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.
Build: NetBeans IDE Dev (Build nbms-and-javadoc-3789-on-090821) VM: Java HotSpot(TM) Client VM, 14.2-b01, Java(TM) SE Runtime Environment, 1.6.0_16-b01 OS: Windows XP, 5.1, x86 Maximum slowness yet reported was 66658 ms, average is 66658
Created attachment 86567 [details] nps snapshot
Reassigning to performance for re-evaluation, I think that the problem is in unusually long time spent in filesystem operations.
Slowness of native NTFS call (WinNTFileSystem.canonicalize0). Reported many times, but there it is still not obvious why some of filesystem is so slow (broken links?, corrupted fs?)
I think I learned how to read these snapshots. Look at the # of invocations in http://statistics.netbeans.org/exceptions/exception.do?id=248491 There is one call to: RegistryImpl.topComponentOpened then there are two calls to: PropertyChangeSupport.firePropertyChange which end up in one call all: org.nbheaven.sqe.codedefects.core.api.install.OpenTopComponentListener From here the number of calls is growing, at the end it reaches 100 calls to File.canonicalize(). At least 100, possibly few thousands as we do sampling and the number of calls is informative and always just bottom boundary. So the problem is not that one call to getBooleanAttribute would be slow, the problem is that org.nbheaven.sqe.codedefects.core.api.install.OpenTopComponentListener does too much I/O activity in AWT thread. The question is where the nbheaven.sqe comes from... CCing Jirka and Marian.
sqe - that would be me
Ok. Seems you are absolutely correct here. Will try to fix this in SQE. Thanks for spotting. Should this be reassigner to 3rd party / sqe?
Thanks Sven.
Doing more work in a BG thread: sqe #55a84cc8f48e
Created attachment 93483 [details] nps snapshot
Created attachment 93530 [details] nps snapshot