W3C home > Mailing lists > Public > public-rdf-comments@w3.org > December 2013

Re: ISSUE-148: IRIs do *not* always denote the same resource

From: David Booth <david@dbooth.org>
Date: Thu, 19 Dec 2013 15:26:22 -0500
Message-ID: <52B3566E.9070705@dbooth.org>
To: Guus Schreiber <schreiber.guus@gmail.com>
CC: Public RDF comments list <public-rdf-comments@w3.org>
On 12/18/2013 09:33 PM, David Booth wrote:
> On 12/18/2013 05:47 PM, Guus Schreiber wrote:
>> David,
>> The RDF WG resolved today [1] ISSUE-148:
>  > Resolve ISSUE-148 by changing the "IRIs have global scope"
>  > bullet point in section 1.3 in Concepts to "By design,
>  > IRIs have global scope. Thus, two different appearances
>  > of an IRI denote the same resource. Violating
>  > this principle constitutes an IRI collision
>  > [WEBARCH]." http://www.w3.org/TR/webarch/#URI-collision
>  >
>  > We very much hope you can live with this outcome. It might
>  > not be exactly what you wanted, but I hope it is at least
>  > very close to it.
> Indeed, this is close, but it still does not address my concern,
> because it still says "two different appearances of an IRI
> denote the same resource" as though that is always true, in
> spite of fact that the next sentence acknowledges that this
> can be violated (though it gives no clue about how).
> I gather that someone didn't like the word "should" or the
> phrase "is intended to" that were proposed earlier.   As a
> compromise, how about inserting "RDF assumes that":
>    "By design, IRIs have global scope. Thus, RDF assumes
>    that two different appearances of an IRI denote the same
>    resource. Violating this principle constitutes an IRI
>    collision [WEBARCH]."
>    http://www.w3.org/TR/webarch/#URI-collision
> David

P.S. According to the meeting minutes,
the WG **did not even consider** any wordings that would have fully 
addressed the concern that caused me to raise this issue.  The only two 
options that were considered were *identical* in still having the 
problem described above.

I know it isn't easy to craft verbiage that is acceptable to many 
different parties with different perspectives, and I have offered to 
help do that.  But the way the working group has handled this -- 
simultaneously prohibiting discussion on the public-rdf-comments list 
while also prohibiting non-member subscription to the working group 
discussion list -- has left me only an extremely blunt instrument for 
communicating with the working group.  In essence, all I can do is raise 
a formal objection.  And friends, that's rather broken.  It sure ain't 
very efficient.

This is the time to get this fixed.  We should not go another 5 years 
with wrong/misleading statements about the uniqueness of IRI reference, 
leading new RDF users down a garden path only to be surprised and 
disappointed later when merged graphs don't work as expected.  One might 
assume that, because this text is a non-normative introduction, it isn't 
very important.  But it *is* important because it affects how the reader 
thinks about the whole specification.  Readers need to know that URI 
collisions can happen **even with RDF that is 100% compliant with the 
RDF specification**.

The currently adopted verbiage *almost* addresses my concern, because it 
acknowledges that the principle of unique URI reference *can* be 
violated.  But alas, as currently phrased, readers are likely to assume 
that such violations are violations of the RDF specifications, rather 
than violations of a higher level architectural objective.  Readers need 
to know that RDF makes the *assumption* of unique URI reference, and 
hence if this assumption is violated, there may be problems due to URI 

I have spent a *lot* of time over the past several years -- probably 
*thousands* of hours cumulatively -- trying to get to the bottom of the 
resource identity issue.  In fact, I may have spent more time on this 
issue than all other members of the RDF working group *combined*, 
excluding Pat Hayes!  But that is mere idle speculation.  The point is 
that although this may seem like a trivial issue to some, it is 
important to me, as to my mind it ultimately underlies nearly 
*everything* we do in the Semantic Web that we're trying to build.

I have done my best to work as efficiently as possible with the RDF 
working group -- moving lengthy discussions to other lists, trying hard 
to reach common understanding with other individuals before proposing 
fixes that might reach consensus, and offering to help craft acceptable 
wording -- but as explained above, the constraints imposed by the 
working group have made the process frustratingly difficult and awkward.

So, I guess I will rephrase my suggestion above.  Would anyone strongly 
*object* to inserting "RDF assumes that" in the currently adopted bullet 
phrasing, as described above?  If so, why?

Received on Thursday, 19 December 2013 20:26:52 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:59:44 UTC