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.
Scenario: 1. Run IDE with opened UTS project (OpenSolaris kernel from 77 build) 2. Close project 3. Create new UST project from 80 build. Note: OpenSolaris should be build. 4. After successfully created project (IZ#131697 should be fixed) parsing starts. 5. In log I see assertion. I believe that bug is hard reproduced.
Created attachment 59466 [details] assertion
Created attachment 59467 [details] full log
Though rarely reproducible, the issue is quite serious. I believe we should at least create a workaround for this for 6.1
This may be a P1. I've escalated it, but I have not yet set this to must fix.
I hadn't seen this for quite some time, but this morning I was getting it pretty much after every change. I exited the IDE, and after a restart, no longer got it. I remember it being like this before. Once you get it, you're pretty much hosed until you restart!
Here is a workaround that I pushed into main: http://hg.netbeans.org/main/rev/19aab9dc947c It allows you starting Netbeans with -J-Dcnd.repository.hardrefs=true option. In this case persistence is replaced with in-memory map, which works fine if you have enough memory. For project like Open Solaris kernel, which has 4-5K files, it's sufficient to have 600Mn heap to work with this option.
I reviewed the workaround and it is correct.
Integrated in HG-pull #8 (See [http://hudson.czech.sun.com:8080//hudson/job/HG-pull/8/] / [Changeset 5220a1767a02]) Link: http://hg.netbeans.org/main/rev/5220a1767a02
The last comment from "qa" user is not related to this issue. I am really sorry for that.
I've put a workaround: http://hg.netbeans.org/main/rev/e3ae83ea3c90 http://hg.netbeans.org/main/rev/833fe4b6ef83
I reviewed the problem and can summarize that "fix" is not possible in a safe way. So, I agree that workaround is the necessary step for 6.1 and the proposed workaround (3 commits) is correct and safe.
One more fix was recently added. A jar to test against release61 has been set to QA
This most recent fix concerned the situation when the persistence is broken, but this fact couldn't be discovered at project opening. After applying this fix, language model processed situation smoothly. The project was reparsed, after that code assistance worked ok.
Created attachment 59931 [details] the mentioned above fix
The attached fix is reviewed by Vladimir Voskresensky.
The fix is verified in NB6.1RC1 (build 200804082140) with provided jar file.
All changed pushed into release61: http://hg.netbeans.org/release61/rev/bbfd88659244 http://hg.netbeans.org/release61/rev/e6aba9752fca http://hg.netbeans.org/release61/rev/2e1bc9002d09 http://hg.netbeans.org/release61/rev/521fa45e0573
See previous entry regarding push into NB61.
verified in NB6.1RC1 build