Full Coverage of DMN FEEL 1.2 Starting with Camunda BPM 7.13

We are happy to announce, that Camunda BPM 7.13 (scheduled for the end of May) includes full coverage of FEEL 1.2 – for more DMN notation elements than before:

• Input Expressions NEW!
• Input Entries
• Output Entries NEW!
• Literal Expressions NEW!

We achieved these improvements by adding the former Community Extension FEEL Scala Engine, written by Philipp Ossler, to the official Camunda Stack.

Let’s have a look at the following example to get a better understanding of how FEEL 1.2 integrates with Camunda BPM.

We want to calculate the maximum credit sum that we can grant to a customer based on the requested credit type and the customer’s credit score.

When the customer’s credit score is…

• … weaker than 80% of the average credit score:
  no credit is granted
• … between 80% and 100% of the average credit score:
  the average credit sum from the past is granted
• … better than the average credit score:
  1.5 times of the average credit sum from the past is granted

Based on the business context, I created a simple Decision Requirements Diagram.

First, let’s have a closer look at the FEEL expressions used in the child decision “Calculate Credit History Key Figures”.

The variable credit_type represents the customer’s credit type, and the variable credit_history the history of already granted credits from the past.

Context

The JSON-like structure is called Context. It consists of the properties avg_score and avg_granted_sum with expressions as values. The Context is assigned to the variablekey_figures so it can be accessed in the root decision.

Filter Expression

The expression credit_history[type = credit_type] filters the variable credit_history for all credit types which equal the variable credit_type. The evaluated result is a list of Contexts structured as follows:

[ {"type": "personal-loan", "score": 505, "granted_sum": 1554.30}, ..., ... ]

Read more about the Filter Expression.

Path Expression

With the help of .score and .granted_sum, the values of the Context can be accessed.

Read more about the Path Expression.

Built-in Functions

The function mean(...) calculates the average and is one out of many more Built-in Functions
you can use in your expressions.

Second, let’s get to know the FEEL expressions used in the root decision “Calculate Max Grantable Credit Sum”.

Mathematical Operators

FEEL allows using Mathematical Operators. You can find an example of multiplication in the first Input Entry: key_figures.avg_score*.8.

Unary Test Comparison

In the first and third Input Entries, the value of the Input Expression credit_score is implicitly compared with the subsequent expressions.

Read more about Unary Test Comparison.

Special Variable

In the second Input Entry, no implicit comparison is performed. Instead, the Special Variable ? is used in conjunction with the x between y and z keyword. The Special Variable represents the value of the Input Expression credit_score.

Try it out!

To try out the example yourself, follow the step-by-step guide below.

Step 0: Run Camunda BPM

Make sure to download Camunda BPM 7.13.0-alpha2 (click on “Preview Release”), unpack it, and start the platform.

Step 1: Deploy Decision Model

Download the decision model calc-credit-sum.dmn (right click & save link as…) and deploy it to Camunda BPM. You can deploy the model quickly with the help of Camunda Modeler.

Step 2: Evaluate Decision

Evaluate the decision by performing the following REST API request:

POST /decision-definition/key/calc-credit-sum/evaluate

{
  "variables":{
     "credit_history":{
        "value":"[{\"type\":\"personal-loan\", \"score\": 505, \"granted_sum\": 1554.30}, {\"type\":\"mortgage\", \"score\": 931, \"granted_sum\": 600900.43}, {\"type\":\"mortgage\", \"score\": 754, \"granted_sum\": 210000.00}, {\"type\":\"personal-loan\", \"score\": 437, \"granted_sum\": 1900.44}]",
        "type": "Json"
     },
     "credit_score":{
        "value": 800
     },
     "credit_type":{
        "value": "mortgage"
     }
  }
}

Step 3: Review Decision History

Go to Cockpit and open the decision instance view to see the result:

You can find the complete example prepackaged as a Spring Boot application here:
FEEL 1.2 Example on GitHub

More Enhancements

• Native FEEL Types represent Spin-Based JSON & XML Variables
• Register Custom Functions Programmatically

Further Reading

• You can read more about DMN FEEL 1.2 in the FEEL Scala Documentation
• Please also see the User Guide to learn more about the Camunda BPM integration
• When you use a Camunda BPM version <= 7.12, learn how to migrate your FEEL expressions with the help of the Migration Guide

Please give us Feedback!

Do you like what we’ve built, or do you want to share some feedback with us?

You can ping us on Twitter @Camunda or reach us out on the Forum.