RE: A present for Guy Fawkes night

-------- Original Message --------
> From: Dan Connolly <>
> Date: 5 November 2004 14:28
> 
> On Fri, 2004-11-05 at 13:39 +0000, Seaborne, Andy wrote:
> > Simon and I were discussing interoperability testing after the
> > telecon. 
> > 
> > It is traditional on Bonfire Night (Guy Fawkes night) in the UK to
let
> > off fireworks to celebrate a failed attempt to blow up the houses of
> > Parliament.  Fireworks include SPARQLers.  So it is with some
> > trepidation, I have put up a SPARQL server based on some rather
> > unproven code.
> 
> Nifty...
> 
> > What works: (and I use "works" in the loosest possible sense)
> > + SELECT (RDF results only),
> 
> Ah... RDF descriptions of the result set, not the matching
> subset? Oh...
> 
> >  CONSTRUCT (maybe)
> 
> Is that how to get the matching subset?

Yes - CONSTRUCT * should give you that.  Should work with cwm
log:semantics (definition:"should" => an earlier version using RDQL
did).

See attached for issing a query from cwm and listing book titles (two
different ways).

> I guess I can try
> it and see... ugh... escaping queries by hand is a pain...

:-)

I see from the logs you are pushing the limits of my MIME type and
charset negotiation code as well ... it doesn't handle q= stuff in
charsets (anyone know of good libraries or example code for this?)

> do you offer a forms-based interface? I don't see one...
> but I can make one easily enough (attached).

Great!  I'll add that.

If you type the query into the address bar of your browser, it will do
the escaping for you (well, at least Firefox and IE do).  There is a
command line tool in CVS that at least does the escaping and building
the query string for you.

Anyone got a good logo?  You should see the number of requests for
/favicon.ico!

> 
> Sure enough... I can get...
> 
> <http://example.org/book/book5>
>       <http://purl.org/dc/elements/1.1/title>
>               "Harry Potter and the Order of the Phoenix" .
> 
> with
> 
> CONSTRUCT *
> WHERE ( $x $p $v) AND $v =~ /Phoenix/
> 
> 
> This is kinda fun!

DESCRIBE <http://example.org/book/book5> should also work.

> 
> 
> > Protocol:
> > 
> > HTTP request must specify the language (lang=) and the query string
> > (query=) Languages are "SPARQL" or
> > "http://jena.hpl.hp.com/2003/07/query/SPARQL" 
> > 
> > In a browser, the results will come back text/plain for simple
> > display. You can ask for application/rdf+xml as well.
> 
> Now there's a front-side-of-one-page protocol spec if I've ever
> seen one!

Room for the status, copyright notices, logos and revision control logs.

Seriously, this isn't supposed to be a proposal - I am just resuing what
I had lying around and guessing where Kendall is going.  I have no
further information.

	Andy

> 
> 
> > The code is in
> > cvs.sourceforge.net:/cvsroot/joseki/Joseki3
> >   and
> > cvs.sourceforge.net:/cvsroot/jena2/BRQL (but the query engine is
> > called ARQ). 
> > 
> > 	Andy
> --
> Dan Connolly, W3C http://www.w3.org/People/Connolly/
> D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E