Eclipse just crashed so I’m going to file a bug now…
Thursday, June 26th, 2008Actually, wait a second there, cowboy (or cowgirl). Before you file a bug, let’s take a look see at those logs of yours. You can find it as a workspace/.metadata/.log log file or as an hs_err_pid log file somewhere (possibly where Eclipse is installed) if you’re using Sun’s JVM. It is quite possible that the bug you are about to file is a duplicate of an existing bug that is already in Eclipse’s Bugzilla database. There’s been a ton of duplicates filed recently and the same questions keeps coming up on IRC over and over again that it’s almost driving me crazy. So for the sake of my own sanity and your own personal time (so you don’t need to waste time asking for help if it’s already a known issue), let’s take a look at three of the common crashers that have risen up in the past couple of weeks and months.
Do you see something like…
org.eclipse.core.internal.dtree.DataTreeNode?
org.eclipse.jdt.internal.compiler.lookup.ParameterizedMethodBinding?
This one is really famous. In fact, this bug has been filed so many times I sometimes wonder if it’ll have more duplicates than what the 64-bit Windows SWT bug did. That one was rather unfortunate since it was fixed but resurfaced later…but I digress. Anyway, common symptoms for this crasher include “Eclipse crashes when I save…”, “Eclipse crashes when I open a project…”, “Eclipse crashes when I…”. I mean, these are such common operations that you just have to go think “Oh wow, either my setup got hosed or Eclipse is a little too flaky for my liking here!”.
To resolve this issue, I suggest you take a look at bug 214092 (a number that I have now committed to memory for bug triaging and IRC support purposes). This is actually not an Eclipse problem. There is a VM bug in Sun’s Java 6 VM that is causing this crasher. I’ve been told by Ismael that this has been fixed in the Java 7 builds so that’s a good start. There’s also another bug that’s for backporting the fix to Java 6 so we should see that soon enough.
If you don’t see the above, are you seeing something like…
org.eclipse.swt.SWTError: XPCOM error -2147467262?
This one has grown in Bugzilla activity recently due to the advent of Firefox 3. You can take a look at bug 213194 for more information. I’ve heard some users comment that Eclipse is still crashing on the final Firefox 3 release. This contrasts with the original problem that was caused by users using a Firefox 3 beta build that was older than what SWT was expecting. In any case, if removing Firefox 3 doesn’t help (someone on IRC just removed XULRunner 1.9 and it helped in that case), I suggest commenting on the bug with as much information about your system as possible.
If you’re not seeing either of the two above, perhaps you’re seeing something like…
Java_org_eclipse_swt_internal_mozilla_XPCOM_NS_1NewLocalFile?
This one is rather new compared to the other two and I haven’t really seen this one brought up too often on IRC myself. This problem is captured in bug 236724 with a workaround suggested by a user (yay for community involvement!). The workaround has supposedly worked for another person I was talking to on IRC, so hey, it might work for you too!
For those of you that aren’t having any problems with Eclipse, I hope this post was still educational to you and hope that you can spread this information to your peers and coworkers so that they can resolve such problems if it arises for them.
For those of you that are hitting one of the bugs I described here, I hope you can find a solution or workaround in the bug reports I mentioned.
Lastly, for those of you that are having problems with Eclipse that don’t seem to match any of the ones I’ve described above, I’m sorry I wasn’t able to be of any help. :( Please feel free to file a bug and/or ask about it on the newsgroup or on IRC.
Update:
Do you see something like…
gtk_enumerate_printers?
I forgot to mention this one. Some users have reported that Eclipse seems to hang indefinitely (or for a very long time, like over a minute) whenever you try to open a file in the editor. This is a slightly uncommon problem that hits users on Linux systems using gtk+ 2.10 and up. This is a bug caused by gtk+ wherein the aforementioned function hangs the UI thread. Originally, the bug was fixed in one of the maintenance releases of gtk+ 2.10 but it has somehow resurfaced again in the 2.12 line. The solution is to start Eclipse with the following method:
eclipse -vmargs -Dorg.eclipse.swt.internal.gtk.disablePrinting
If that does help, please add your system’s information to bug 215234 so that the SWT team can collect all of this information and try to fix it. I know this isn’t a crasher but it is pretty much annoying enough to be a crasher since once it hangs you pretty much have to kill Eclipse and restart it or wait for the function to return (who knows when that’s going to happen?).


