RE: Proposed agenda item (no global datatyping)

> -----Original Message-----
> From: ext Jeremy Carroll [mailto:jjc@hplb.hpl.hp.com]
> Sent: 29 August, 2002 12:43
> To: w3c-rdfcore-wg@w3.org
> Subject: Proposed agenda item (no global datatyping)
> 
> 
> 
> 
> If the chair wishes, I would be prepare to propose the 
> following resolution:
> 
> 
> The WG resolves:
> + that the RDF Core Specs will include a local datatyping mechanism
> + that the RDF Core Specs will not include a global 
> datatyping mechanism

I believe it is important to differentiate between global datatyping
assertions versus global implicit datatyping of inline literals. They
are similar but not identical. 

> + to rescind the stake-in-the-ground decision
>     DECISION 2002-02-22/2:
>     DECISION 2002-02-22/3:
>   see
> http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0656.html
> 
> 
> 
> If we were to consider this I would suggest three straw poll questions
> before hand:
> 
> Qu1: who cannot live with a tidy model theory and no global 
> datatyping?
> Qu2: who cannot live with an untidy model theory and no 
> global datatyping?
> Qu3: who cannot live with no global datatyping?

Please see the Implications section of the restructured DT spec
(http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Aug/0257.html)

I don't see things as being so simple. Either we must simple
leave the MT silent about the meaning of inline literals or
we must choose tidy vs. untidy. 

If we actually make a decision for tidy or untidy inline literal
semantics (which has been the actual point of deadlock) then
the particular mechanisms for global implicit datatyping fall
clearly out of that decision, either way.

If we choose tidy semantics for inline literals, then there is
no global implicit datatyping semantics at all. If we choose
untidy semantics for inline literals then the global datatyping
assertion machinery defined in Part 1 of the restructured 
specification essentially work as-is for global implicit
datatyping.

The real question here is, who cannot live with leaving the
meaning of inline literals completely undefined by the MT
and outside the scope of RDF Datatyping, as that is the real straw poll
which will determine whether the local datatyping only solution
is palatable to the WG.

It's not about whether to provide for global implicit datatyping. 
It's about whether inline literals will be ascribed any semantics
at all. Giving inline literals any formal semantics essentially
decides the global datatyping question one way or another.

Patrick

Received on Thursday, 29 August 2002 06:16:52 UTC