Hi.

Suppose I have two pools:  "Participant A" and "Participant B".

Suppose I have a task called "Transmit data to participant B" within pool "Participant A".  An outgoing message flow from the task indicates that data is being passed to something in the "Participant B" pool.  The process in process in the "Participant A" pool is now finished, indicated by an outgoing sequence flow leading to a end event.

What if the "Transmit data to participant B" task can fail due to timing out?  I'd include an intermediate timer event on the "Transmit data to particpant B" task, and have an outgoing sequence flow leading to the aforementioned end event.

So here's my question:  If the process for "Participant B" can not continue without receiving the message from the task in the "Participant A" pool, is it assumed that the message is not sent from task "Transmit data to participant B" if the time-out occurs?  It's not clear whether the message is sent when the task completes, or when the time-out occurs.

I welcome any alternative modeling suggestions.

Thanks in advance,

vinny

Thanks a lot for the comments, larimar.    I hope my simplified example model conveyed what it was supposed to convey.

1)  I generally use UML for modeling systems processes prior to their development; but in this case, the system exists so I tried to apply BPMN.  Additionally, the BPMN Quick Reference Guide (QRG) on this site, as well as many other examples that I've seen on the Web from reputable BPMers, use BPMN to document general systems processes.  I initially modeled this process using a Sequence diagram, which seemed to be the best UML modeling tool to use.  The Activity diagram, in my opinion, is lacking, too simplistic, and doesn't, in this case, at least, have what I need to convey the truth about the process.  Plus our stakeholders, as well as developers and management, are really starting to see the benefits of BPMN for accurate modeling.

2)  BPMN specifies that each pool contains separate processes.  Participant B, in my real-world model, is a third-party vendor, not part of our organization.  Our means of communication are HTML post-backs (i.e. messages).  Again, the BPMN QRG on this site shows messages transferring from the Travel Agency to the Funding Institution:  This is very similar to my process.  The fact is that the Participant A pool process is finished when it attempts to transmit the data to Participant B.  If it can't transmit the data, then the Participant A process is still complete, only the message is not sent to Participant B to trigger the event necessary for the Participant B process to continue.  There is no parallel processing occurring here, by the way; every task is sequential and depends on receiving those post-back messages in order to proceed.

3)  Great point, and I had thought about making that a collapsed looping subprocess.  I still might do that...

4)  Technically, the process for Participant A is complete, whether it times-out or not.  A trigger is required for the Participant B process to proceed.

This is a rather challenging prospect, in my opinion, and I'm sure I'm not the first one to have to deal with it.  Sure, I could put both participants in separate lanes in the same pool to make things simpler, but really don't think that's conveying accurately what's occurring.

Again, thanks for your comments.

vinny