RE: [ALL] RDF/A Primer Version


> Indirect identification in [4], is followed by section 2.3 
> "URI comparisons" ... 
> ... starting this way: "URIs that are identical, 
> character-by-character, refer to the same resource."
> But previous section clearly states that the same URI can 
> refer "indirectly" to different things in different 
> (so-called local) processing contexts. So maybe this sentence 
> should read more exactly :
> "URIs that are identical, character-by-character, refer to 
> the same resource in the same processing context."
> We tend to focus on the most frequent processing context, the 
> Web client-server interaction which aims to retrieve a 
> representation of the resource. But, Alistair, all the VM 
> cookbook shows clearly that the representation that you 
> retrieve in this specific processing context (http GET) 
> depends on devilish details of both client and server 
> configuration, so even in this case where the resource is 
> supposed to be "directly" referenced, actually the referent 
> is not uniquely defined by the URI. 

Your overall point may be correct -- I don't know, I'm still trying to
figure it out! -- but what you've stated above seems to fly in the face
of RFC 3986[8] (the definition of URIs).   RFC 3986 is very clear that a
URI identifies a resource as a whole -- not a specific representation of
that resource.  

The idea that it may be okay for a URI to have multiple referents also
seems to conflict fundamentally with the principle that "By design, a
URI identifies one resource"[8] (regardless of context).  If there
really is no conflict between these views, then I think the community as
a whole is badly in need of greater education to understand why not.  I
know I am!

[4] WebArch on URI collision:

[7] RFC 3986:

[8] WebArch on indirect identification:

David Booth

Received on Monday, 30 January 2006 18:42:36 UTC