Maybe some of you can help me solving a very strange issue with java and JComboboxes. I tried several hours to trace down the problem but I cant find a solution. I don't want to paste a huge code here, but this simple loop demonstrates it:
JComboBox cb;
for(int i=0;i<1000;i++)
{
cb=new JComboBox();
}
I can run this code wherever I want, the 1000 ComboBoxes are never GCed and I do not understand, why???
A JComboBox creates a DefaultListModel with a listener to the combobox. So garbage collection on such an object cluster is postponed. However after the fourth run with me either the garbage was collected or the JIT found it did not need to create those objects.
Maybe your problem was that calling explicitly System.gc() not cleaning it up? That I can imagine.
*How to trace the problem
I tried the following to exclude components.
private static class ReducedJComboBox<T> extends JComboBox<T> {
@Override
public void setEditor(ComboBoxEditor anEditor) {
}
}
public static void main( String[] args )
{
System.out.println("; " + Runtime.getRuntime().freeMemory());
long t0 = System.currentTimeMillis();
ComboBoxModel model = new DefaultComboBoxModel();
ComboBoxEditor editor = new ComboBoxEditor() {
public Component getEditorComponent() {
throw new UnsupportedOperationException("Not supported yet.");
}
public void setItem(Object anObject) {
}
public Object getItem() {
throw new UnsupportedOperationException("Not supported yet.");
}
public void selectAll() {
}
public void addActionListener(ActionListener l) {
}
public void removeActionListener(ActionListener l) {
}
};
JComboBox cb;
for (int i = 0; i < 1000; i++) {
cb = new JComboBox();
cb.setModel(model);
cb.setEditable(false);
cb.setEditor(editor);
}
long dt = t0 - System.currentTimeMillis();
System.out.println("dt=" + dt + " ms; " + Runtime.getRuntime().freeMemory());
System.gc();
System.out.println("finally " + Runtime.getRuntime().freeMemory());
}
I tried with a simple testcase using the NBTestCase#assertGC method. The benefit of this method is that is prints out the strong references to the object you are checking in case of a memory leak. It also triggers GC by filling the heap with a dumb byte array, forcing the GC to start.
The very simple test case I used
public class ComboBoxMemoryLeak {
public static void main( String[] args ) {
EventQueue.invokeLater( new Runnable() {
@Override
public void run() {
List<WeakReference<JComboBox>> references = new ArrayList<WeakReference<JComboBox>>( );
JComboBox comboBox;
for (int i = 0; i < 1000; i++ ){
comboBox = new JComboBox( );
references.add( new WeakReference<JComboBox>( comboBox ) );
}
comboBox = null;
for ( int i = 0, referencesSize = references.size(); i < referencesSize; i++ ) {
System.out.println( "i = " + i );
WeakReference<JComboBox> weakReference = references.get( i );
NbTestCase.assertGC( "Combobox", weakReference );
}
System.out.println("No memory leak found");
}
} );
}
}
leading to the following trace on my Mac running with JDK1.6
i = 0
Exception in thread "AWT-EventQueue-0" junit.framework.AssertionFailedError: Combobox:
private static sun.awt.AppContext sun.awt.AppContext.mainAppContext->
sun.awt.AppContext@30f7f540-table->
java.util.HashMap@c324b85-table->
[Ljava.util.HashMap$Entry;@770fba26-[8]->
java.util.HashMap$Entry@63adf08f-value->
java.beans.PropertyChangeSupport@4f1b8540-children->
java.util.Hashtable@2305454a-table->
[Ljava.util.Hashtable$Entry;@4a9a4ba3-[0]->
java.util.Hashtable$Entry@6aed0f19-value->
java.beans.PropertyChangeSupport@23597cac-listeners->
sun.awt.EventListenerAggregate@2f39c244-listenerList->
[Ljava.beans.PropertyChangeListener;@2e2e06bd-[0]->
javax.swing.JViewport$1@2a72cf60-this$0->
javax.swing.JViewport@2b9c1dc4-parent->
javax.swing.JScrollPane@b99f7c6-parent->
com.apple.laf.AquaComboBoxPopup@6699166f-comboBox->
javax.swing.JComboBox@3bc634b9
at junit.framework.Assert.fail(Assert.java:50)
at org.netbeans.junit.NbTestCase$4.run(NbTestCase.java:1351)
at org.netbeans.junit.internal.NbModuleLogHandler.whileIgnoringOOME(NbModuleLogHandler.java:143)
at org.netbeans.junit.NbTestCase.assertGC(NbTestCase.java:1309)
at org.netbeans.junit.NbTestCase.assertGC(NbTestCase.java:1285)
at ComboBoxMemoryLeak$1.run(ComboBoxMemoryLeak.java:32)
at java.awt.event.InvocationEvent.dispatch(InvocationEvent.java:209)
at java.awt.EventQueue.dispatchEventImpl(EventQueue.java:677)
at java.awt.EventQueue.access$000(EventQueue.java:85)
at java.awt.EventQueue$1.run(EventQueue.java:638)
at java.awt.EventQueue$1.run(EventQueue.java:636)
at java.security.AccessController.doPrivileged(Native Method)
at java.security.AccessControlContext$1.doIntersectionPrivilege(AccessControlContext.java:87)
at java.awt.EventQueue.dispatchEvent(EventQueue.java:647)
at java.awt.EventDispatchThread.pumpOneEventForFilters(EventDispatchThread.java:296)
at java.awt.EventDispatchThread.pumpEventsForFilter(EventDispatchThread.java:211)
at java.awt.EventDispatchThread.pumpEventsForHierarchy(EventDispatchThread.java:201)
at java.awt.EventDispatchThread.pumpEvents(EventDispatchThread.java:196)
at java.awt.EventDispatchThread.pumpEvents(EventDispatchThread.java:188)
at java.awt.EventDispatchThread.run(EventDispatchThread.java:122)
So yes, on my machine I would conclude the JComboBox is still present in memory through a reference in the AppContext. Searching on that leads us to this SO question. I tried connecting JConsole to a main method with a while loop with your statement. Unfortunately my main program threw an OutOfMemoryException before JConsole could connect, so I couldn't generate a nice picture like trashgod did in his answer.
i can do this code where ever i wont the 1000 ComboBoxes are never GCed
and i do not get it, why
answer is maybe very simle Object could be GC'ed if there doesn't exist another reference to this Object, or another issue could be that you referenced from/to the static Object (never GC'ed only can take null value), or yours Objects are static
