W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > June 2001

XML:Base - impact on RDF (first pass)

From: Jan Grant <Jan.Grant@bristol.ac.uk>
Date: Tue, 5 Jun 2001 16:04:40 +0100 (BST)
To: RDFCore Working Group <w3c-rdfcore-wg@w3.org>
Message-ID: <Pine.GSO.4.31.0106051528560.2880-100000@mail.ilrt.bris.ac.uk>
I've made a cursory "this is what changes/needs thinking about" skim of
xml:base and the M+S document; I'd encourage people to contribute
anything else that I've missed here.

RDF is serialised as an XML document. Thus, xml:base would appear to
have an impact. You'd think so, anyway, at least if an RDF parser is
built on top of an xml:base-compliant XML tokeniser.

With the inclusion of an interpretation for xml:base attributes:

- relative URIs appearing in rdf:about, rdf:resource, rdf:type
  attributes should probably be resolved relative to the xml:base
(according to the xml:base spec and normal resolution of relative URIs)

This is probably a good thing and what most people would expect.

- fragment identifiers and rdf:ID attributes should probably also be
  interpreted relative to this base, (depending on what people decide to
  do with rdf:ID).


- other occurrences of URI-references in the grammar occur in
  productions [6.7] (aboutEach) and [6.8] (aboutEachPrefix). The latter
  clearly doesn't matter any more :-) but the former (if aboutEach
  stays around) would be affected.

Again, this is probably a good/obvious/expected behaviour.

- Thus far I've not paid too much attention to what RDF does with
  XML markup in a property with rdf:parseType="Literal" - but xml:base
  would probably be inherited in this case (according to the spec) and
  this would need to be sorted out.


Without the inclusion of the recommended interpretation:

- in the following

<rdf:Description rdf:about="http://random.ioctl.org/#bletch"
	xml:base="http://other.ioctl.org/#zub"
	 />

the xml:base part matches the propAttr production in the grammar of
section 6 (production [6.10]) and will (according to section 6 thus far)
emit something like:

<http://random.ioctl.org/#bletch> <http://www.w3.org/XML/1998/namespacebase> "http://other.ioctl.org/#zub".

...which looks a bit broken to me. (I'm using the NS-concatenation rule
together with the standard binding for the xml namespace here).

Other benefits of xml:base are the ability to stick relative URIs in
test cases and ensure that they always resolve to the same absolute URIs
although I don't consider this to be sufficient motivation in itself to
include xml:base.

-- 
jan grant, ILRT, University of Bristol. http://www.ilrt.bris.ac.uk/
Tel +44(0)117 9287163 Fax +44 (0)117 9287112 RFC822 jan.grant@bris.ac.uk
Bolstered by my success with vi, I proceeded to learn C with 'learn c'.
Received on Tuesday, 5 June 2001 11:06:13 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:37:01 EDT