How to properly use observe to track region changes

ok, I have to admit, that for Java the docs are not really helpful - I will optimize that ;-)

Here is a working example for an onChange observation:

a helper for the formatted printout:
  private static void p(String msg, Object... args) {
    System.out.println(String.format(msg, args));
  }

the example:
      Region reg = new Region(0,0,80,80); // define the observed region
      reg.highlight(2);
      reg.onChange(new ObserverCallBack(){ // define the handler
        @Override
        public void changed(ObserveEvent evt) {
          if (evt.getCount() > 3) {
            p("in handler: %d - stopping", evt.getCount());
            evt.stopObserver();
          } else {
            p("in handler: %d", evt.getCount());
          }
        }
      });
      reg.observeInBackground(); // start observation in background forever
      int n = 0;
      while (reg.isObserving()) { // do something while observe is running
        p("%d - observing", n);
        reg.wait(1.0);
        n++;
      }

which prints something like this:
[log] highlight R[0,0 80x80]@S(0) for 2.0 secs
0 - observing
in handler: 1
1 - observing
2 - observing
3 - observing
4 - observing
in handler: 2
5 - observing
in handler: 3
6 - observing
in handler: 4 - stopping

--- some comments
- if you want to track some motion or some continuous change, then you have to either loop around an observe(time) or use a construct like the above example with an observe in background.
- in your main workflow, you have to track wether anything happens at all and if it does, when nothing happens anymore (which is not implemented in the example), but which is needed in your case.

come back if you need more help

BTW: you have to use the 1.1.2 nightly build in about 2 hours

With the code you provided, reg.isObserving() always evaluates to false on my end.
However, without the while loop, the event handler works fine.

Also, observeInBackground() requires a parameter of type double. The call without a parameter does not work for me.
Is there another way to run it forever, other than declaring a huge double value?

Is there a way to tell the % difference between changes? I've seen other Sikuli functions return a similarity parameter.
I am calling :
         characterRegion.onChange(new ObserverCallBack(){ // define the handler
  @Override
  public void changed(ObserveEvent evt) {
   List<Match> characterChanges = evt.getChanges();
   for(int i = 0; i < characterChanges.size();++i) {
    System.out.println(characterChanges.get(i));
   }
  }
 });

evt.getChanges() is the only function for ObserveEvent that I've tried that doesn't return null.

This is a small sample set of my moving character:
M[956,521 8x18]@S(S(0)[0,0 1920x1080]) S:-1.00 C:960,530 [-1 msec]
M[956,531 8x8]@S(S(0)[0,0 1920x1080]) S:-1.00 C:960,535 [-1 msec]
M[956,529 8x10]@S(S(0)[0,0 1920x1080]) S:-1.00 C:960,534 [-1 msec]
M[956,521 8x18]@S(S(0)[0,0 1920x1080]) S:-1.00 C:960,530 [-1 msec]

And this of my "still" character:
M[956,521 8x9]@S(S(0)[0,0 1920x1080]) S:-1.00 C:960,525 [-1 msec]
M[956,521 8x8]@S(S(0)[0,0 1920x1080]) S:-1.00 C:960,525 [-1 msec]
M[956,521 8x7]@S(S(0)[0,0 1920x1080]) S:-1.00 C:960,524 [-1 msec]
M[956,522 8x8]@S(S(0)[0,0 1920x1080]) S:-1.00 C:960,526 [-1 msec]

Even my still character slightly shifts in place. So I need to find a way to differentiate.
The only thing i see is the H of M, and the second variable of C.
But is there a way to find how much of the image actually changed during the event handler call?

Also, where do I find this 1.1.2 nightly build? the SikuliX page still says 1.1.1 is the current version.

Thanks RaiMan, that solved my question.