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

Re: Datatypes Convergence

From: Daniel W. Connolly <connolly@w3.org>
Date: Wed, 6 Feb 2002 15:07:33 -0500 (EST)
To: Brian McBride <bwm@hplb.hpl.hp.com>
cc: Sergey Melnik <melnik@db.stanford.edu>, patrick Stickler <patrick.stickler@nokia.com>, RDF Core <w3c-rdfcore-wg@w3.org>
Message-ID: <Pine.LNX.4.30.0202061453390.10098-100000@tux.w3.org>

Hmm... Brian caught me in IRC and cleared up some confusion;
in particular, I got the impression that the B4 age/title
inference didn't work with this converged proposal.

I'll reserve judgement until I get time to read more.

Please include lots of examples in any new write-up.

In particular, I hope the proposal allows

	dc:date rdfs:range xsdr:date.

to be consistent with
	<dc:date>2000-12-23</dc:date>
and inconsistent with
	<dc:date>not the lexical form of any date<dc:date>

I'd like to see examples from PRISM and cc/pp too.

On Wed, 6 Feb 2002, Daniel W. Connolly wrote:

> On Wed, 6 Feb 2002, Brian McBride wrote:
>
> > I'm really glad to see the progress made on datatypes this week.  I think
> > we may have got to the point where can have a single proposal with some
> > unresolved issues.  This would be magnificent progress from having two
> > proposals.
>
> Hmm... I'm not so sure...
>
> > I therefore suggest that Sergey and PatrickS take an action jointly to
> > prepare a converged datatyping proposal document and circulate it to the WG
> > by Wed, 13th Feb 2002.  That document should be based on:
> >
> >    - using rdf:dType for datatyping
>
> ? why?
>
> >    - that S-P/TDL is used for global typing
>
> Why? Let's please promote S-B. It's evidently acceptable
> to both PRISM and CC/pp.
>
> >    - using Graham's doublet suggestion, as adopted by Pat to fix
> >      the model theory
>
> Sigh... the simplest model theory, as Jos suggested, just
> has literals denote themselves. Let's please just stick
> with this simple design.
>
> >    - using the simpler of the two approaches suggested by Pat for the
> >      model theory
> >    - that <foo> <bar> "10" is untyped, as in current m&s, but a
> >      syntactic transform can be used to turn it into a typed structure
> >      with a b-node
>
> ??? Why?
>
> It looks like more unmotivated complexity; it seems to fail
> the B4 case just as badly as other non-S proposals.
>
> This looks like a huge step backwards to me.
>
> I don't like earlier TDL proposals, but I think I understand it now.
> Likewise, I understand folks don't like S, but I think they understand it.
> Please let's not explore alternatives that are so complex that nobody
> understands them well enough to object.
>
> > They key thing about this document is lock in the progress we have made, so
> > if there are outstanding issues that are still unresolved, note them in the
> > document and bring them back to the WG for comment.
> > Of course, I'd love it if there weren't any.  I wouldn't expect this
> > document to necessarily cover the model theory at this stage.
>
>

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
Received on Wednesday, 6 February 2002 15:08:38 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:45:06 EDT