Re: Issue http://www.w3.org/2000/03/rdf-tracking/#rdf-ns-prefix-confusion

Art,

Thanks for this.  Its good to provide a rationale.  I suggest it would
be good practice for us to try to provide a written rationale for all
the decisions that we make.

The main difference I see between this proposal and the earlier one is
the handling of legacy.

The m&s grammar is clear; attributes such as ID, about etc are
unqualified.  Both proposals suggest that this is changed to require
qualified attributes.

If the grammar were changed as in this proposal, all strictly conforming
RDF 1.0 would not be legal RDF 1.1 (well maybe you could argue that its
legal, but it wouldn't mean the right thing).

The earlier proposal mandates that an RDF 1.1 processor will process 
'correct' RDF 1.0 correctly.  This proposal leaves that decision up to
the implementor.

Do we have any good reasons for preferring one approach to the other?

Will this change impact existing data and specifications based on RDF?

Brian


Art Barstow wrote:
> 
> Hi,
> 
> We discussed:
> 
>  http://www.w3.org/2000/03/rdf-tracking/#rdf-ns-prefix-confusion
> 
> at a Team meeting and we propose:
> 
> 1. The grammar in section 6. of the M&S spec:
> 
>   http://www.w3.org/TR/REC-rdf-syntax/#grammar
> 
>  should be changed so that all of the RDF attributes (e.g. about,
>  resource, etc.) are qualified with a RDF namespace.
> 
> 2. Text along the lines of the following should be added to the M&S spec:
> 
>   [[
>     All elements and attributes in RDF must be namespace qualified
>   ]]
> 
> Among the reasons for this position:
> 
> o It is a myth that unqualified attributes inherit the namespace of
>  their enclosing element.   See the Namespace Myths, particularly
>  Myth #4:
> 
>    http://www.xml.com/pub/a/2000/03/08/namespaces/index.html?page=2
> 
> o XPath support is becoming more pervasive and by using it, finding
>  qualified attributes is fairly easy
> 
> o Search engines (such as google) would have trouble following pointers
>  from one RDF document to another and a simple XPath expression to find
>  pointers would help them without having to process all of RDF
> 
> o Prefixed attributes work today with most of the deployed code
> 
> Art
> ---

Received on Wednesday, 25 April 2001 08:02:58 UTC