Re: name of rdf:nodeID attribute

>>>Garret Wilson said:
> 
> The addition of rdf:nodeID is handy, especially from the standpoint of 
> end users new to RDF who balk at creating universal URI identifiers for 
> all their resources.
> 
> However, using the name "rdf:nodeID" both for the description of the 
> resource and for the reference is inconsistent and confusing. XML has a 
> long history of using "ID" and "IDREF" as names of an identified XML 
> node and the node that refers to it, respectively. Similarly, RDF uses 
> "rdf:about" and "rdf:resource" for the similar concepts based upon 
> resource reference URIs.
> 
> Can't we make RDF more internally consistent and consistent with XML by 
> using "rdf:nodeID" and "rdf:nodeIDRef", respectively?

That makes more sense when you are in XML defining an identifier
with ID and then refering to it with IDREF (or IDREFS).  

In RDF you never define identifiers, you are always using them for
nodes (or arcs) in triples of the RDF graph.  The identifiers are
URIs as used in rdf:about or rdf:resource and blank node identifiers
which in the RDF/XML syntax are used with the rdf:nodeID attribute.

(There is also rdf:ID in the RDF/XML but that is seen as another way
to do URIs, along with a little checking).

I think we considered making two attributes for rdf:nodeID but given
the experience that people cannot seem to remember when to use
rdf:about or rdf:resource, it made more sense and was a smaller
change to add a single attribute.   This seems to have worked out
fine.  nodeID and nodeIDref would bring no benefits and would likely
just repeat the rdf:about/rdf:resource confusion.

(I note in amusement, the last draft of the RDF primer had the latter
confusion.  It is fixed now ;)

Dave

Received on Monday, 13 January 2003 06:39:11 UTC