractRuleBase.populateTypeDeclarationMaps() (AbstractRuleBase.java:336).
The latter is not called before the NPE. Execution in addPackages() do not
get to the classTypeDeclaration.put() because newPkg.getTypeDeclarations()
is empty. So, apparently, the TypeDeclaration is never put into the Map.
Attached is the minimal testcase that exhibits the same behavior.
http://drools.46999.n3.nabble.com/file/n3683063/TestRules.zip
TestRules.zip
--
View this message in context:
http://drools.46999.n3.nabble.com/NPE-for-any-role-event-tp3683063p368306...
Sent from the Drools: User forum mailing list archive at Nabble.com.
6:59 a.m.
(1) Your Java code fails to populate the Knowledge Base with the
package generated by the builder.
base.addKnowledgePackages( builder.getKnowledgePackages() );
(2) Do not qualify the class name after declare. TestEventObject is in
the same package; it has even been imported (which you needn't do).
-W
On 23/01/2012, capn_ed <cncosper(a)capn-ed.net> wrote:
I am using Drools version 5.4.0.Beta1. My rules are defined in a drl
file.
I cannot get a type with @role(event) to run correctly. In every case I've
tried, I get the following error at runtime:
---------------------
java.lang.NullPointerException
at
org.drools.reteoo.ReteooFactHandleFactory.newFactHandle(ReteooFactHandleFactory.java:56)
at
org.drools.common.AbstractFactHandleFactory.newFactHandle(AbstractFactHandleFactory.java:80)
at
org.drools.common.AbstractFactHandleFactory.newFactHandle(AbstractFactHandleFactory.java:65)
at org.drools.common.NamedEntryPoint.createHandle(NamedEntryPoint.java:740)
at org.drools.common.NamedEntryPoint.insert(NamedEntryPoint.java:286)
at org.drools.common.NamedEntryPoint.insert(NamedEntryPoint.java:122)
at org.drools.common.NamedEntryPoint.insert(NamedEntryPoint.java:52)
at test.DroolsTest.main(DroolsTest.java:62)
--------------------
Looking at the source:
ReteooFactHandleFactory.java:56:
if ( type.getTimestampExtractor() != null ) {
"type" is null, hence the NPE. Tracing through in Eclipse, I find that in
ReteooFactHandleFactory.java, on line 54, is the following:
TypeDeclaration type = conf.getTypeDeclaration();
getTypeDeclaration() returns null. Stepped in, goes to
AbstractRuleBase.java:745:
public TypeDeclaration getTypeDeclaration( Class< ? > clazz ) {
TypeDeclaration typeDeclaration = this.classTypeDeclaration.get(
clazz );
if ( typeDeclaration == null ) {
// check super classes and keep a score of how up in the
hierarchy is there a declaration
TypeDeclarationCandidate candidate = checkSuperClasses( clazz );
// now check interfaces
candidate = checkInterfaces( clazz,
candidate,
1 );
if ( candidate != null ) {
typeDeclaration = candidate.candidate;
}
}
return typeDeclaration;
}
On line 746,
TypeDeclaration typeDeclaration = this.classTypeDeclaration.get( clazz );
this.classTypeDeclaration is type Map<Class< ? >, TypeDeclaration>. Map
is empty (size = 0), so the get() returns null. checkSuperClasses() and
checkInterfaces() also attempt to retrieve a TypeDeclaration from this same
(empty) map. So, a null typeDeclaration is returned.
Checking classTypeDeclaration references, the put() occurs in
AbstractRuleBase.addPackages() (AbstractRuleBase.java:476), and in
AbstractRuleBase.populateTypeDeclarationMaps() (AbstractRuleBase.java:336).
The latter is not called before the NPE. Execution in addPackages() do not
get to the classTypeDeclaration.put() because newPkg.getTypeDeclarations()
is empty. So, apparently, the TypeDeclaration is never put into the Map.
Attached is the minimal testcase that exhibits the same behavior.
http://drools.46999.n3.nabble.com/file/n3683063/TestRules.zip
TestRules.zip
--
View this message in context:
http://drools.46999.n3.nabble.com/NPE-for-any-role-event-tp3683063p368306...
Sent from the Drools: User forum mailing list archive at Nabble.com.
_______________________________________________
rules-users mailing list
rules-users(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users
11:36 a.m.
Thank you very much. That seems to do the trick.
I was going to say that the call to add knowledge packages wasn't in the
Drools Expert User Guide example, but looking back at the User Guide, I see
that it IS there in the example setup. I also see that in my previous
project, which did not use events, I had that call, but commented it out for
some reason. As far as I am able to determine, though, for the fact-based
setup, that call didn't make any difference. It worked just as well without
it. And the Drools Fusion User Guide doesn't give an example for setting up
the knowledge base/builder/session/etc.
For the benefit of those who might be searching for a similar problem:
When I started with an event-based drl file, I began by copying my existing
(working) project and modifying it to use the stream processing mode. There
was some hints about how to do this in the Drools Fusion documentation, but
it wasn't complete. The implication is that the setup that is shown in the
Drools Expert User Guide example should be used. That example was:
Create a knowledge builder.
Add resources to the knowledge builder.
Check for errors.
Create a knowledge base from the builder.
Add knowledge packages to the knowledge base, (which step I skipped, but
didn't notice.)
The Drools Fusion User Guide says to create a KnowledgeBaseConfiguration
object, and set the options desired, but doesn't give a clue about what to
do with it after that. I initially assumed that it would be passed in an
overloaded form of the KnowledgeBuilder getKnowledgeBase() method, but no
such overload exists. By examining the available methods, I found an
overload of KnowledgeBaseFactory.newKnowledgeBase() that took a
KnowledgeBaseConfiguration object, but then I had to re-order my
initialization steps, because I had the KnowledgeBase before I created the
KnowledgeBuilder:
Create a KnowledgeBaseConfiguration object.
Set the options desired.
Create a KnowledgeBase using KnowledgeBaseFactory.newKnowledgeBase(),
passing in the config object.
Create a KnowledgeBuilder using
KnowledgeBuilderFactory.newKnowledgeBuilder() passing in the KnowledgeBase
object.
Add resources to the builder.
Check for errors.
And, the step that I forgot, Add knowledge packages to the knowledge base.
Without the final step, things seemed to work correctly as long as I didn't
include @role( event ).
--
View this message in context:
http://drools.46999.n3.nabble.com/NPE-for-any-role-event-tp3683063p368545...
Sent from the Drools: User forum mailing list archive at Nabble.com.
2:06 p.m.
This is indeed very confusing.
The methods
public static KnowledgeBuilder newKnowledgeBuilder(KnowledgeBase kbase)
public static KnowledgeBuilder newKnowledgeBuilder(KnowledgeBase kbase,
KnowledgeBuilderConfiguration conf)
are - although in the stable API - without any documentation.
Apparently it is an alternative way for populating a KBase, but who
knows?
Why just a
@role(event)
is disregarded when these forms of builder creation is used looks like
a bug to me.
-W
4:15 p.m.
Sorry, I mis-spoke (typed).
I meant that when I remove @role(event), I can execute the same rules
without the null pointer exception, with no other change to the rules or the
Java code. I don't think that's a bug. Because the event role was not
assigned to the type, it skipped over attempting to assign a timestamp
during the fact insertion.
--
View this message in context:
http://drools.46999.n3.nabble.com/NPE-for-any-role-event-tp3683063p368621...
Sent from the Drools: User forum mailing list archive at Nabble.com.
4694
days inactive
4695
days old
4 comments
2 participants
participants (2)
-
capn_ed -
Wolfgang Laun