Community
Participate
Working Groups
If a LogService is present and a bad EventHandler listens for the org/osgi/service/log/LogEntry/* topic then endless event dispatching can occur if the EventHandler continually throws runtime exceptions. The event admin implementation should probably try to handle this.
Created attachment 186317 [details] possible fix This fix will detect a possible endless LogEntry event posting by wrapping the EventHanders exception in a special exception when the event topic starts with "org/osgi/service/log/LogEntry". This way we can detect a second exception from a handler if the log call causes another exception from an EventHandler.
Created attachment 186319 [details] possible fix The previous patch prevented endless events but had a bug. It tried to cast the event property "exception" to Class which would cause a cast exception. The property "exception" is not a Class but the actual exception. This patch fixes that.
Created attachment 186321 [details] proposed fix OK, lets try this again. Same idea but with a simple instanceof check.
patch released.