One of the main reasons I went to OMG Think Tank was to hear a detailed discussion of the two BPMN 2.0 proposals on the table, and the roadmap for adoption. But guess what? Not a single word about it on the agenda. Unbelievable. Heckuva job, Program Committee!
Nevertheless I was able to get a feel for where things stand by talking to some of the protagonists. To review, there are two competing submissions. The one from IBM, SAP, Oracle and others is basically BPMN 1.1 with a defined XML serialization, formal execution semantics, a few new event types, expanded data modeling, and a brand new choreography model that nobody understands (or cares about). The other one, from OMG's MDA old guard, is BPDM, a formal metamodel for any process modeling language, with (in theory) a mapping to BPMN as one such. I admit I can't make heads or tails of it myself. There is also an XML serialization of sorts.
The September submissions from both camps were incomplete. The date for the final complete versions is now November, to be discussed at the OMG technical meeting in Santa Clara in December, and both groups are planning to submit. Attempts to reconcile the two approaches have not gone too well, but the groups do not want a competitive vote, i.e. choose A or B. So the plan (or maybe, "a" plan) is to redirect BPDM to become a new Business Modeling Framework, a metamodel that interconnects the myriad of modeling standards in a formal way - process (BPMN), rules, organizational, value chain, etc etc. BPMN 2.0, i.e. the notation and XML, would follow the IBM et al proposal. And there would be a mapping between BPMN and the framework. So instead of 2 competing proposals, there would be 3 specs that in principle work together, not at cross purposes. I think the IBM et al group is happier about this plan than the other side, so we'll see if it happens or not. It's hard to see how all that gets done by November, in any case.
If it does, the best case scenario as I understand it would be a vote-to-have-a-vote at the December meeting and the vote itself at the March 2009 meeting, at which time it would move to finalization phase and tool vendors could "begin work." So the earliest we could expect to see tools compliant with a semi-finalized spec would be midyear 2009.