I’ve updated my little dotNet demo – the one that is a socketlistener with an SSL protocol handler From “Mentalis” combined with a webid validation agent in the onCertReceive event). Given a webid from a cert provisioned by ODS presented by an https client, the agent will now have uriburner’s sparql endpoint perform the ASK query, for a reference indicated in the FROM clause. Going beyond that ASK and seeking a higher assurance is now the goal.
The ASK ask query seems of quite limited value (now it works).
Noting how others’ validators work (or block), I now check the DNS authority of the webid URI (to see if it resolves in Xms or less). The point is, it doesn’t wait 15s (as it might do). And, then, I similarly check the status of the remote resource (which must return a 200 status message in YmS or less). If these liveness checks work (building in dependency on the DNS and “the web” as set of endpoints), the code then concocts a wrapper URI for the resource, inviting a translation service to always stream out the XML form of triples.
Such a translation service would be hosted locally, in any operational deployment. Thus, I see no need add no timeouts. In demo land, the code can be become blocked if the translator does not respond.
The concocted URI we formed above now goes into the FROM clause of the sparql query, along with some uriburner service requirement flags that induce the sparql server not to consume its own triple store but to go out and populate a temporary triple store on the fly. it does it for the purposes of just this transaction – by parsing the identified resource documents. And, if there are multiple FROM clauses it does this several times.