This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
[jdk1.3.1_01, NetBeans IDE 3.4 Beta 2 (Build 200206201555)] I'm not able to reproduce it reliably but from time to time it occurs. See attached stack trace.
Created attachment 6433 [details] stack trace
Can you attach a thread dump, please ? The attachment is the ide logfile -- interesting errors from the RequestProcessor (Petr, what does that ISE mean ?), but no deadlock clue at all.
Created attachment 6434 [details] Sure, here is the thread dump
The ISE should not happen. It is from java.util.Timer we use for delayed schedules. It is thrown when Timer$TimerThread have its die boolean set. The boolean is set to true only when: a) If Timer is gc()ed (can't be - private static variable in RP) b) Timer.cancel() is called (can't be - we don't call it) c) TimerThread.run() finishes - possible by exception in our TimerTask code, but there shouldn't be any exception and it would be logged, wouldn't it? d) Someone killed the thread - the most possible case. In what circumstances have that happened (stopped compilation or something like that?)
I'm not sure about circumstances but definitely not stopping a compilation. Maybe umounting a filesystem with opened files in editor or just closing these files. I'm really not certain about an action before the freeze.
I am confused; there's no java parsing thread in the thread dump at all ? That's interesting. I am positive that I only create the RequestProcessor and never call stop() on it. Particular tasks posted into the RP may be cancel()ed, but the RP itself is never stop()ed. Petre, is there any reason (e.g. unhandled exception, ...) why the RequestProcessor could terminate its thread ?
It has nothing to do with RP.stop nor Task.cancel() The only possible thing that comes to my mind is somebody calling Thread.stop on the timer thread. Maybe there is a problem with initialization of RP from inside internal execution but I can't believe RP can be initialized that late. Yes, it can also be possible if exception occurs in the TimerTask but look at the task I'm posting there from the RP's schedule... I'll provide a logging patch on monday to test which of the cases have occured if Jan will be able to reproduce it.
Unfortunately thread dump does not provide enough information to locate and fix the problem. Closing as works-for-me. Please provide more info how to reproduce it.
Resolved for 3.4.x or earlier, no new info since then -> closing.