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.
| Summary: | ClassNotFoundException: org.eclipse.mylyn.internal.bugzilla.core.BugzillaRepositoryConnector | ||
|---|---|---|---|
| Product: | platform | Reporter: | Tomas Danek <musilt2> |
| Component: | Launchers&CLI | Assignee: | Jiri Rechtacek <jrechtacek> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | davehun, jglick, jrechtacek, jtulach, mkleint, mmirilovic, muellermi, rrhegde |
| Priority: | P2 | ||
| Version: | 7.2 | ||
| Hardware: | All | ||
| OS: | All | ||
| Issue Type: | DEFECT | Exception Reporter: | 187653 |
| Attachments: |
stacktrace
logs with desired logging possible patch on netbinox side proposed patch of ide/launcher/unix/netbeans |
||
|
Description
Tomas Danek
2012-05-04 08:39:24 UTC
Created attachment 119068 [details]
stacktrace
after deleting user & cache dir I am not able to reproduce again :( I'm able to reproduce after there steps: - delete user and cache dir - install beta javase - run it (shortcut from spotlight), do not import settings - Team|Find issue... - push "New" -> bugzilla dialog appears, close it - Tools|Plugins -> install all modules from Beta UC - confirm ide restart - create new maven javase project - close ide - start ide again -> CNFE during startup something rotten is in /Users/tomas/Library/Caches/NetBeans/7.2beta/ (after steps in comment 3) when i remove it, it starts to work. and just to narrow down even more, problem is in /Users/tomas/Library/Caches/NetBeans/7.2beta/netigso (as expected) (In reply to comment #4) > something rotten is in > /Users/tomas/Library/Caches/NetBeans/7.2beta/ > (after steps in comment 3) > when i remove it, it starts to work. Single report from ten days ago. Closing. (In reply to comment #6) > Single report from ten days ago. Closing. Tomas, if you are able to reproduce - please REOPEN immediately. here you have newer log if you wish.... dev build, using steps in comment 3 *** Bug 212633 has been marked as a duplicate of this bug. *** Does this happen on all platforms or only on Mac. Can you demonstrate the problem for me? (In reply to comment #11) > Can you demonstrate the problem for me? Feel free to stop in my office. *** Bug 212606 has been marked as a duplicate of this bug. *** *** Bug 212610 has been marked as a duplicate of this bug. *** *** Bug 212786 has been marked as a duplicate of this bug. *** *** Bug 212799 has been marked as a duplicate of this bug. *** BaseFile of some of the bundles is not correct. It should be relative to Resources/NetBeans directory on Mac, but it is not. It contains the "Resources/NetBeans" prefix again, thus (when the final path is being constructed), this part is duplicated. The JAR file then cannot be found. Turning on more logging ergonomics#95ee511df2dc - this will print the value of "installArea" which is then used to make all paths relative. A message like INFO [org.netbeans.modules.netbinox]: Install area set to file:/apps/netbeans/ shall appear in every message.log. It will be interesting to find out if it differs somehow in a run preceding the one that throws the exception. So on a successful run (a rarity these days), I get INFO [org.netbeans.modules.netbinox]: Install area set to file:/space/nb/curr/ which is in fact the IDE install dir. Will report back with log if I see the problem again. Same message printed in a run in which I got SEVERE [org.netbeans.modules.netbinox]: framework event org.eclipse.equinox.registry type 2 java.lang.NullPointerException at org.eclipse.core.internal.runtime.ResourceTranslator.getResourceBundle(ResourceTranslator.java:69) ... Integrated into 'main-golden', will be available in build *201205230300* on http://bits.netbeans.org/dev/nightly/ (upload may still be in progress) Changeset: http://hg.netbeans.org/main-golden/rev/95ee511df2dc User: Jaroslav Tulach <jtulach@netbeans.org> Log: #212061: Printing out the install area location *** Bug 212847 has been marked as a duplicate of this bug. *** Created attachment 119810 [details]
logs with desired logging
attached /var/log/* logs with issue reproduced.
The updater is using paths like INFO: File /Applications/NetBeans/NetBeans Dev 201205240002.app/Contents/MacOS/../Resources/NetBeans/bin/../cnd/modules/org-netbeans-modules-cnd-debugger-gdb2.jar.pack.gz deleted. May 24, 2012 1:27:23 PM org.netbeans.updater.ModuleUpdater unpack e.g. with two "..". That is pretty nasty! The installation directory (from messages.log.2) is however absolutized properly: /Applications/NetBeans/NetBeans Dev 201205240002.app/Contents/Resources/NetBeans/ and the install area is also correct: Install area set to file:/Applications/NetBeans/NetBeans Dev 201205240002.app/Contents/Resources/NetBeans/ However for some reason, the install area for messages.log.1 is different: Install area set to file:/Applications/NetBeans/NetBeans Dev 201205240002.app/Contents/ The last execution (messages.log) contains proper install area again, but obviously the paths are wrong as the subsequent error shows: Install area set to file:/Applications/NetBeans/NetBeans Dev 201205240002.app/Contents/Resources/NetBeans/ opening reference:file:Resources/NetBeans/ide/modules/org-eclipse-equinox-registry.jar because of resource needing org/eclipse/core/internal/registry/OffsetTable.class Error creating delegate for /Applications/NetBeans/NetBeans Dev 201205240002.app/Contents/Resources/NetBeans/Resources/NetBeans/ide/modules/org-eclipse-equinox-registry.jar The problem might have some relation with launcher. Maybe the passed in parameters are different when the IDE is restarted. (In reply to comment #25) > The problem might have some relation with launcher. Maybe the passed in > parameters are different when the IDE is restarted. By the marked dups (issue 212606, issue 212610, issue 212786, issue 212799) it doesn't seem the problem is connected to Mac or with restarting IDE after installed updates. OK, the error must be in a way we compute the install area. Could you please re-run the test with follwing patch: http://hg.netbeans.org/ergonomics/rev/a934e5673116 Enough to have log from second run, I guess. (In reply to comment #24) > Install area set to file:/Applications/NetBeans/NetBeans Dev > 201205240002.app/Contents/Resources/NetBeans/ This is wrong, BTW; should be: file:/Applications/NetBeans/NetBeans%20Dev%20201205240002.app/Contents/Resources/NetBeans/ (In reply to comment #27) > http://hg.netbeans.org/ergonomics/rev/a934e5673116 > Enough to have log from second run, I guess. See http://statistics.netbeans.org/analytics/exception.do?id=582279 from bug #211733 which I guess is related (?). > > Install area set to file:/Applications/NetBeans/NetBeans Dev
> > 201205240002.app/Contents/Resources/NetBeans/
>
> This is wrong, BTW; should be:
>
> file:/Applications/NetBeans/NetBeans%20Dev%20201205240002.app/Contents/Resources/NetBeans/
That would very likely broke Equinox - Equinox is well known for not understanding %20 and requiring " ".
I fall into this problem too. There is output of 'jps -v': * 1st process is instance of IDE after restart by AU * 2st process after my manually start IDE from icon dhcp-prague08-second-floor-10-163-21-197:core-main lion$ jps -v 84856 Main -Djdk.home=/System/Library/Java/JavaVirtualMachines/1.6.0.jdk/Contents/Home -Dnetbeans.default_userdir_root=/Users/lion/Library/Application Support/NetBeans -Dnetbeans.system_http_proxy=PAC http://wpad/wpad.dat http://wpad/wpad.dat -Dnetbeans.system_http_non_proxy_hosts= -Dnetbeans.dirs=/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/Resources/NetBeans/nb:/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/MacOS/../Resources/NetBeans/bin/../ergonomics:/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/Resources/NetBeans/ide:/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/Resources/NetBeans/java:/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/Resources/NetBeans/apisupport:/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/MacOS/../Resources/NetBeans/bin/../webcommon:/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/Resources/NetBeans/websvccommon:/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/MacOS/../Resources dhcp-prague08-second-floor-10-163-21-197:core-main lion$ jps -v 85701 Main -Djdk.home=/System/Library/Java/JavaVirtualMachines/1.6.0.jdk/Contents/Home -Dnetbeans.default_userdir_root=/Users/lion/Library/Application Support/NetBeans -Dnetbeans.system_http_proxy=PAC http://wpad/wpad.dat http://wpad/wpad.dat -Dnetbeans.system_http_non_proxy_hosts= -Dnetbeans.dirs=/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/Resources/NetBeans/nb:/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/MacOS/../Resources/NetBeans/bin/../ergonomics:/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/Resources/NetBeans/ide:/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/Resources/NetBeans/java:/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/Resources/NetBeans/apisupport:/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/MacOS/../Resources/NetBeans/bin/../webcommon:/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/Resources/NetBeans/websvccommon:/Applications/NetBeans/NetBeans Dev 201205140400.app/Contents/Resources/NetBeans The different seems to be: .app/Contents/MacOS/../Resources (first start) .app/Contents/Resources/NetBeans (second start) I am not sure why these directories are added to the list of clusters. In my opinion none of them should be there. The caching system relies on launcher to not mutate the netbeans.dirs value if there are no changes in installed modules. I am glad I can pass the bug to somebody else... (In reply to comment #32) > The different seems to be: > > .app/Contents/MacOS/../Resources (first start) > .app/Contents/Resources/NetBeans (second start) I saw the difference too (that's the reason why I post it here :). I have no clue why the paths are there and why differs. I try figure out it. Anyway, I'm afraid this problem is not shared in all of duplicates of this one. > I am not sure why these directories are added to the list of clusters. In my > opinion none of them should be there. The caching system relies on launcher to > not mutate the netbeans.dirs value if there are no changes in installed > modules. > > I am glad I can pass the bug to somebody else... (In reply to comment #30) > Equinox is well known for not understanding %20 and requiring " ". Filed upstream? Created attachment 119993 [details]
possible patch on netbinox side
Created attachment 120014 [details]
proposed patch of ide/launcher/unix/netbeans
Jesse, Jardo, please look on my patch. I can confirm it solves this problem. I'm going to integrate it tomorrow. Thanks (In reply to comment #37) > Jesse, Jardo, please look on my patch. I can confirm it solves this problem. > I'm going to integrate it tomorrow. Thanks I meant the proposed patch of ide/launcher/unix/netbeans. Making sure absolute and canonical paths are used all the time is good idea. In readClusters I'd use: "$basedir"/etc/netbeans.clusters. "grep -v platform" is missing in readClusters. It should be there, imho. There is no point in including the platform cluster among the -Dnetbeans.dirs=... (In reply to comment #39) > "grep -v platform" is missing in readClusters. It should be there, imho. There > is no point in including the platform cluster among the -Dnetbeans.dirs=... Right, ${netbeans.dirs} must _not_ include the platform cluster (that is given by ${netbeans.home}). The removal of the test for absolute cluster names in readClusters also looks wrong. This will cause "/other/cluster" to be converted to "/opt/netbeans//other/cluster". I think the only things that should be changed in readClusters are "$progdir/../$X" -> "$basedir/$X" and "$progdir"/../etc/netbeans.clusters -> "$basedir"/etc/netbeans.clusters. I guess the problem with the existing absolutize_paths was that it did not work for nonexistent clusters, right? (*) Is there a reason why nonexistent clusters are even included in ${netbeans.dirs} to begin with? (*) http://stackoverflow.com/questions/284662/how-do-you-normalize-a-file-path-in-bash/10808025#10808025 (In reply to comment #40) > (In reply to comment #39) > > "grep -v platform" is missing in readClusters. It should be there, imho. There > > is no point in including the platform cluster among the -Dnetbeans.dirs=... > > Right, ${netbeans.dirs} must _not_ include the platform cluster (that is given > by ${netbeans.home}). > > The removal of the test for absolute cluster names in readClusters also looks > wrong. This will cause "/other/cluster" to be converted to > "/opt/netbeans//other/cluster". I think the only things that should be changed > in readClusters are "$progdir/../$X" -> "$basedir/$X" and > "$progdir"/../etc/netbeans.clusters -> "$basedir"/etc/netbeans.clusters. Okay, I'll change the patch this way. > I guess the problem with the existing absolutize_paths was that it did not work for nonexistent clusters, right? (*) Right. > Is there a reason why nonexistent clusters > are even included in ${netbeans.dirs} to begin with? AU can add new cluster (one of nonexistent clusters yet) thus it will be needed after restart IDE. core-main/rev/2da99c8faadb *** Bug 213310 has been marked as a duplicate of this bug. *** Integrated into 'main-golden', will be available in build *201206010001* on http://bits.netbeans.org/dev/nightly/ (upload may still be in progress) Changeset: http://hg.netbeans.org/main-golden/rev/2da99c8faadb User: Jiri Rechtacek <jrechtacek@netbeans.org> Log: #212061: ClassNotFoundException: org.eclipse.mylyn.internal.bugzilla.core.BugzillaRepositoryConnector *** Bug 211840 has been marked as a duplicate of this bug. *** *** Bug 214808 has been marked as a duplicate of this bug. *** *** Bug 214811 has been marked as a duplicate of this bug. *** *** Bug 214806 has been marked as a duplicate of this bug. *** *** Bug 214809 has been marked as a duplicate of this bug. *** |