Re: How/Why to implement Reification efficiently...

I have a (fairly definite :-) opinion on this:

> Another example:
> 
> <?xml version="1.0"?>
> <rdf:RDF
>   xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
>   xmlns:s="http://description.org/schema/">
>   <rdf:Bag ID="BAG_ID">
>      <rdf:li resource="xyz"/>
>      <rdf:li resource="abc"/>
>   </rdf:Bag>
>   <rdf:Description ID="MyID">
>    <s:documents>
>     <rdf:Description aboutEach="#BAG_ID">
>       <s:creator>Karl</s:creator>
>     </rdf:Description>
>    </s:documents>
>   </rdf:Description>
> </rdf:RDF>
> 
> If you use one of the four parser available online (hope
> I didn't forget one?)
> [1],[2],[3],[4], you get four different results.
> This problem has been discussed a while ago (see [5],[6]
> or the question in [7]:Which should we implement?),
> but there is still no decision. In my opinion the
> reason for this is that there is no authority who
> could make such a decision (or recommendation).

Although the RDF M+S "allows" this, it is in fact illegal to write an
"aboutEach" construct anywhere except on the top level inside the
rdf:RDF element. Parsers should report an error.

So you may ask, "why is this illegal". Here's my explanation: the said
"aboutEach" construct does not result in a single node as its "value"
(when "evaluated"), and hence causes problems when the parser attempts
to build a graph.

Regards,

	- Ora

-- 
Ora Lassila <daml@lassila.org> +1 (781) 993-4603

Received on Thursday, 30 November 2000 11:10:30 UTC