[ebxml-iic] ebXML IIC TC -- connection with ebXML BPSS

[Philippe De Smedt <pdesmedt@agentisinternational.com>]

All,

Word from ebXML BPSS that they would like to have IIC's involvement. See
notes below.

-Philippe


-----Original Message-----
From: Philippe De Smedt [mailto:pdesmedt@agentisinternational.com]
Sent: Monday, October 22, 2001 12:19 PM
To: bhaugen
Cc: John Yunker (E-mail); Brian Hayes (E-mail); James Clark (E-mail);
Paul R. Levine (E-mail); Welsh, David; 'Duane Nickull'; James Bryce
Clark; ebxml-iic@oasis-open.org
Subject: RE: A man with two watches never really knows what time it is,
even if one of them is a Patek Phillippe and the other is, um,
object-oriented


Bob,

I'm sure the OASIS ebXML IIC TC would be delighted to include BPSS in its
activities. So far, we've only been interacting with the ebXML TC's that are
under OASIS, but we do intend to expand to cover other TC's (inside and
outside of OASIS, both ebXML and non-ebXML) as well.

The IIC TC is copied on this message so hopefully we'll get some discussion
going on how to approach this. Are there any members of the IIC TC
participating in BPSS, and BPSS members participating in IIC? Let's identify
those, if any, and then see if we can find volunteers to act as liaisons.

Thanks for bringing this to my attention.

-Philippe

-----Original Message-----
From: bhaugen [mailto:linkage@interaccess.com]
Sent: Saturday, October 20, 2001 5:56 AM
To: Welsh, David; 'Duane Nickull'; James Bryce Clark;
pdesmedt@agentisinternational.com
Cc: John Yunker (E-mail); Brian Hayes (E-mail); James Clark (E-mail);
Paul R. Levine (E-mail)
Subject: Re: A man with two watches never really knows what time it is,
even if one of them is a Patek Phillippe and the other is, um,
object-oriented


Jamie, Philippe and all,

Why would it not be possible to develop conformance tests
for ebXML business processes and execution software?
I know that does not solve all the problems you describe
below, but it would help.

I wonder if the OASIS IIC TC will cover ebXML-BPSS or
just the stuff that migrated to OASIS?  (That's why I copied
Philippe De Smedt on this message.)

Also, I very sure that conformance levels would help here, too.
It will be much easier to prove conformance at the Business
Transaction level than the complex Business Collaboration
level.  The transaction level is basically the same as RosettaNet,
there's lots of prior practice.

-Bob Haugen

----- Original Message -----
From: James Bryce Clark
To: Welsh, David ; 'Duane Nickull'
Cc: John Yunker (E-mail) ; Brian Hayes (E-mail) ; Bob Haugen (E-mail) ;
James Clark (E-mail) ; Paul R. Levine (E-mail)
Sent: Friday, October 19, 2001 7:15 PM
Subject: A man with two watches never really knows what time it is, even if
one of them is a Patek Phillippe and the other is, um, object-oriented


> If both BPS's are valid processes,  will there be problems from my end
> (or do I???).

could still very well be if what someone (of course not Duane Co. but maybe
Little Co. and Big Co. - of course) publish in the RegRep is
factual/supported by their internal business operations effectively. No one
ever claimed what was published in the RegRep was true * * *
And no-one will, unless you pay them for it.  Everything in a registry is
user beware -- caveat emptor -- unless someone undertakes to you to assure
you otherwise.

Users of a deposited artifact (whether in UML, XML or BackBaconML) face
several risks:

   1.  "Do I need an analyst"   You must either verify for yourself that it
does what the description says it does, or trust someone else's description.
You may choose blind trust if you wish.  It is likely in my view that some
market makers will offer binding detailed business-process descriptions for
a price -- i.e., I can safely use this catalog-order implementation because
I get an English-language explanation of its terms and consequences from XYZ
Co.  For a price.

   2.  "Do I need an outside integrator"   You must either verify for
yourself by testing that it works with your back end, EAI, market portal or
whatever, or trust someone else's assurance.  You may choose blind trust if
you wish.  It is likely that some vendors and integrators will sell detailed
IG's -- not "ebXML supply chain v2.1 for dummies", but "ebXML supply chain
v2.1 for people running QuickBooks v.xx on Win2000," or a particular SAP
build, or whatever.

     3.  "What's the lingua franca"    Every artifact in a registry will be
expressed in a definitive original reference schema.   That is to say,
whether you regard a particular text as being better in French, English or
Latin, the fact is that every representation of it is in ONE language, and
SOME ONE expression of it has to be deemed exclusively authoritative in a
particular use case.  This is true until you have replicable perfect
roundtrip transforms.  I would be delighted but surprised to hear that we
are planning on shipping that one this quarter.

Crystal ball says:
Yes, there probably will be UML-XML pairs of allegedly identical processes
in the registry.  But I do not expect that authors will offer an open
guarantee of roundtrip logical identity, other than perhaps to paying
customers.   (And given how roundtripping works today, what additional
constraints might they wish to place on XML style issues, in order to offer
such assurances on a repeatable basis?)
Prospective trading partners may wish to haggle over whether they are, in a
given transaction, taking the UML or the XML representation as
authoritative, when the registry offers both.  Probably depends on which one
they trust more as their own native tongue.  Different users of different
systems will reach different conclusions.
Dave's later description suggests that we (ebtwg) will be able to ship
UML/XML pairs with absolutely certain logical identity.  That would be
great.   Someday, no doubt.  But what if early offerings of transform
routines, when repeated, produce stylistically different variations of XML
code that HUMANS can tell are identical, but MACHINES might not?  Are XML
strict production techniques sufficiently standardizable at this time?
Would settling on one style, over others, differentially advantage various
tool makers?

Regards  Jamie