From: Olivier M. <om-...@ti...> - 2005-11-09 17:44:44
|
Le 09-nov.-05 =E0 15:30, Milan Babuskov a =E9crit : > 2. Then I remove some events (using the old method: Clear + Define =20 > all those left). That's the only thing that seems a bit problematic. When removing them all this way (yeah I know this was the only way to =20= do it until 2.4.5), the whole vectors which used to store the counts =20 are destroyed. It is very difficult to know for sure what you will =20 get triggered or not when you re-define them all. Because of the =20 delays between : - the time the commit happens - the FB client is notified - the FB client consider to notify IBPP by calling the call back I would recommend a new test (but no hurry) where you drop those =20 events that you need to drop by using the new Database::DropEvent. It =20= will drop a single event, and the "context" for the others should not =20= be lost. --=20 Olivier |