[Fwd: Re: REPEAT: RDF Plain Literals and xsd:string]

First response

Brian

-----Forwarded Message-----

From: Dave Peterson <davep@iit.edu>
To: Brian McBride <bwm@hplb.hpl.hp.com>, w3c-xml-schema-ig@w3.org, cmsmcq@acm.org
Cc: Pat Hayes <phayes@ai.uwf.edu>
Subject: Re: REPEAT: RDF Plain Literals and xsd:string
Date: 24 Jul 2003 09:39:01 -0400

At 2:02 PM +0100 030724, Brian McBride wrote:

>RDFCore is being asked whether RDF plain literals [*] and xsd:string
>denote the same thing.  Intuitively it seems that they ought, but we
>held of stating this because we have been a little confused about how to
>interpret the XML Schema datatypes specification.
>
>For example, whilst the value space of xsd:string is defined to be a
>sequence of characters, and the value spaceof xsd:anyURI is defined to
>be the set of URI references, which are defined by RFC 2396 to be
>sequences of characters, it has been stated that the value spaces of
>xsd:anyURI and xsd:string are disjoint.

Two points should be made:

   o	The descriptions of the relations on the value spaces of Schema
    	datatypes are normative for schema processing but are advisory
    	for other applications.

   o	The WG is aware that some of our primitive datatypes have
    	values that could reasonably be identified with values from
    	other primitive datatypes.  For 1.0, at least, we chose to
    	make those values (artificially?) unequal *for schema
    	processing purposes*.  This decision gets occasional discussion;
   	some would like to change it.  The WG hasn't yet thought out
    	all the ramifications of such a change.

Personally, I see nothing wrong if other applications choose to
treat such values as identical.  I *hope* that the WG will decide
to make such a change for schema processing, but there are many
numeric datatypes for which the corresponding ramifications are
more complicated, and it may take awhile to get them sorted out.
-- 
Dave Peterson
SGMLWorks!, for IDEAlliance

davep@iit.edu

Received on Thursday, 24 July 2003 10:14:20 UTC