Dynamically creating task * nodes * for each group of actor

[elifarley]

Consider the following pseudo code:

// List whose elements are arrays of actor ids.
// Its contents are not known in advance
List<String[]> listOfArrays = { {"actor-A", "actor-B"}, {"actor-C", "actor-D"} };

for (String[] actorIds : listOfArrays ) {

 ! create a task-node assigned to all actors in actorIds;
 ! wait until any of those actors completes his/her task by selecting either YES or NO;
 ! If the actor selected "NO", abort. Otherwise, continue.

}

Example of execution:

List<String[]> listOfArrays = { {"actor-A", "actor-B"}, {"actor-C", "actor-D"} };

! create task node assigned to {"actor-A", "actor-B"};
! wait for the task to be completed by either actor;
(Suppose actor selected "YES")
! create task node assigned to {"actor-C", "actor-D"};
! wait for the task to be completed by either actor;
(Suppose actor selected "YES")
! Done.

What would be the best way to implement this with jBPM ?

Maybe there's a way of achieving the same goal without creating task nodes...

Thanks for any ideas you may have.

[elifarley]

Let me clarify the last post with an imaginary use case (close to the real one).

A node in my workflow loads a text file describing stages through which an order request must pass in order to be approved:

--
actor-A, actor-B;
actor-C, actor-D, actor-E;
actor-F;
--

Each order request has its associated text file (otherwise I could simply design a fixed workflow with the appropriate task nodes).

For each line, a task should be assigned to the group of actors listed on that line.

I tried to parse the text file and create a transient, dynamic process definition based on it, so that I could persist it, call it using a process-state node, and then purge it from the database - all from the same master process.

Unfortunately, it doesn't seem to work like that.

I'll try to create this transient process definition and post a JMS message to be later processed. At least it sounds more promising...

[elifarley]

There's no need to create one task node for each line of the input file.

I've created only one task node and a decision node pointing to each other in a loop. They've allowed me to iterate through each input line, reusing the single task node.

Cheers,
Elifarley