RE: Updated portTypeExtension proposal

Hi Pete,

Answers ( hopefully ) inline

Gudge

> -----Original Message-----
> From: Peter Vanderbilt [mailto:pv@nas.nasa.gov] 
> Sent: 09 October 2002 23:57
> To: www-ws-desc@w3.org
> Subject: Re: Updated portTypeExtension proposal
> 
> 
> 
> What's the current thinking about the following questions 
> related to portType inheritance:
> 
> Assume portType C derives from portTypes A and B.  What 
> appears as wsdl:port elements in wsdl:service?  Is there one 
> port element or three?

You could list a port address and binding to C ( which would also
include A and B )

You could list a port address and binding to C along with different
addresses ( and bindings ) for A and B. In this case A and B would both
still be available through the 'C' binding on the 'C' port.

> 
> If one, is its binding structured in a way that allows 
> independent binding mechanisms for A and B?  Different 
> addresses?  Can a client that operates on services containing 
> (wsdl:) port A (and doesn't know about C) operate on a 
> service containing port C?  In other words, is each C port 
> also an A port?

Yes, I think a C port is also an A port.

> 
> If the example above results in three ports, how are the 
> operations allocated to the ports?  

One port.

> Presumably there are 
> ports for A and B separately.  Does the port for C reference 
> operation bindings for just those directly defined in 
> portType C?  Or does it include operation bindings for A and 
> B?  Is it legal for a service to have a C port without both A 
> and B ports?

With the binding stuff, as it stands today, a binding for C would
typically bind the operations of C and those of A and B. It is also
possible to define a binding which ONLY binds the operations of C. Or
only those of A and/or B for that matter. It seems that currently the
binding model allows all the combinations.

Received on Wednesday, 9 October 2002 19:09:59 UTC