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.
Using today's Q-build candidate: #2003-02-12-0100 jdk1.4.1, RH7.2 ===================================================== I was hunting some bug in Editor or some of VCS's modules If you're interesting in this stuff like E.'s stacktraces I'll add a link to it;-) So, I have opened a file in Editor (displaying me wrong content, see the bug link about it). And click on the Cross (X) button to close whole editor component. And the deadlock occured:-( I guess it is in CloneableEditorSupport.java But you'll se it yourself;-)))
Created attachment 8914 [details] FTD
Hi, I was able to reproduce this deadlock, using this steps: 1. Mount a (testing) CVS filesystem 2. Create a file on it. 3. Do an external modification. 4. Go back to the IDE. Some exceptions are thrown (Dan is filling bug about them). 5. Close the editor. 6. Open the created file again 7. Do another external modification 8. Go back to the IDE. Exceptions ... . 9. Close the editor --- deadlock occured. I was not able to reproduce it on local filesystem (both exceptions and deadlock).
Thank you Honzo, BTW: the bug is issue #30989 and I can't say that one of them depends on the other, but they are clously related and without issue #30989 this one wouldn't happend (probably:)
QA folks please always add THREAD keyword to deadlock bugs!!! Thx
Created attachment 8985 [details] openide patch which should solve the problem
Honzo, could you try the attached patch please. Thanx.
I cannot reproduce this issue on W2000 on today's dev build. From the attached thread dump I can see a strange order of calls which my patch should prevent. However being able to reproduce the issue would help to verify also the patch.
Thanks David, but you can't realy reproduce it in todays build 'cause Mila already fixed the issue #30989. So if we wish to verify your fix/patch we must take the build on which it was discovered. That's mean: #2003-02-12-0100 I'd like to test it, unfortunately i'm working wrom home and downloading build using modem conection isn't fun stuff, you know:-/ thans
Hi, I tested the patch on 200302120100. I was able to reproduce the deadlock on un-patched IDE, but it is random. I was not able to reproduce it with the patch, only the attached exception occured (according to Mila, it is issue #30989), and I needed to close the editor again (I had to close it twice, for the first time the exception was thrown). It seems that the deadlock is fixed by the patch. Thank you.
Petr Zavadsky reviewed the patch and while it may solve the problem it can also break something else. Without being able to reproduce the problem and test the patch and behaviour I hesitate to commit it. I also talk with Jan about this issue and how to reproduce it. It only appears directly after the problem reported as issue 30989 and so it might be sideeffect of this issue. I propose to first solve the issue 30989 (Jan told me he is going to reopen it) and then continue with this one. The code I'm patching was not modified recently (there was not any modification in text package recently) and so I do not see any reason for regression. It is of course possible that problem is there for long time and now revealed, but to fix it and do not cause some other regression I need reproducible scenario. I have also my doubts about whether this is Q-build stopper when it is not that easy to reproduce and when it looks like sideeffect of some other issue.
After some other discussions I'm closing as WONTFIX. Please read my previous comment. At the moment this issue looks as sideeffect of issue 30989 - Jan told me that only time when he saw this error was directly after the exceptions from issue 30989. Once the 30989 is fixed (it is fixed at the moment, but problem was reproduced again - ask Jan L. for more details) and this problem was reproduced, please reopen and provide details. Thanx.
agree with David