Object-oriented serialization (Was Re: Some questions)

[Rick Jelliffe]

Yes, except XLinks are specified in instances, not as a schema per se.
I hope the XML Schema will have some extension mechanism to
allow these kinds of thing, but who knows.

It is true that sequence and containment relations between elements
in a content model could be treated of as some kind of extended link.

<xlink:extended  >
    <xlink:locator href="http://xxx?xpointer=//boy/dirt"  role="cause"
/>
    <xlink:locator
href="http://xxx?xpointer=//boy/dirt/followingSibling::mud"
role="effect" />
</xlink:extended>

(You could modify schematron-rdf to generate these kinds of
XLinks pretty easily.)

But the trouble with attempting to use XLinks to directly declare
some part of an XML Schema is there is no way to nicely interact
with content models using maxOccurs--if a dog has two eyes and
we want to link between them and the two eyes are declared using
a single <type name="eye" maxOccur="2" /> then we are sunk.
We really want to link in the instance not in the schema.

And we cannot use hrefs to the instances because we don't know
what the instance document URI is: a URI identifies a particular
resource not a class of resources.  XLinks are not designed for
us as schema declarations.

So I think there needs to be first-class support for this in the
schema language itself: in the case of XML Schemas, probably
the most possible thing would be a role attribute (or some equivalent)
on groups.  There is not much there to hook onto. Any ideas on this
would be welcome, even if just to help me think through the
issues for Schematron.

Rick Jelliffe

From: W. E. Perry <wperry at fiduciary.com>

>Is this not precisely the reason that 'behaviour' (or whatever we are
eventually to call it)
>of XLinks is indispensable? Not as a replacement for the
document-centric assumptions that
>text order is meaningful or that the implicit parent-child relationship
of element containment
>is significant, but as the mechanism for specifying (granted, to a
perhaps more data-oriented
>audience) either where these relationships should be explicit, or where
they are replaced by
>explicitly presented alternatives.
 >
>Rick Jelliffe wrote:
>
>> Perhaps a major part of the problem is that sometimes the document
order
>> is meaningful and other times just an artifact of there being no "&"
>> connector in XML content models, and there is no way to decide.  And
>> when the order is important, there is no way to label what its
>> significance is; indeed, the same thing is true of every axis
including
>> the children and parent axes.



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)