Re: Request to publish HTML+RDFa (draft 3) as FPWD

On Sep 22, 2009, at 15:53, Sam Ruby wrote:

> My conclusion is that defining RDFa in HTML in terms of a DOM or an  
> Infoset are but two of the possible ways of achieving the desired  
> result, namely being precise as to what triples MUST be produced  
> from a given input.

The issues raised so far demonstrate that the failure to use the  
conceptual model(s) of the underlying specs lead to things ending up  
leaving unspecced cases. Also, I believe what I've said about XOM  
earlier on the RDFa TF list and what James said recently about lxml on  
this list show that the mismatch of the definitional model is a  
problem when off-the-shelf tools have been crafted to fit the  
definitional model of the underlying specs.

> If the current draft doesn't contain that level of precision, that's  
> the basis for one or more bug reports.  But in my opinion defining  
> RDFa in terms of the underlying processing model -- while it may end  
> up being the most convenient way to achieve precision or ultimately  
> end up being a virtual necessity -- is not itself a fundamental  
> requirement.


My point is that it's a fundamentally bad idea to use a definitional  
model that gives rise to bug reports, lack of precision and issues  
with off-the-shelf libraries that wouldn't arise if the definitional  
models of the underlying specs were used.

That it's possible to use a mismatching definitional model doesn't  
make it good spec writing to fail to use the model of the specs you  
are layering a new spec over. (I'm not suggesting the use of the  
cumbersome Infoset terminology directly. HTML5 uses DOM terminology,  
and that's OK. Atom defines shorthand [http://tools.ietf.org/html/rfc4287#section-1.3 
] that's OK, too.)

-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/

Received on Tuesday, 22 September 2009 13:28:54 UTC