Per axis updates in the LogitechExtremeJoystick

[cal-pratt]

The LogitechExtremeJoystick now stores the last polled values of heave sway surge and yaw. If an axis event happens it will not cause the Joystick to emit the invalid values by polling data that has not yet been prepared by the joystick. Fixes #263

EDIT looking at this more closely I was able to remove the createAxes method in the Joystick source completely. We can get all the data we need from events. Same thing can be done for the buttons; just have to verify what all the component id's are for them.

[ConnorWhalen]

Does isolating polling to a specific axis fix the problem? Are the invalid axis updates not going to be processed like every other axis update? Seems like behaviour will be unchanged

[cal-pratt]

Does isolating polling to a specific axis fix the problem? Are the invalid axis updates not going to be processed like every other axis update? Seems like behaviour will be unchanged

Actually now that I think of it no... hahaha. will update this

[cal-pratt]

Alright @ConnorWhalen I think I fixed it; does this make sense? On top of what was already changed I added the following logic:

        .skip(INITIAL_INPUT_DELAY, TimeUnit.MILLISECONDS, scheduler);

Now the initial events won't be seen by the joystick controller, and when events do happen, the -1 values are never polled because each axis is polled individually.

[ConnorWhalen]

The logic is sound. I'll review the code then we can test it on thursday

[cal-pratt]

Rabble Rabble Rabble!

Just tested this. The change prevents -1 being displayed, but it's still weird.

When the joystick reattaches we create a new LogitechExtremeJoystick object, meaning we lose the last values after a disconnect.. all the values go to zero. ie) if you set the slider to -1 then unplug, reconnect, and move the another axis, then only that axis value updates and it reports the slider value as 0.

At the same time it's still a lot safer than if the values go to -1. Additionally if we keep the old values we still don't know if the joystick was moved while it was unplugged. ie) Set the slider to -1. unplug, set the slider to 0. reconnect. move a different axis. slider would still be incorrect reporting -1 ...

So moral of the story is that joydev is lame and we should rewrite the kernal /s

[ConnorWhalen]

I think this is as good as it's going to get. I'm good with pushing this

[ConnorWhalen]

Since there's no way to know the re-connected joystick's initial state setting every axis to zero is the safest option

[cal-pratt]

There was a small bug on thursday, Axis RZ for yaw was set to be Axis RX. .. my bad
Am paranoid so I'd like to test this one more time before we push.