W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > June 2001

Re: (tentative) container model proposal

From: Dan Brickley <danbri@w3.org>
Date: Tue, 12 Jun 2001 15:53:30 -0400 (EDT)
To: Aaron Swartz <aswartz@upclink.com>
cc: Graham Klyne <Graham.Klyne@baltimore.com>, <w3c-rdfcore-wg@w3.org>
Message-ID: <Pine.LNX.4.30.0106121548020.20596-100000@tux.w3.org>

hi Aaron,

On Tue, 12 Jun 2001, Aaron Swartz wrote:

> Dan Brickley <danbri@w3.org> wrote:
>
> > Containers are also, as you point out, syntactically privileged in the RDF
> > syntax specification. But then, so is the rdf:type construct: we can write
> > <wn:Person foaf:name="dan"/> instead of a more verbose piece of XML,
> > because the RDF syntax provides sugar for common idioms. The container
> > machinery in the syntax is in the same category...
>
> I disagree, I do not see them in the same category. The typedNode construct
> is clearly very useful for a number of purposes. It makes things clearer and
> less labor-intensive. Furthermore, there is unlikely to be need for an
> alternate version of "type" so providing one in the spec is a reasonably
> safe bet. Last of all, the typedNode syntax is difficult to confuse with
> other syntaxes, where as the rdf:li construct appears like just another
> propertyElt (in fact, it seems as if Mozilla treats it this way).
>
> My goal is to reduce the complication and number of "exceptions to the
> rules" in the RDF spec. I feel this is important for wide adoption of the
> spec, and best benefits the HTML authors I represent on behalf of the HWG.
> And I do not feel that:
>
> <rdf:Seq>
>     <rdf:li>a</rdf:li>
>     <rdf:li>b</rdf:li>
>     <rdf:li>c</rdf:li>
> </rdf:Seq>
>
> is simpler or more intuitive than:
>
> <rdf:Seq>
>     <rdf:_1>a</rdf:_1>
>     <rdf:_2>b</rdf:_2>
>     <rdf:_3>c</rdf:_3>
> </rdf:Seq>
>
> Nor do I feel that it is intuitive for:
>
> <rdf:Bag>
>     <rdf:li>a</rdf:li>
>     <rdf:li>b</rdf:li>
>     <rdf:li>c</rdf:li>
> </rdf:Bag>
>
> to generate:
>
>   _:genid rdf:type rdf:Bag .
>   _:genid rdf:_1 "a" .
>   _:genid rdf:_2 "b" .
>   _:genid rdf:_3 "c" .
>
> Now, as a WG bound by backwards compatibility as we are, I don't think
> there's much that we can do about this. However, I think that it's important
> to make my feelings on this subject clear.

your feelings are noted!


I don't really think we disagree much. When you say...

> I disagree, I do not see them in the same category. The typedNode construct

...I take you as making a claim about the relative merits of the
special container syntax, auto-numbering mechanism etc. contrasted with
the (rather handy) rdf:type / typedNode mechanism.

My claim was pretty modest: just that both rdf:type and rdf:_n constructs
are similarly privileged in RDF's XML syntax, but that neither deserve
any special architectural privilege w.r.t. the basic formalities of the
triples model. Whether we feel the one is more/less useful, intuitive etc
is a separate issue, and one that you're right to postone to future work
on syntax beautification.

Dan
Received on Tuesday, 12 June 2001 15:53:53 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:37:06 EDT