W3C home > Mailing lists > Public > www-international@w3.org > January to March 2005

Re: XMLLiterals and language

From: Martin Duerst <duerst@w3.org>
Date: Wed, 19 Jan 2005 15:24:00 +0900
Message-Id: <6.0.0.20.2.20050119150809.06348798@localhost>
To: Reto Bachmann-Gmuer <reto@gmuer.ch>, Jeremy Carroll <jjc@hplb.hpl.hp.com>
Cc: www-rdf-interest@w3.org, www-international@w3.org

At 21:32 05/01/18, Reto Bachmann-Gmuer wrote:

 >I think it is generally possible to model language-information without 
xml:lang, personally I'd probably have renounced on promoting this 
xml-heritage to rdf in favour of a general practice like in your example

xml:lang support in the RDF Model and Syntax REC was chosen because
one of the goals of RDF/XML was to allow XML documents to also be
interpreted (possibly with very few changes such as additions of default
elements,...) as RDF. And the other way round, xml:lang is the
natural way to express language information (on the actual text)
in XML.

I think using some graph structure and properties rather than
some 'magic' in the RDF model would have worked fine.

But please note that this still leaves you with the possible
difference between the 'intended' language (this is the
title/message/... used in German) and actual text language
(this title/message/... is actually in Latin).

 >(possibly using rdf:value instead of rdfs:label and - if legal - 
rdfs:Literal instead of ex:Message). In practice there are often situation 
where a literal may be plain as well as contain XHTML,

Yes indeed. That's why the current state of xml:lang in RDF
is not very helpful at all.

 >real world usage of RSS shows that even properties designed to have a 
plain-literal value are often used containing encoded XHTML. The widely 
used RSS mod:content [1] mandates the encoding of XHTML in a plain literal

I originally thought about using this module. But now I see that they
indeed have used escaping. This is just simply not the way combination
of namespaces is supposed to work in XML. And to me personally, I'm
sad to say, but it simply looks disgusting. I'm glad to know that at least
one feed format, namely Atom, at least allows inline content without
escaping.

 >but it would probably be easy for parsers and producers to switch to 
parseType="Literal".

Yes, this was the original intent of parseType="Literal" to provide
a smooth transition from simple text to text with markup. Unfortunately,
very few people understand that need; XML Schema has a very fundamental
distinction between simple types (including string) and complex type
(everything including markup), and other specs don't do much better.

Regards,    Martin.

 >In contrast changing the range of the property from Literal to something 
like "ex:MessageWithPossibleLanguage" is likely to be far too complicated 
to be adopted.
 >
 >reto
 >
 >1. http://web.resource.org/rss/1.0/modules/content/
 > 
Received on Wednesday, 19 January 2005 07:12:52 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 19:17:04 GMT