RE: DESCRIBE ENDPOINT

> -----Original Message-----
> From: public-rdf-dawg-request@w3.org [mailto:public-rdf-dawg-request@w3.org]
> On Behalf Of Kjetil Kjernsmo
> Sent: 01 July 2009 14:03
> To: public-rdf-dawg@w3.org
> Subject: Re: DESCRIBE ENDPOINT
> 
> On Wednesday 01 July 2009 14:50:05 Gregory Williams wrote:
> > Doesn't this assume that you have control over the entire server? This
> > would be a non-starter for any endpoint that didn't, e.g. a directory-
> > based apache handler, or a CGI.
> 
> Yup, that's partly why I suggested it as an optional feature. To support a
> feature that says "this is what this server can do", you would necessarily
> need to have some influence of the server config anyway, but indeed, not
> everyone can be expected to support this. I don't know how big that problem
> would be in reality though?
> 
> > And there's still the tool support issue of using OPTIONS, compared to
> > the much more prevalent GET/HEAD/POST verbs.
> 
> Yeah, it is a real problem, but nevertheless, I think it is better than
> nothing for those who can use it, and I see no obvious drawbacks for those
> that do.
> 
> Kind regards
> 
> Kjetil Kjernsmo

In the same vein as DESCRIBE SELF would be a new verb in the protocol not the QL:

http://example/service?options


(c.f. http://example/service?query=)

By putting it here, there are no issues of "DECRIBE SELF" independent of the protocol (the API case) and also that the request is, must likely, routed to the query protocol processing code (e.g. servlets) so does not assume control of anything more than was the case for the ?query in the first place.  Is SOAP compatible (ish).  

 Andy

Received on Friday, 3 July 2009 20:12:49 UTC