Bugs item #2901257, was opened at 2009-11-20 10:54
Message generated for change (Settings changed) made by hansonr
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=379133&aid=2901257&group_id=23629
Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: v11
>Status: Closed
Resolution: None
Priority: 5
Private: No
Submitted By: Dirk Walter (dirkwalter)
Assigned to: Nobody/Anonymous (nobody)
Summary: Random NullPointerException
Initial Comment:
I occasionally randomly get a nullpointer exception, I think it is a threading
issue since it ussualy happens if I change the molecule I am showing before it
finishes displaying another one. I am using jmol 11.8.9.
Stacktrace:
java.lang.NullPointerException
at org.jmol.viewer.Viewer.getModelFileNumber(Unknown Source)
at org.jmol.viewer.Viewer.setStatusFrameChanged(Unknown Source)
at org.jmol.viewer.AnimationManager.setStatusFrameChanged(Unknown
Source)
at org.jmol.viewer.AnimationManager.setAnimationOff(Unknown Source)
at org.jmol.viewer.AnimationManager.setAnimationOn(Unknown Source)
at org.jmol.viewer.AnimationManager.clearAnimation(Unknown Source)
at org.jmol.viewer.AnimationManager.clear(Unknown Source)
at org.jmol.viewer.Viewer.zap(Unknown Source)
at org.jmol.viewer.Viewer.zap(Unknown Source)
at org.jmol.viewer.Viewer.createModelSetAndReturnError(Unknown Source)
at org.jmol.viewer.Viewer.createModelSetInline(Unknown Source)
at org.jmol.viewer.Viewer.openStringInline(Unknown Source)
----------------------------------------------------------------------
Comment By: Bob Hanson (hansonr)
Date: 2009-11-24 23:05
Message:
OK, this error I know. It's coming from the renderer methods because some
other thread has done a ZAP command (accompanying any new model), and that
version of Jmol has a bug I just fixed that sets the modelSet to null for a
short period of time. That's why we have an intermittent bug. So the first
thing to do is try that again with the latest version of Jmol I have
checked in (11.8.11). In this new version I have the ZAPping thread send a
cancel message to the renderer and wait until the renderer stops. This
should completely fix that issue.
----------------------------------------------------------------------
Comment By: Dirk Walter (dirkwalter)
Date: 2009-11-24 11:47
Message:
Two more random nullpointer exceptions, not sure if they are related, still
using 11.8.9, but now all data is loaded using the data command, so
threading should not be an issue.
They also happened right after each other so they are probably related.
java.lang.NullPointerException
at org.jmol.viewer.Viewer.getVisibleFramesBitSet(Unknown Source)
at org.jmol.modelset.ModelSet.setModelVisibility(Unknown Source)
at org.jmol.viewer.Viewer.setModelVisibility(Unknown Source)
at org.jmol.viewer.Viewer.renderScreenImage(Unknown Source)
at org.jmol.viewer.Viewer.renderScreenImage(Unknown Source)
at
gov.nih.nhlbi.icbl.qbt.biochemistry.gui.editors.panels.JmolPanel.paint(JmolPanel.java:83)
at javax.swing.JComponent.paintToOffscreen(JComponent.java:5124
java.lang.NullPointerException
at org.jmol.viewer.Viewer.setShapeProperty(Unknown Source)
at org.jmol.viewer.Viewer.hoverOff(Unknown Source)
at org.jmol.viewer.MouseManager.hoverOff(Unknown Source)
at org.jmol.viewer.MouseManager.mouseEntered(Unknown Source)
at org.jmol.viewer.MouseManager11.mouseEntered(Unknown Source)
at java.awt.Component.processMouseEvent(Component.java:6272)
at javax.swing.JComponent.processMouseEvent(JComponent.java:3
*-*-*-*-*
The relevant code in the custom JPanel:
public void paint(Graphics g) {
if(hide || queue.hasQueued()) {
super.paint(g); //paint a blank panel while waiting for data
} else {
getSize(currentSize);
g.getClipBounds(rectClip);
viewer.renderScreenImage(g, currentSize, rectClip);
}
}
----------------------------------------------------------------------
Comment By: Dirk Walter (dirkwalter)
Date: 2009-11-23 15:09
Message:
That could very well by happening, I fire a event that goes into the Event
Dispatch thread to update the JMol panel, then returns, at which point the
next thread might be going into the EDT to fire another update, such as
when the user is scrolling through the list.
----------------------------------------------------------------------
Comment By: Bob Hanson (hansonr)
Date: 2009-11-23 12:22
Message:
That's supposed to work fine. The applet, for example, does exactly that to
apply a script after a load of string data.
The curious thing here is that the zap method is finding a null modelSet.
That implies there is a command being given by another thread prior to the
zap command that is clearing out the model set. But as far as I can see,
the only command that would do that would be zap itself. How two threads
got going like that is still a mystery to me.
I've made some changes that should make it completely impossible to ever
hit this exception again, but of course it would be more satisfying to know
why it happened in the first place. About the only thing I can imagine is
that you issue a load or zap command within
jmolviewer.script("....")
and then, shortly thereafter you also issue
jmolviewer.openStringInline(...)
The first would be running queued by the script processor thread, and the
second would be excuted by the Event thread.
Curious....
----------------------------------------------------------------------
Comment By: Dirk Walter (dirkwalter)
Date: 2009-11-23 11:37
Message:
I am following it with another command, so that would be the source of the
problem.
We have 2 types of molecules we are showing and we have different display
scripts for each, so I do a openStringInline followed by a evalString to
set the display properties, I made sure the commands are done in the Event
dispatch thread and called in that order. I will try the data command and
see if it solves the problem, unless there is a better way?
----------------------------------------------------------------------
Comment By: Bob Hanson (hansonr)
Date: 2009-11-23 07:20
Message:
Dirk, openStringInline should work fine provided that is all you are doing
and not following it up with some other command. The alternative is a DATA
command within a script. I'm still puzzling over why the ZAP command from
that is finding no model set. The only think I can think of is that you are
issuing two commands in a row, first a jmolViewer.script() or
jmolViewer.evalString() command, which gets queued and executed by an
independent thread, and then this openStringInline command.
Bob
----------------------------------------------------------------------
Comment By: Dirk Walter (dirkwalter)
Date: 2009-11-23 06:06
Message:
if openStringInline is not the way to do it how would I do this better?
----------------------------------------------------------------------
Comment By: Bob Hanson (hansonr)
Date: 2009-11-22 13:27
Message:
I believe this is a bug that goes WAY back to the origins of Jmol. I have
checked in fixes for Jmol 11.9.11 and Jmol 11.8.10. The problem arises from
the repaint renderer thread continuing during the zap command. Possibly...
----------------------------------------------------------------------
Comment By: Bob Hanson (hansonr)
Date: 2009-11-20 12:59
Message:
hmm, the zap is only being called from createModelSetAndReturnError because
there was an error in the data so that an atom set was not created. It's
trying to clear the (already cleared) model with a zap. That could be the
problem.
----------------------------------------------------------------------
Comment By: Bob Hanson (hansonr)
Date: 2009-11-20 12:54
Message:
I see you are using openStringInline. That's not queued, and it could be
causing the problem. I'll look into that. It would be nice to get this to
be reproducible.
----------------------------------------------------------------------
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=379133&aid=2901257&group_id=23629
------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Jmol-developers mailing list
Jmol-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/jmol-developers
