Most math and backend storage was floats anyways, so most of these changes just half the data copied during function calls without reducing precision.

Improved UI when no joysticks are connected. Also removed hotplugging support for joysticks as it wasn't working, though SDL2 should fix that.

Vastly improved the joystick documentation, removed some redundant/unnecessary code, and fixed a couple of small bugs. This code should be ready for live testing.

All UI elements in the JoystickWidget window are now updated properly when switching UASes, which was the last major issue with this code. A single small bug remains involving the UI not updating the axis values when switching joysticks until the axis is moved.

Have settings loaded/stored correctly for all joysticks based on their UAS type and autopilot type. Switching between UASes doesn't work right yet, however.

The settings for each joystick are loaded and saved on switch. The state of the enable joysticks checkbox is not saved yet, however, though it should really be tracked by JoystickInput so that it can sleep more when joysticks are disabled.

Restored range limiting to the throttle channel. Now if the UAS can't reverse, the throttle can be set to only use the positive range of the axis (useful with auto-centering control sticks) or to use the full range (for sticks that hold position).

The joystick code now understands if a UAS can reverse or not (based on UASInterface::systemCanReverse) and scales the joystick input accordingly. This isn't ideal on auto-centering axes and so the limit-range code should be brought back as an option for the throttle channel.

Joystick axes can now be specified as being exclusively-positive in the range [0:1.0] for the throttle. This should really be a vehicle-specific setting.

Added an inversion checkbox for each joystick axis so they can be individually inverted or not. This will be a joystick-specific configuration and independent of the autopilot used. Also refactored the code so that the JoystickAxis objects connect directly to the JoystickInput thread for updating the backend. The JoystickWidget is now just a facilitator and doesn't do any redirecting itself.

The button/axis/hat readings are now compared to the previous readings and signals are only emitted if they changed. The joystickChanged signal is still emitted every sample period.

The hat readings from the joystick now works correctly. Also moved some signals over to using an enum from an int type.

Moved the JOYSTICK_*_MAPPING enum into JoystickInput where it belongs. Also renamed some local variables.

Refactored some variables and added some documentation. Also, now the joystickChanged signal emits NaN for axes that aren't mapped, so consumers should handle this properly.

Refactored much of the joystick interface with the largest changes being the removal of dead/unused code including broken event-based SDL reading code. Additionally the mapping code should now work, though it hasn't been tested with controller switching or a real UAS. Finally the existing settings code has been commented-out as it is not compatible with the current state of this code and needs to be able to support different joysticks based on name.

The joystick settings window has now been reworked to be a fixed minimum size and dynamically scale with both the number of buttons and axes on a given joystick/controller. Mapping functionality is incomplete at this point still, though the UI exists for it.

The Joystick Settings window now supports selecting between multiple joysticks for control input and updates its UI accordingly. Much work is left to be done: light/dark styling, settings management, control mappings.

Joystick axes can now be specified as being exclusively-positive in the range [0:1.0] for the throttle. This should really be a vehicle-specific setting.

Added an inversion checkbox for each joystick axis so they can be individually inverted or not. This will be a joystick-specific configuration and independent of the autopilot used. Also refactored the code so that the JoystickAxis objects connect directly to the JoystickInput thread for updating the backend. The JoystickWidget is now just a facilitator and doesn't do any redirecting itself.

The button/axis/hat readings are now compared to the previous readings and signals are only emitted if they changed. The joystickChanged signal is still emitted every sample period.

The hat readings from the joystick now works correctly. Also moved some signals over to using an enum from an int type.

Moved the JOYSTICK_*_MAPPING enum into JoystickInput where it belongs. Also renamed some local variables.

Refactored some variables and added some documentation. Also, now the joystickChanged signal emits NaN for axes that aren't mapped, so consumers should handle this properly.

Refactored much of the joystick interface with the largest changes being the removal of dead/unused code including broken event-based SDL reading code. Additionally the mapping code should now work, though it hasn't been tested with controller switching or a real UAS. Finally the existing settings code has been commented-out as it is not compatible with the current state of this code and needs to be able to support different joysticks based on name.

The joystick settings window has now been reworked to be a fixed minimum size and dynamically scale with both the number of buttons and axes on a given joystick/controller. Mapping functionality is incomplete at this point still, though the UI exists for it.

The Joystick Settings window now supports selecting between multiple joysticks for control input and updates its UI accordingly. Much work is left to be done: light/dark styling, settings management, control mappings.

3DMouse working on Windows and Ubuntu