Musing over Namespaces

[Rick Jelliffe]

 From: Arjun Ray <aray at q2.net>


>On Sat, 18 Dec 1999, Len Bullard wrote:
>> Dan Brickley wrote:
>>
>> > By defining schema languages in instance syntax, we implicitly
>> > promote the idea that there will be some big payoff for doing so
>> > (otherwise, lets stick with DTDs).
>
>How about improving DTDs? (Just a thought.)

Also, there are mixed strategies that are worthwhile ocnsidering too:
in particular, regex syntax (i.e., (),|+* ) is very convenient.  I think
it would be great for XML Schema to allow a simplified syntax where
we could go
    <element name="boy" type="(frog+, snail+, puppyDogTail+)">
XSchemas and XLink and RDF include simplified syntax, so there
is precedent.

ISO has a correction to SGML in the wings to allow alternative
schema syntaxes to DTDs: I think it has been waiting for 2 years.
I think the problem with extending DTDs is that you have to create
new declarations (unless you use PIs): also, DTDs  are much more
aimed at parsing while what is needed is at a different level:
structures, datatypes, semantics.

>> There is some payoff:
>>
>> 1.  Political.  XML can finally dissolve the XML to SGML parentage.
>
>Done deal already: XML is a W3C trademark, and AFAICT, the XML 1.0
>Recommendation does not reference SGML (ISO8879+TCs) normatively.

But that only means that you don't need to read ISO 8879 in order to use
the XML spec. SGML is referred to in the abstract and at the start.
XML 1.0 is also referred to by ISO 8879 Annex L.

>> 2.  Technically.  It is a stronger schema model.
>
>That's the hope.  The latest XSchema stuff has plenty of changes.

The new draft needs a lot of scrutiny to test that it *is* a stronger
model.
Certainly, it approaches schemas from a more abstract perspective rather
than the nuts and bolts approach of  DTDs with PEs.

>Isn't *some* (form of) schema indispensible?  The real issue seems to
be
>the content of schemas (i.e. what they provide for in the way of
asserted
>and verifiable constraints) rather than the syntax, although sheer
>clumsiness of instance-syntax could become an issue too.

There is also the issue that different classes of languages have
different
families of constraints.  XML Schemas look like being specially good
for relational database interaction, forms, fun with keys, commerce, and
serializing (single inheritence?) objects.

But we need to test it.  I think we may have XML-Schemas with us for
quite afew years, so it is in all developer's interest to go through it
and
objectively think "does this solve my problems" or at least "does this
prevent solution of my problems".  For example, are you happy that
XML Schemas make infoset contributions?  That means that processing
software must be "schema-aware" before using the data. Should there
be mechanisms in place by which a document can say "I use infoset
contributions: if you don't have a full XML processor, don't accept me!"

Or, what should the criteria for validity be: structures, structures +
datatypes,
structure+ datatypes + encoding-checking?  Or should it implement
an ANY like XML DTDs (any element that is defined) or like WF XML
(accept anything)?

There are hundreds of these questions that XML Schemas 1.0 is
having to face, and hundreds more that must be deferred till later.
The new drafts are said to be "feature complete", so this is a good time
to start reading and thinking "could I actually use this thing?"

Rick Jelliffe


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 unsubscribe, mailto:majordomo at ic.ac.uk the following message;
unsubscribe 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)