[rules-users] Behavioral differences between session.insert() and entryPoint.insert()?

I am seeing different behavior whether I insert events directly into session or into a named entry-point. The below rule and test is greenbar'ed: ---- rule "How_do_I_correlate_events_arriving_in_2_or_more_streams?" when $e : WithdrawalEvent($accountNumber : accountNumber, $amount : amount) over window:time(30s) from entry-point "stream" $f : FraudWarningEvent(accountNumber == $accountNumber) over window:time(30s) from entry-point "stream" then results.put("accountNumber", $accountNumber); results.put("amount", $amount); end ---- def insert(fact) { session.getWorkingMemoryEntryPoint("stream").insert(fact) } @Test def void How_do_I_correlate_events_arriving_in_2_or_more_streams() { // Not the same account insert new WithdrawalEvent(accountNumber: "AAA", amount: 100) insert new FraudWarningEvent(accountNumber: "BBB") fireAllRules() assert results.isEmpty() // Not within the 30s window advanceTime 31, SECONDS insert new FraudWarningEvent(accountNumber: "AAA") fireAllRules() assert results.isEmpty() .... } ---- But if I do not specify a stream in the rules (ie, remove the 'from entry-point "stream"') and instead insert directly into the session (like many of the the drools integration do)... def insert(fact) { session.insert(fact) } ...the test fails. It does not matter how far I advance the clock before inserting the (second) FraudWarningEvent, the rule always fires. I only found this difference after looking at the integration tests and seing that I did not need to use a named entry-point for events. Did I understand incorrectly? (I was looking at /drools-compiler/src/test/resources/org/drools/integrationtests/test_CEP_DelayingNot.drl and its associated java test.) -- View this message in context: http://www.nabble.com/Behavioral-differences-between-session.insert%28%29... Sent from the drools - user mailing list archive at Nabble.com.