I had this bug fire off on me last night. Unfortunately, after reviewing the error log, I managed to delete it without saving or submitting it
However, I did glean some useful information from the additional log info I added:- The offending DragGesture was generated 10 milliseconds after the previous DragGesture, but was based on exactly the same set of events.
It seems pretty clear to me that a bug in the JVM is occasionally generating a spurious DragGestureRecognized event.
I believe we can detect this case and ignore it, but I eagerly await any more bug submissions from people who see thison beta 6.
I wonder if this problem would go away if we switched to the Swing DnD
classes, from the AWT DnD classes we use now? Iām not suggesting we do this
before 3.1, but I might give it a try after that. (The DnD code could use
some refactoring anyhow.)
Ah, that rings a bell. This bug has been reported before when using AWT DnD classes with Swing components.
Yes, I think your suggestion is a good one, but definitely not in 3.1.0. It would be good for a 3.1.x release perhaps.
In the meantime, I am fairly confident that just quietly catching and ignoring the Exception and returning with no action will solve our problem until we do refactor.
Ok. Because this is such a hard problem to reproduce, I think we should do
this toward the beginning of 3.2 development, in order to give us maximum
time to see if our changes were successful.