Build: NetBeans IDE 6.9.1 (Build 201011082200)
VM: Java HotSpot(TM) Client VM, 10.0-b22, Java(TM) SE Runtime Environment, 1.6.0_06-b02
OS: Windows XP
Maximum slowness yet reported was 75577 ms, average is 39760
Created attachment 107945 [details]
From the last snapshots it seems that ActionStateUpdater.actionPerformed() invoke update of all node actions in EDT which is blocked by project save by its closing - saveProject() acquires writeAccess on the same mutex as is needed for reading by update of all actions. I will talk to t_h who contributed on this place and eventually reassign this issue.
I'm reassigning the issue since it looks like Explorer API problem.
May I get an explanation why the issue is supposed to be problem in explorer? I see 65s wasted in EJB.
Right, the delay seems to be rising in EJB, but actually it's coming between ExplorerActionsImpl.updateActions which leads to BaseActionProvider.isActionEnabled and ProjectManager.saveProject.
Especially now it looks to me as badly reassigned as well, sorry for that. :/
It is related to Projects. I took a look quickly into WEB, EJB projects and in both of them it could happen. Generally it can happen for any BaseActionProvider subtype.
Jesse, as designer of ActionProvider.isActionEnabled and other project code, shall advice you how to make the call non-blocking.
According to the snapshot, 64 seconds is being spent in a single call to ClassLoader.getResourceAsStream - which would seem to indicate that the user's JVM is out of file handles, at max heap, swapped out, or otherwise pretty much busted. If so, there is nothing to be done in IDE code.
Ok, totally agree. Closing as WONTFIX for now but jocke feel free to reopen if you would be able to reproduce it without exhausted resources. Thanks.