Yeah, I agree. This seems like something many people would need to build otherwise. I'm not expert enough in our current l1/l2 timeout settings to know whether we sometimes have this timeout info now. If so, then it seems we should use it to definitively send node left events.
On Jan 27, 2009, at 4:40 AM, Geert Bevin wrote: > > On 27 Jan 2009, at 06:14, Taylor Gautier wrote: > >> I think while that might be helpful, it's not clear that all edge >> cases can be eliminated and even if there is one edge case left >> over, it means the app developer will still have to handle the >> situation. >> >> I suppose in the final analysis, we could possibly invent a "hard >> timeout" which would say that no matter what x seconds after a >> disabled event, leave the cluster and send the node left event, >> meaning TC is ultimately responsible for the events and the >> programmer in such a scenario simply has to write code to respond to >> the events. I could see arguments for this as it leaves the event >> timing/delivering to TC such that if we figure out a better way to >> deliver more reliable events in the future, in theory the app code >> doesn't have to change. > > I would like that, I kind of dislike the fact that a programmer still > has to handle timeouts themselves and basically setup their own > callback mechanism. Imho there should be a unified approach to this. > > -- > Geert Bevin > Terracotta - http://www.terracotta.org > Uwyn "Use what you need" - http://uwyn.com > RIFE Java application framework - http://rifers.org > Flytecase Band - http://flytecase.be > Music and words - http://gbevin.com > > _______________________________________________ > tc-dev mailing list > tc-dev@lists.terracotta.org > http://lists.terracotta.org/mailman/listinfo/tc-dev _______________________________________________ tc-dev mailing list tc-dev@lists.terracotta.org http://lists.terracotta.org/mailman/listinfo/tc-dev
