Re: RDF concepts

At 07:26 25/10/2002 +0200, Jeremy Carroll wrote:


>The version for review is ready; although I have not allowed Graham a chance
>to check over my latest changes (we have discussed the essential direction of
>these changes).
>
>The current URL is:
>
>http://sealpc09.cnuce.cnr.it/jeremy/RDF-concepts/2002-10-25/rdf-concepts.html
>
>I expect Graham will want to copy it onto his site -
>(a tar ball for you, with the this version URL fixed is:
>http://sealpc09.cnuce.cnr.it/jeremy/RDF-concepts/2002-10-25.tgz
>)
>After he has done this, we should copy something to the archive.
>
>My release notes, over and above the version Graham circulated last week are
>as follows:
>
>- *all* literals are typed
>- two new predefined types for what were previously untyped literals.
>- literal is a pair (datatypeURI, lexical form)
>     - lexical form is also a pair (string, language-identifier)
>- datatype mappings (see section 2.4.4) can be from string=>value or from
>lexical form=>value
>
>
>I believe I have slightly overstepped

This proposal extends the XSD datatyping model to allow the lang string to 
take part in the mapping to the value.  Our charter states:

[[Specifically, the RDF Core Working Group is not chartered to develop a 
separate data typing language that duplicates facilities provided by XML 
Schema data types.]]

Whilst what is proposed is not separate, it is different.  Bearing in mind 
that this is not exactly a tranquil area, I suggest the WG should prefer an 
acceptable alternative which does not extend the xsd datatyping model.

>the appropriate editorial freedom, and I
>am prepared to backpeddle if the WG is unhappy with where I have gone.
>
>Rationale for change to all typed literals:
>- language id in typed literals, appeared to be the path to consensus at last
>telecon

We were influenced largely by the desire not to rule Nokia's data 
illegal.  Despite that, the majority were in favour of not having the lang 
code on the literal, but Nokia's willingness to dissent carried the 
day.  Since then we have learned that Nokia's data is illegal anyway and 
Patrick has said:

   [[So I guess the WG can omit lang tags from literals entirely.
It's looking like it won't matter to us one way or another.]]

   http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Oct/0277.html

Which I take to be a withdrawing of that dissent.


>- all three forms of literal then were very similar, the change to making 
>them
>all typed is a rationalization

Yes - its neater.

>- this change has been suggested some times on the e-mail list [1], [2], with
>no explicit opposition

   http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Oct/0266.html

That's hardly a reason for doing it.  And I think you are perfectly aware 
that there are folks who are not comfortable with this.


>- the two most extensive external reviews of the abstracrt syntax in its 
>first
>WD from TimBL [3] and Massimo [4] both suggest changes along these lines
>(neither of them felt confortable with the status given the XML literals).

Yea, legacy is like that.  As I recall, it was the W3C team rep who 
strongly advocated XML literals being different.

[...]

>The principle risks is:
>- removing the langauge identifier from typed literals is now harder, if the
>call to include them was in error.
>Two possible reasons why it might be in error are locale vs langauge
>confusion, and XSD conformance. I have tried to phrase the document to show
>appropriate deference to these concerns.

Can we run it past Misha asap and get a quick view.

Brian

Received on Friday, 25 October 2002 03:54:50 UTC