- From: Assaf Arkin <arkin@intalio.com>
- Date: Wed, 23 Apr 2003 17:09:19 -0700
- To: "Cummins, Fred A" <fred.cummins@eds.com>
- CC: "Monica J. Martin" <monica.martin@sun.com>, public-ws-chor@w3.org
> > >This seems like a rather imprecise definition. How do you >expect to use "service type?" If it is used within a choreography >specification, I would expect it to refer to a WSDL interface, or >something more specific, such as a real estate seller using a >seller interface. > > >Fred > I agree. I think the choreography specification should refer to WSDL interfaces. But that's just a means to achieve some goal, I don't think it describes the goal. Let's say that we only care about using WSDL artifacts when expressing choreography. WSDL has interfaces so a choreography could use WSDL interface if nothing else than to justify their existence. It could also use WSDL services directly. Why one and not the other? If you want to define a choreography once and then use that definition with any combination of applicable services you need to define it not based on specific services, but rather in terms of more generic service types. Using WSDL interfaces is in my opinion the only interesting way to achieve this goal. But if my goal was to write the choreography in terms of services, not service types, I would probably not be using WSDL interfaces. What I need is that generic concept that allows me to rationalize why the choreography should be expressed using WSDL interfaces. The term role is an alternative, but in my opinion overloaded. It also makes it easier to write use cases using this more generic term. As a buyer my business requirement is to be able to use any entity offering a service that implements the generic seller service type. Using WSDL interfaces is technically interesting but not a business objective. arkin
Received on Wednesday, 23 April 2003 22:14:46 UTC