CamundaLibrary 1.1: One library for all and long polling

Libraries Camunda
1

CamundaLibrary 1.0 and 1.1 are released.

1.0 Version: one library for all

Previously we had CamundaLibrary.ExternalTask and CamundaLibrary.Deployment and so on as sub libraries. That was very inconvenient. Only to experts it might have been obvious that the structure of sub libraries was equivilant to the Camunda REST API documentation. Even we often got confused, if we had to use for starting a process instance the sub library ProcessDefinition or ProcessInstance. Don’t bother anymore: The sub libraries are gone and there is only CamundaLibrary:

*** Settings ***
Library    CamundaLibrary    http://localhost:8080

*** Task ***
Demo task
    # former: CamundaLibrary.Deployment
    deploy model from file    bpmn/demo_process.bpmn

    # former: CamundaLibrary.ProcessDefinition
    start process    demo_process

    # former: CamundaLibrary.ExternalTask
    ${variables}    fetch workload    my_topic
    do some processing
    complete task

Also some keywords got deprecated throwing warnings.

1.1 Long polling

Since tasks have individual processing durations some might end too soon, while there predecessor are still busy.

Assume this:

*** Task ***
Slow task 1
    fetch workload     my_topic1
    sleep   2s
    complete task

Super fast task 2
    fetch workload    my_topic2
    # complete task immediately
    complete task

Because second task finishs immediately, it won’t catch any workload from task 1 until task 2 is restarted.

With long polling you can now tell a fetch workload keyword to negotiate a timeout (in milliseconds). If a workout does arrive within timeout at the camunda acitivity, Camunda calls back the robot client delievering the new workload. Long polling is activated with àsync_response_timeout setting the timeout in milliseconds (30 minutes maximum):

*** Task ***
Slow task 1
    fetch workload     my_topic1
    sleep   2s
    complete task

Super fast task 2 waiting for task1
    # wait up to 3000 milliseconds for new workload passed on by task1
    fetch workload    my_topic2    async_response_timeout=3000
    # complete task immediately
    complete task

Enjoy.

1 Like