Re: The RDDL challenge

[Patrick Stickler, Nokia/Finland, (+358 40) 801 9690,]

----- Original Message ----- 
From: "ext Tim Berners-Lee" <>
To: "Tim Bray" <>; <>
Cc: "WWW-Tag" <>
Sent: 09 December, 2002 23:26
Subject: Re: The RDDL challenge

> I have some examples of RDDL  in RDF at
> with some discussion.
> Tim BL


In your RDDL example, you associate a stylesheet (among other things)
with a namespace:

    <Namespace rdf:about="">
        <presentation rdf:parseType="Resource">
            <nature rdf:resource="whgatever2#CSSStyleSheet"/>
            <related rdf:resource=""/>
            <title>Color me pink</title>

which seems incorrect to me considering the use of vocabularies
for modular documentation models. A given term (element) may have
a different presentation depending on the actual document model
(schema) in which it occurs. A <name> element may have a complete
different presentation depending on whether it occurs in an 
instance of an <invoice> versus an instance of a <calendarEvent>,
and the proper place to define the presentation of particular
elements is at the model (schema) level, not the termset level.

Furthermore, how is a given CSS stylesheet part of the representation
of a "collection of names". The information provided in your RDDL
examples do not, IMO, constitute in any way knowledge that can be
construed as part of the resource denoted by the namespace URI.
Rather, it is external metadata about how elements or other 
constructs named with terms from that namespace are to be processed
in a given context. The schemas, stylesheets, etc. are not part
of the resource that is that "collection of names". And in fact,
those schemas can even conflict in how they use those names (e.g.
xhtml:html for Strict/Transitional vs. Frameset). They are external,
and as such have no business being in a representation of the
resource accessible from dereferencing the URI denoting the resource.

The cheapest airline fares to Paris may be useful information, but
they are not part of any valid representation of Paris. Right?

Do you really consider your RDDL instance to be a valid representation
of the "collection of names" denoted by the namespace URI? Or is
this being treated as an official "look the other way because it's
so much easier than a more correct solution" exception to
the REST architecture?



Received on Tuesday, 10 December 2002 04:49:03 UTC