OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   Re: extensibility in XSchema?

[ Lists Home | Date Index | Thread Index ]
  • From: Peter Murray-Rust <peter@ursus.demon.co.uk>
  • To: <xml-dev@ic.ac.uk>
  • Date: Fri, 19 Jun 1998 19:35:43

Welcome to the club, Mark.

At 09:47 19/06/98 -0700, Mark D. Anderson wrote:
>I hesitate to ask this question because (a) I haven't been
>perusing either the discussion or the specs in detail, and
>(b) I don't know SGML very well, which seems to be a
>prerequisite to participation.

No, it isn't :-). You do have to know XML, of course - and you clearly do.
I don't believe that all the current participants have a strong formal
background in SGML. And I have welcomed the injection of new ideas from
non-SGML people.

I think the only pre-requisite is a strong familiarity with information
management and ability to think hard and take time to work things out

>But I'd like some reassurance that when XSchema is done,
>I will *not* ever have to parse comment structures to retrieve
>metadata that the XSchema authors didn't choose to include
>in a fixed repertoire of XSchema metadata declarations.

This  is certainly the vision that some people have. I think we'd all agree
that any parsing of comment structures represented a broken model. However
we are all discovering what the potential and limits of metadata are at
present. Thus I think we shall find that there are problems that are very
difficult and we may need other approaches. For example, XML cannot, per
se, transmit behavior. 
>Currently, if I want to write a tool that helps users create
>new instances from a schema, I've been using xml
>to declare the "template", not a dtd, because I found it so
>objectionable that I would have to read dtd comments to get
>the information I wanted.

We sympathise. My latest DTD (VHG) has precisely this problem. But I don't
want to hack my own template before XSchema is done.
>No fixed metadata will accomodate everything someone might want to know
>about an element: suggestions for what GUI control to use for
>entry and read-only reporting, a short and a long help text,
>special validation regexps, complex conditional relationships
>across elements, and so on.

Agreed. I think agreement on interactive behavior is a very challenging
area - if we could agree on some aspects it would be a major advance.

>Right now, writing a tool for creation of new xml documents
>from only a dtd for information is like trying to write a 
>database entry tool from only the rdbms-level schema information.
>The information that a DTD provides is a joke, from the perspective
>of an authoring tool.

Well, I wouldn't use the word 'joke', but I agree you have to go outside
the DTD. Nonetheless the DTD can be quite useful for *some* applications.

>There seems to be a constant undercurrent of religion in xml-dev
>about validation vs. parsing vs. semantics, whose portent I
>don't apprehend. 

Yes. It's healthy that there are a variety of views. This is a very
difficult problem because it is the first time that behavior (I use
'semantics') is being discussed in a global - and therefore partailly
unregulated - context. I think that there is a very wide spectrum of
opinion, depending on where you come from.

Note that for the first year a large number of the participants in XML had
an SGML background. That's changing, especially on XML-DEV. Lots of people
have strong formal backgrounds from other disciplines. And - in my view -
that's healthy. But don't write SGML off - it is a very powerful approach
with proven success. However it's not always easily accessible.

>The information necessary for validation is 
>typically insufficient for an authoring tool. If the data 
>structure can't accomodate the other information which the authoring
>tool needs, then it will be left with putting information in
>comments, or in some completely different file. 

Yes. This is one of the motivations for XSchema. I should caution that
XML-data and RDF also are involved in this sort of activity. Thus XSchema
may need to qualify its proposal with things like:
	- we propose [a patter/rule language], but may later revise this to be
compatible with XSL
	- we propose [semantic validation criteria] but may later revise these to
be compatible with RDF.

>I would expect that everything necessary for validation would
>be an explicit part of XSchema, but that the schema for XSchema
>itself (metametadata) would allow arbitrary elements in a valid
>XSchema schema.

I look forward to the next draft :-)


xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev@ic.ac.uk
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To (un)subscribe, mailto:majordomo@ic.ac.uk the following message;
(un)subscribe xml-dev
To subscribe to the digests, mailto:majordomo@ic.ac.uk the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa@ic.ac.uk)


News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS