RE: UMM, modeling and alignment (was Example RN PIP 3A4)

[Jean-Jacques Dubray <jjdubray@exceloncorp.com>]

As Kurt pointed out RN and OAG do not distinguish intend by distinct message
formats. Hence it will be hard to come by with a realistic example that does
not use XPath predicates to identify the intent.

My intent in this string of email was just to point out the spec is indeed
open and support both models. IMHO, not using Xpath wherever possible is
preferable as intend change much less often than document formats.

Salute,

JJ-

-----Original Message-----
From: James Bryce Clark [mailto:jbc@lawyer.com]
Sent: Monday, May 07, 2001 3:52 PM
To: jj
Cc: kkanaskie@lucent.com; plevine@telcordia.com; ebxml-bp@lists.ebxml.org;
ebxml-core@lists.ebxml.org; ebxml-discussion@lists.ebxml.org;
ebXML-PoC@commerceone.com
Subject: UMM, modeling and alignment (was Example RN PIP 3A4)

JJ, your comment is an excellent and important point.

For now we have deliberately kept the BPSS object model open, with a
variety of alignment paths at the expense of simplicity.

XPath probably facilitates half-hearted partial adoption of ebXML business
processes by some, and may be misused by others.   The risk is that some
legacy users, or legacy-driven vendors, will make the minimal alignment and
them re-freeze themselves at a low level of interoperability.   I don't
think that is avoidable;  we must facilitate wide adoption.   XPath plays a
useful role there.   My hope is that the upgrade path to logically modelled
transactions will exert an upward pull on users, once they dip their feet
in the pool.

UMM/UML expressions of reusable transactions are the "bait" for this
upgrade path to greater interoperability.   I think that is why so many of
us strongly support ebXML's bias towards modelling and UMM.  That metamodel
is not perfect, but it's pretty good, reasonably vendor-neutral, in use in
commerce, and within reach.

After Vienna one of our continuing projects for discussion should be a set
of best practices for use of the specs, that nudge people in that "right"
direction.

Our job is to be the good guys in e-commerce.   After we crack open a few
bottles of Gewurztrammer this week, the next order of business is to blow a
bunch of attractive, reusable business processes and components into
repositories, make them discoverable through logical expression, and see if
the world bites.

Best regards    Jamie

- - -
James Bryce Clark
Spolin Silverman & Cohen LLP
310 586 2404    jbc@lawyer.com

At 07:55 AM 5/7/2001,  Jean-Jacques Dubray wrote:
>Kurt:
>
>Thanks for this great work.
>
>I would like to emphasize one point: we should avoid using XPath
> expression on the message document because this makes
>collaborations much less reusable across standards. I understand
>that this is not feasible in the case of RN but moving forward, a good
>design principle would be to use distinct document formats to
>express different intend. The binding at the format level would
>happen in the DocumentSpecification.
>
>This might look like a little step one way or the other, however,
>converging at the "business collaboration" level is even more
>important than converging at the format level. Formats can most
>often be mapped to each other, it can quickly become un-
>manageable to map and adapt business processes to a slew of
>nearly identical collaborations. Once a collaboration is in place an
>agreed upon by an industry or an eco-system, it is relatively simple to
>manage N different document formats flowing through the same
>infrastructure as they are nicely decoupled from each other.
>
>Jean-Jacques Dubray, Chief Architect
>eXcelon Corp.
>
>-----Original Message-----
>From: Kanaskie, Kurt A (Kurt) [mailto:kkanaskie@lucent.com]
>Sent: Friday, May 04, 2001 12:51 PM
>To: 'ebxml-bp@lists.ebxml.org'
>Cc: 'ebxml-ccbp-analysis (E-mail)'; 'ebXML-PoC'
>Subject: Example RN PIP 3A4 in BPSS
>
>Attached is an ebXML Business Process Specification XML
>document for RosettaNet PIP 3A4 version 1.4 that I did as a personal
>POC.    I think it is correct, but if someone finds a mistake or I left
>something out, please let me know. Some interesting points:
> * Use of Packages to represent Cluster, Segment and PIP
> * Use of guardExpression and XPath to determine the response
>  type, since RosettaNet uses the same message for both positive
>  and negative responses.
> * Use of XPath to reference other elements.
> * Only 74 lines of XML for the entire PIP!

* * *

------------------------------------------------------------------
To unsubscribe from this elist send a message with the single word
"unsubscribe" in the body to: ebxml-bp-request@lists.ebxml.org