W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > October 2002

Re: Typed literals text

From: Patrick Stickler <patrick.stickler@nokia.com>
Date: Wed, 16 Oct 2002 15:14:56 +0300
Message-ID: <002501c2750d$a3ee18f0$544516ac@NOE.Nokia.com>
To: "ext Jeremy Carroll" <jjc@hpl.hp.com>, <w3c-rdfcore-wg@w3.org>

In the majority version, in the comparision of typed literals, I wonder
if the test

a.. The two language identifiers compare equal (case insensitive comparison). 

should be removed. I think it has been generally agreed that language
identifiers have no affect on datatyping equality and while you are
clearly avoiding addressing any datatyping semantics in the abstract
syntac doc, the simple equality test should nevertheless reflect that
language has no relevance to equality for typed literals.

Then again, if the equality test is the basis for tidyness and node
merging, then of course that test couldn't be removed. Hmmmm....

> Concrete textual issues are:
>
> + lang tag - in or out

Definitely in. I'm not sure why my earlier examples have not
proven to be compelling that this is necessary. But datatyping
and language are disjunct qualifications. E.g.

   xsd:Name"Finland"-en
   xsd:Name"Suomi"-fi

where a presumed type of xsd:string for

   "Finland"-en
   "Suomi"-fi

won't meet the datatyping expectations of

   foo:countryName rdfs:range xsd:Name .

as xsd:string is not (necessarily) a valid xsd:Name and clearly
the language qualification is not determined by the datatype
and just as relevant as the datatype.

> + do we talk about the typed value here or only in the model theory

The abstract syntax should indicate, as the majority version does,
that the typed literal denotes the value, but which value it is
(the mapping) is an issue to be left to the model theory, I think.

> + do we want to restrict to the canonical form

Absolutely not. RDF Datatyping has no conception of canonical
forms whatsoever and there is no requirement, even by XML Schema
that all simple datatypes have canonical lexical spaces.

Yes, I know canonical forms makes hacking inline literals
easier for the DPH, but one simply cannot presume that any
canonicalization has taken place for arbitrary RDF knowledge
(and on the SW, *most* knowledge will be arbitrary and outside
the scope of control of particular SW agents).

Patrick
Received on Wednesday, 16 October 2002 08:14:58 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:52:26 EDT