On 11/03/13 18:07, Melanie Courtot wrote:
> I did see that and wondered if it may cause the issue. When I copy/paste the query that is passed as parameter from the console, I did also notice that the extra parameter
> // This string is appended the 'endpoint' variable and the query to it again to give a link to the "raw" query results.
> 'endpoint_query_url': "?output=text&query=",
>
> is not present. I thought maybe this specific parameter is not supported by Sesame, and therefore ignored. If I add it "manually" it still returns an attachment. (I also tried output=json, format=text to mimick virtuoso, no success)
The value of this parameter/variable is not used when querying the
endpoint for sgvizler visualization, but when an error message is given
to the user. Then a link to the "raw" results are given using this
variable, so that the user can check that the results are as expected.
This parameter is apparently badly explained, others have misunderstood
it before. I added it because Joseki, which I had installed locally when
I started writing sgvizler, has this feature.
> The part that bothers me is that the day before the queries were working fine against the same server, and as far as I know nothing has been updated nor parameter changed. (It is locally installed on my machine, and was not even restarted in between)
Strange, but difficult for me to help here. What do could do is try to
run the endpoint on localhost:80 or serve the html pages on the port
that the endpoint runs on. Or setup some proxy that translates the
requests between the ports. What will get you out of CORS trouble.
I remember testing a PHP proxy script, that worked fine. I don't have it
now, but a search gave me, e.g.,
https://github.com/cowboy/php-simple-proxy/blob/master/ba-simple-proxy.php
which looks promising, but I have not tested it.
> Ok - it seems support for jsonp is in the works for Sesame and should be out fairly soon. If I can't find any solution I'll either wait or switch triplestore - I am just testing locally, so not a big deal (though it seems a bit of a shame)
Yes, I looked around and found this, and include it for others:
http://www.openrdf.org/issues/browse/SES-1019
Please report back if there are news.
Thanks,
Martin