The bad syntax was my mistake. I was editing his example in Eclipse, and while
testing with SELECT instead of DESCRIBE as he suggested, I probably introduced
that typo and forgot to put DESCRIBE back there.
Thanks for the detailed explanation Andy! :)
Bruno
From: Andy Seaborne <[email protected]>
To: [email protected]
Sent: Thursday, November 27, 2014 6:35 AM
Subject: Re: Valid JSON keys in SPARQL endpoint response for Describe queries
On 27/11/14 02:02, Bruno P. Kinoshita wrote:
>
> Hello Andy!
>
> The more I learn, the more I realize how little I know :)
:-)
> I think his code's gist expired, but FWIW here is another gist from my local
> copy [1].
> Thank you!Bruno
> [1] https://gist.github.com/kinow/f36c7cf989a4ee1b5478
?? Bad syntax!
"selectan" not "select"
and it calls execSelect, not execDescribe. May be that is why they
force "format="
The original had "describe *", does not mention ?s ?p ?o and got back a
result set with (?s, ?p, ?o) so it looks like the query was issued with
the override: format=application/sparql-results+json so the far did do
something plausible, although asking for a graph and setting "accept" to
a result set format has a HTTP status code of 406. (How does the caller
know it's (?s, ?p, ?o) otherwise?)
The default accept header for a model (CONSTRUCT or DESCRIBE)
text/turtle, application/n-triples, application/rdf+xml;q=0.9,
application/turtle;q=0.8, application/x-turtle;q=0.8, text/rdf+n3;q=0.7,
application/n3;q=0.6, text/n3;q=0.6, text/plain;q=0.5, */*;q=0.1
which does not mention JSON but does have */*. The format= overrides that.
(note to selves - add JSON-LD, remove N3).
Andy
