Re: Understanding 'chaining'

Thanks Mark, I'm only keeping half an eye on developments so
explanations like this are very welcome.

On 26/11/2007, Ivan Herman <ivan@w3.org> wrote:

> My first, initial reaction is that the parsing/generation process might
> become somewhat more complex (putting my implementor's hat on).

That was my first impression too - it brought to mind the case of
running XSLT over arbitrary RDF/XML (nightmare or non-starter,
depending on your tolerance levels :-)

While tree-oriented models are rarely first choice for RDF developers
they can be handy for presentation and are likely to play a big role
in (X)HTML consumers at large. So I would imagine it fairly important
to be straightforward to access the RDF in the markup via these views.

I for one can't tell by observation whether these latest suggestions
might impact parsing in any significant fashion. But perhaps an ad hoc
perspective/metric on the complexity might be something like: how many
lines of code does it actually take to do -

RDFa -> XSLT -> Turtle
RDFa -> DOM (->treewalker) -> Turtle

Or to flip things around, how straightforward would it be to rewrite
(some of) SPARQL to XPath/XQuery/Javascript-on-DOM.

I'm not suggesting any of these approaches as a full alternative to a
dedicated RDFa parser (or parser+model+sparql engine), only I think
the introduction of conveniences should be balanced against impedance
mismatch with traditional (X)HTML tools, to avoid unnecessary pain
further down the road.

Cheers,
Danny.

-- 

http://dannyayers.com

Received on Tuesday, 27 November 2007 09:45:21 UTC