Choreography semantics and WSDL

Hi Steve,

Near the end of today's meeting we had a conversation about semantics, and I proposed that choreography relied, at least in part, on semantics expressed at the WSDL level. You objected to that idea using the argument that a choreography could be defined at a time when the Web services that would use it do not exist yet, and the choreography itself would actually be used to define them.

I think that argument brings up the usual confusion between abstract WSDL and concrete WSDL. If choreography relies on WSDL (which, I know, has not been decided yet), then abstract WSDL interfaces would need to be defined at the time a particular choreography is being developed (by definition), regardless on whether the associated Web services exist or not at that time. Additionally, if a mechanism is available to associate semantics with WSDL abstract interfaces, then choreography would also be able to leverage the same mechanism (even though additional semantics information might be required which goes beyond the semantics of the individual Web service). When the same choreography is used to actually build concrete Web services, then the semantics associated with the WSDL abstract interfaces would be used to determine the semantics of the concrete Web services being implemented.

So I think the real objection should relate to the case where a choreography is not expected to rely on abstract WSDL (which is still being debated), not to the case where a choreography is being developed in the absence of concrete Web services.

Ugo

Received on Tuesday, 2 December 2003 19:50:07 UTC