RE: FW: Action Item 2004-07-01 Solution to 168/R114


> AIUI an agent now has to examine all of the operations in an 
> interface, divine which of them accept the same input 
> message, and anticipate any one of a series of actions 
> occurring and output messages in response. Hmmm.

A consumer of a service has to understand the contract it will be bound
by, yes. One always reads a whole contract no?

> So if it were possible to describe an operation as accepting 
> InMessage and returning either OutMessage or OtherOutMessage, 
> wouldn't that be good enough?

Sure, I'm not too fussed about individual MEP, the WG will define those
that it sees as being fundamental. I guess the people who come from
messaging vendors or folks like yourself with real world consultancy
practice will have much to say on that.

> What happens if OtherOutMessage is from the second operation 
> that has a different MEP? The sender may be expected to 
> process and send additional messages, but as the operation, 
> and therefore the MEP is unclear.
> It doesn't know if it's dancing to "Strauss" or "The Pogues".

I'm not sure I follow this. Whatever the consumer receives has been set
out in its contract or the contract of the service. Once the message
arrives its content and structure can be used to do the right thing -
the right thing will be helped by the fact that the consumer's state
machine gets hints from the MEPS in the WSDL.


Received on Thursday, 15 July 2004 08:09:51 UTC