Skip to content

Multi-instance Tasks (loops):#

Using Modeler

    <bpmn:scriptTask id="scriptTask" name="Script Task">
      <bpmn:incoming>Flow_159xzcz</bpmn:incoming>
      <bpmn:outgoing>Flow_0t7z2os</bpmn:outgoing>
      <bpmn:multiInstanceLoopCharacteristics isSequential="true" 
        camunda:collection="$(data.records)" />
      <bpmn:script><![CDATA[this.log('testing from the inside: '+data.loopKey);]]></bpmn:script>
    </bpmn:scriptTask>

Any Task can have multi-instance either Sequential or Parrallel

Loops have a unique data structure to support multiple items for same node with each of unique data

We will use an example to illustrate the use of loops: - We would like to conduct a poll on various departments about new projects coming up - We will retrieve department list from a DB using Get Department List service task - For each department will initiate a subProcess department with a multi-instance

Defining loop collection#

  • Input Collection is defined as an Array, defines the multipicity of the node.
  • Each value in the collection becomes a key to item

Design

  • Since we like to have various departments vote at any time, we make the multi-instance parallel
  • Inside the subprocess instance will assign users and designate more information about the vote in Prepare for a vote
  • A User Task vote to allow desigated users/user groups to vote with some input fields
  • After all department complete their vote, another task Summarize Votes will process the results

Design

Data structure#

  • Each loop instance stores its own unique data elements, for subProcess loops data is shared among all node items inside the subprocess

So the data structure looks like:

Loop name

Key

data 

department: {
    IT: {
      scriptLog: 'added by script event,key:IT',
      votedBy: 'IT-User',
      vote: 85
    },
    HR: {
      scriptLog: 'added by script event,key:HR',
      votedBy: 'HR-User',
      vote: 100
    },
    Billing: {
      scriptLog: 'added by script event,key:Billing',
      votedBy: 'Billing-User',
      vote: 80
    }
  }
Any node inside the loop will share same data, so inside 'HR' 
  console.log(item.data.vote,'for ',item.itemKey);
Will produce: 100 for HR.

However, Summarize Votes task is outside the subprocess and at the root, so it has the instance data structure So once you have access to instance data: 

  Object.keys(item.data.department).forEach(key=>{
    console.log('Vote for:',key,item.data.department[key]['vote']);

  });
will produce: 
Vote for: IT 100
Vote for: HR 85
Vote for: Billing 80

  • During the execution, either through an event or a script/service task 

        item.token.execution.getItems().filter(itm=>(itm.elementId=='Activity_vote')).forEach(voteItem=>{

        let vote =voteItem['data']['vote'];
        let dept=voteItem.itemKey;

        console.log("vote Item",voteItem.seq,'dept:',dept,'vote',vote);

    });

  • In a detached fashion, like a batch job or remote access 

        let items=await api.data.findItems({"items.elementId":'Activity_vote',"data.caseId":1050});
    items.forEach(voteItem=>{

        let vote =voteItem['data']['vote'];
        let dept=voteItem.itemKey;

        console.log("vote Item",voteItem.seq,'dept:',dept,'vote',vote);

    });
    In Addition, item.itemKey is populated with the key value to facilitate data search:

  • Search for a particular item by key: 

        let items=await api.data.findItems({"items.elementId":'Activity_vote',"items.itemKey":"HR"});

Standard loop#

Parallel or sequential have the loop collection defined at start of the task.

Standard Loop hower is different in that is executed multiple times until an explicit call to end the loop

alt text

To end the loop: 

this.token.loop.end();
is done in either insided the task or as trigger end-event

alt text