RE: [ebxml-cppa] BPSS to WSDL mapping

[Cory Casanave <cory-c@enterprise-component.com>]

Jean-Jacques,
Yes and no.

You can import WSDL into a BPSS specification, (note that you will loose the
endpoint - these would have to go into the CPP).  The WSDL generated BPSS
would, of course, not be as rich as a full BPSS but it does define a service
that is equivalent to the WSDL.  This "reverse engineering" is useful for
adapting to existing systems (Yes, WSDL just became legacy!).  It is also
useful as a starting point to create a more expressive BPSS.  

And (Switching directions), while it is true that WSDL produced from BPSS
would not have choreography (for example), so what!  That is not the job of
WSDL.  The BPSS semantics specify this so why reproduce it in WSDL?  WSDL is
low-level endpoint semantics.  You say that we will get this when we have
WSFL - but BPSS is already filling that role, we don't need yet another way
to say the same thing (We will probably get it anyway - so W3C can invent it
:).  You can produce WSDL+WSFL from a BPSS just like you can produce the
WSDL.

We do mappings between lots of technologies, the idea is to map from as high
a level model as you can and produce the set of specifications, code (or
whatever) that captures those semantics.  You don't need to map to just one
thing and you don't need every target to capture every semantic (which is
the job of the higher level model).  So the higher level model should be
considered a constraint on whatever is behind the WSDL (What we call the XML
component).

If WSDL is someday extended to have the BPSS semantics, then we don't need
BPSS.  It is the semantics, not the representation that is essential.  We
can map representations but can't invent semantics.

Cory

> -----Original Message-----
> From:	Jean-Jacques Dubray [SMTP:jjd@eigner.com]
> Sent:	Tuesday, March 12, 2002 7:45 AM
> To:	'bhaugen'; 'OASIS ebxml-cppa'; ebtwg-bps@lists.ebtwg.org
> Subject:	RE: [ebxml-cppa] BPSS to WSDL mapping
> 
> WSDL does not have the appropriate semantics to map to BPSS as Bob
> Haugen explained in this email thread. However, you can certainly create
> a number of operations that will support the BPSS protocol such that you
> can run on top of a web service infrastructure. But again, the set of
> WSDL specification created is not enough to map isomorphically to a BPSS
> definition. Just by looking at the WSDL produced, you would still lack
> the ability to enforce a particular sequence of invocation (at least
> until WSFL is ready), and also lack the ability to unambiguously declare
> that you have reached a given business state when a particular operation
> is invoked. 
> 
> Hope that helps.
> 
> Jean-Jacques Dubray____________________
>  
> 
> 
> >>-----Original Message-----
> >>From: bhaugen [mailto:linkage@interaccess.com]
> >>Sent: Tuesday, March 12, 2002 7:28 AM
> >>To: OASIS ebxml-cppa; ebtwg-bps@lists.ebtwg.org
> >>Subject: Re: [ebxml-cppa] BPSS to WSDL mapping
> >>
> >>Some people have touched on this angle, but
> >>just to make it explicit:  there is a big difference
> >>between one-shot messaging or RPC use cases
> >>for Web services on the one hand, and longer
> >>business conversations on the other.
> >>
> >>Most of the Web services gurus I know of
> >>understand that there are problems with long
> >>conversations, although their solutions vary
> >>from replacing HTTP (Don Box) to an explicit
> >>model for long conversations that works over
> >>many mechanisms (ebXML).
> >>
> >>One problem with long B2B conversations
> >>is business state alignment.  For example,
> >>was that offer accepted or rejected?  Was
> >>that order fulfilled correctly?  Did you
> >>receive that payment?  Is the claim settled?
> >>Etc.
> >>
> >>So you are building a business protocol
> >>stack over the technical protocol stack.
> >>The business protocol stack starts with
> >>the business transactions (offer-acceptance,
> >>notify-confirm, etc.) and builds other
> >>business protocols like commitment-
> >>fulfillment and claim-settlement on
> >>top of them.
> >>
> >>WSDL is a puny mechanism for the
> >>business conversation protocols.
> >>
> >>BPSS is a good start in the correct
> >>direction, in my biased opinion.
> >>
> >>-Bob Haugen
> >>
> >>P.S. I think the conversations apply
> >>to B2C as well as B2B - don't you
> >>want your order to be fulfilled?
> >>But the B2C people have worked out
> >>a set of patterns using standard HTTP
> >>methods that seem to be approaching
> >>defacto standard status.
> >>
> >>
> >>----------------------------------------------------------------
> >>To subscribe or unsubscribe from this elist use the subscription
> >>manager: <http://lists.ebtwg.org/ob/adm.pl>
> 
> 
> 
> ----------------------------------------------------------------
> To subscribe or unsubscribe from this elist use the subscription
> manager: <http://lists.ebtwg.org/ob/adm.pl>