Re: Comments on "SPARQL 1.1 Uniform HTTP Protocol for Managing RDF Graphs"

On Sun, 2011-03-20 at 21:36 +0100, Kjetil Kjernsmo wrote:
> On Friday 18. March 2011 22:45:33 David Booth wrote:
> > 2. Because this kind of ambiguity of reference is inescapable (though
> > the example is an extreme case), so we have no choice but to learn to
> > deal with it.  
> 
> So, what you're saying is "relax, whether it is an RDF Graph or an RDF 
> Document is not important to specify, ambiguity is inescapable and can be 
> useful"? 

Not quite.  Yes, "ambiguity is inescapable and can be useful".  But no,
don't relax.  :)

> 
> I can see that point, and I appreciate the arguments that you make. However, 
> it still seems that accepting this means accepting what webarch calls a URI 
> Collision. 

That is correct, because ultimately it is *impossible* to avoid them.

HOWEVER, the webarch admonishment against URI collision *is* a good goal
or guiding principle, because as a URI owner it is good to think beyond
your immediate needs and consider how you and others would likely wish
to use your URI in the future.  If the way you are defining it is likely
to cause ambiguity (a/k/a "URI collision") in the class of applications
that you wish to support, then you should make your URI declaration more
specific, such as differentiating between the movie called "The Sting"
and a forum discussion about that movie:
http://www.w3.org/TR/webarch/#URI-collision

Ultimately this is a judgement call about the range of applications that
you want your URI to support.

> Also, I very much appreciate Sandro's clean sheet terminology. This new 
> terminology is unfortunately not currently available as normative reference, 
> and for such a simple protocol as we currently discuss, I feel that we should 
> be able to specify it clearly enough for all practical purposes without.

+1

I think Sandro's terminology will be a *big* help in reducing confusion.

> 
> There are two points that I feel has not been answered yet, and that's
> whether the apparent conflict that I think I see between the usages of
> RDF Document and RDF Graph are real, is the ambiguity real? And if so
> would a URI collision be acceptable? 

I don't know exactly what conflict you mean, but in the context of the
RDF working group, I think it is important to be very precise, to avoid
confusion.

> 
> Whatever may be the case, I think that a discussion of these muddy
> waters does not belong in the specification in question. This is
> something that could be implemented by thousands of developers, and we
> need to make sure we do not scare anyone away (as hard as that may
> be).

I'm not sure exactly which "muddy waters" you mean, so I cannot comment.


-- 
David Booth, Ph.D.
http://dbooth.org/

Opinions expressed herein are those of the author and do not necessarily
reflect those of his employer.

Received on Monday, 21 March 2011 16:53:41 UTC