[JBPM-9774] Incorrect object type returned in a DMN business-rule task of a BPMN

[elguardian]

Jira: https://issues.redhat.com/browse/JBPM-9774

[tarilabs]

some questions, please?

[tarilabs]

not resolved.
#1947 (comment)
please leave the editor usage standard.

[tarilabs]

looks good to me, thank you @elguardian ! 🚀

[sonarcloud]

Kudos, SonarCloud Quality Gate passed!

0 Bugs
0 Vulnerabilities
0 Security Hotspots
0 Code Smells

No Coverage information
No Duplication information

[jomarko]

@elguardian thank you for the PR. It seems to be working fine. I found just a single issue related to years and moths duration/java.time.Period.

I was not able to successfully map years and months duration from DMN into java.time.Period in BPMN/Java.

See commented parts of my PR elguardian#6

Is that something I should report separately, or I just selected wrong BPMN/Java type for years and months duration?

[tarilabs]

re #1947 (review)

maybe just needs/missing something along the lines of

https://github.com/kiegroup/kogito-runtimes/blob/caa2a93186d856dbdb53047c9289a54568d0d7ec/kogito-codegen-modules/kogito-codegen-decisions/src/main/resources/class-templates/DecisionRestResourceQuarkusTemplate.java#L44-L46

?

[jomarko]

@tarilabs thank you, that works perfect for object serializing!

See:

• elguardian@7071687
• https://github.com/kiegroup/drools/pull/3669

However the testBuildDMNPeriodProcess test, where single value is returned, returns org.kie.dmn.feel.lang.types.impl.ComparablePeriod, not a java.time.Period

What is your opinion about it. It is a separate issue, as it not anymore about returned object type?

[tarilabs]

I don't understand the question @jomarko , sorry?

When a DMN model returns a ComparablePeriod as part of the decision results, this would allow the underlying Jackson to serialize it as a usual Period in the ISO format. I believe once deserialized by Jackson to the Java Pojo type of the process variable, the Pojo would have the Period as a type of the field.

[jomarko]

@tarilabs ok, let me try put it differently.

ComparablePeriod variable = (ComparablePeriod) processInstance.getVariable("period");

works, while

Period variable = (Period) processInstance.getVariable("period");

Throws an error:

java.lang.ClassCastException: class org.kie.dmn.feel.lang.types.impl.ComparablePeriod cannot be cast to class java.time.Period (org.kie.dmn.feel.lang.types.impl.ComparablePeriod is in unnamed module of loader 'app'; java.time.Period is in module java.base of loader 'bootstrap')

What is for me hard to decide if:

• it is an issue
• it is expected behavior

[tarilabs]

I would expect a process variable would be of type Period, and not using internal classes of DMN (ComparablePeriod).

[elguardian]

@jomarko the info about deserialization is related to the java class dependant on how you map those clases so @tarilabs is right

[jomarko]

@elguardian hello, sorry, this PR was not ready for merge from my point of view. There is no ack from QE side. Also please notice I prepared PR where I fix some cases of mapping DMN types indo BPMN/Java types. #1947 (comment)

Without addressing them or reporting them separately, we can not mark JBPM-9774 as done.

[elguardian]

@jomarko provide a proper test plz. The test you provided is not using the new code. For the code make it work you need to provide class loader and runtime manager.

[jomarko]

@elguardian working on tests updates as #1952

However I would like to ask update your added tests. I checked the newly created asset businessRuleCollectionTaskProcess.bpmn2, and it can not be opened in our latest VS Code tooling. See the screenshot below. I think it is importnat we use assets that are compliant with current tooling.

[elguardian]

@jomarko This was requested by DMN support for generic collections. the problem is in stunner in this case. please file a jira to stunner team so they can allow this sort of mapping.

[jomarko]

@elguardian I got the same for businessRuleTaskProcess.bpmn2. Let me check my understanding. Does a JBPM-9744 brings a support for a DMN <-> BPMN mapping, that can not be modeled using our current tooling? Just manually?