RE: OWL 2 Struct. Spec specification problem? - "absolute IRI"

Hi WG!

According to RFC 3987, Section 2.2, the ABNF for the non-terminal "IRI" is
given by (after some non-terminal substitution not made in the spec but by
myself):

     IRI ::= absolute-IRI [ "#" ifragment ]

So it seems reasonable to me to only do a very small change to the
Structural Spec of the form:

   Each IRI MUST be absolute
   with an optional fragment.

Btw, the RDF-Based Semantics has this as a topic as well:

 
<http://www.w3.org/2007/OWL/wiki/RDF-Based_Semantics#topic-ont-noteiriref>

Michael

From: public-owl-comments-request@w3.org
[mailto:public-owl-comments-request@w3.org] On Behalf Of Barclay, Daniel
Sent: Monday, August 31, 2009 7:59 PM
To: public-owl-comments@w3.org
Subject: OWL 2 Struct. Spec specification problem? - "absolute IRI"

In the OWL 2 Web Ontology Language Structural Specification and
Functional-Style Syntax draft at
http://www.w3org/TR/2009/CR-owl2-syntax-20090611/, section 2.4, IRIs,
says:

   Ontologies and their elements are identified using Internationalized
   Resource Identifiers (IRIs) [RFC3987]; thus, OWL 2 extends OWL 1,
   which uses Uniform Resource Identifiers (URIs).  Each IRI MUST be
   absolute.

That wording probably should be adjusted so that readers don't
accidentally assume that those IRIs must be absolute in the sense in
which RFC 3986 uses the term (that is, not having fragment identifiers).


That OWL specification is based on RFC 3987, and RFC 3987 is based
on RFC 3986.

Unfortunately, RFC 3968 doesn't use "absolute" in the normal sense
(meaning not relative to something); it uses "absolute" to refer to not
having a fragment identifier.

Its section 4.3 (see http://tools.ietforg/html/rfc3986#section-4.3)
says:

   4.3. Absolute URI

     Some protocol elements allow only the absolute form of a
     URI without a fragment identifier.  For example, defining
     a base URI for later use by relative references calls for
     an absolute-URI syntax rule that does not allow a fragment.

         absolute-URI  = scheme ":" hier-part [ "?" query ]

      ...

Although that first sentence could be explained as first referring to
URIs that are absolute in the usual sense and then only referring
to the subset of those without fragments, it easily sounds like it's
defining "absolute" to mean not having a fragment component.

Regardless of that first sentence, the name of the absolute-URI
production strongly implies that "absolute URI" refers to the strings
matching that production.  Given the lack of anything saying that
the English phrase "absolute URI" does not in fact refer to what the
absolute-URI grammar production generates, the only reasonable
interpretation is that RFC 3986 defines "absolute URI" to refer to
URIs without fragment components.

Since allowing only non-fragment IRIs is _not_ what the OWL
specification means to specify (right?), the OWL spec.'s text should
also refer to not being relative or not being an IRI reference (or
should refer to allowing fragment portions) to be clear.


Daniel
--
(Plain text sometimes corrupted to HTML "courtesy" of Microsoft Exchange.)
[F]

--
Dipl.-Inform. Michael Schneider
Research Scientist, Dept. Information Process Engineering (IPE)
Tel  : +49-721-9654-726
Fax  : +49-721-9654-727
Email: michael.schneider@fzi.de
WWW  : http://www.fzi.de/michael.schneider
=======================================================================
FZI Forschungszentrum Informatik an der Universität Karlsruhe
Haid-und-Neu-Str. 10-14, D-76131 Karlsruhe
Tel.: +49-721-9654-0, Fax: +49-721-9654-959
Stiftung des bürgerlichen Rechts, Az 14-0563.1, RP Karlsruhe
Vorstand: Prof. Dr.-Ing. Rüdiger Dillmann, Dipl. Wi.-Ing. Michael Flor,
Prof. Dr. Dr. h.c. Wolffried Stucky, Prof. Dr. Rudi Studer
Vorsitzender des Kuratoriums: Ministerialdirigent Günther Leßnerkraus
=======================================================================