Thanks David:
<Quote>
This then allows you to configure a catalogue of business process
components into which you put - CPA, Context, BPSS and CAM templates -
and when discovery occurs - the new partner can configure the Context
and load the templates into their software and start using it.
</Quote>
I understand that this configuration would be a design-time, not a
run-time - is that so? If so, how can such configuration take place at
run-time? How can a discovering service "know" at run-time how the
database the underlies the service should be mapped (table/field level)
to the interface definitions of the newly discovered service? Is this
even possible in April 2004?
Joe
David RR Webber wrote:
>
> Joe,
>
> The later - dynamic interface configuration - is the domain
> of CAM - acting as a service of registry.
>
> We can do one part of this already. Given a context
> configuration - you apply your context values - and
> CAM will adjust the interchange content accordingly.
>
> Example - if you are a supplier in Canada to GM in
> Detroit - you need to provide cross-border shipment
> information - while if you are in Michigan - you don't.
>
> That's the point of the work on Context for BPSS
> V2.0 that we have in place.
>
> This then allows you to configure a catalogue of
> business process components into which you put -
> CPA, Context, BPSS and CAM templates - and
> when discovery occurs - the new partner can
> configure the Context and load the templates into
> their software and start using it.
>
> The second part of your scenario - automatic
> interchange linkage is also facilitated by CAM
> templates - and here is where the UID mechanism
> is vital. This allows you to have agent software
> that matches UIDs in target and source nouns.
> Then there are the noun definitions themselves -
> we're working on that in SCM at the moment.
> That is the other piece of the puzzle needed by
> CAM and the agent - to be able to automatically
> pull the facet and predicate information for a
> given UID of a noun - and then apply those
> rules to the mapping. The OWL work in SCM
> is also very important to be able to classify and
> relate like / equivalent / and similar parts so the
> software agent can determine that a UID from
> one domain is actually related to a UID from
> another.
>
> The peices are there - what is needed as ever
> is projects and funding to validate it and complete
> the detail implementation.
>
> DW
>
> ----- Original Message -----
> From: "Chiusano Joseph" <>
> To: <>
> Sent: Tuesday, April 20, 2004 2:29 PM
> Subject: [regrep] Dynamic Discovery/Binding/Invocation
>
> > The following question might sound silly coming from someone who knows
> > our spec so well, but it's a usage scenario that I just had to begin to
> > think about.
> >
> > Is anyone using ebXML Registry for dynamic discovery/binding/invocation
> > of Web Services? More specifically:
> >
> > - A Web Service would dynamically discover another Web Service;
> >
> > - It would use the discovered Web Service's interface definitions (e.g.
> > WSDL document) to dynamically construct message definitions that conform
> > to the discovered Web Service's interface definitions;
> >
> > - It would then map its data to these messages definitions and invoke
> > the discovered Web Service;
> >
> > If no one is using an ebXML Registry this way, how close are we to this?
> > It seems to me that the "dynamically constructed messages definitions",
> > and - even more so - the dynamic mapping, would be challenging. Does
> > this fall more under semantic technologies? If so, will it ever be
> > possible?
> >
> > Thanks,
> > Joe
> > --
> > Kind Regards,
> > Joseph Chiusano
> > Associate
> > Booz | Allen | Hamilton
> >
> > To unsubscribe from this mailing list (and be removed from the roster of
> the OASIS TC), go to
> http://www.oasis-open.org/apps/org/workgroup/regrep/members/leave_workgroup.php.
> >
> >
--
Kind Regards,
Joseph Chiusano
Associate
Booz | Allen | Hamilton
