RE: Bindings : a quick cut at an abstract definition

Hi Highland!

> I have compared Glen's abstract approach with my "proposed sections"
> approach.  It seems that my detailed sections are what is 
> missing from 3, at
> least in part.

Right.  See comment below under #3.

> >>2. A specification of the binding's "private" facets. 
> 
> An example would be a specific use of MDN's as receipt 
> acknowledgements,
> which could include SOAP content as part of the returning MDN.  
> Am I correct in assuming that unique usages/behaviors of 
> features listed
> under (3) would be captured here, as well?

Only inasmuch as the usage is really limited to this particular binding, and
not expected to be either implemented by or referred to by other
extensions/bindings.

> >>3. A specification of the binding's "public" SOAP facets.
> 
> MEP - Message Exchange Patterns Supported
> Message Addressing
> Message Correlation
> SOAP Encapsulation - MIME, etc
> Compression
> Security 
> 
> Reliable Messaging
> 	Sending Message Behavior
> 	Receiving Message Behavior
> 	Generating an Acknowledgement Message
> 	Resending Lost Messages and Duplicate Filtering 
> 	Duplicate Message Handling
> 	Failed Message Delivery
> 	Message Order Preservation
> 	Message Delivery/Failure Notification
> 	Split Messages
> 
> Error Reporting and Handling
> 	Error Types (Bound Protocol Delivery) and Handling
> 	SOAP Fault Types and Handling
> 	Reporting Errors

Yup - the key here is that you do not actually have to talk about ALL these
things in your particular binding, and in fact we may not (and hopefully
will not) even define all these things in the context of the WG's output.
The binding framework gives you a place to hang connections between your
binding and specified facets such as the ones you mention in your list.  We
as a WG take on defining a very small number of these facets, and a very
small number of bindings, and then let other groups or communities do the
rest.

> To map this to my binding outline, am I correct in making 
> these comparisons:
> 
> 1)Named components are analogous to features (like security, 
> compression,
> correlation)
> 2)Pieces of data would be the meta data or infoset data 
> associated with
> these features
> 3)Communication patterns are message exchange patterns (?)

Seems right.

> Thoughts?  Shall we discuss this at today's TBTF?

I hope so!

Thanks,
--Glen

Received on Monday, 30 July 2001 13:17:58 UTC