W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > April to June 2005

URI/IRI concrete and abstract syntax

From: Dan Connolly <connolly@w3.org>
Date: Thu, 21 Apr 2005 16:17:49 -0500
To: RDF Data Access Working Group <public-rdf-dawg@w3.org>
Message-Id: <1114118269.3004.284.camel@localhost>

We currently have:

  "let RDF-U be the set of all RDF URIs"

  http://www.w3.org/2001/sw/DataAccess/rq23/#defn_RDFTerm
  $Revision: 1.319 $

(I had to do view-source to get that fragid. While perhaps
my browser should do better, I'd like to have an index
of abstract syntax terms at the end or something.)

The term at the other end of the link is actually
"RDF URI reference"...

[[
A URI reference within an RDF graph (an RDF URI reference) is a Unicode
string [UNICODE] that:

      * does not contain any control characters ( #x00 - #x1F,
        #x7F-#x9F)
      * and would produce a valid URI character sequence (per RFC2396
        [URI], sections 2.1) representing an absolute URI with optional
        fragment identifier when subjected to the encoding described
        below. 
]]
  -- http://www.w3.org/TR/rdf-concepts/#dfn-URI-reference

The discrepancy is probably due to a comment from me.

The good news is that definition of the term "RDF URI reference"
turns out to be the same as the slightly more modern term, "IRI"
from RFC3987:

[[
   IRI reference: Denotes the common usage of an Internationalized
      Resource Identifier.  An IRI reference may be absolute or
      relative.  However, the "IRI" that results from such a reference
      only includes absolute IRIs; any relative IRI references are
      resolved to their absolute form.  Note that in [RFC2396] URIs did
      not include fragment identifiers, but in [RFC3986] fragment
      identifiers are part of URIs.
]]
 -- http://www.ietf.org/rfc/rfc3987.txt


So I suggest replacing RDF-U with just I ala...

 Definition: RDF Term

 let I be the set of all IRIs
 let RDF-L be the set of all RDF Literals
 let RDF-B be the set of all blank nodes

 The set of RDF Terms, RDF-T, is I union RDF-L union RDF-B.

followed, outside the definition box, by:

  Note that all IRIs are absolute; they may or may not include fragment
  identifiers[RFC3987, section 1.3]. Also note that IRIs include URIs
  (and URLs). This definition matches the
  definition of RDF URI Reference from [12].

I suggest striking

  An RDF Term is anything that can occur in the RDF data model.

or move it outside the box and phrase it as:

  This definition of RDF Term collects several basic
  notions from the RDF data model.

In the concrete syntax, i.e. the grammar, it's all IRI references.
So this is pretty good...

[3]   BaseDecl   ::=   'base' QuotedURIref

probably better as QuotedIRIref.

But...

[12]   SourceSelector   ::=   URI

should be

[12]   SourceSelector   ::=   IRIref

and

[63]   URI   ::=   QuotedURIref | QName

should probably be

[63]   IRIref   ::=   QuotedIRIref | QName

likewise
[41]   VarOrURI   ::=   Var | URI 
should be
[41]   VarOrIRIref   ::=   Var | IRIref

Also...

[69]   <Q_URIref>   ::=   '<' ([^> ])* '>' /* A URI relative reference :
RFC 3869 */

should be...

[69]   <Q_IRIref>   ::=   '<' ([^> ])* '>' /* A IRI reference : RFC 3987
*/

and let's move that comment into the prose of the spec...
how about under "Query Term Syntax".

We should probably be more clear about whether this
is a sparql query or not:

  SELECT ?x WHERE { <foo###bar> dc:title ?x }.

REQUEST FOR TESTCASE.

I suggest that yes, it's a SPARQL query as defined by the
grammar, but it's erroneous; i.e. it's in the same category
as queries that don't obey the limitations on where variables
can go when using OPTIONAL.

So we probably need a new kinda of test case.


I dunno if we should rename sop:isURI .

Huh? what's this?

  RDF terms are identified by r:term and the constituant subclasses:"

  types of RDF terms with the prefix r:. SPARQL operators are named
  with the prefix sop:.

what are r: and sop: bound to?


Under 11.2.3.4 sop:isBlank it says
  This query matches the people with a name and an mbox which is a URI:
but the query seems to be about matching blank nodes.
(the data should have some non-bnode people to make the distinction
more clear).

This is sorta use-mention-buggy:

 (foaf:mbox should only have a URI as its object)

suggest:

 the range of foaf:mbox is mailbox, not literal.



Scanning for occurences of "URI" in the document, I find:

  When undeclared, the prefixes below stand in place of the URIs given:

which sounds like part of the specification of the syntax of SPARQL,
but it's in a "Document conventions" section. I suggest changing that
to something like:

  In this document, many examples assume the following namespace
  prefix bindings:


next I find:

  The terms delimited by "<>" are relative URI references [RFC 3986].
  After parsing, these are resolved to give URIs.

which is pretty good, but (a) the can be absolute too, and (b) they
should be internationalized and (c) we should mention a base IRI
somehow. Is a base IRI an input to a parser? I suppose that's one
way to look at it. But I suggest:

  The terms delimited by "<>" are IRI references [RFC 3987].
  They stand for IRIs, either directly or relative to a base IRI.

next:

  or an optional datatype URI

s/URI/IRI/

 Because URIs can be long, SPARQL provides an abbreviation mechanism.

s/URI/IRI/ . not critical, since this is informative text. Hmm... maybe
"IRIs are just like URIs (and URLs even) but internationalized" should
go somewhere earlier.

Also: I'd say

 Because URIs can be long and repetetive, ...

my mailer says that's spelled wrong. Oh well. never mind.


next...

  The base URI for the resolution of relative URIs [RFC 3869] can
  be explicitly declared with the BASE keyword.

s/URI/IRI/ . And put that in a separate para. Maybe add an example,
even.


next...


 Turtle allows URIs to be abbreviated with prefixes:

s/IRI/URI/.
umm... didn't we just cover that?

 the subject, which is an RDF URI reference or a blank node
 the predicate, which is an RDF URI reference
 the object, which is an RDF URI reference, a literal or a blank node

s/RDF URI reference/IRI/ 3x

next...

  and datatype URI both match exactly

s/URI/IRI/
hmm... be explicit about the fact that it's the resovled IRI
we're comparing, not the possibly relative or qname form?

next...

  zero or more named graphs, identified by URI reference.

s/URI reference/IRI/

next...

  and each ui is a URI
  the GRAPH keyword allows access to the URIs naming the graphs
 bound to URIs of the graph being matched

s/URI/IRI/ 3x

next...

  The query below uses the graph with URI
http://example.org/foaf/aliceFoaf


s/URI/IRI/ optional. It's also a URI.

next...

 Each time a graph is read into the aggregator, it is given a URI
 by the local system.
 ... a different URI by the system

s/URI/IRI/ x2

next...

  # Graph: locally allocated URI: urn:x-local:graph1

Please don't use x- nor unregistered urn namespaces.
Use the example domain:

  # Graph: locally allocated URI: cid:rnd123@example

next...

  The URI for the date datatype has been abbreviated

s/URI/IRI/ (optional. it's also a URI)

next...

 g is a URI where g

s/URI/IRI/

next...

  The FROM clause gives a URI that ...
  A query processor may use these URIs ...
  The FROM clause a single URIs that indicates ...
  A query can supply URIs for the named graphs
  3. URIs

s/URI/IRI/ x5

next...

  with URI label

s/URI/IRI/ (optional. it's also a URI)

next...

  directly named by URI,

s/URI/IRI/

and so on (I stopped auditing at Explicit URIs...)



-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E
Received on Thursday, 21 April 2005 21:17:52 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:15:23 GMT