RE: i0001: EPRs as identifiers (why XML?)

Hey Steve,

> WSDL is designed for extensibility and allows for multiple 
> bindings. It can be used to provide a service contract 
> language that's independent of the underlying protocol, 
> transport, and transfer syntax. We successfully use this 
> extensibility in production today with a number of customers. 
> Whether you like the fact that WSDL is extensible is neither 
> here nor there -- it's in the spec.

Right, and I don't disagree with the things you do with WSDL (and even
WS-IF :-) - they make sense in that domain.

> Why should WS-Addressing take it upon itself to undo the 
> abstractions provided by the existing WSDL specification by 
> supporting only a single protocol?

I'm not sure I'm asking WS-Addressing to harm WSDL in any way. In my
mind while there is useful overlap between SOAP and WS-Addressing,
introducing WSDL into the picture is a red herring - services exist
independent of whether or not there is any WSDL description of them (and
in some cases in spite of!).

Actually while I'm on the subject: why does WSDL keep cropping up in
this WG? Addresses and interfaces are chalk and cheese.

> You also seem to assume that a service is reachable over only 
> a single protocol.
> Why?

No, I don't think that at all, though I do like the notion of everything
using the same transfer mechanism (SOAP + WS-Addressing). All I am
saying is that SOAP doesn't have a standard addressing element as part
of the message format. Other protocols do have such elements and
therefore probably don't need this augmentation.

Perhaps I wasn't clear about this: to me the EPR stuff in WS-Addressing
is the least interesting bit. The ability to have a standard address to
format that we can all exchange is so, well, unexciting. The sexy bit of
WS-Addressing is that it makes SOAP transport neutral (yay!).

Jim

Received on Wednesday, 24 November 2004 04:23:48 UTC