XML is broken (was Re: Why Doesn't IE5 use the DTD toValidate?)

[Simon St.Laurent]

At 12:33 PM 4/2/99 -0800, Sara Mitchell wrote:
>James Clark's responses on the issue have cleared up the issue
>from
>my perspective. I agree that the XML spec is not as explicit as 
>it should be on what forces a validating parser to validate and
>that
>has allowed Microsoft to slide. But please don't suggest a whole
>new set of rules!

When the old rules don't work, it's time for a new set.  Chuck the old,
build the new, and don't feel bad about the transition.  Accomodating the
old is fine, but keeping the old at the expense of the new is going to cost
XML a lot.  Politically, I can see why the W3C wants XML to be stable, but
at some point the cost of stability is higher than the cost of change.

>We don't need another set of rules.  I understand that much of
>this
>is awkward to people who are new to XML and it may not be clear 
>why some things have to be as complicated as they are. But adding 
>on a new set of rules just makes it more complicated, not less
>so.

I'm hardly new to XML; it's just taken me about two years to come to the
conclusion that some things are irreparably broken.  The
validating/non-validating external resources/no external resources and
namespaces/no namespaces issues are poison at the very heart of XML, not
just little symptoms that can be brushed away.

>Again, this is part of the strength of SGML and XML. Information
>needs 
>to identify how it should be handled, don't stuff it in a
>separate
>application! There are certainly cases where it's advisable for a 
>receiving application to demand validation -- and other cases
>where the
>author needs to demand validation. But having that information in
>the document itself is important. 

Part of the strength of document-oriented SGML and XML, but a disaster for
data-oriented XML.  The past is holding us back, as an old document-centric
model denies us the ability to create schemas that have control over the
document rather than the other way around.

>> * Describe more than just text and elements.
>
>People are working on this area and it's appropriate for some
>things. 
>But don't demand that documents with information for human
>consumption
>fit into a more rigid requirement needed for processing data.
>There
>are two audiences here, and the requirements for the information
>should
>fit the audience.

Describing data more precisely seems like a win to me, whatever the
application area.  I have no problem with letting document structures
remain just structures of text; I don't demand that every document identify
its floating points and currencies, by any means.  It does seem, however,
that a tighter set of rules would better accomodate both document and data
(and mixed) applications.

>> * Allow supporting tools (like XSL and XLink, which benefit greatly from a
>> validating environment) to demand validation of documents against schemas
>> before attempting processing.
>> [SNIP] 
>
>This could be done quite simply by clarifying the XML spec to
>make it 
>explicit that any presence of an ELEMENT declaration means that a 
>validating parser must validate. Then Microsoft can either step
>back 
>up to the bar or make it clear that IE5 is not a validating
>parser.

That would be a start, but it still leaves many ugly problems unanswered.

Simon St.Laurent
XML: A Primer
Sharing Bandwidth / Cookies
http://www.simonstl.com

xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at ic.ac.uk
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/ and on CD-ROM/ISBN 981-02-3594-1
To (un)subscribe, mailto:majordomo at ic.ac.uk the following message;
(un)subscribe xml-dev
To subscribe to the digests, mailto:majordomo at ic.ac.uk the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at ic.ac.uk)