Name: rm29839 Date: 10/02/97


I've found that Java is actually fast enough for
*action* games on some JITs. (I've gotten
upwards of even around 90 frames per second.)

But I need keyboard/mouse/joystick *polling*
(equivalent of GetAsyncKeyState()/joyGetInfoEx()
on Win32) to make the games playable at all
(as action games, that is).

Please add them. Those games depends on them.

Thank you.


company - Acorn Associates , email - ###@###.###
======================================================================

Name: krT82822 Date: 12/30/99


java version "1.2.2"
Classic VM (build JDK-1.2.2-W, native threads, symcjit)

I need the ability to poll to find the mouse position. I need this to support
an autoscroller, and there are plenty of people who would like to use it for
other purposes as well; see bug 4083501, for instance.

Here is the reason I need this for an autoscroller. My autoscroller currently
triggers when the mouse is in a certain area. Because there is no way to poll
the current position of the mouse I must use a Timer, and in that Timer I must
repost mouseDrag events. When I do this for a view that takes a long time to
draw, the timer reposting mouse events takes precedence over actual mouse
motion.

(I think that is a bug also anyway, and I will submit a seperate report for it;
the bottom line, IMHO, is that your input event loop should be running in a
seperate thread from the paint event system; since it's not, the UI feels very
sluggish and unresponsive; I'm aware of the reasons you didn't do that, but I
think in the long run it was the wrong choice. But if I could poll it wouldn't
matter as much.)

The result is that autoscroll events pile up and are handled before "real"
mouse drag events, so that when I move the mouse out of the autoscroll area to
stop autoscrolling, it takes 30sec-1min to stop scrolling, which is not very
user friendly.

If I could poll the mouse position, I wouldn't have to post a new mouse event
and my autoscroller would stop scrolling immediately.

As pointed out in bug 4083501, there are many other reasons people might want
to poll hardware input devices, and they all come down to the same reason:
eliminating innacuracy due to long event queue length. But special support for
mouse/keyboard/joystick polling without having to do something radical like JNI
would be very nice.
(Review ID: 99448)
======================================================================