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.

The UASInterface now defines a systemCanReverse() function for indicating if the UAS is capable of commanding a reverse thrust.

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.

The UI now updates properly for when a detected joystick is lacking either buttons or axes (as in the case of an in-built accelerometer).

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 JoystickWidget now has a parent so that it gets cleaned up when the MainWindow is closed preventing zombie-dialog syndrome.

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.

The QGCUnconnectedInfoWidget now remains square and has more rounded edges so that it looks more like a button.

Finished merging QGCVehicleConfig.ui, which I hadn't done properly last night. Things are now in a usable state again.

Changed many functions in the abstract *Link classes to be const. This facilitates setting other functions as const. The metadata for specific links are reset when the links are added.

The details on the success/errors for the MAVLink protocols are now shown for each link as tooltips that are updated every 5s. Values are currently wrong and unsure why as of yet.

Removing UASes now returns the UI to an almost pristine state. This was done by altering the activeUASSet signal so that it can emit a NULL UAS. Crashes may still exist in autopilot-specific code.

Partway through refactoring of UAS deletion. This is in order to actually support removing UASes from QGC, which is currently impossible.