Geir, I absolutely agree with your point that this would be a stronger
community and a better software stack if we could bring all together
into one project. And that was exactly the point of my initial
proposal. I was just expressing the fact that a web-service (WSDL and
XML messaging) based engine may have different constraints than more
"traditional" state-based workflow engines as it must be rather
tightly coupled with the WS layer (I think that was Sanjivas's point).
But here I'm talking about communication (interface) constraints, not
purely execution constraints (nowadays very few specifications don't
rely on WSDL anyway).

Now in terms of execution, no standard cover all workflow patterns as
defined by Pr. Wil van der Aalst. However a robust implementation
should support most of them. Many "pure web service" oriented
specifications do not consider workitems either, even it's often very
useful for end users. I think that's the sort of things you have in
mind Geir. Am I wrong ?

An implementation that wouldn't accept extension would of course be a
very bad one. Supporting BPEL was just a first goal when we
implemented Twister (we actually implemented a worklist) but any
robust product should be able to evoluate to accept other
specifications for both orchestration and choreography as well as
others "nice to have" features not directly covered by WS-*
specifications.

Speaking about WS-CDL, this specification covers choreography which is
different from orchestration (sometimes opposed as private and public
processes). BPEL tried to address both calling them executable process
and abstract process. But as the specification is strongly oriented on
executable (as in bpEl) it somewhat failed to properly address
abstract ones. I think the Oasis TC is even talking about dropping
that part (Sanjiva?). So WS-CDL now covers this ground and does it
quite well.

So in short (finally), I believe it's possible to have an
implementation that would be definitely web services oriented AND that
would support most BPM use cases. The first relates to communication,
the second to execution model which are two different things.


On Thu, 21 Oct 2004 23:59:48 -0700, Geir Magnusson Jr
<[EMAIL PROTECTED]> wrote:
> 
> On Oct 21, 2004, at 10:41 PM, Uijin Hong (íìì) wrote:
> 
> >
> >
> > I hope to see Agila(or Apache-BPM-engine)
> > could run both BPEL and WS-CDL.
> 
> +1!
> 
> --
> 
> 
> Geir Magnusson Jr                                  +1-203-665-6437
> [EMAIL PROTECTED]
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
> 
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to