[
https://issues.redhat.com/browse/DROOLS-6030?page=com.atlassian.jira.plug...
]
Gabriele Cardosi commented on DROOLS-6030:
------------------------------------------
[~mario.fusco1]
As per description, the invocation itself is present in two classes, with slightly
different reasons:
1) PMMLLoaderService.loadPMMLRuleMappers is called (mainly) starting from Kogito, where
the org.drools.model.Model are instantiated from (compiled) classes, a new
KnowledgeBuilderImpl is built out of such models, and then the newly built packages are
added to the parameter KnowledgeBuilderImpl (see
[
method|https://github.com/kiegroup/drools/blob/706f219d1877152d885a1acbef...])
2) inside PMMLRuntimeFactoryInternal, a new "clean" KieBase is created out of
the provided one, and populated only with packages relevant for the evaluated model (see
[
here|https://github.com/kiegroup/drools/blob/706f219d1877152d885a1acbef3d...]
and
[
here|https://github.com/kiegroup/drools/blob/706f219d1877152d885a1acbef3d...])
2)
: replace it with the preferred one and do thorough tests could take one or two days.
But the real issue is: what is the preferred one ? As far as I know, the use case by
itself has never implemented/requested before, and I have not the required knowledge of
drools-core to estimate how much time it could need to create a "nice"
implementation.
Evaluate "KieBase.addPackage" replacement effort from PMML
----------------------------------------------------------
Key: DROOLS-6030
URL:
https://issues.redhat.com/browse/DROOLS-6030
Project: Drools
Issue Type: Task
Reporter: Gabriele Cardosi
Assignee: Gabriele Cardosi
Priority: Major
Currently, the method KieBase.addPackage(s) is used inside PMML in two classes
1) PMMLLoaderService.loadPMMLRuleMappers
2) PMMLRuntimeFactoryInternal.getPMMLRuntime(String, String, KieBase)
3) PMMLRuntimeFactoryInternal.createKieBase(KnowledgeBuilderImpl)
In all those cases the needs is to create a "clean" kiebase that contains only
the model-specific packages, to avoid name/session clashing.
Scope of the current ticket is to evaluate an alternative, since the method itself is
going to be deprecated.
--
This message was sent by Atlassian Jira
(v8.13.1#813001)