On Mon, Aug 3, 2015 at 6:07 PM, Anshum Gupta <[email protected]> wrote: > Guess bypassing that check and always returning true instead would help > confirm that. If that's the case, we should just initialize the timeout to > Long.MAX_VALUE and check for that to short-circuit?
Another possibility is to run queries without a time limit on a reader that isn't wrapped. That could possibly be friendlier to branch prediction with mixed queries (some without timeouts and some with). -Yonik --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
