Re: Dropping WSDL RDF mapping of unknown extensions?

* Jacek Kopecky <jacek.kopecky@deri.org> [2007-03-12 22:10+0100]
> Hi Eric,
> a long time ago, you raised the comments that the WSDL RDF mapping [1]
> should use canonicalization for the parts that stay XML, i.e.
> documentation and optional extensions, both element and attribute. After
> all the time, I'm now considering whether those parts are really (in any
> way) useful parts of the RDF form of WSDL. 
> 
> Both documentation and unknown optional extensions can be happily
> ignored in WSDL, and I'm concerned now that canonicalization would be
> undue burden on RDF mapping implementations, with no discernible benefit
> (AFAICS).

My first reflex to the request is "why drop it", but my first reflex
to your suggestions is "he must have a good reason".

Does the mapping actually produce much burden? I think sawsdl is a use
case *for* it. We defined a graph representation of the SAWSDL
attribute, but if we had relesed the mapping first, folks might have
implemented sawsdl entirely in RDF, counting on the generic mapping to
give them what they need to access these new-fangled SAWSDL
attributes.

OTOH, maybe this use case does not justify the added cost. What *is*
that cost?

> What do you think about just plain dropping these parts altogether? 
> In the WSDL RDF mapping document this would mean that the second half 
> of section 2.2.1 and the whole section 2.2.2 would disappear.
> 
> Please let me know of your opinion, (and anybody else on the WS-Desc WG
> list will be welcome to chime in),
> 
> Jacek
> 
> [1] http://www.w3.org/TR/wsdl20-rdf/
> 
> 

-- 
-eric

office: +1.617.258.5741 NE43-344, MIT, Cambridge, MA 02144 USA

(eric@w3.org)
Feel free to forward this message to any list for any purpose other than
email address distribution.