RE: [ws-caf] Mt Everest and WS-CF

["Furniss, Peter" <Peter.Furniss@choreology.com>]

Doug,

Not everything, but more than just WS-Context.

What you are describing seems to me to be a particular use of
WS-Context, with its own set of rules, context-type and with (as you
suggest) some implications on propagation. The rules of this one are
that any one receiving the context can add anything it likes to it, can
process or remove anything it understands in it, but must either update
the base (if by reference) or propagate (if by value) the context with
these modifications, and containing exactly anything not understood. And
there have to be some definition of "related messages" for propagation -
other than responses, I'm not sure it would work by-value, unless
parallel invocation was forbidden.

An implementation has to understand all that, though since the rules
allow for opaque child elements, it doesn't have to understand
everything.

Peter


> -----Original Message-----
> From: Doug Bunting [mailto:Doug.Bunting@Sun.COM]
> Sent: 27 May 2004 19:02
> To: ws-caf
> Subject: Re: [ws-caf] Mt Everest and WS-CF
> 
> 
> Peter,
> 
> You seem to be arguing from the position that all who see the
> WS-Context 
> structure must understand everything in that structure.
> 
> Instead, the WS-Context structure is a shared information
> space into which 
> either party may put information it considers important (and 
> allowed to be 
> shared) about the current activity.  All participants may use 
> and reuse 
> some subset of the shared information.  However, it is very 
> important that 
> those who add to the context (or tie internal information to 
> the context 
> identifier) see their information later in related 
> messages[1].  A subset 
> of those who receive a particular context may understand the 
> dependent 
> specification or application semantic responsible for its 
> content but all 
> should include it in related messages.
> 
> thanx,
> 	doug
> 
> [1] Yes, "related messages" was somewhat ill-defined the last time I
> looked.  The base set of messages understood to fall within the same 
> activity (perhaps just direct responses) should likely be 
> described in the 
> WS-Context specification and extended, as needed, in 
> dependent specifications.
> 
> On 27-May-04 10:28, Furniss, Peter wrote:
> 
> > Jim,
> > 
> > 
> >>The point is, I want it to be understood and a standard helps that. 
> >>Not at the SOAP level where the mechanics of mustUnderstand help, 
> >>but by the _people_ building the service. I can get them to support 
> >>a standard (hey, they can reuse the work with their other partners) 
> >>but I can't get them to use my own ad-hoc structure.
> > 
> > 
> > But you can get them to understand your own ad-hoc overload of the
> > ws-context identifier ?
> > 
> > Do you accept that they will have to implement something
> you define ?
> > You seem to think WS-Context alone will do it, but it has no
> > semantics.
> > 
> > Peter
> 
>