W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > August 2003

whitespace in XSD and RDF

From: Jeremy Carroll <jjc@hpl.hp.com>
Date: Thu, 14 Aug 2003 12:04:58 +0300
To: w3c-xml-schema-ig@w3.org, w3c-rdfcore-wg@w3.org
Message-Id: <200308141204.58020.jjc@hpl.hp.com>


Background:
XML Schema raised this comment:
http://www.w3.org/2001/sw/RDFCore/20030123-issues/#xmlsch-02

RDF Core added this note:
[[
In [XML-SCHEMA1], white space normalization occurs during validation according 
to the value of the whiteSpace facet. The lexical-to-value mapping used in 
RDF datatyping occurs after this, so that the whiteSpace facet has no effect 
in RDF datatyping. 
]]

RDF implementors did not like it, e.g:
http://lists.w3.org/Archives/Public/www-rdf-comments/2003JulSep/0076

[[
** Failures - could fix:
NegativeEntailmentTest xmlsch-02/Manifest.rdf#whitespace-facet-2 - FAIL
NegativeEntailmentTest xmlsch-02/Manifest.rdf#whitespace-facet-1 - FAIL
  These test non-mutual entailment of a valid literal with an invalid 
  literal that differs only by whitespace. Unfortunately our XSD 
  handling library is happy with the whitespace and doesn't 
  treat " 3 " as an invalid int. 
  This could be fixed if that is indeed how XSD is supposed to work,
  though the current behaviour seems more useful in practice.
]]

The RDF Core WG asked me to have a go at drafting something that conceded 
somewhat to the last phrase above. 

The proposal is to delete the above note, and add the following:

[[
Implementation Note: (normative) 
In [XML-SCHEMA1], white space normalization occurs during validation according 
to the value of the whiteSpace facet. The lexical-to-value mapping used in 
RDF datatyping occurs after this, so that in RDF datatyping the whiteSpace 
facet formally has no effect. However, in line with the principle of being
liberal with what is accepted and strict with what is produced:

+ During input processing of XML Schema Datatypes within RDF, software
SHOULD apply the appropriate whitespace normalization immediatly before
the lexical to value mapping, and MAY produce a warning if any whitespace
is changed in this normalization. 

+ Software generating RDF data SHOULD NOT add additional whitespace to the 
lexical forms produced. 

(The keywords MAY, SHOULD, SHOULD NOT are defined in [RFC ????])
]]

(The keywords RFC is a new normative reference in concepts - off the top of my 
head it is 2117)

Dan, depending on the e-mail traffic can we take that as a formal proposal to 
reopen xmlsch-02 and close it as addressed by the above text. I imagine there 
will be time at tomorrows telecon.

Jeremy
Received on Thursday, 14 August 2003 06:05:28 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:59:38 EDT