- From: Gary Brown <gary@enigmatec.net>
- Date: Tue, 13 Jul 2004 19:28:00 +0100
- To: "Monica J. Martin" <Monica.Martin@Sun.COM>
- Cc: <public-ws-chor@w3.org>, <steve@enigmatec.net>
> mm1: The conversation continues, on the 'perform' evaluation on > choreography composition. These questions and issues you raise touch on > my previous questions about our recognition of, visibility to, and > dependencies evident in choreography and sub-choreographies. We > discussed this in the F2F in Cannes where we only acknowledged the outer > not the inner box [1]. In your proposed solution, I have two questions: > > * You bind the roles in the relationship between the inner and outer > choreographies. Therefore, if bound in a relationship over a > channel, the partner cannot change roles in the sub-choreography. > This would require a different composed choreography if the buyer > assumes another role in the inner (sub-)choreography correct? gb: This depends on whether the new role is related to the interface with the outer choreography - and/or whether the bound relationship's interactions are affected by the change. If the sub-choreography is simply being enhanced to provide more functionality, but the outer choreography still only requires its existing functionality from the sub-choreography, then everything should still be ok - hope this is answering your question, if not we can discuss on the conf call. > * Payment Process and Goods Distribution are separate transactions > that relate to exchanges that may be bounded by business > agreements (thus separating fulfillment responsibilities). Can we > provide traceability and error functions when this compositional > capability occurs? Perhaps this will be discussed in Tony's > submission. I think these boundaries would partially be based on how the sub-choreography was defined. If the sub-choreo includes the notion of transaction support between the participant (that will be bound with the outer choreo) and the other participants - then I guess the outer choreo could extend this transactional boundary in whatever ways it requires. Something to think about with Tony's transaction submission...... > > Thanks and I look forward to the discussion. > > [1] I think we discussed blue and red boxes but I don't remember which > was inner or outer. > > >
Received on Tuesday, 13 July 2004 14:28:07 UTC