RE: High-level comments on datatyping draft

> -----Original Message-----
> From: ext Jeremy Carroll [mailto:jjc@hplb.hpl.hp.com]
> Sent: 27 August, 2002 23:52
> To: w3c-rdfcore-wg@w3.org
> Subject: Re: High-level comments on datatyping draft
> 
> 
> 
> 
> Sergey:
> >Finally, I think that global datatyping (Sec. 3.2) is out of 
> scope of the
> >current document. At this point of time we do not have a mechanism
> >robust enough to standardize on global datatyping.
> 
> Brian:
> > I interpret these comments in the light of your preference for tidy
> literals.
> 
> I think not ...
> 
> my understanding (which dates from before my holiday and 
> during Brian's!) is
> that the WG appeared to have a consensus (minus at least 
> Patrick) forming
> around not doing a global idiom.
> 
> I am not up to date on what has happened while I have been relaxing.

I know the feeling ;-)

> My understanding of the motivations are:
> - the WG has had a mild majority in favour of tidy global idiom

Well, as Brian points out, the last straw poll actually showed
a mild majority in favor of untidy literals.

> - the community process showed a majority in favour of untidy global
> - many members of the WG, including myself, concluded that we 
> cannot reach
> *consensus* on a global idiom

I don't think that is presently the majority view of the WG. I
thought the "round-the-bar" discussions last week were very
encouraging that we may in fact be able to reach concensus
about global datatyping. But then, I've been known to hallucinate ;-)

> Thus, I have argued that even with the community backing, 
> while I still
> believe in the untidy global idiom, I do not think it is ready for
> standardization.

I don't see how it would not be ready for standardization, since
it is little more than current rdf:type and rdfs:range semantics
and it is precisely what most folks seem to have been expecting.

It also is a technically sound solution. The proponents of tidy
literals are of course free to disprove that assertion, if they
can.

If using rdfs:range to assert the rdf:type of untidy literals
is unsound, then general rdfs:range and rdf:type semantics are also
unsound and should be removed from RDF.

> This discussion is deliberately punting on tidiness an issue 
> that Patrick at
> no point let us forget is model theoretic in nature.
> 
> My sense at the last telecon I attended was that if we had 
> been asked the
> following questions:
> 
> * we resolve not to do a global datatype idiom
> * we agree that we all can live with both a tidy or an untidy 
> model theory

Recent comments from Pat indicate that OWL cannot live (at
least healthily) with a tidy MT. Also recent communications with 
the CC/PP community reflect a clear desire for untidy global 
datatyping.

> * we will have an e-mail discussion and then a straw poll on tidiness
> 
> that, with the probable exception of Patrick, many of the WG 
> would have gone
> along with that. For instance, my position of favouring 
> untidiness is a lot
> less rigid once we have dropped the ambition to do global 
> datatyping. I
> suspect that those who favour tidiness likewise could live 
> with untidiness
> if it purely in the model theory and not part of datatyping.
> 
> Jeremy

The present situation, as I see it, is that

1. OWL wants untidy literal semantics (Pat, please correct if not so)
2. CC/PP wants untidy global datatyping (e.g. rdfs:range with datatypes
   asserting that datatype for inline literals)
3. The original expectation of the first RDF Core WG was that there
   would be something like rdfs:range based global datatyping, as
   DanB recently pointed out
4. The community clearly favors untidy literals
5. The last WG straw poll favored untidy literals
6. Tidyness must be decided irregardless of datatyping
7. Choosing a tidy MT precludes any future untidy semantics as the RDF
   layer

I consider all of the above, combined, to be more than sufficiently
compelling to adopt untidy literals and global datatyping as
defined in the latest working draft

http://www-nrc.nokia.com/sw/rdf-datatyping.html

No, that doesn't mean that every member of the WG is in favor
of untidy literals, but I think that the above 7 points clearly
reflect a very strong preference and need overall for untidy literals
(and global datatyping).

The WG should review the latest working draft, and base discussion
and decisions using that as a basis, IMMHO.

Regards,

Patrick

Received on Wednesday, 28 August 2002 03:44:21 UTC