xml-dev - (XML is SGML, part 1,002) Re: RELAX to ISO

[Rick JELLIFFE <ricko@geotempo.com>]

KenNorth wrote:
> 
> > In other words you are not using "XML" you are using "SGML":
> 
> Don't you think Schema and other specifications move us away from this
> model?

No, why?  

A WebSGML document doesn't need to have a DTD.  There is nothing in SGML
that
deprecates or prevents any kind of processing: schema-processing for
example.
There is nothing that prevents a GROVE for an SGML document from having 
additional non-standard properties attached. An XML Schema is a
specifation 
of a transformation (augmentation) of an information set and some
outcome results: processing.

In fact, quite the reverse.  The stated aim of SGML is for rigorous 
description of document notations: the more rigour the better.  
(And ISO itself has also defined technologies "Lexical Typing"
and "Architectural Forms" which provide many of the things that
XML Schemas will be used for (in fact, one way to implement
XML Schemas could use those facilities: I thought Extensibility 
used architectures in some part of their implementation, for
example.)

People tend to think of SGML as being what they encountered rather than
what the standard is about: a technology for standard rigorous
description of
a comprehensive languages for generalized markup.   So people who 
came in from FrameMaker will expect SGML to involve detailed document
analysis and fixed DTDs while people who came in using OmniMark will
expect their systems to allow very fast on-the-fly changes to DTDs on
a completely casual basis.

> > it is an industry profile which belongs to W3C not to ISO--we wanted to
> give them
> > the freedom to maintain it (and indeed, we cannot take it from them!)
> > and develop it.
> 
> Why is submitting a specification to ISO "taking it away" from W3C? There
> nothing that takes away the W3C's mission to incubate technology or its role
> in evolving new versions of XML-related specifications.

ISO specs have different formats and constraints and politics to W3C. 
A spec can only have one controller, human nature being what it is.
For example, ISO has its ISO HTML (a subset of HTML 4), but its links
to W3C HTML are by initial design and the continued will of SC34 to
keep compatability with W3C HTML.
 
> Will we no longer see new ISO standards because technology incubators feel
> something is taken away when they contribute to a standard?  (I missed it
> when IBM closed its doors after having SQL and SGML "taken away" by the ISO
> ;-)

Yes. We see it all the time.  ISO votes are by nation; W3C votes are by
member
organization; why is SOAP W3C not ISO?  Because the stakeholders
negotiating 
are companies and the views of national standards bodies are irrelevant
to
the action.  A better example is Java: not ISO standardized because ISO 
standardization is not rubber-stamping.

Now there is a process possible at ISO to recognize standards made by
other 
bodies: I think CCITT and IETF are included, but I forget the details. 
I am
not sure whether these actually become ISO standards or whether they
simply
can be referenced by normative standards. 

Cheers 
Rick Jelliffe