This topic applies only to the IBM Business Process Manager Advanced configuration.

Stand-alone tasks

Stand-alone tasks follow the service-oriented architecture (SOA) pattern and they are loosely coupled with the components that invoke them (to-do tasks), or the components that are invoked by them (invocation tasks). They can be wired to another component using the Service Component Architecture (SCA) infrastructure.

Stand-alone tasks have an autonomy setting of either peer or child. Stand-alone tasks with peer autonomy communicate with their partner components exclusively by SCA means. That is, to-do tasks receive input messages and return output or fault messages, and invocation tasks send input messages and receive output or fault messages. No further information exchange or lifecycle control happens.

Stand-alone tasks are made available as SCA components in the following ways:

To-do tasks have an interface that can be wired to a client component.
Invocation tasks have a reference that can be wired to the service to be invoked.
Collaboration tasks are self-contained SCA components. Although collaboration tasks are stand-alone task components they have no SCA references or SCA interfaces and therefore cannot be wired to other service components. Instead they provide interfaces so that people can start them and work with them using the Human Task Manager APIs.

When you are first planning your human task, you should model it as a stand-alone task if any of the following conditions are present:

The task provides just another service
You intend to replace the stand-alone task later on and don't want to change the component to which it is wired.

Because stand-alone tasks are modeled separately, they can be reused. Stand-alone tasks always emit their audit log events as human task events.