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.

Bug 103027 - AE(assert support.getContainer().listInvalid().isEmpty()) when install a module
Summary: AE(assert support.getContainer().listInvalid().isEmpty()) when install a module
Status: RESOLVED WORKSFORME
Alias: None
Product: platform
Classification: Unclassified
Component: Plugin Manager (show other bugs)
Version: 6.x
Hardware: Sun Solaris
: P3 blocker (vote)
Assignee: Jiri Rechtacek
URL:
Keywords: RANDOM
Depends on: 103064
Blocks:
  Show dependency tree
 
Reported: 2007-05-01 00:18 UTC by _ pcw
Modified: 2007-08-08 11:56 UTC (History)
1 user (show)

See Also:
Issue Type: DEFECT
Exception Reporter:


Attachments
example nbm module project (4.64 KB, application/x-compressed)
2007-05-01 00:22 UTC, _ pcw
Details
NBM that demonstrates the restart bug. (2.58 KB, application/octet-stream)
2007-05-01 00:22 UTC, _ pcw
Details
messages.log from clean userdir on XP (23.75 KB, text/plain)
2007-05-01 10:18 UTC, _ pcw
Details
java.util.concurrent.ExecutionException trace (1.98 KB, text/plain)
2007-05-01 10:18 UTC, _ pcw
Details
console log + thread dump after ExecutionException. (53.40 KB, text/plain)
2007-05-01 10:20 UTC, _ pcw
Details

Note You need to log in before you can comment on or make changes to this bug.
Description _ pcw 2007-05-01 00:18:05 UTC
NBM's that require restart cannot be installed via the plugin manager either
when downloaded from their update center or installed from NBM on local disk.

The install dialog proceeds normally, but when after choosing "Restart Now" on
the restart dialog, the IDE shuts down and does not restart.  Upon manually
restarting, the NBM's are still not installed.

NBM's that do not require a restart work ok.

This is a critical bug for the AJAX (jMaki and Phobos) NBM's, as they have the
restart flag enabled to avoid classloader issues.

See attached NBM project (trivial NBM w/ restart flag set to true) and NBM build
from said project.
Comment 1 _ pcw 2007-05-01 00:22:15 UTC
Created attachment 41991 [details]
example nbm module project
Comment 2 _ pcw 2007-05-01 00:22:50 UTC
Created attachment 41992 [details]
NBM that demonstrates the restart bug.
Comment 3 _ pcw 2007-05-01 00:23:58 UTC
This bug is present in the NB 6.0 M9 candidate for JavaOne as well as the
current nightly builds of the trunk.
Comment 4 Jiri Rechtacek 2007-05-01 09:12:05 UTC
Product Version         = NetBeans IDE Dev (Build 070429)
Operating System        = Linux version 2.6.20-15-386 running on i386
Java; VM; Vendor; Home  = 1.6.0_02-ea; Java HotSpot(TM) Client VM
1.6.0_02-ea-b02; Sun Microsystems Inc.; /space/shared/java/jdk1.6.0_02/jre
System Locale; Encoding = us (nb); UTF-8

The attached NBM is installed correctly. Worksforme.
Comment 5 _ pcw 2007-05-01 10:17:25 UTC
I originally received this on Solaris x86.  I have also reproduced this on
Windows XP with a clean user dir, messages.log from that invocation attached.

Also, when attempting to reproduce this (using the attached NBM) on XP while the
background source scanning was running, I received the attached assertion
failure caused by a java.util.concurrent.ExecutionException.  The IDE locked up
tight after that, thread dump attached.
Comment 6 _ pcw 2007-05-01 10:18:24 UTC
Created attachment 42000 [details]
messages.log from clean userdir on XP
Comment 7 _ pcw 2007-05-01 10:18:58 UTC
Created attachment 42001 [details]
java.util.concurrent.ExecutionException trace
Comment 8 _ pcw 2007-05-01 10:20:23 UTC
Created attachment 42002 [details]
console log + thread dump after ExecutionException.
Comment 9 _ pcw 2007-05-01 10:21:40 UTC
One more thing -- both machines that reproduced the restart problem (and the
exception) are dual core.
Comment 10 rmatous 2007-05-01 17:02:55 UTC
I think that you use the old way how to address you userdir: "-userdir
/space/nbudir/my2/". 
You are supposeed to use this one now: "--userdir /space/nbudir/my2/"


This call ./netbeans -userdir /space/nbudir/my2/ is probably source of your
problem and I can verify(tesetd on your restart.nbm) that it cause exactly the
same problem that you originally described in this issue. Please, reopen if I'm
wrong.
Comment 11 _ pcw 2007-05-01 21:04:27 UTC
1. You are correct, switching to --userdir corrected the problem.  Jiri filed
issue 103064.

2. I reopened this issue as P3 to discuss the ExecutionException I received
(attached).  Is this related to the original problem or is it a different bug? 
If different, it needs to be fixed.  On the surface, it looks like a race
condition caused by invalid threading + background source parsing + dual core.
Comment 12 rmatous 2007-05-02 09:56:27 UTC
I think that the only suspicious assertion that has something to do with plugin
manager code is in trace4.txt(InstallSupportImpl.java:112):

assert support.getContainer().listInvalid().isEmpty();
Comment 13 Jiri Rechtacek 2007-05-21 10:35:52 UTC
Renamed subcomponent 'module manager' to 'plugin manager' consistent with
current naming.
Comment 14 Jiri Rechtacek 2007-08-08 11:56:28 UTC
This problem hasn't appeared again and no-one reported it for long time. Closed as worksforme.