> Peter, > > Are you able to implement a pattern similar to RelFinder where the > SPARQL endpoints are configurable? We also do something similar re. our > iSPARQL tool (i.e., have a list of known endpoints but leave the user > with option to add others). > > I've played around extensively with Pivot (your stuff and ours), so I > am > very interested in a loosely coupled solution. >
Yes, in our system one can define ad hoc federations across multiple SPARQL endpoints. However, the performance in such a loosely coupled scenario is - not surprisingly - orders of magnitude worse than in a centralized scenario, especially when the queries involve joins across multiple SPARQL endpoints. We are working on improving this situations as well ;-) Regards, Peter