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

[Bob Haugen <linkage@interaccess.com>]

Arofan Gregory wrote:
>My feeling is that we need to model the core components in a syntax-free
>way, and publish them as such, fully documented. This seems to be the
>approach the group is taking. 

I think some of about syntax-free is because some of us don't think
there is such a thing, that you must publish in *some* syntax, 
although that would not necessarily need to be XML.
If what you mean is "XML-free", but some other syntax -
UML, Conceptual Graphs, KML, etc etc - then the question
is which alternative syntax do you mean, and is it one more
syntax, toolset, etc. that everybody needs to learn?

If you mean "English prose" or something like that, then ok,
I'm with you, but it may lack precision for some people.

> In doing the eCo
>semantic recommendations work, it became very clear that decisions about
>what is an attribute and what is an element entirely depend on
>implementation considerations, and cannot be standardized in a way that is
>useful to all. (Not to mention that huge pain that an argument about these
>issues would entail in a group laregly made up of people coming from the EDI
>world). 

I understand those issues all too well.  

>I think we have identified a powerful new approach to these issues,
>and that it is incumbent on us to make it simple enough to be usable.

That sounds terrific and I am mightily interested.  I look forward to the 
details.

>I speak from experience - with CBL, there are many, many compromises that I
>have to make in order to create single schemas for single message types, and
>I am never happy with the result, because there is no awareness of context
>in schema the way it exists today.

It would be really interesting to read a summary of the lessons
you learned from that experience.  eCo and CBL both really advanced
the state of the art.

Thanks,
Bob Haugen