W3C home > Mailing lists > Public > www-ws-arch@w3.org > November 2002

RE: Proposed Draft Charter for Choreography WG

From: Stephen White <swhite@SeeBeyond.com>
Date: Fri, 8 Nov 2002 16:01:52 -0800
Message-ID: <C513FB68F8200244B570543EF3FC65370B0ED4AD@MAIL1.stc.com>
To: www-ws-arch@w3.org
One of the ideas that BPMI is trying to get across is that there exists a
need for interoperability at two levels: at the level of the machines
running the process and the level of the humans that will design and manage
the processes. The XML languages that provide the interoperability at the
machine level (e.g., BPEL4WS and BPML) are "programming languages" that will
not serve well to be visualized for the management interoperability level. A
one-to-one mapping of the execution language constructs would produce a very
confusing notation. Also, the formal structure of these lower-level
languages would provide for a disjointed configuration of process
information when dealing with complex models, such as a set of complex
loops. The complex models can be displayed in a single flow-chart model
level, but will appear as multiple levels (even multiple files for BPEL4WS)
at the lower level. Such configurations are fine for machines, but would be
difficult for a human to understand at dev-time or run-time.
I am considering the abstract specifications (WSCI, abstract BPEL4WS) and
executable specifications (BPML and BPEL4WS) to be basically equivalent in
this context. Both types of specifications model process behavior the same
way and the main difference between them is the formal way that they do or
do not handle process data. This difference will not typically show up in
the graphics of a flow-chart notation that a business analyst would use.
There is a long history of business processes being modeled by business
analysts with simple flow-chart notations (even simpler than UML). There has
always been a technical gap between these models and the technology to
implement those processes. Business Process Modeling Notation (BPMN) is
designed to bridge that gap by providing a flow-chart notation and the
technical mappings to the XML executable languages. BPMN is also designed to
graphically handle both abstract (public, B2B) and executable (internal,
private) processes and will map to any current or future specification that
fits within these two types of processes.
A working draft (0.9) of BPMN will be available to the public very soon.
-----Original Message-----
From: Assaf Arkin [mailto:arkin@intalio.com]
Sent: Friday, November 08, 2002 1:39 PM
To: Champion, Mike; www-ws-arch@w3.org
Cc: Stephen White
Subject: RE: Proposed Draft Charter for Choreography WG
I'm intrigued by Edwin's comment "BPML learned the hard way that the
notation language was a very important aspect of the usability and therefore
could not be an after thought."  Can someone elaborate on that and what
BPML's experience suggests to us? 
We never imagined that people would be sitting down and writing XML
documents to define their services and processes.
Out of the two possibilities we looked at, English (or any other human
readable language) and visual notation, we preferred the visual notation as
the more useable means to expressing processes, and so we started working on
Business Process Modeling Notation (BPMN) alongside with BPML.
BPMN provides the visual notation that talks to normal people at a language
they can understand, which is then transformed to BPML (or BPEL or WSCI, or
any other XML language) to produce a document that is applicable
specifically to the software that has to operate on it.
UML provides a good foundation for modeling, but as most vendors have found
out, it order to use it for the definition of processes that software can
operate on, one must put some constraints on these definitions and add
additional stereotypes. BPMN achieves just that. It starts with a familiar
visual notation, but describes the precise semantics that would allow a
piece of software to operate on that definition and provides transformations
to other XML languages.
I have forwarded this e-mail to Stephen White who chairs the BPMN working
group, so he can provide more detailed information.
You may want to take a look at the BPMN working group's Web site:
Received on Friday, 8 November 2002 19:04:03 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:41:00 UTC