Bug 412873

Summary:	[regression] ArrayIndexOutOfBoundsException when starting Eclipse with a review open
Product:	z_Archived	Reporter:	Tomasz Zarna <tomasz.zarna>
Component:	Mylyn	Assignee:	Miles Parker <milesparker>
Status:	RESOLVED FIXED	QA Contact:
Severity:	normal
Priority:	P2	CC:	milesparker, sam.davis, steffen.pingel
Version:	2.0
Target Milestone:	2.0.1
Hardware:	PC
OS:	Windows 7
Whiteboard:

Description Tomasz Zarna

2013-07-12 12:13:35 EDT

It's a reincarnation of bug 408761, still occurring in 2.0. Running of master [1], starting Eclipse with a review already open:

java.lang.ArrayIndexOutOfBoundsException: 2
	at org.eclipse.emf.common.util.BasicEList.assign(BasicEList.java:118)
	at org.eclipse.emf.common.util.BasicEList.addUnique(BasicEList.java:418)
	at org.eclipse.emf.common.util.AbstractEList.add(AbstractEList.java:301)
	at org.eclipse.emf.common.notify.impl.BasicNotifierImpl$EAdapterList.add(BasicNotifierImpl.java:193)
	at org.eclipse.mylyn.reviews.core.spi.remote.emf.RemoteEmfConsumer.<init>(RemoteEmfConsumer.java:124)
	at org.eclipse.mylyn.reviews.core.spi.remote.emf.AbstractRemoteEmfFactory.getConsumerForLocalKey(AbstractRemoteEmfFactory.java:183)
	at org.eclipse.mylyn.internal.gerrit.core.remote.PatchSetContentIdRemoteFactory.pull(PatchSetContentIdRemoteFactory.java:40)
	at org.eclipse.mylyn.internal.gerrit.core.remote.PatchSetContentIdRemoteFactory.pull(PatchSetContentIdRemoteFactory.java:1)
	at org.eclipse.mylyn.reviews.core.spi.remote.emf.RemoteEmfConsumer.pull(RemoteEmfConsumer.java:164)
	at org.eclipse.mylyn.reviews.core.spi.remote.JobRemoteService$1.run(JobRemoteService.java:60)
	at org.eclipse.core.internal.jobs.Worker.run(Worker.java:54)

[1] https://git.eclipse.org/c/mylyn/org.eclipse.mylyn.reviews.git/commit/?id=c603cfc349a6deed355ea6388166665d71c8fb96

Comment 1 Steffen Pingel

2013-07-12 13:06:06 EDT

This doesn't surprise me considering that the model is accessed from a job. Miles, how was this "fixed" last time?

Comment 2 Miles Parker

2013-07-15 12:57:50 EDT

Its not the same issue if you look at the stack trace. The other one was driven by a call to pull user. Really, we shouldn't need true remote acess at all. I'll take a look today.

Comment 3 Miles Parker

2013-07-15 14:14:28 EDT

https://git.eclipse.org/r/#/c/14571/

Unfortunatly, I can't reproduce issue so I can't effectivly test, but this should do it. Tomek, can you verify that this addresses your issue?

Comment 4 Miles Parker

2013-07-16 18:53:41 EDT

Ok, *that* wasn't the right way to solve it. See bug 413124. Back to the drawing board...

Comment 5 Miles Parker

2013-07-16 19:19:36 EDT

https://git.eclipse.org/r/#/c/14605/

Everyone: Given prior issue (which wasn't easy to reproduce) we'll want to verify this one carefully. While it shouldn't carry any risk deadlock, because we're no longer adding the adapters synchronously, it's *possible* that we might introduce subtle updating issues with this change. I haven't seen anything, so just a note to be on the lookout for that.

Comment 6 Miles Parker

2013-07-19 13:16:14 EDT

I've gone ahead and submitted this change. That will give us a chance to evaluate behaviour prior to 2.0.1 release. We can back this one out as well is we see issues, but I don't anticipate any problems.