Community
Participate
Working Groups
When I expand the "this" variable, I see the following errors on every step. If the tree isn't expanded, I don't see the errors. Don't know if there is anything that can be done or not. Signature cmd: exception java.lang.Exception: Couldn't get ClassFile object for signature cmd Interfaces cmd: exception java.lang.Exception: Couldn't get ClassFile object for Interfaces cmd Signature cmd: exception java.lang.Exception: Couldn't get ClassFile object for signature cmd Interfaces cmd: exception java.lang.Exception: Couldn't get ClassFile object for Interfaces cmd Signature cmd: exception java.lang.Exception: Couldn't get ClassFile object for signature cmd Interfaces cmd: exception java.lang.Exception: Couldn't get ClassFile object for Interfaces cmd Signature cmd: exception java.lang.Exception: Couldn't get ClassFile object for signature cmd Signature cmd: exception java.lang.Exception: Couldn't get ClassFile object for signature cmd Signature cmd: exception java.lang.Exception: Couldn't get ClassFile object for signature cmd Signature cmd: exception java.lang.Exception: Couldn't get ClassFile object for signature cmd Signature cmd: exception java.lang.Exception: Couldn't get ClassFile object for signature cmd Signature cmd: exception java.lang.Exception: Couldn't get ClassFile object for signature cmd
Could you provide some steps to reproduce this problem. I tired with the "SmallBall" demo, and could not reproduce the problem.
Ah - the error message are in the WTK console. I am not able to see any errors in the Eclipse console, so I'm not sure there is anything we can do about this. It seems to be related to an "<unknown value>" we are getting for the "canvas". When I debug it - I see that the canvas has been garbage collected, which does not make much sense to me.
Darin, When this is run via my EclipseME plugin, the WTK console and Eclipse console are one in the same thing. So, maybe it looks worse to me. On the other hand, it is just an annoyance.
This appears to be causing more than just annoyance for some users. See http://sourceforge.net/tracker/index.php?func=detail&aid=904266&group_id=86829&atid=581073 for a few examples. I took a look at the referenced bugs and related bugs on Sun's Bugzilla. It really appears to me that they are expecting for ClassPrepare events to occur before things will work. http://developer.java.sun.com/developer/bugParade/bugs/4459954.html http://developer.java.sun.com/developer/bugParade/bugs/4459390.html http://developer.java.sun.com/developer/bugParade/bugs/4459388.html http://developer.java.sun.com/developer/bugParade/bugs/4385248.html http://developer.java.sun.com/developer/bugParade/bugs/4463162.html http://developer.java.sun.com/developer/bugParade/bugs/4449649.html Is there any chance that the class prepare event code that I suggested for Eclipse bug 30816 could be incorporated to fix this problem?
I just tested the change to request class prepare events for KVM connection as I documented in 30816. It seems to completely solve this problem.
We will attempt to do something for this for 3.0. Perhaps we will have to add a preference such that we can allow users to turn this on for KVM.
I would guess that EclipseME (and other) users would be happy to have some way to set this value as soon as possible. I think eventually, it might be necessary to provide some type of debugging "type" that can specify all of the debugging values and then be flipped for individual projects. That is definitely not necessary for 3.0.
I commented on the solution to this in bug #50531. Sun has a new WTK that works (I've confirmed it), but the only way to get at it is to download Sun ONE 5, Micro Edition, and rip it out. Works flawlessly with the latest Eclipse I20040309. Will probably work with M7.
Err, sorry, bug #50530
Although it sounds like the KVM is making steps in the right direction, this will not help users attempting to debug using Eclipse for quite some time. I would still like to stress the importance of doing whatever we can for current Eclipse and WTK users. I don't really want to maintain a patched version of the Eclipse debugging plugin, but I think this particular fix is important enough that I would do that if necessary.
Created attachment 8729 [details] Patch to M7 org.eclipse.jdt.debug for KVM debugging I'm attaching a diff to the M7 code for org.eclipse.jdt.debug with the changes I mention in bug 30816. These changes seem to greatly improve the usability of debugging against the KVM. Any chance we will see these changes in M8?
It is unlikely this will be included in M8. Your patch showed up just as we were locking down for the M8 build week and test pass.
I have just verified that this patch still applies and functions as expected in the M8 code from CVS. Any chance this patch (or something similar) will be applied for M9?
Planning to have this fixed in M9.
Applied patch. Looks like the simplest change we can make for KVM users. Changes to VirtualMachineImpl and EventRequestManagerImpl. The change will have no effect on non-KVM users.
please verify, Luc.
Verified.