RE: The initial ebXML business vocabulary - A call for a list ofcandidat

-----Original Message-----
From: Cole, Wavell (GXS) [mailto:Wavell.Cole@gxs.ge.com]
Sent: Monday, June 25, 2001 10:25 AM
To: John Evdemon; ebxml-dev@lists.ebxml.org
Subject: RE: The initial ebXML business vocabulary - A call for a list of candidates

When ebXML was in its embryotic state, there was a general feeling that using existing EDI element identifiers would be undesirable, they would not reflect the directions towards business process solutions. However, now that ebXML approaches a mature stage and universal element names appear light years from resolution, I am not so sure about this.

The UNTDED (UN Trade Data Elements Directory) which is the core of all UN/EDIFACT messages, contains descriptions of all elements by number and short description plus attributes. As an example, DE1004 is a "Document/Message Number". A similar identification in X12 is 324 "Purchase Order Number".

The advantage of the UN over X12 was that DE1004 could be applied to any business document to number it. The type of document is identified clearly in an XML document so the more generic element name is justified. If I was sending a EDI DESADV (despatch advice) in response to a EDI ORDERS message I would identify the original purchase order using again, DE1004 "Document/Message Number". No confusion and universal applicability.

I note that the UCC-EAN GCI usage is element name = Document, Attribute of ID. Fairly close to the UNTDED.

XML schemas developed by various organisations have used the existing EDI UNTDED conventions and have been able to accelerate their releases. The outstanding variation to this is RosettaNet which has been developed by a specific Industry although with some leaning on X12. Other industry members have expressed interest in using the RosettaNet element identifiers.

If I was developing my own schema my logical choice would be to fall back on the years of effort put into the UNTDED. XML purists however would pull their hair out no doubt with cries of "..but it does'nt matter!". After all is'nt this the very essence of XML, use any element name you like but define it clearly.

Having said that, the argument for a universal approach for element names in electronic business messages has been widely accepted. As there are numerous schema developments for the business community taking place every day and everybody confident in their approach, it would appear that a 'new' universal vocabulary will present difficulties in bringing to fruition - should we persevere?

With the capability of extensibility within the XSD schema, we could namespace to a UNTDED vocabulary, or other vocabulary for that matter. What I am suggesting here is re-writing, or use as is, the existing UNTDED for ebXML use. History has shown that no matter how close we get to a universal development, there is always some differentiation required by a user that has not been foreseen. EDI was difficult to change, XML is more sympathetic.

I can look back on at least 12 years of waiting for application software developers to interface to EDI messages and with little response. I do not believe that there will be a rush to support XML. There are around 1000 different back-office and ERP applications in use here downunder within our comparitively small business community. Middleware software providers can ease the burden and accelerate the uptake of XML particularly with the use of any-to-any translation software now firmly in place.

-----Original Message-----
From: John Evdemon [mailto:jevdemon@vitria.com]
Sent: Friday, 22 June 2001 9:55
To: ebxml-dev@lists.ebxml.org
Subject: RE: The initial ebXML business vocabulary - A call for a list
ofc andidates

On Abid Farooqui: Wednesday, June 20, wrote:
>
> In my opinion these needs are tackled even today by systems like
traditional
> EDI but I think that traditional EDI's weakness is in its complexity, lack
> of similarity between document standards across verticals, its non-tagged
> nature and its price (both the purchase price and the maintenance price
> because you literally need to employ and EDI person to keep the maps
> up-to-date, keep the software up-to-date and deal with new partners that
> come onboard etc.).

Interesting statement - EDI is, indeed, implemented differently across
verticals. I believe you'll see the same thing happen with whatever markup
recommendation comes out of ebXML.

> You need an EDI consultant basically and that is simply
> not going to work in the 21st century.

Somebody better tell this to the Global 1000 using EDI to conduct e-business
today. Whatever happened to the concept of leveraging and extending
existing investments in technology? The best approach is usually to
preserve what works and extend it to integrate with others.

> Moving on to XML based standards ... XML is simple and it provides
> flexibility. However, flexibility requires responsibility. What I mean by
> that is that if we over engineer the solution (business document standards
> etc.) even with XML then you end up just like with EDI after everything is
> said and done. Too complex, requires a consultant almost full-time and so
> on. That is no improvement, its just change of hands.

Many companies that have already adopted existing XML "standards" have had
to replace or extend them because the "standards" didn't reflect their
internal semantics. A realistic example is a large EDI-enabled corporation
using XML to communicate with SMEs. Most XML "standards" do not support all
of the fields in a typical EDI Purchase Order. Further, most existing XML
"standards" only support a few business transactions (e.g. Purchase Order,
Invoice, Shipping Notice). Other transactions (such as Freight Receipt and
Invoice or Materials Safety Data Sheet) are simply unavailable.

Someone suggested that ebXML adopt XEDI to begin defining markup - XEDI
supports every X12 and EDIFACT transaction from any version - including the
examples just mentioned. See www.xedi.org for more info.

> If we are really
> serious about solving the business problem then the requirements are
> simple,
> make a solution that is not necessorily perfect but applies to 90% of the
> scenarios, make it simple to use, administor and add partners with and
make
> it work across industry verticals because business does not happen in a
> vaccum. If we should ever learn something from recent history it should be
a
> lesson from Microsoft. When they first came out with Windows, the
operating
> system was hardly even a contendor for even close to perfect (it isn't now
> even) but it was simple to use. It provided the end user an interface that
> is much more intuitive (visual, GUI) and it had software running on it
that
> solved most of the common problems. This provided a marked advantage
> over
> Unix like operating systems even though they were and perhaps even today
> are
> more robust than windows. Yet you can see that windows has taken off.

Microsoft was building on top of an already established "standard" called
DOS. DOS wasn't pretty, but it worked (much like EDI itself). Using
XML-based representations of EDI transactions makes them easier for non-EDI
companies to create and consume.

ebXML seems to be looking to adopt an XML standard that is already prevalent
in the business community. I suggest that we should be looking at broader
e-business standards - restricting ourselves to custom XML-based initiatives
will be much more difficult (since most XML-based initiatives are rather
immature when compared to EDI).

There are currently far more companies using EDI than XML (or, more
correctly, XCBL) for e-business. That said, ebXML should define its markup
using the dominant e-business standard (EDI) as a foundation.

> In essense in this particular situation with ebxml, if we are to develop a
> solution, we need to develop it so that it solves most of the problems not
> all (the ROE ... Return Of Effort, in going from most incrementally to all
> is too large to justify even trying) and that it is asstandard across
> verticals as possible to reduce complexity and is easy enough to use that
> any decent computer user can administor and maintain it.

Agreed - again, EDI is already a standard across many different verticals.
Defining markup for these verticals should leverage the metadata associated
with existing standards. XEDI would seem to provide a great starting point
for markup development since it already represents over 3000 business
transactions (all of which are derived from the X12 and EDIFACT metadata
standards).

Its already available - its called an 850 Purchase Order and an 855 Purchase
Order Acknowledgement (or, in the case of EDIFACT, Orders and OrdersP).
Granted these are EDI standards, but a quick look at the associated metadata
helps you realize that these documents contain the fields that most
companies use to conduct business.

> Specific
> extensions
> to these universal document(s) can be applied by each industry group if
they
> are really really needed (I discourage these from the bottom of my heart)
> and as that happens software patches (perhaps even optional components)
> can
> be released by vendors to deal with these extensions. If things are kept
> simple I don't think moving from today's systems to the new ones will be
as
> painful as you might think.

The real value that EDI can provide to ebXML is in the metadata - ignore the
cryptic syntax and associated VAN requirements. Leverage EDI's
comprehensive set of metadata to define the schemas for ebXML and avoid
re-inventing the wheel.

John Evdemon
Regional CTO/Director of Engineering
Vitria Technologies
www.vitria.com
www.xmls.com

------------------------------------------------------------------
The ebxml-dev list is sponsored by OASIS.
To unsubscribe from this elist send a message with the single word
"unsubscribe" in the body to: ebxml-dev-request@lists.ebxml.org

ebxml-dev message