Hi,
I've been doing some JCK 6b testing, and a DnD failure appears in the
testsuite. I've debugged the problem down to something that I cannot
rationalize, and I was hoping someone here could enlighten me.
The failing JCK test essentially calls
"DragSourceContext.updateCurrentCursor(1, 1, 1)" and then calls
DragSourceContext.getCursor, comparing this cursor to an expected cursor.
The problem is in updateCurrentCursor, which ignores the target action,
and simply hard codes it to DnDConstants.ACTION_NONE:
protected synchronized void updateCurrentCursor(int sourceAct, int
targetAct, int status) {
// if the cursor has been previously set then dont do any defaults
// processing.
if (useCustomCursor) {
return;
}
// do defaults processing
Cursor c = null;
--> targetAct = DnDConstants.ACTION_NONE;
switch (status) {
case ENTER:
case OVER:
case CHANGED:
int ra = sourceAct & targetAct;
[This is both openjdk 6 & 7.]
As a result of this override of targetAct, "ra" will _always_ evaluate
to false, and the code will never return a droppable cursor.
Does this override serve a purpose that I am unable to guess?
FWIW, I have run the JCK 6b api/java_awt/dnd tests with the highlighted
line deleted, and they all now pass.
Keith
