W3C home > Mailing lists > Public > www-rdf-interest@w3.org > July 2002

Re: Input sought on datatyping tradeoff

From: Brian McBride <bwm@hplb.hpl.hp.com>
Date: Mon, 15 Jul 2002 19:39:29 +0100
Message-Id: <5.1.0.14.0.20020715193251.029a8ad8@0-imap-1.hpl.hp.com>
To: "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>
Cc: www-rdf-interest@w3.org

At 06:50 13/07/2002 -0400, Peter F. Patel-Schneider wrote:

[...]

>Well, actually, I remain unclear as to just exactly what the current specs
>define a literal to be, partly because the current WD specs and WG
>discussion has two different notions of what a literal is.  The MT says
>that a literal is a string, but other documents and discussion say that a
>literal is a three-part structure.

The model theory has not yet been updated to reflect the decision that a 
literal is a three part structure (thanks for that phrase by the way).  Pat 
is thinking about whether the model theory can work sensibly with this 
decision.


>Why is this important?  Well it appears to me that two literals in XML/RDF
>with the same lexical representation can end up being different, for
>example if their language (French, English, ...) is different.

That is not the case for n-triples.  If there were a language specifier in 
the examples, it would be represented in the n-triples.  There is none for 
these examples.

>   How this
>impacts datatyping is totally unclear to me. For example, is the literal
>"10,5" a valid lexical representation of a floating point number when the
>language is French?  Perhaps "1,000" denotes 1 when the language is
>French and 1000 when the language is US/English.

These are all issues we could get into.  However, my understanding is that 
xml datatypes have no notion of different langauge representations of the 
same value, so lets assume the answer is no, for now.

Brian
Received on Monday, 15 July 2002 14:40:55 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:54 GMT