james mamel Jan. 7, 2010, 12:24 p.m.

Hi everybody,

i found many interesting points according to collapsing pools and show the in- and outcoming message flows to a black box (http://en.bpmn-community.org/tutorials/5/). My question is, it is although possible to collapse lanes, which are neasted in a pool. But what happens with the sequenz flow? I do not mean a collapse sub-process.

Thanks a lot in advance James

Alexander Grosskopf Feb. 4, 2010, 10:12 p.m.

That's not possible. I can also not think of a desirable case. If a role is involved in a process and is worth mentioning in the model then put him in together with his contribution (performed activity).

Indeed, the sequence flow semantics would raise a lot of issues and make all our nice Model Verifikation tools cry:)

Sometimes, you might want to hide information to reduce complexity. Fair. I propose to decompose the processes and build a highlevel model to glue them together.

Hope that helps, alexG

Vlad Malik Aug. 13, 2010, 3:32 p.m.

Without an example, it's hard to be specific.

Consider how you want to represent that lane. You say it's not a subprocess, but it depends on what you consider a process. What the business will call a process is usually a very high level process, like "Processing claims". But that really is several sub-processes: "Receiving claims", "Investigating claims", etc, which may not necessarily even be sequential.

Say in your case a lane represents a participant. Well, often process participant can be thought to have their own process, which might be modeled separately. For instance, I go to a bank and the bank gives me money. My process (using the service) is separate from their process (delivering service). Sometimes, you might want for various reasons to split even a "process" performed by the same person.

Sometimes, you may want to split a lane into its own pool, creating an orchestrated "multi-pool" process. Here is a good short article on the subject: Link to Article. If so, you could collapse or exclude that pool if it is not relevant to a given discussion. But not all tools support multiple pools yet or understand their semantics.

Bruce Silver, in his book, also mentions something called "gray boxes", which means you represent only the relevant steps in a pool. Not sure how this would work or how tools (if any) implement this. Some tools link messages to pools, some to tasks within the pool. Some hide the message flow when you collapse the pool, some do not. Intalio for instance requires that the message link to a task in a pool, which I suppose requires you to have "gray boxes". It also hides message paths when you collapse the pool. I definitely think there are ways of doing it in theory, and this sort of flexibility around viewing the models is sorely needed. Not away of any tools that have matured to that point yet.

Please login to reply.