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

Re: Followup on rdf-ns-prefix-confusion

From: Dave Beckett <dave.beckett@bristol.ac.uk>
Date: Sun, 13 May 2001 18:43:47 +0100
To: Aaron Swartz <aswartz@upclink.com>
cc: Art Barstow <barstow@w3.org>, w3c-rdfcore-wg <w3c-rdfcore-wg@w3.org>
Message-ID: <1867.989775827@tatooine.ilrt.bris.ac.uk>
>>>Aaron Swartz said:
> Dave Beckett <dave.beckett@bristol.ac.uk> wrote:
> 
> > 2.  Existing parsers MUST accept the about, id etc. attributes in
> >   unqualified form on input but MUST NOT emit them on output.
> 
> Looking over this again, I think this should be changed to:
> 
>     Parsers MAY accept the about, id, etc. attributes in
>     unqualified form on input but MUST NOT emit them on output.
> 
> Why would we want to make an additional rules for "existing parsers"?

OK, 'existing' was too much.  But the MUST should stay.  The core of
existing syntax MUST continue to be accepted and thus this wording,
which is a conformance declaration of RDF/XML syntax processors (and
hence new to RDF) should apply to ensure the existing syntax remains.


> Also, do we define how parsers should interpret unqualified attributes? If
> so, how should one process this testcase (as raised by Jason Diamond on
> rdf-interest):
> 
> <eg:Class about="http://foo">
>   <eg:property>bar</eg:property>
> </eg:Class>

The wording 
  "parsers MUST accept the about, id, etc. attributes in unqualified form"
applies thus this (continues to be) a legal typedNode production.
Existing parsers that don't accept that are broken, IMHO since not
only does the RDF M&S document contains examples of both, but the
existing grammar uses the about attribute without a prefix.

So taking Jason's example in
  http://lists.w3.org/Archives/Public/www-rdf-interest/2001May/0098.html
and fleshing it out - see first attachement [cloaked as text/html to get
it stored in mail archive]

In future we will add words that require that all processors of the
above MUST return two statements (s,p,o form):

  (http://foo,  rdf:type, http://example.org/Class)
  (http://foo,  http://example.org/property, "bar")

and is equivalent to the rdf:about form as given in the second
attachment.  On output, conformant processors MUST emit RDF/XML like
the second attachment.  Please ignore issues of serialising ordering,
namespace prefixes that haven't been considered yet.


Is this clearer?

(Note: online SiRPAC gets this wrong with first example and needs
rdf:about form of second example.  I haven't tested other parsers)

Dave



Received on Sunday, 13 May 2001 12:43:59 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:35:48 EDT