Re: Choreography and REST

Bob,

I don't disagree. However, the pre/post condition semantics need to
(somehow)
be expressed. REST itself doesn't address this. Where would we find the
declarative pre/post
conditions expressed not only for a resource, but for a class of resources?
for classes
of related resources?

What this ends up looking like is a distributed Makefile and how many
people (especially
business types!) can read, understand and internalize a Makefile? In my
expefrience,
very few.

Personally, I like declarative languages, I think it is a superior
approach.
But, most people have a *really* hard time with them. XSLT is a good
example.
Most people I know throw their hands up in despair when faced with having
to write, or worse debug, an XSLT stylesheet:)

Cheers,

Christopher Ferris
Architect, Emerging e-business Industry Architecture
email: chrisfer@us.ibm.com
phone: +1 508 234 3624


                                                                                                                                  
                      bhaugen                                                                                                     
                      <linkage@interacc        To:       www-ws-arch@w3.org                                                       
                      ess.com>                 cc:                                                                                
                      Sent by:                 Subject:  Re: Choreography and REST                                                
                      www-ws-arch-reque                                                                                           
                      st@w3.org                                                                                                   
                                                                                                                                  
                                                                                                                                  
                      08/09/2002 12:32                                                                                            
                      PM                                                                                                          
                                                                                                                                  
                                                                                                                                  




Christopher Ferris wrote:

>Paul's analysis doesn't take into account the fact that
> there are real business issues to be considered w/r/t
> the order of things that need to be understood
> *and agreed upon*, a priori to beginning an exchange
> of messages.

Chris,

I agree with every word of your paragraph,
but Paul's articles have caused me to
rethink how I might satisfy the business
requirements.

As you know, I'm coming from the ebXML
Business Process group where we put
a lot of emphasis on choreography.

But since then, as we got deeper into the problem
(in UN/CEFACT eBTWG), it became more clear
that choreography decisions are business decisions
(or decisions of whatever is the domain of discourse)
and not just programming logic e.g. forks and joins
and state transitions.

So we started to use business object states
(from a semantic business model) as the basis
for choreography expressed as pre and post
conditions for document exchanges.

Now I'm opening my mind up to the direction
Paul Prescod is proposing and thinking that
it fits a lot with where we went in eBTWG, if you
think of the business objects as REST
resources.

Another way to think of it is as speech acts
in a structured conversation.  One speech
act, a commitment, promises a delivery
of goods or services; a reciprocal speech
act promises payment.  Subsequent
speech acts notify of the delivery and
payment events.

Each speech act becomes a REST
resource, which can then be used
as a posting address or condition for
subsequent speech acts.

There can be published rules (other
REST resources), and there is a kind
of choreography embedded in the
speech acts.  But it's very different
from the procedural workflow style
of choreography.

Think about it.  I don't have it all worked
out in my mind yet, and don't know if
Paul has either, but don't dismiss it
out of hand.

(P.S. I don't know if my speech-act
scenario is anything near what Paul
has in mind.)

-Bob Haugen

Received on Friday, 9 August 2002 14:20:15 UTC