RE: Parsing Containers

Hello.

For what it's worth, I'd like to offer my strong support for your proposal.
I'd tried bringing up the same issues but lack the clear and concise voice
that you do.

Both repat [1] and RDF.NET [2] already handle containers in almost exactly
the manner in which you propose. During development, I couldn't justify the
extra code to enforce the needless (in my opinion) restrictions. I'm glad
that there's others who feel the same way that I do.

I suspect that the alternate syntax for containers was specifically to
prevent constructs like this:

<rdf:Description rdf:about="http://foo" rdf:_1="1">
  <rdf:li>2</rdf:li>
</rdf:Description>

Productions 6.25-27 all say that containers with member attributes must be
empty elements. Typed nodes (production 6.13) don't have that restriction
with regards to property attributes.

I can see how some people might view this as being ambiguous but the loss of
flexibility hardly seems worth it. And with clear rules and examples such as
the ones you've provided, there should be no ambiguity.

By the way, maybe you should include the above example in your proposal.
It's the only case that I thought was missing. I believe it should generate
these triples:

[http://foo, rdf:_1, "1"]
[http://foo, rdf:_2, "2"]

With regards to Issue #1, since neither of my parsers treats containers as
anything other than typed nodes, about, aboutEach, etc, are permitted. I
couldn't fathom a reason for that particular restriction, either. Enforcing
this restriction would require that parsers be notified somehow of what
classes are subclasses of rdfs:Container. This is a complication that
doesn't buy us anything.

What does everyone else think? I hope the responses are all positive and
that your propsal is included in the next revision of the recommendation.

Thanks,
Jason Diamond.

[1]
http://injektilo.org/rdf/repat.html
[2]
http://injektilo.org/rdf/rdf.net.html


> -----Original Message-----
> From: McBride, Brian [mailto:bwm@hplb.hpl.hp.com]
> Sent: Friday, December 15, 2000 1:50 AM
> To: 'barstow@w3.org'; 'begeddov@jfinity.com'; 'christop@ics.forth.gr';
> 'connoly@w3.org'; 'eric@w3.org'; 'jan@swi.psy.uva.nl';
> 'jason@injektilo.org'; 'rjc@netscape.com'; 'melnik@db.stanford.edu'
> Cc: Dave Beckett (E-mail 2)
> Subject: Parsing Containers
>
>
> Dave Beckett and I recently sent a proposal on how RDF parsers
> should handle
> RDF container syntax.  This proposal can also be found at the
> following URL:
>
>
> http://www-uk.hpl.hp.com/people/bwm/rdf/issues/containersyntax/current.htm
>
> We are trying to see whether the community of folks in RDF
> Interest can make
> progress in clarifying some of the outstanding RDF issues.
>
> For that to happen, as a writer or maintainer of a current RDF
> parser, your
> input is crucial.  So I'm writing to ask you to respond to the proposal,
> whether positively, negatively or neutrally.
>
> Brian McBride
> HPLabs
>

Received on Saturday, 16 December 2000 01:08:43 UTC