W3C home > Mailing lists > Public > www-xml-linking-comments@w3.org > July to September 2002

Re: proposal for an xlink data model

From: Erik Wilde <net.dret@dret.net>
Date: Wed, 28 Aug 2002 12:31:00 +0200
Message-ID: <3D6CA664.1010203@dret.net>
To: www-xml-linking-comments@w3.org


ignoring item 1 (which was about xpointer), here my thought on item 2:

Henry S. Thompson wrote:
 >   2) There is a long-standing issue between XLink and XHTML (and SVG adn
 >      ...) wrt link recognition.  At the moment XLink defines linking
 >      semantics wrt constellations of elements and attributes based on 
 >      namespace and relative location in a document.  This means that
 >      without breaking backward compatibility XHTML cannot be
 >      straight-forwardly XLink compatible, or make use of XLink's 
 > What I would like to explore is the possibility of separating out the
 > signal from the consequences, by specifying a new Infoset property or
 > new Infoset properties in each case, and
 >  a) Making the presence of that property/those properties the signal;
 >  b) Making sure everything needed to determine the semantics is
 >     available from the value(s) of that property/those properties.
 >   2') This one is a bit harder, because multiple properties are
 >       involved, or even, as I think about it, a new InfoItem as well.
 >       Suppose we define a new property [links], whose value may be a
 >       set of Link InfoItems.  XLink 1.1 would then factor itself into
 >       a) Link recognition, which constructed a [links] property and
 >       b) Link semantics, which interpreted Link Info Items.  Then
 >       XHTML could just specify how the various attributes of e.g. IMG
 >       turn in to two Link InfoItems and it all just works :-)

this extension of the infoset is what i would like to do. in a short
report (http://dret.net/netdret/docs/tikrep148.pdf) i have outlines some
of the basic ideas. i brought up the subject on xml-dev, but quickly was
drawn into the issue of xlink vs. xhtml syntax, which is not the core of
the proposal. i think the following need to be done:

- define a framework for extending the infoset
- define the xlink data model as one such extension
- define xlink 1.1 as an xml serialization of the data model
- define additional specs on top of the data model, for example
    - a dom module for accessing links
    - css selectors for selecting links
    - xpath mechanisms to access links
    - xhtml as another syntax for the same data model

it is my opinion that the whole architecture of web standards could
benefit from this approach, and we have seen similar approaches in other
areas such as dom and css, where the creation of interdependent modules
has become the standard way for modelling.

we (ie, the swiss federal institute of technology) have just joined w3c,
and from what i understand, we should now have better leverage to
propose ideas like this. how should i proceed to have this considered by
the w3c? currently i have the impression that the linking wg is mainly
concerned with finalizing xpointer, and the xml core wg would require to
be re-chartered to cover this topic. complicated...


erik wilde  -  tel:+41-1-6325132  -  fax:+41-1-6321035
            mailto:net.dret@dret.net -  http://dret.net/
            computer engineering and networks laboratory
            swiss federal institute of technology  (eth)
            * try not. do, or do not. there is no try. *
Received on Wednesday, 28 August 2002 06:32:18 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:32:24 UTC