Yes, that is correct. The task should have been explicitly spawned as a daemon (which would not prevent app from exiting), or canceled during graph shutdown. It has been fixed in trunk, and we're adding integration tests to check this kind of problem. We totally missed this with normal unit testing. Sorry about that.
Also, we had some dangling RMI in neo4j-shell. Fixes for both should be available in snapshot artifacts and will be released with the next milestone, as Mattias mentioned. Cheers, Andreas On Oct 13, 2010, at 5:11 PM, Chris Gioran wrote: > On Wed, Oct 13, 2010 at 5:48 PM, Rick Bullotta > <rick.bullo...@burningskysoftware.com> wrote: >> Lucky guess. ;-) > > It seems that the load() method of the UdcExtensionImpl creates a > java.util.Timer to do its thing but I cannot find anywhere a call to > the timer's cancel(). In that case (as per the javadoc) the Timer will > take arbitrarily long to terminate. I think this explains the observed > behavior. > > cheers, > CG > _______________________________________________ > Neo4j mailing list > User@lists.neo4j.org > https://lists.neo4j.org/mailman/listinfo/user _______________________________________________ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
