[osx] Unexpected behavior for right stick of modern Xbox controller

[connorjclark]

Xbox Wireless Controller

This controller has three sticks: 0 and 1 are the left and right sticks, and 2 is the DPAD (hat switch).

According to the allegro joystick interface on OSX, these switches have these number of axis:

Stick 0 has 3 axis (kHIDUsage_GD_X, kHIDUsage_GD_Y, kHIDUsage_GD_Y)
Stick 1 has 1 axis (kHIDUsage_GD_Rz)
Stick 2 has 2 axis (the hat switch)

I noticed that when processing stick 1 in static void add_elements, two HID usage interfaces are ignored with decimal values 196 and 197. These values are not defined in the HID header.

Listing the odd things:

• Why does Stick 0 have 3 axis instead of 2? I know it's a totally different stack, but https://hardwaretester.com/gamepad only lists two axis for both Stick 0 and 1.
• Why don't Sticks 0 and 1, which are identical physical controls, have an identical number of axis?
• What are usages 196 and 197? Did this Xbox controller make up their own usage ids?
• When moving the right stick (Stick 1) to the right, I get a mix of events for Stick 1, and for Stick 0 Axis 2, which is super broken. When moving that right stick Up I don't get any events from Stick 0.

[connorjclark]

I found a similar issue on Windows for DirectInput wjoydxnu.cpp w/ the Stadia controller. Same exact stick layout, same exact number of axis as the example described above.

Also found issues with secondary stick for PS4 dualshock controller. Allegro discovers the controller supports Rx/y/z, but actually the inputs that come through when moving the second stick are for Z and Rz.

Perhaps a good solution: https://www.gamedev.net/forums/topic/613913-directinput-identifying-second-thumbstick/4878869/

[connorjclark]

the mac issue was fixed w: #1698

the windows issue is fixed with #1463