RE: [wsbpel-implement] Portability Definition -- informal

["Ugo Corda" <UCorda@SeeBeyond.com>]

>There are some open questions here. Potentially, when we speak of observable message traffic,
>we can mean "abstract" WSDL messages, or we can mean binding-level stuff, encoded for
>shipment on the wire. It really depends on how much of the WSDL we want to pay attention to.
>By worrying about binding-level issues, we are treading on the ground covered  by
>interoperability, but are staying closer to the "intuitive" definition of process portability,
>where we'd like to be able to swap BPEL implementions without disturbing the rest of the
>system. I think we touched on this overlap in the call today.

 

This is an interesting point. Since a BPEL process is defined independently of any binding, it seems that portability should also abstract from the particular bindings used in a deployed process. In principle the same process should behave the same regardless of the fact that one deployment binds to SOAP over HTTP and another one binds directly over JMS/MQSeries.

 

Or we could take an approach similar to WS-I, and say that at least for the first "version" of our portability concerns we assume that all bindings are done to SOAP over HTTP.

 

I am not sure which major problems we would encounter if we tried to pursue the former type of portability definition. Has anybody gone through that experimentation already?

 

Ugo