This seems a lot stronger in the end-- you get the Fuseki integration desired, but this would also be useful outside that context. And the code is going to be rather a lot easier to write. Much of Fuseki's code concerns managing HTTP action, which isn't really of the essence to your goal.
--- A. Soroka The University of Virginia Library > On Sep 29, 2016, at 11:34 AM, Andy Seaborne <[email protected]> wrote: > > > > On 28/09/16 23:59, Jason Koh wrote: > ... >> Questions: >> 1. What is the functions related to SPARQL query processing? I am looking >> at SPARQL_Query.java, ActionSPARQL.java, etc., but this kinda >> reverse-engineering does not work well until now. I will look into the >> JavaDoc more, but I would appreciate if I can get an enlightment here. > > There is another way to hook in that may be useful. > > Instead of the HTTP request lifecycle, hook into the transaction lifecycle on > the dataset. > > > A DatasetGraph wrapper means you can add the functionality to any existing > dataset. > > This is in jena-arq. > > All Fuseki requests on the data are performed inside a transaction so if you > catch a write transaction, and when it commits, trigger SPIN processing. > > Fuseki uses Jena's assembler mechanism to describe datasets so they can be > setup and configured with needing application Java code > > Andy
