In response to my previous blog post, Sebastian Stein identifies the areas of improvement for BPMN 2.0. One very important area he has identified is execution semantics besides the well known lack of exchange format. And I would like to reassure the wider BPM community that the same is a very important target for the BPMN 2.0 RFP and consequently for the submissions. The approach 1 as outlined and characterized by Sebastian and Bruce attempts to address both of these concerns using the inherent metamodel of BPMN 1.0 and 1.1. There are some great facets of the second submission as well, and there are merger talks in progress.
Going back to what is (or should be) good about BPMN 2.0:
Users get the flexibility to "underspecify" a business activity, e.g. as a business user I know that "Create Order" step is to be followed by an "Audit Step". I may not know exactly how these steps will be executed. If I know it or if there is an IT specialist refining the model, he can specify the execution details of each of these steps.
The execution semantics is an area of focus and ironing out the ambiguous IOR gateways and resulting patterns is high on the agenda. This is also an area for learning from user experience. e.g. in this diagram - how many times does a business user expect the activity "Step X" to be executed?
You will get a different answer from different people. How about the case when the activity Step X is after another similar construct - how many times will the activity X be executed in that scenario.
Stay tuned for the next blog post where I will discuss token flow semantics.