Community
Participate
Working Groups
I20110111-0800. After importing all required binary plug-ins I had tons of errors after plug-in import. It looks like it failed to update the PDE container correctly regarding the SWT bundles. The importer used to work pretty well for quite some time up to now. I suspect that this might be caused by recent changes either for bug 332765 or bug 330047.
> I suspect > that this might be caused by recent changes either for bug 332765 or bug > 330047. Bug 330047 was in HEAD for some time, so this one is out of the loop.
Tried again on a simpler workspace: when drying to re-import I got: !ENTRY org.eclipse.core.jobs 4 2 2011-01-12 09:40:52.086 !MESSAGE An internal error occurred during: "Import Plug-ins and Fragments". !STACK 0 java.lang.NullPointerException at java.io.File.<init>(File.java:222) at org.eclipse.pde.internal.core.PDERegistryStrategy.getContributionsTimestamp(PDERegistryStrategy.java:300) at org.eclipse.core.internal.registry.ExtensionRegistry.<init>(ExtensionRegistry.java:682) at org.eclipse.core.runtime.RegistryFactory.createRegistry(RegistryFactory.java:58) at org.eclipse.pde.internal.core.PDEExtensionRegistry.createRegistry(PDEExtensionRegistry.java:85) at org.eclipse.pde.internal.core.PDEExtensionRegistry.getRegistry(PDEExtensionRegistry.java:77) at org.eclipse.pde.internal.core.PDEExtensionRegistry.getExtensionPoint(PDEExtensionRegistry.java:134) at org.eclipse.pde.internal.core.PDEExtensionRegistry.findExtensions(PDEExtensionRegistry.java:205) at org.eclipse.pde.internal.core.SourceLocationManager.processExtensions(SourceLocationManager.java:342) at org.eclipse.pde.internal.core.SourceLocationManager.getExtensionLocations(SourceLocationManager.java:192) at org.eclipse.pde.internal.core.SourceLocationManager.searchExtensionLocations(SourceLocationManager.java:273) at org.eclipse.pde.internal.core.SourceLocationManager.findSourcePath(SourceLocationManager.java:60) at org.eclipse.pde.internal.ui.wizards.imports.PluginImportOperation.getSourceManager(PluginImportOperation.java:706) at org.eclipse.pde.internal.ui.wizards.imports.PluginImportOperation.importSourceArchives(PluginImportOperation.java:774) at org.eclipse.pde.internal.ui.wizards.imports.PluginImportOperation.importAsBinary(PluginImportOperation.java:380) at org.eclipse.pde.internal.ui.wizards.imports.PluginImportOperation.importPlugin(PluginImportOperation.java:342) at org.eclipse.pde.internal.ui.wizards.imports.PluginImportOperation.runInWorkspace(PluginImportOperation.java:147) at org.eclipse.core.internal.resources.InternalWorkspaceJob.run(InternalWorkspaceJob.java:38) at org.eclipse.core.internal.jobs.Worker.run(Worker.java:54)
Trying a second time (without restart) works without exception.
Created attachment 186604 [details] Test workspace
The NPE is probably unrelated since it's there since at least 3.6 (filed bug 334061) and I did not get the NPE in my dev workspace where I first encountered the issue. Also, I now tried several times in my dev workspace and could not reproduce. It would still be good to review the recent (e.g. since M4) PDE UI code changes.
(In reply to comment #5) > The NPE is probably unrelated since it's there since at least 3.6 (filed bug > 334061) and I did not get the NPE in my dev workspace where I first encountered > the issue. > > Also, I now tried several times in my dev workspace and could not reproduce. It > would still be good to review the recent (e.g. since M4) PDE UI code changes. Import is working fine for me in I20110118-0800 The import operation code hasn't changed since August. I can't see how Bug 330047 would change anything related to this (same state is created, if the contents of the array differed then the new array list, you would get an NPE). Can we mark this as WORKSFORME?
Couldn't reproduce in recent builds, including I20110124-1800.
I just had this again with I20110308-2000: Error Tue Mar 15 09:43:07 CET 2011 An internal error occurred during: "Import Plug-ins and Fragments". java.lang.NullPointerException at java.io.File.<init>(File.java:222) at org.eclipse.pde.internal.core.PDERegistryStrategy.getContributionsTimestamp(PDERegistryStrategy.java:303) at org.eclipse.core.internal.registry.ExtensionRegistry.<init>(ExtensionRegistry.java:682) at org.eclipse.core.runtime.RegistryFactory.createRegistry(RegistryFactory.java:58) at org.eclipse.pde.internal.core.PDEExtensionRegistry.createRegistry(PDEExtensionRegistry.java:85) at org.eclipse.pde.internal.core.PDEExtensionRegistry.getRegistry(PDEExtensionRegistry.java:77) at org.eclipse.pde.internal.core.PDEExtensionRegistry.getExtensionPoint(PDEExtensionRegistry.java:134) at org.eclipse.pde.internal.core.PDEExtensionRegistry.findExtensions(PDEExtensionRegistry.java:205) at org.eclipse.pde.internal.core.SourceLocationManager.processExtensions(SourceLocationManager.java:342) at org.eclipse.pde.internal.core.SourceLocationManager.getExtensionLocations(SourceLocationManager.java:192) at org.eclipse.pde.internal.core.SourceLocationManager.searchExtensionLocations(SourceLocationManager.java:273) at org.eclipse.pde.internal.core.SourceLocationManager.findSourcePath(SourceLocationManager.java:60) at org.eclipse.pde.internal.ui.wizards.imports.PluginImportOperation.getSourceManager(PluginImportOperation.java:709) at org.eclipse.pde.internal.ui.wizards.imports.PluginImportOperation.importSourceArchives(PluginImportOperation.java:777) at org.eclipse.pde.internal.ui.wizards.imports.PluginImportOperation.importAsBinary(PluginImportOperation.java:383) at org.eclipse.pde.internal.ui.wizards.imports.PluginImportOperation.importPlugin(PluginImportOperation.java:345) at org.eclipse.pde.internal.ui.wizards.imports.PluginImportOperation.runInWorkspace(PluginImportOperation.java:145) at org.eclipse.core.internal.resources.InternalWorkspaceJob.run(InternalWorkspaceJob.java:38) at org.eclipse.core.internal.jobs.Worker.run(Worker.java:54)
The plug-in where the import crashed was org.eclipse.core.filesystem.linux.x86. A plug-in test was running before, but it had been terminated for a few minutes before I started the import. But it's possible that a Synchronize Workspace operation was still running in the background.
Even if we can't reproduce, it would still be good to at least fix the NPE: ISharedPluginModel#getInstallLocation() can return null, so callers should protect against this.
Curtis, can you +1 the bug for RC1 and review my patch?
Created attachment 194568 [details] Fix
+1. Patch applied for RC1. Thanks for the patch Dani, didn't realize it was so trivial of a fix otherwise I would have looked at it sooner.
(In reply to comment #13) > +1. Patch applied for RC1. > > Thanks for the patch Dani, didn't realize it was so trivial of a fix otherwise > I would have looked at it sooner. Curtis, you should only REVIEW the patch. The committer who prepared it should commit it then to HEAD. But thanks anyway.
(In reply to comment #14) > Curtis, you should only REVIEW the patch. The committer who prepared it should > commit it then to HEAD. But thanks anyway. The style we have followed in PDE is as follows: 1) Set the milestone 2) Attach your patch 3) Use the flags to request a review (or multiple reviews). 4) Check that all reviewers are cc'd so they receive bug updates 5) Comment the bug asking for review 6) Assign the bug to yourself 7) Once all reviews are +1, the final reviewer can commit (commit comment must include bug number for tracking) In addition for contributed patches: 1) Bug is assigned to the person who commits the code 2) Bug is marked with keyword 'contributed' and iplog updated as necessary. If you would like this changed, we should discuss on Monday.
(In reply to comment #15) > (In reply to comment #14) > > Curtis, you should only REVIEW the patch. The committer who prepared it should > > commit it then to HEAD. But thanks anyway. > > The style we have followed in PDE is as follows: > > 1) Set the milestone > 2) Attach your patch > 3) Use the flags to request a review (or multiple reviews). > 4) Check that all reviewers are cc'd so they receive bug updates > 5) Comment the bug asking for review > 6) Assign the bug to yourself > 7) Once all reviews are +1, the final reviewer can commit (commit comment must > include bug number for tracking) > > In addition for contributed patches: > > 1) Bug is assigned to the person who commits the code > 2) Bug is marked with keyword 'contributed' and iplog updated as necessary. > > If you would like this changed, we should discuss on Monday. Yes, please. I personally find it somewhat offending if my work gets "stolen" from someone.
This bug could not be reproduced, but I verified that the null check has been released into the build.
