Represents a special kind of task result which should be handled as a structured output with fields that haven't completed yet.

Structured tasks results can serve as a placeholder result for a task and make their data partially available.

Sometimes tasks divide their work into multiple sub-tasks. In this case the result of a task can include values from the sub-tasks. This would require from the parent task to wait all of its child worker tasks and then construct an appropriate object that contains these results. This behaviour can hinder dependent tasks in their execution as this can result in waiting for tasks which are unrelated to the work of the dependent task.

Take the following example:

Task A is a complex task that divides its work into separate partitions. Task A will return a complex object as a result, with fields of X and Y, each corresponding to the result of its worker tasks.
Task B has an input dependency on task A, and will use the field X of its result. This means that task B will be stalled until task A finishes completely.

This is an undesirable scenario, as task B only needs the result of the sub-task X, while the long running of sub-task Y can stall the execution of task B.

In order to avoid this, structured task results were introduced which are able to separate the concerns of unrelated results contained by their parent.

With structured results, task B in the example above will not be stalled, as task A can return before the completion of its sub-tasks. Task A returns a structured task results with the fields X and Y, which have the values of the sub-task task identifiers. Task B will receive the structured result from task A, and see the task identifier for sub-task X. It will then wait for sub-task X, and proceed with its execution. We can see that in this case it was not necessary for task B to wait for the completion of sub-task Y to get the result from sub-task X.

Tasks can choose to retrieve the actual objectified, non-structured result by calling toResult(TaskResultResolver).

If a task doesn't want to handle structured task results (but it is still encouraged to to so), it can choose to call getActualTaskResult(TaskIdentifier, TaskResultResolver) which converts any possible structured results to object representation.

The TaskContext instance for the tasks can be used for the TaskResultResolver parameter of the methods in this interface.

Implementations are not required to implement Object.hashCode() and Object.equals(Object), but if they do, they must ensure that if two results equal, then they semantically provide access to the same results. The actual end results are not required to equal, but the semantics along which they are accessed must. E.g. Two StructuredObjectTaskResult can equal if their enclosed task identifier equal, however, it is not required that the task outputs of the associated task identifiers are equal.

Note: When constructing strucuted task results, make sure to not have circular references in them.