Re: XML for Structured Data

In message <78DFE33066ABD0118B9200805FD431BA05919C@RED-16-MSG.dns.microsoft.com> Jean Paoli writes:
> As Jon said in a previous e-mail, the XML ERB have decided to open the
> namespace debate. Andrew Layman from Microsoft and myself have been
> - with the approval of the XML ERB- discussing with a lot of groups in
> the 
> W3C and in the IETF the subject of using XML for Structured Data.
> 
> The mail below describe the issues we were discussing with all of these
> groups (Digital Signature, DAV, Labels) and the set of limitations of
> XML
> that have been clearly identified by them.
> 
> At this time a lot of these groups expressed their interest in using XML
> 
> for structured data and I think personnally that this is a huge win for
> XML
> if they do choose to use it: This will represent a huge range of
> distributed
> applications and protocols which will use XML on the Web.
> This is why the XML ERB agreed to open officially these subjects to
> discussion
> and asked me to post 5 questions to this mailing list for revising the
> XML-syntax.
> 
[...]
This is a first reaction to the hyper-proposal and isn't a fully considered 
reply.  The overall approach seems very valuable, if somewhat overwhelming
and I think it will be important that the discussion has some structure.

My main question is to ask for more detailed clarification on the impact 
on the current status of XML-lang, because it seems that there are 
significant changes proposed not only to the language but the requirements
of processing software.

I'm mainly addressing 'it shall be easy to write software that processes 
XML documents'.  At first sight some of the proposals seem quite forbidding
even to a mythical CS student :-)  I'm also ignorant of parts of SGML and am
not clear how some of what is proposed fits into SGML (I assume it's all
legal SGML).

SD1.  No problem.  I found myself doing this all the time when I write on
paper serviettes at mealtimes in Barcelona and no-one complained.  I think
it's a selling point for human readers of XML files.

SD2. Please can we have more clarification?  I have found myself wishing to
use something like this in the past, but cannot see how it fits into the
SGML language.  So questions:
	- is this valid SGML?
	- if so, can it be represented in the ESIS stream?  If so, I assume
		it is a post-parsing process, rather than a parsing process.

<INTERMEZZO>
As a greater number of XML-legal constructs are hardcoded into the language,
the requirement for a post-parser (or pre-application) grows.  There is now
a substantial amount of processing required for XML-LINK (e.g. is the 
containment correct, how are attributes inherited) and SD[1-5] puts another
substantial burden on the processing.
</INTERMEZZO>

SD3.  Fully supportive of this as a starting point.

SD4. This appears not to require any changes to XML-lang, and is in essence 
a meta-application.  I think there is a role for identifying such areas
which can perhaps be worked on in parallel to the syntax development.
The scale of this looks like being beyond the ability of the CS grad to
do in less than a few weeks.

SD5.  This is almost identical to something I was going to post when the 
floodgates were opened for discussion.  I love it. It solves one part of the
namespace problem - identifying the semantics.  To solve the namespace
collision there has to be some processing at parser level, because
presumably the DTDs have the shorter names and need to know they are
aliased in the document.  So running the following through (n)sgmls
would presumably throw the error that mml.VAR and cml.VAR were not 
declared int the DTDs.  Or is this only intended for WF documents?
<MML>
<*XML-SCHEMA XML-REF="http://www.w3.org/pub/WWW/TR/WD-math" xml-as="mml"/>
<mml.VAR>x</>
</>
<CML>
<*XML-SCHEMA XML-REF="http://www.w3.org/pub/WWW/TR/WD-math" xml-as="cml"/>
<cml.VAR TYPE="FLOAT">1.2</>
</CML>
In any cases I see a significant amount of processing required and it's 
important to highlight where it needs to take place.

	P.

<GENTLE_PLEA>
All my XML work is unfunded and I have to pay for incoming mail, so is it
possible to post 70Kbyte *.html file on the WWW somewhere rather than 
including them in messages?
</>

-- 
Peter Murray-Rust, domestic net connection
Virtual School of Molecular Sciences
http://www.vsms.nottingham.ac.uk/

Received on Friday, 16 May 1997 05:12:31 UTC