W3C home > Mailing lists > Public > www-rdf-comments@w3.org > October to December 2002

Re: Comments on abstract syntax

From: Jeremy Carroll <jjc@hpl.hp.com>
Date: Mon, 18 Nov 2002 16:29:40 +0100
To: timbl@w3.org
Cc: www-rdf-comments@w3.org

Hi Tim

thanks for your comments on the RDF abstract syntax of 29th July.


There is a new version


which addresses many of your issues.

This message concerns those to do with the abstract syntax:

Can we call theis "Graph Abstract Syntax"? It is bad enough calling abstract
syntax abstract syntax when it isn't syntax -- but calling it syntax is
stretching it to far. Most folks would expect a syntax here.


about URI encoding etc.
I don't think this is well explained. The distinction missing is betwen the
string which identifies something (a URI with or without a "#") and a string
in a document which makes a reference by being either that URI or an
encoding or base-relative version of it. The TAG will clear this up I hope
soon. It should be enough to point out that the abstract syntax uses URIs
(with or without #) and any doucemnt refering to the resources in questoin
can of course n a real syntax use a relative URI where a base URI is shared
beteween writer and reader.
Not much change there.
I am waiting on IRI-Everywhere at the TAG.
Note that the WG has decided to use absolute URIs. Relative URIs are only
used within specific syntaxes (e.g. RDF/XML) as abbreviations for absolute

I have to say I have a problem with RDF being tied to always have to have an
XML literal as a base type. This breaks layering - and level breaking
features should I believe be left for another layer. You should not require
any RDF machine to have to include an XML infoset system. The choice of XML
syntax was supposed to be an enginering but arbitrary choice.

If the XML included can be represented by a cannoncial tstring then this
makes it easier. I don't see the cannoncialization as having been completely
defined here. An RDF engine needs to be able to say whether two literals are
This is largely fixed, XMLLiteral is now a datatype.
The canonicalization isn't completely defined, by intent ...
The current round of specs seems to have lost the relevant text.
Somewhere we should say that from RDF/XML to the abstract syntax you take
the exclusive canonicalization *with or without* comments and *with any
InclusiveNamespaces PrefixList* paramater.

The undefinedness on the issue of comments and namespaces that are not
visibly used is because the WG felt there was insufficient implementation
experience to make a call on these.

Surely if you specify exactly a tag, then it makes the job of xcomparing XML
things that much cleaner. I suggest you specify some tage like rdfxml:lit or
something. It gets much easier eg to make tests for equality, run tests on
parsers etc, if this is defined once and for all.
The tag is "rdf-wrapper" (no namespace); the value space of the datatype is
Canonical XML documents with root element rdf-wrapper.

You can't say "Many nodes are blank", as some applications will have many
but some will have none at all!

Suggest you do not introduce the concept "tidy". It is a waste of breath and
will make people think there are two RDFs, tidy and untidy. Later on you sat
that the set of nodes in an ARDF graph is tidy. Pleasse just roll that into
the definition here or there: No two nodes in an RDF graph have equal

Add: A "boring^H^H^H^H^H^Hclassic" RDF graph is one in which there is a
label for every arc.
]] and
In general, s/p/I(p)/ in last bullet except for classic graphs.
Rejected. Our charter does not permit such a dramatic change to RDF graphs.

For more detail about how we have accepted your proposed changes see the
current WD. Please respond if you wish us to officially consider your
preference for allowing blank nodes in predicate position. Currently you
have just had an editor reject it without consulting the WG.

Received on Monday, 18 November 2002 10:30:52 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:44:01 UTC