Thanks for the details
I see:
&format=XML
format= and output= do the same thing.
(Worse, the code already forces the lookup map to contain lowercase keys
so this looks like it was intended)
Now fixed in svn and in tonight's dev build of Fuseki.
Andy
On 18/09/13 17:53, Jacek Grzebyta wrote:
I have launched fuseki manually and what RelFinder does:
17:42:35 INFO Server :: Fuseki 0.2.8-SNAPSHOT
2013-07-25T18:33:01+0100
17:42:35 INFO Server :: Started 2013/09/18 17:42:35 BST on
port 3030
17:43:16 INFO Fuseki :: [1] GET
http://xxxxxxxx:3030/phibase/query?&format=XML
&query=SELECT%20%3Fs%20%3Fl%20WHERE%20%7B%20%20%3Fs%20%3Fsomeprop%20%3Fl%20%2E%20%20%7B%20%20%3Fs%20%3Chttp%3A%2F%2Fwww%2Ew3%2Eorg%2F2000%2F01%2Frdf%2Dschema%23label%3E%20%3Fl%20%20%7D%20UNION%20%7B%20%3Fs%20%3Chttp%3A%2F%2Fpeople%2Ebrunel%2Eac%2Euk%2F%7Ecsstlls2%2Fontologies%2Fphi%2Dbase%2Drdf%2Eowl%23gene%3E%20%3Fl%20%7D%20%2E%20FILTER%20regex%28%3Fl%2C%20%27ge%27%2C%20%27i%27%29%2E%20FILTER%20%28lang%28%3Fl%29%20%3D%20%27%27%20%7C%7C%20langMatches%28lang%28%3Fl%29%2C%20%27en%27%29%29%2E%20%7D%20LIMIT%2020%20
17:43:16 INFO Fuseki :: [1] Query = SELECT ?s ?l WHERE { ?s
?someprop ?l . { ?s <http://www.w3.org/2000/01/rdf-schema#label> ?l }
UNION { ?s <
http://people.brunel.ac.uk/~csstlls2/ontologies/phi-base-rdf.owl#gene> ?l }
. FILTER regex(?l, 'ge', 'i'). FILTER (lang(?l) = '' ||
langMatches(lang(?l), 'en')). } LIMIT 20
17:43:16 INFO Fuseki :: [1] exec/select
17:43:16 INFO Fuseki :: [1] 200 OK (228 ms)
The problem is labeled yellow. I guess "format" (not output) shouldn't be
key-sensitive. And my patch made it. But you are right that not all
parameters should be key-insensitive so my patch is not valid now.
Regards,
Jacek
On 18 September 2013 17:34, Andy Seaborne <[email protected]> wrote:
On 18/09/13 15:23, Jacek Grzebyta wrote:
Dear All,
Hi I try to launch fuseki with RelFinder RDF visualiser. I found that
fuseki is key sensitive for parameters value in sparql request. My patch
makes fuseki unsensitive. Please could you put it into the code if you
think it's good and correct.
Regards,
Jacek
Some parameters must be case sensitive (e.g. stylesheet name, javascript
call back name). I take it we're talking about the names used in
"output=", the query string parameter to set the content-type. (please
confirm - you not talking about the "Accept" header in the HTTp request).
I've recorded this as;
https://issues.apache.org/**jira/browse/JENA-534<https://issues.apache.org/jira/browse/JENA-534>
and it seems to me no reason not to do it; output= is a Fuseki extension
and not part of the standard (and not uncommon elsewhere).
But why is RelFinder sending output= at all, and why isn't it in the right
case? (seems to be a minor bug in relFinder?)
Andy
