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.
Not sure what code is to blame here. Datasystems? Lookup?
Created attachment 16997 [details] (Beginnings of) log + a thread dump
Continues to stop the startup sequence until you delete folder-lookup.ser. 040820 custom, JDK 1.4.2, test user dir without much in it, one project open, one Java file open. Nothing remarkable that I can see, just bad luck?
Happened to me again (same user dir).
Happened on my regular user dir too. Getting to be pretty annoying.
Created attachment 17004 [details] Thread dump
Yarda, please investigate, possibly fix it
Is likely related to Friday's fix of nonfunctional lookup cache (issue 46726). Maybe disable the cache completely? But we have to be aware of possible perf implications.
"We (rmatous and jtulach) think that we know the root case of 47588 and that we have simulated it in this test" Checking in unit/src/org/openide/loaders/InstanceDataObjecIssue47707Test.java; /cvs/openide/test/unit/src/org/openide/loaders/InstanceDataObjecIssue47707Test.java,v <-- InstanceDataObjecIssue47707Test.java initial revision: 1.1 Sorry for the name of the test, the right issue number is 47588, I'll leave a note in the issue 47707.
Tested and fixed. /cvs/openide/loaders/src/org/openide/loaders/InstanceDataObject.java new revision: 1.16; previous revision: 1.15
> Maybe disable the cache completely? > But we have to be aware of possible perf implications. I have measured startup time with three cases yesterday (first three times are vanilla starts, next three are complex starts with project open and a few files open in editor): trunk 8/18 - 23.7s, 17.6s, 17.8s; 27.5s, 26.8s, 26.9s (before the fix, with lookup cache on) trunk 8/22 - 22.3s, 18.3s, 18.1s; 27.7s, 27.7s, 28.0s (with yarda's fix of issue 46726, with lookup cache on) trunk 8/18 with the lookup cache turned off - 22.5s, 17.6s, 17.4s; 27.1s, 27.4s, 27.3s So, turning the lookup cache off would be also an option. It seems that efficiency of the lookup cache is very limited.
*** Issue 47789 has been marked as a duplicate of this issue. ***
*** Issue 47794 has been marked as a duplicate of this issue. ***
*** Issue 47812 has been marked as a duplicate of this issue. ***
I see it in 200408241800 (JDK 1.5.0 b63, linux RH9). It occures very often.
Created attachment 17108 [details] threaddump
What you see is not a problem in instance data object. That is fixed. The last stacktrace looks like the one from issue 47772.
Does this persist in 4.2's builds?? I haven't seen it. Closing, reopen if you will.