RE: Dubray paper comments + questions

> -----Original Message-----
> From: public-ws-chor-request@w3.org
> [mailto:public-ws-chor-request@w3.org]On Behalf Of bhaugen
> Sent: Wednesday, February 26, 2003 8:00 AM
> To: Jean-Jacques Dubray; public-ws-chor@w3.org
> Subject: Re: Dubray paper comments + questions
>
>
>
> Jean-Jacques Dubray wrote:
> > I think that what Assaf is expressing is that BPSS and the paper I
> > published on ebPML.org describe message exchange with no specific
> point
> > of view (from a party perspective), therefore appearing to be at the
> > "center".
>
> I understand, but think there are a number of interesting issues
> arising from this difference.
>
> > So instead of introducing this notion of "centralized sequence of
> > exchange" I think it is fair to say that BPSS offers a "neutral" view
> of
> > the choreography while BPEL/BPML and WSCI offers and "interface-bound"
> > or "party-specific" view of the collaboration.
>
> I agree with that formulation to some extent.
> But one could also describe the party-specific models
> as focusing on one party's internal procedural workflow
> and expecting the other party to conform.

It's the chicken and egg problem and neither spec is free of that.

In BPML you would describe the internal procedural workflow. But we are not
positioning BPML as a choreography language, we are positioning it as a
language to describe - in loose terms - the internal procedural workflow.

Then it's up to you to decide whether to begin with a choreography (in BPSS
or WSCI) and build a workflow, or whether to decide that the choreography
should support your existing workflow. You can make that decision then write
a BPSS to support your workflow, can't you?

It's really a matter of best practice, and I've heard at least once the
claim that since some system has an implementation the interface could be
expressed using BPSS in order to communicate with it, in effect binding the
choreography to the workflow.

Let's not confuse best practices with models.


> > In all cases, the goal is more or less to achieve state alignment
> > between collaborating parties.
>
> Is it?  Are the internal activities really aimed at state alignment
> between collaborating parties?

No. But where do you read about 'internal activities'?

arkin

> Yet another way to think about it is the difference between
> procedural and conversational workflow models -
> where procedural = command-and-control and
> conversational = collaboration.
>
> -Bob Haugen
>
>

Received on Wednesday, 26 February 2003 16:26:02 UTC