The Service Interface

["Jeffrey A. Estefan" <jeffrey.a.estefan@jpl.nasa.gov>]

Folks,

I'm in the process of crafting a UML component diagram for the Service 
Description artifact, but we need to nail down the notion of Service 
Interface before I proceed.  I don't want to indulge into endless debate 
here, I just want to know what is the minimum set of service 
implementation-independent things needed in order for a consumer agent to 
communicate with a provider agent (i.e., service).

I happen to like Wikipedia's general definition of interface which states 
that "An interface defines the communication boundary between two entities, 
such as a piece of software, a hardware device, or a user."  Of course for 
the RA, we're talking about software interfaces that exist between separate 
software components that provide a mechanism by which the components 
communicate (consumer agents and provider agents/services) and not user 
interfaces per se.  We should perhaps extend that notion to physical 
interfaces as well, i.e., interfaces between hardware components because 
there is an evolving world of things like XML gateways out there.

The WSA defines a service interface as follows:  "A service interface is the 
abstract boundary that a service exposes. It defines the types of messages 
and the message exchange patterns that are involved in interacting with the 
service, together with any conditions implied by those messages."  In terms 
of relationships to other elements, it goes on to say a service interface 
defines "the messages relevant to the service."  The explanation states "A 
service interface defines the different types of messages that a service 
sends and receives, along with the message exchange patterns that may be 
used."

So is only a subset of the Information Model and Behavior Model enough, 
i.e., messages and MEPs?  What about service bindings (), endpoints, 
operations, faults, etc.  Do they have a role in the service interface.  In 
other words, are these examples of service implementation-independent things 
that are necessary the consumer agent to communicate with the provider 
agent/service?  In terms of the RM and RA, are there elements of Service 
Reachability at play?

Whatever we agree on, we should be able to map to concrete examples (just to 
check our own sanity of course and not to include in the actual RA spec.) 
say one from the Web services world using WSDL 1.x and 2.0, and perhaps one 
from the CORBA IDL world.  Note that in the WSDL 2.0 world, the service 
interface defines what abstract "functionality" a Web service provides and 
the binding describes how to access the service.   In the RM and RA, we 
currently model Service Functionality separate from Service Interface.

Interested in your thoughts.  Pragmatic ones, please!

Cheers...

 - Jeff E., JPL