Re: call for agenda items

[Patrick Stickler, Nokia/Finland, (+358 40) 801 9690, patrick.stickler@nokia.com]


----- Original Message ----- 
From: "ext Dave Beckett" <dave.beckett@bristol.ac.uk>
To: "Graham Klyne" <GK@NineByNine.org>
Cc: "Brian McBride" <bwm@hplb.hpl.hp.com>; "RDF Core" <w3c-rdfcore-wg@w3.org>
Sent: 05 December, 2002 14:11
Subject: Re: call for agenda items 


> 
> >>>Graham Klyne said:
> > 
> > At 09:40 AM 12/5/02 +0000, Brian McBride wrote:
> > 
> > >Sorry this is late.  I've been buried in other stuff and my network at 
> > >home went down last night.  I'm catching up this morning.
> > >
> > >Suggestions for agenda items for this weeks telecon welcome.  Please let 
> > >me know what, if any, key issues you feel we need to discuss.
> > 
> > Should rdf:value be deprecated?
> 
> No.
> 
> We owe the community not to arbitrarily chuck out stuff

I wouldn't consider a decision to deprecate rdf:value as "arbitrary".
Clearly there's been alot of thought and discussion about it, and
the fact that it has no actual semantics in the MT suggests that it
has no place in the RDF vocabulary.

> that has
> previously been recommended and being used.  

The problem is that it is used in different ways by different
folks, and no RDF processor has a clue what it means apart
from application specific knowledge which may conflict between
different cases. That's exactly what our specs are supposed to
eliminate.

> In particular, Dublin
> Core is using this right now for the purpose it was suggested in RDF
> M&S - multi-value properties.  This turns up in DC to implement the
> "dumb down" rule as they call it, which you can consider as similar
> the units/value example in recent messages.

Fair enough. And DC can continue to use it, as a deprecated term,
until such time as a future version of DC provides a more satisfactory
treatment by migrating to a term with clear semantics; e.g. dc:value

> This loses for me on the following grounds:
> 
>    * no new information - we previously agreed to keep rdf:value and
>      to explain it in the primer.

The problem is explaining it in a way that is (a) useful to users
and (b) reduces ambiguity/conflict/confusion. The lack of any
real semantics seems to be making that difficult at best to
accomplish.

If we're not going to actually say what it means, then we should
deprecate it. It's no different than introducing some term
rdf:blargh which folks are then free to assign any arbitrary 
semantics to.

The same is true of every other term that doesn't have a proper
definition in the MT, not just rdf:value.

The role of the Primer, as I understand it, is to provide an easy
to understand presentation of the more formal content of the 
specifications. Having the Primer suggest uses for a term that
has no formal semantics seems rather dubious.

>    * no big changes in the last versions of documents
> 
>      - this was in particular where some of the sticky parts of
>        rdf/xml came from first time round, as I understand it.

This doesn't seem to me to be a huge change. It's a choice between
adding verbage that tries to clarify the legacy usage of rdf:value
(without actually suggesting any actual semantics) or simply saying
it's deprecated. 

I'd think the latter option is less change overall.

Patrick

Received on Thursday, 5 December 2002 07:35:01 UTC