Re: Datatypes [Was: Minutes telecon 26th July 2002]

On 2002-07-28 10:49, "ext Jeremy Carroll" <jjc@hplb.hpl.hp.com> wrote:

> 
> 
> 
>> 12) Datatypes
>> 
>> Proposed as main theme for next week.
> 
> Eric is eager for us to make substantial progress.
>
> I can think of one possibility that might work.

Fine, but let's not rush and make a poor choice for the sake of
making "some" choice. Sorry, Jeremy, but I don't see your proposal
as constituting "substantial progress". It's a step backwards.
 
> We have substantial, probably unaminous, consensus on datatyping.
> 
> Well, on the local idiom for datatyping.
> 
> Let's just stick with that and not attempt to address the global idiom.

I think we have pretty strong consensus on the global idiom as well.

Of those that chose to explicitly respond as requested to Brian's
inquiry, there is overwhelming preference for untidy semantics
(almost unanimous).

Most importantly, the tidy/untidy issue is as much about the model theory as
it is about datatypes, so punting on the global idiom now is not an option
since this is really about whether literals in the model theory have tidy or
untidy representation).

We have an important decision to make about the MT, not just datatyping.

* In Bristol, there was a weak majority preference for untidy literals.

* In the response from the community, there is a strong preference for
  untidy literals.

* The WG has already decided to go with one of the two options put forth
  in Brian's inquiry, with the goal of the inquiry to help the WG choose
  one of those two options based on community preference/intuitions.

We should go with untidy literals, both syntactically and semantically,
and modify the MT accordingly.

Punting on the global idiom is not an option, IMO, given the MT issues
regarding tidyness of literals.

> ...
>
> In my view we have done excellent work on the global idiom,

I agree, and I think that work is pretty much completed.

> but there are
> still three or four viable candidates:
> 
> - untidy
> - tidy
> - two properties

This has already been rejected by the WG. I don't see how it can
be considered still on the table.

> - something more linked to XML Schema (I don't believe this has been
> sufficiently articulated yet, but it appears to be Patel-Schneider's
> preference.)

We don't want anything that depends on XML Schema specifically, though
we definitely want a solution that works well with XML Schema. We need
the RDF datatyping solution to be framework-neutral. Thus, this is
not, IMO, an option.

--

The WG has already decided that datatyping should work by one of the
two proposed options, based on the tidyness of literals, or lack
thereof. The WD agreed to solicit comments from the community as
to which of those two options are preferable. The community has expressed
clear and strong preference for untidy literals. Thus, we should now be
*done*. We should now be able to make a confident decision in favor of
untidy literals and get the WD out the door.

You seem to be proposing that we invalidate all of the progressive consensus
decisions the WG has made to date. We're 99% done and you want us to
open up everything again to debate.

The last final choice, for the last open issue, was whether literals
are tidy or untidy. That decision should now be clear, given the
feedback from Brian's inquiry, and we should technically be *done*
and able to close that last open issue -- leaving the rest as actions to
the editors of the datatyping WD.

If Eric want's substantial progress, then that should take the form
of adopting the untidy literal option and tasking the editors to get
the WD out ASAP.

*That* would be the fastest and most substantial progress, IMO.

Cheers,

Patrick
 
--
               
Patrick Stickler              Phone: +358 50 483 9453
Senior Research Scientist     Fax:   +358 7180 35409
Nokia Research Center         Email: patrick.stickler@nokia.com

Received on Sunday, 28 July 2002 04:39:01 UTC