Re: [Fwd: FW: [ebxml-bp] Proposal for correlation mechanism in ebBP / WSDL mapping]

[David RR Webber <david@drrw.info>]

Hmmm,

I just wondering if this is the same 'use case' as a master BPSS calling child
BPSS processes.

In that case - its just a BTA step in the overall BPM.

However in the web service case - I'm seeing it could be interaction within a
BTA.

In either case - the key thing is that the outcomes are deterministic - so we
have the normal BPSS behaviour of known outcomes, failure/success conditions as
a minimum, along with time-to-perform.  Obviously we can also build some
bespoke beginsWhen / endsWhen checks to ensure meaningful outcomes too.

So long as the invoked component has a clear way of supporting these - and we
can reference those - a non-normative technical note on BPSS-binding comes to
mind - its seems that we can accommodate them.

That's my initial thoughts on this.

Thanks, DW


Quoting "Monica J. Martin" <Monica.Martin@Sun.COM>:

> Forwarded for Rajesh who has had problems sending to our list. JJ Dubray 
> is on vacation until next Tuesday, so we may have a delayed response. 
> Other responses from the team encouraged. Thanks.
> 
> > -----Original Message-----
> > *From:* Rajesh Pradhan
> > *Sent:* Monday, August 02, 2004 2:50 PM
> > *To:* 'ebXML BP'
> > *Cc:* 'Jean-Jacques Dubray'
> > *Subject:* RE: [ebxml-bp] Proposal for correlation mechanism in ebBP / 
> > WSDL mapping
> > *Importance:* High
> >
> > Jean-Jacques,
> >  
> > My apologies for piping up this late in the mapping process, but I 
> > have a few points apropos the context :) ..
> >  
> > I agree that WS CTXT  is the best of the lot. However , is making  it 
> > normative the best way to proceed ?There exist other mechanisms like 
> > WS-Addressing ( I'm not sure if they have licensing restrictions ) . 
> > There are also other cases where opaque correlation sets might make 
> > more sense ( correlation sets managed by the engines with just their 
> > mechanism schemas referenced in the correlation set definition ) . 
> >  
> > Is there a way by which we can allow the CSets to have different 
> > schemes ? The biggest advantage I see is that if the correlation 
> > scheme is already defined, then the modeler does not need to define 
> > one whilst mapping.
> >  
> >  
> > Just my two bits,
> >  
> > Thanks,
> > Rajesh.
> 
> 
> 
> 
> 


http://drrw.net