Re: Syntax Free Models - was: [Fwd: Oracle Input for Core ComponentsWG]

["Martin Bryan" <mtbryan@sgml.u-net.com>]

Betty

> Sorry you weren't there last week.  The syntax you have described
> to me reminds me of a 'poor man's XML database'.  I have known many
> companies who have been contracted to provide an XML database and use the
> file system as a physical database.  It also looks a lot like the BSR. The
> BSR has the hierarchy in the name, i.e., ContractOrganizationName.  The
> problem I see with this method is where do you show branching?

It is based on the BSR concept of semantic object naming rather than on an
XML database, but expressed in a form that is relevant to querying an XML
database. In my message to you I did not include the section on Process
Chain Selection, which is where the branching comes in (once you've selected
a route down a process chain branching should not be necessary).

> I have to go back to my original questions.  What is the goal?
> Who are we doing this for?  If we are looking 10-15 years into the future
> when something else has replaced XML, then yes, maybe I can see this
> approach.  However, in the IETM (Interactive Electronic Technical
> Manual) world, we always thought we would have holograms to
> represent the data by now |-).

The goal is to be able to tie any piece of information in an ebXML data
stream to an unambiguous definition of the meaning of that piece of data.
(i.e. it is the BSR goal, not the XML one of defining where in the message
structure the data needs to reside.) The point is that the meaning should be
independent of any ordering of processes needed to create or transmit the
data.

> I have looked at X12 (I haven't looked at EDIFAC but I assume
> that they are similar).  There is an implicit hierarchy in the X12
> standards.  The XML can fall directly from the EDI.  I don't believe
> the hierarchy is difficult.

The hierarchy of standards is actually a hierarchy of processes - we need to
clearly distinguish this hierarchy from the hierarchy of message components
(unfortunately these seem to be mixed in many cases)

>The modeling isn't difficult.  A lot
> of the hierarchy is just plain old common sense.  Inventing yet
> another way to show the hierarchy will expend more time and energy
> than a lot of volunteers have.  If we aren't going to model XML in XML,
> then there are lots of tools to available, UML, EXPRESS, IDEF, etc.

The problem with things like UML, EXPRESS, IDEF, etc, is that they allow you
to express nets as well as hierarchies. What I suspect is needed is a
language that only allows the definition of hierarchies.

> There are a lot of manpower issues revolving around this effort.
> We have a diverse group with varied backgrounds coming together. This
> effort requires that:
>
> 1. XML professionals be trained and have an understanding
>    of EDI.
> 2. EDI professionals be trained and have an understanding
>            of XML.
>       in addition
>
>         3. Training in whatever modeling methodology is selected.

This is what I've been trying to do this side of the pond for the last
couple of years!

> I am attaching a graphic that shows an XML model for 'goods'
> just as an example.
>
> Looking at X12, here are some of the areas that I believe
> are going to be difficult to come to agreement in XML:
>
> 1.  When to use elements and when to use attributes.
> 2.  Naming conventions
> 3.  Implicit vs. explicit naming conventions.
> 4.  Reuse components
>         5.  Standard abbreviations

I don't have time to answer these at present, but for 1-3 I would refer you
to the draft CEN Working Agreement prepared as part of the ISIS European
XML/EDI Pilot Project last year.

> Martin, you and I both know that 18 months isn't a very long time
> to get a consensus standard out the door, in fact it may be impossible.

Agreed, though there are many who think it must be substantially completed
in 6 months!

> I have only seen it once with the railroad industry and we were only
> developing one DTD for a parts catalog.  This effort is much broader than
> a single DTD.  This issue needs to be resolved before we can move on.

Hopefully the ebXML Core Components group will be the place to do this.
>
> In "Field of Dreams" the voice said "If you build it, they will
> come".  That may be true for baseball fields but if we don't "build it"
> for the common business, they will not come.

Oh they will still come, but then go off along their own path rather than
following our too-narrow highway.

> I was told about the first commercial that mentioned XML last week while
> in Orlando.  I finally saw the commercial last evening.  The Microsoft
> commercial revolved around understanding eBusiness and XML.  Either we do
> this in Internet time or we don't do it at all.

Have you not noticed how slow Internet time is getting? Its 2 years since I
asked for a meeting on XQL - the requirements for which finally surfaced
last week. Seems that building concensus within W3C is as slow a process as
within ISO and UN once real problems start to be addressed!

Martin