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.
The test fails w/o any obvious reason (seems like it passes on some machines but fails on continual builder). I will remove it from the tests set until it gets fixed.
The test is not broken. It fails because IDE throws an exception. I don't know yet why the exception is thrown. I can't reproduce it any way. I noticed that JDK on which build runs was changed from j2sdk1.4.2_09 to jdk1.5.0_06. Is it intentional?
Created attachment 28996 [details] Stack trace
Jiri, NB 5.5 daily and continuous and builds run on Solaris 9 SPARC, on JDK 1.5.0_06. The change of JDK for build is intentional. Let's meet later today and watch the NB 5.5 c/b what happens.
Sorry, I did not want to say that the test is broken. I just see it failing and do not see how to fix it so I think it makes little sense to send "broken build" messages all the time.
OK, neither I don't know exact reason why it is failing. I want to ask core guys for help. What we know about it so far: - ArrayIndexOutOfBoundsException (see attachment) is thrown when opening help window. The test clicks main menu item 'Help|Help Contents'. - it started to happen when we switched from j2sdk1.4.2_09 to jdk1.5.0_06 - it can be reproduced only on multiprocessor SPARC with Solaris 9 From above it seems to be a threading issue when concurrent threads are accessing the same data. Do you have an idea how to fix it or better investigate?
It looks like problem in JavaHelp. AFAIK there was some change in 2.0_02 to improve performance - some parsing was moved to background thread. Ihave no idea if there is any workaround. I will ask JavaHelp if there is any solution.
Ruda is building machine multiprocessor?
Marek, yes, pytlik is NB 5.5 buildmachine and it's SunFire V880 with 4x UltraSPARC III+ CPUs. More details about this hardware is available at http://www.sun.com/servers/midrange/v880/ Output of 'uname -a': SunOS pytlik 5.9 Generic_117171-05 sun4u sparc SUNW,Sun-Fire-880. Well this hardware was EOLed some time ago, so we may try to apply lates Solaris 9 patches and see what happens, but don't expect major changes.
Thanks for info. How often does it happen? I tried to reproduce it manually on our testing 2 CPU Sparc machine with Sol 10/JDK 1.5.0_06. I tried to start IDE and open Help window 5 times but problem does not appear. (No I do not blame RE HW :))
It happenned now. Maybe 10% cases.
It happens almost everytime (but not 100%) in commit validation suite. Impacted test case was removed from the suite for now. But I was not able to reproduce it manually. On RE HW we reproduced it this way: cd nb_all55/ide/test ant -Dxtest.attribs=validation -Dnetbeans.dest.dir=<NB55home> -Djdkhome=<jdk_home>
I filed issue #6394203 against Javahelp.
I temporary disabled JavaHelp test in Commit Validation (trunk), do not forget enable it again after fixing this issue.
We agreed on lowering priority to P3 and leave this issue opened for tracking. Problem is in JavaHelp NOT in NetBeans. We must somehow convince them to fix it.
No progress on Javahelp issue.
New JavaHelp version 2.0_04 is integrated to main trunk => this issue is fixed.
I added testHelp to the commit validation suite again.
*** Issue 90871 has been marked as a duplicate of this issue. ***
*** Issue 69011 has been marked as a duplicate of this issue. ***