W3C home > Mailing lists > Public > xmlschema-dev@w3.org > February 2001

Re: does RDF require understanding all 82 URI schemes?

From: Ralph R. Swick <swick@w3.org>
Date: Mon, 12 Feb 2001 08:06:38 -0500
Message-Id: <200102121307.IAA18647@tux.w3.org>
To: ht@cogsci.ed.ac.uk (Henry S. Thompson), "Sean B. Palmer" <sean@mysterylights.com>, "Jonathan Borden" <jborden@mediaone.net>, "Dave Beckett" <dave.beckett@bristol.ac.uk>, "David Megginson" <david@megginson.com>, <www-rdf-interest@w3.org>, <xmlschema-dev@w3.org>
At 08:27 AM 2/12/2001 +0000, Henry S. Thompson wrote:
>Indeed, the one with a # on the end is a perfectly good namespace URI, 
>it's just not _the_ namespace URI for XML Schema, and documents with
>that as their namespace will _not_ be recognised _by XML Schema-aware
>processors_ as defined in the XML Schema spec. as being XML Schemas.

yes, I concur.

It seems to be a common misperception that RDF 1.0 _requires_
a specific character in the right-most position of a namespace URI.
This was not the intention of the RDF designers and a careful
reading of  http://www.w3.org/TR/1999/REC-rdf-syntax-19990222/
will reveal that neither "#" nor "/" is required at the end.  Indeed,
there was much discussion about whether RDF should choose a
specific delimiter when concatenating the namespace name with
the local name and it was decided to leave the full URI entirely
within the scope of the namespace owner to define.

At the present time, the XML language specifications do not
endorse any use of a namespace URI other than string
comparison with other namespace URIs.  The RDF designers
asserted from the beginning that we wanted to use URIs to
their fullest , not just as strings to be compared.  That is, RDF
_expects_ that any URI can be presented to the Web for resolution.
But RDF still treats URIs as "opaque" in the sense that RDF does
not require any parsing of the URI.

I acknowledge that, strictly speaking, opacity and concatenation
are in conflict.  However, I still do not think it was a mistake for RDF
to rely on other characteristics of URIs as specified by RFC 2396.

Sean wrote
>you *cannot* restrict people
>to using a few special characters at the end of their RDF namespaces,

I disagree.  We _could_ place such restrictions.  But the primary
reasons for requiring a specific delimiter eventually came back
to an implication of endorsement on the act of parsing a property
or class URI to extract its namespace.  We want to make it clear
that RDF does *not* endorse parsing URIs to extract RDF semantics.
That is why RDF Schema proposes

So a proposed restriction to end all RDF namespace names with
"#" or "/" was thought to be a gratuitous (unnecessary, unwarranted)
constraint.  I do feel a tiny bit of remorse for the confusion we have
added by giving namespace name designers this extra freedom.
Received on Monday, 12 February 2001 08:07:57 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:55:51 UTC