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

RE: 'Peter proposal' on typed literals

From: <Patrick.Stickler@nokia.com>
Date: Fri, 4 Apr 2003 14:44:25 +0300
Message-ID: <A03E60B17132A84F9B4BB5EEDE57957B01B90C71@trebe006.europe.nokia.com>
To: <dave.beckett@bristol.ac.uk>
Cc: <w3c-rdfcore-wg@w3.org>



> -----Original Message-----
> From: ext Dave Beckett [mailto:dave.beckett@bristol.ac.uk]
> Sent: 04 April, 2003 13:52
> To: Stickler Patrick (NMP/Tampere)
> Cc: w3c-rdfcore-wg
> Subject: Re: 'Peter proposal' on typed literals 
> 
> 
> 
> >>>"Patrick.Stickler" said:
> > 
> > To clarify, I am proposing that
> > 
> >    <rdf:Description rdf:ID="foo" xml:lang="en">
> >       <some:property rdf:parseType="Literal">
> >          <h1>Blargh</h1>
> >       <some:property>
> >    </rdf:Description>
> > 
> > would result in the triple
> > 
> >    #foo some:property XML"<h1>Blargh</h1>"@en .
> 
> This is reversing a previous decision on XML literals being made into
> typed literals with the rdf:XMLLiteral URI as the datatype URI.

I fully appreciate that fact.

> Please can you explain the new information that was brought up
> and the reason for this particular solution.

Simple. As Peter points out, and I agree, having a single "special"
datatype that does not work the same as all other datatypes is a
problem.

> > Where the interpretation of such XML literals would be akin
> > to that of M&S along with the new considerations of 
> canonicalization,
> > and such.
> 
> I wouldn't cite M&S as particularly specific on this point.  We
> discussed this many times before and parseType="Literal" was too
> vague in M&S.

Well, we certainly would be following the intent of the charter if
we clarified things accordingly.

I think the WG would have a very strong and fast concensus of how
non-datatype XML literals would be defined and the editorial effort
would not be prohibitive.

> > Thus parseType="Literal" would not result in a typed literal of
> > any kind, and the datatype rdfs:XMLLiteral would be removed
> > from all RDF specs.
> 
> OK, I'm clear on what you want to do.  I want to know why.

So that *all* rdfs:Datatype's would have identical treatment.
 
> > As a second part of this proposal, lang tags would simply be
> > ignored for typed literals and removed from the graph syntax
> > for typed literals entirely. Thus
> > 
> >    <rdf:Description rdf:ID="foo" xml:lang="en">
> >       <some:property rdf:datatype="&xsd;int">10<some:property>
> >    </rdf:Description>
> > 
> > would result in the triple
> > 
> >    #foo some:property "10"^^xsd:int .
> > 
> > and not
> > 
> >    #foo some:property "10"@en^^xsd:int .
> > 
> > --
> > 
> > I guess these really constitute two proposals, but the first enables
> > the second, and both address last call comments.
> 
> Yes and on this second proposal, please explain what new information
> means this change is required.  Which specific last call comments
> (URIs please) does it/they address?
> 
> 
> Both of these will cause significant implementation
> rewritings/reversions - and I *do* know this since I recall changing
> it last time.  Plus there is also having to update/revert and check
> several test cases and WD text changes.

Fair enough. But I think this falls within the scope of "doing it right
this time around".

Patrick
Received on Friday, 4 April 2003 06:44:29 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:56:52 EDT