xml-dev - Re: Foreign Names

[Paul Prescod <paul@prescod.net>]

Steven Champeon wrote:
> 
>...
> 
> I still don't understand why it's so complex. In a document that may
> contain X element types, and which actually contains Y in any given
> instance, you make 0 > Y >= X translations plus possibly more if you
> also change the attribute naming. You would only need to include a map
> of the standard names to those used in the document at hand; you
> wouldn't even need to provide a new DTD if all that changes are the
> identifiers. Just run it through a single pass transformation and
> you're back to the original DTD.

That's fine for the problem of linguistic translations but isn't there a
bigger problem wherein organizations of all sorts need also to
manipulate the underlying structure to normalize different views of the
same information? 

Let's rephrase the question this way: if the Americans, Koreans and
Russians independently invent and implement large scale applications for
the management of hydro-electric systems through-out their countries.
Let's say that ISO decides to implement a standardized system.  Should
they 

 a) choose
 b) standardize something new and force the existing systems to be
rewritten
 c) standardize something new and implement some sort of mapping to the
existing systems

There is only one choice that preserves the investment in existing
systems. 

This is a superset of the foreign name translation problem. Independent
invention makes it harder to solve, but not fundamentally different.

To put dollar signs in your eyes, consider the problem that occurs when
Compaq and DEC merge and need to integrate their XML-based information
systems. It's the same problem and now two billion dollar companies need
a solution.

In my opinion this is the largest open question in XML practice. I think
that the XML world is new enough that people don't understand that true
schema "standardization" is neither possible nor desirable.

Architectural forms, schema subtypes, RDF subtypes and XSLT all point
towards solutions but none are perfect. XSLT is the closest in that it
is the most flexible and it is link-aware (as archforms and schema
subtype are not) but it has a lot of weaknesses also.

I can see the utility of a simple renaming syntax but I think that it is
at least as important to focus on the much larger problem of independent
invention and unification.

-- 
 Paul Prescod  - ISOGEN Consulting Engineer speaking for himself
"Ivory towers are no longer in order. We need ivory 
networks. Today, sitting quietly and thinking is the 
world's greatest generator of wealth and prosperity."
 - http://www.bespoke.org/viridian/print.asp?t=140

***************************************************************************
This is xml-dev, the mailing list for XML developers.
To unsubscribe, mailto:majordomo@xml.org&BODY=unsubscribe%20xml-dev
List archives are available at http://xml.org/archives/xml-dev/
***************************************************************************