Joystick 'get_action_strength' Weirdness

marty

Under Godot 3.1, I have two gamepads attached.

With the first gamepad (Device 0), reading 'get_action_strength' from its left joystick returns rational values (ie. in the -1 to 1 range).

But with the second gamepad (Device 1), reading 'get_action_strength' from its first joystick returns NaN when pressed in any direction (and returns a '0' when not pressed).

I'm baffled.

The gamepads are tested, work fine in the GitHub Godot joypads demo, and continue the described behavior when switched - so it seems to be Godot-centric (ie. not a hardware issue).

Anyone have any idea what I might be doing wrong? Could this be a Godot bug?

marty

Solved!

For anyone else who makes the same bonehead mistake ...

At some point, I set the dead-zones for the second gamepad to 1.0 (ie. from the default 0.5), which apparently results in NaN returns from analog inputs like joysticks.

I still wonder if Godot couldn't handle this better? If nothing else, by passing a code-friendly '0' instead of a code-breaking NaN.

TwistedTwigleg

That is definitely strange. I agree that it returning 0 would make a lot more sense than passing NaN. Or at least it wouldn't be as game breaking, since last I knew there isn't a way to check for NaN in Godot.

Maybe it is a bug? Theoretically speaking, a deadzone of 1 should always return 0 because any input the joystick makes is within a 1 to -1 radius, and if the joystick deadzone is set to a radius of 1, every joystick input should be within the deadzone and therefor should always return 0, or at least that is what I would expect.

This is defiantly good to know though! I imagine debugging something like this could be quite time consuming.

Megalomaniak

I've changed this from discussion into a question so you can mark your second post as the answer. ;)

@marty said: At some point, I set the dead-zones for the second gamepad to 1.0 (ie. from the default 0.5), which apparently results in NaN returns from analog inputs like joysticks.

I still wonder if Godot couldn't handle this better? If nothing else, by passing a code-friendly '0' instead of a code-breaking NaN.

I wonder, might setting the dead-zones to 1.0 result in NaN because the value somehow ends up out of bounds? Might be worth creating a issue report on the tracker maybe.

marty

Indeed and agreed, TwistedTwigleg and Megalomaniak! ;-)

I'm off to GitHub now ...