Re: protocol draft updated

On Sun, Feb 27, 2005 at 12:53:25AM -0500, Dan Connolly wrote:
> 
> The SADDLE stuff looks like a good basis for discussion.
> Hmm... Domain: G, S... is that short
>   rdfs:domain G, S
> or for
>   rdfs:domain [ owl:unionOf (G S ) ].
> ?
> 
> I think you mean the latter. Better say so somehow. Maybe just use 
> RDFS/OWL/turtle like that?

Yeah, good point.

> The "abstract protocol notation" section... hmm... more stuff between 
> our readers and the "hello world" example.

Based on feedback from Pat Hayes that using some of the notation
before explaining it was confusuing. Plus, I'm not intending this to
be in a primer-style, so I don't really care about making "hello
world" come as quickly as possible. You care about that, and I've
heard yr position on this, but I'm unconvinced.

> We've had mixed feedback from within the WG and elsewhere about whether 
> an abstract protocol is the way to go, so I took a crack at taking a 

I haven't really heard that, other than Mark and Jan's comment, which
I discussed extensively in email with Mark before he submitted
it. -shrug-

I think the value of having the abstract bits will go up as time goes
on, or at least once we have n > 1 concrete bindings.

> "hello world" example from UC&R, putting it up front, and following it 
> with as little specification as I think we could get away with, to give 
> us another organization to consider:
> 
>  The SPARQL Service Interface
>  http://www.w3.org/2001/sw/DataAccess/prot26
>  $Id: prot26.html,v 1.1 2005/02/27 05:25:44 connolly Exp $

> I think the interface it specifies is used in a couple actual SPARQL 
> services... hmm... no, http://sparql.org/books complains when I leave 
> the lang=SPARQL out.

I couldn't find a service that did explicit limit or distinct in the
protocol, hence my dropping it from 1.17, especially since it's in
sparql query.

Kendall

Received on Sunday, 27 February 2005 09:14:09 UTC