1. Create a Data Object with one field, type doesn't matter, String is enough. Override the toString() method to include this new field and save it. 2. Create a process with Start Node -> Human Task -> End Node. 3. Define one process variable with the type of the Data Object from the first step. 4. Configure Human Task Input and Output variables, both with the same name, which will reference the process variable as source and target. So the Human Task will act as a way to read and modify the data object process variable using a form. Assing this task to yourself (e.g. wbadmin user). 5. Save the process instance and generate all forms. 6. Deploy the process and run a new instance of it. The starting process form should have one field, provide something as a value. 7. Go to Task Inbox, start the task assigned to you, verify that the value is there, you can modify it. Complete the task. 8. Process is finished. The process variables log correctly shows the object's field value history. 9. Now go back, add one more field to the data object and include this data object in the toString() method. Save the data object. 10. Delete all the forms (there should be 3 of them). 11. Open the process definition and regenerate the forms. 12. Redeploy the project. 13. Run the process instance. Now 2 fields should be presented in the initial form. Fill both of them. 14. Go again to Task Inbox and start the task. Now the newly added field does not have the value from the previous step. Just the first field should be populated. 15. Populate also this second empty field and complete the task. 16. The process should be finished. Now go to the process variables log and open the history of the variable. You will notice 2 things: a) the second field is still null, so neither of the forms actually set its value b) the "middle" value in the history shows the previous implementation of the toString() method.