Re: daml:collection

On Thu, 2002-04-25 at 04:56, Brian McBride wrote:
> I've been catching up on the daml:collection discussion.  Thank you DanC for
> 
>    http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Apr/0334.html
> 
> I don't follow the reasons for:
> 
>    [[* add a 'highest index' property to bags:
> it tells you the highest index that's used
> to relate a collection to one of its members.
> This is only slightly better than a "count" property,
> to my mind.
> ]]
> 
> You reject the count property on the grounds of arithmetic and 
> comparison.  RDF must already have the concept of equality of properties 
> and ordering of the ordinal properties.  I don't see that this proposal 
> requires any more than that.

I'm working out the details... cwm seems to cooperate when
there are two continents but not when there are six...
I hope to get back to you shortly...

> It seems to me, that if we are closing RDF's current containers, then this 
> is the front running option.  I'm a wee bit nervous that it introduces 
> something close to negation by the back door though.  Pat?

Yes, any form of closed list involves negation. The whole
point of a closed list is to be able to conclude that something is
*not* in it.

> The alternative is that the owl folks use daml collection.  They can do 
> that without  any help from us, just as the daml folks did.

But I don't think we can condone it.

i.e. we can't say "no, that file isn't RDF, but you're free
to put <rdf:RDF> at the top of it."

It's sorta like the XSLT-template-in-HTML-document's-clothing
situation that has come up in the TAG. For the sake of
a coherent architecture, it's important that each document's
meaning as a formula is clear, and doesn't depend on
whether you use an RDF parser or a WebOnt parser.

Hmm... I'm not sure how to make this point clearly...


>  Daml was quite 
> happy to define daml:collection as an extension to RDF.  All that is needed 
> is a preprocessor to turn it into legal RDF.  Such a preprocessor may be 
> built into an RDF parser, but that is an implementation matter.

No, it's an interoperability nightmare.

> Between these two choices, should we ask the customer which they prefer?

Not if you want my support.

The latter is unacceptable; I'd object.

The former is ugly; I'd abstain.

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Thursday, 2 May 2002 17:19:30 UTC