Data Objects and Variables

For automating processes, the transfer and storing of data is essential. In an automated process, data is stored as so-called “Data Objects” or “Variables”. There are two location where data is stored: in the process instance or in the global Data Store. The data inside the global Data Store is independent of one process and is managed inside the Management System. Therefore, the definition of this data must not be serialized in the BPMN process. So, this page only describes the serialization of process data created within a process instance.

Process Data Serialization

Process Data Objects can be created in the following ways:

On the process level: within the Automation tab of the Property panel there is a table to create Variables
Inside a UserTask: for using process data, the UserTask editor offers a direct possibility to create new Variables
Inside a ScriptTask: the variable.XXX library functions offers functionality to create new Variables

Only for the first two options, it is possible to serialize the variable specification in the BPMN, because there definition happens at process design-time. (The creation within ScriptTask happens at runtime, so the Variables are not serialized in BPMN. But ScriptTasks can also access defined Variables from the process level or UserTask.)

Variables are serializes as follows:


...
<process id="_09abb321-f819-480a-9827-66264d17b4df" name="PROCEED Main Process" processType="Private" isExecutable="true">
 
  <extensionElements>
      <proceed:variables>
          <!--
          name: required, unique
          dataType: required, one of: string, number, boolean, object, array, null
          description: optional, free text
          requiredAtInstanceStartup: optional, boolean, default false
          defaultValue: optional, set at design time
          enum: optional, array of strings seperated by semicolon, the value must be one of these
          const: optional, boolean, default false, non-changeable value after set for the first time
 
          dataType similar to JSON Schema: https://datatracker.ietf.org/doc/html/draft-bhutton-json-schema-validation-00#section-6.1.1
          - string: "Text" -> optional: textFormat="email|url" (default: not present, means: no string contraints)
          - number: "Number"
          - boolean: "On/Off"
          - object: "Combined Structure" -> no defaultValue and no enum in UI yet
          - array: "List" -> no defaultValue and no enum in UI yet
          -->
 
          <proceed:variable name="Maschinen Typ" dataType="string" textFormat="url" description="Gibt den Typ einer Maschine an" requiredAtInstanceStartup="true" defaultValue="https://machine-type.de/3d-printer" enum="https://machine-type.de/3d-printer;https://machine-type.de/cnc;https://machine-type.de/grinder" const="false" />
 
 
      </proceed:variables>
  </extensionElements>
  ...

The serialization of Process Variables in PROCEED is not yet BPMN conform. There is no itemDefinition, dataObject or ioSpecification created at design time and this elements are also not considered during execution of the process at runtime.

Runtime behaviour

During runtime, a process variable is accessible within the process level and all of its Sub-Processes, Activities, Tasks, Gateways or Events. Tasks usually first copy the process variables into local task variables (the local scope). During execution of the Task, this local data can be modified and after finishing the tasks, the local variable states are written back into the process variable states of the process instance. (Exception: Script Tasks have functions that can directly change the state of a process variable.) If two parallel Tasks write to the same process Variable after the Task finishes, the latest executions wins and overrides the process variable value.