Ouya Controller doesn't fully work

[slime73]

Original report by Landon “Karai” Manning (Bitbucket: karai17, GitHub: karai17).

---

The directional pad seems to work, but other buttons do not. Probably an SDL issue.

[slime73]

Original comment by Pavol Rusnak (Bitbucket: prusnak, GitHub: prusnak).

---

How do you test?

[slime73]

Original comment by Landon “Karai” Manning (Bitbucket: karai17, GitHub: karai17).

---

#!lua

function love.joystickpressed(joystick, button)
  love.graphics.print(joystick .. " - " .. button, 100, 100)
  print(joystick, button)
end

This should print out every recognized button on any controller. If a button does not print anything, love/SDL is not registering said button.

[slime73]

Original comment by Colby Klein (Bitbucket: shakesoda, GitHub: shakesoda).

---

I wasn't getting anything with my Ouya controllers in my testing, googling showed that SDL2 has had some work on controllers (among other things) in Android post-2.0.1 (in HG tip) so this just might be resolvable by bumping the SDL version.

[slime73]

Original comment by Martin Felis (Bitbucket: MartinFelis, GitHub: MartinFelis).

---

I have just updated SDL2 to the latest development version (changeset d4a88c49247e). A new alpha build can be found at http://fysx.org/~martin/shared/love_android_sdl2_alpha5.apk.

The gamepad API for Android was introduced with API level 12, therefore it will only 3.1 and above.

[slime73]

Original comment by Colby Klein (Bitbucket: shakesoda, GitHub: shakesoda).

---

Controllers are working great now! The only issue I've found so far is that the Ouya system button doesn't respond (and it should - it's usually used for pause).

[slime73]

Original comment by Landon “Karai” Manning (Bitbucket: karai17, GitHub: karai17).

---

perhaps it is registering as a hat instead of a button, or some other wonky issue?

[slime73]

Original comment by Martin Felis (Bitbucket: MartinFelis, GitHub: MartinFelis).

---

I fear it requires adjustment of the file https://hg.libsdl.org/SDL/file/2e4f1bd21196/src/joystick/SDL_gamecontrollerdb.h. However I do not have such a controller at hand so I cannot test it. Maybe with help from the SDL2 mailing list a fix could be made.

[slime73]

Original comment by Alex Szpakowski (Bitbucket: slime73, GitHub: slime73).

---

You can add / change game controller mappings in SDL with SDL_GameControllerAddMapping. This is exposed in LÖVE (although a bit differently) with love.joystick.setGamepadMapping.

[slime73]

Original comment by Carlo Cabanilla (Bitbucket: clofresh, GitHub: clofresh).

---

The ouya controller wasn't working for me as a gamepad until I manually mapped the buttons:

gamepads = {
    -- ouya controller
    ["4f5559412047616d6520436f6e74726f"] = {
        {"a", "button", 6},
        {"b", "button", 7},
        {"x", "button", 9},
        {"y", "button", 10},
        {"leftstick", "button", 16},
        {"rightstick", "button", 17},
        {"leftshoulder", "button", 12},
        {"rightshoulder", "button", 13},
        {"dpup", "button", 1},
        {"dpdown", "button", 2},
        {"dpleft", "button", 3},
        {"dpright", "button", 4},
        {"leftx", "axis", 1},
        {"lefty", "axis", 2},
        {"rightx", "axis", 4},
        {"righty", "axis", 5},
        {"triggerleft", "axis", 3},
        {"triggerright", "axis", 6},
    }
}

 for guid, mappings in pairs(gamepads) do
    for i, mapping in pairs(mappings) do
        love.joystick.setGamepadMapping(guid, unpack(mapping))
    end
end

[slime73]

Original comment by Martin Felis (Bitbucket: MartinFelis, GitHub: MartinFelis).

---

How about the system button? Is it reporting something? Or does it do anything?

[slime73]

Original comment by Carlo Cabanilla (Bitbucket: clofresh, GitHub: clofresh).

---

The system button doesn't register as anything. This is how I discovered what everything does: https://gist.github.com/clofresh/8531100

[slime73]

Original comment by Jairo Luiz (Bitbucket: tangzero, GitHub: tangzero).

---

I think the pause button is mapped as a keyboard button named pause...

[slime73]

Original comment by Pavol Rusnak (Bitbucket: prusnak, GitHub: prusnak).

---

I used code from previous comment to create a config for Moga Pro controller.

gamepads = {
    -- moga pro controller
    ["42726f6164636f6d20426c7565746f6f"] = {
        {"a", "button", 6},
        {"b", "button", 7},
        {"x", "button", 9},
        {"y", "button", 10},
        {"leftshoulder", "button", 12},
        {"rightshoulder", "button", 13},
        {"start", "button", 18},
        {"leftx", "axis", 1},
        {"lefty", "axis", 2},
        {"rightx", "axis", 3},
        {"righty", "axis", 4},
        {"triggerright", "axis", 5},
        {"triggerleft", "axis", 6},
        {"dpx", "axis", 7},
        {"dpy", "axis", 8},

    }
}

Digital Pad behaves like two axis 7 & 8, with values -1,0,1. Not sure what the naming conventions for such setup are.

Feel free to adapt and push to SDL_gamecontrollerdb.h in SDL2.

[slime73]

Original comment by Alex Szpakowski (Bitbucket: slime73, GitHub: slime73).

---

For l1/r1 and l2/r2 did you mean to do leftshoulder/rightshoulder and triggerleft/triggerright ?

Feel free to adapt and push to SDL_gamecontrollerdb.h in SDL2.

Also keep in mind if doing this that you'll have to account for button/axis/hat indices starting from 1 in LÖVE's API, and from 0 in SDL's (i.e. you'll need to put button 5 instead of button 6 for the 'a' button in the string used by SDL.)

[slime73]

Original comment by Pavol Rusnak (Bitbucket: prusnak, GitHub: prusnak).

---

Alex: Right, you are correct. I fixed the snippet. What about digital pad naming? Is dpx/dpy OK? I could not find an example where digital pad is represented as two axis. All I found were four buttons.

[slime73]

Original comment by Jairo Luiz (Bitbucket: tangzero, GitHub: tangzero).

---

I just test. The OUYA menu button is mapped to a key named menu. You can access this key with love.keypressed and love.keyreleased callbacks.

[slime73]

Original comment by Landon “Karai” Manning (Bitbucket: karai17, GitHub: karai17).

---

Let's all take a moment to reflect on the thought that someone at OUYA deliberately made the menu button on the controller register as a global key press, instead of a joystick-specific button press. This was a design decision.

-- imnotsorry.lua
function love.keypressed(key, isrepeat)
	if key == "menu" then
		love.joystickpressed(1, key)
	end
end

function love.joystickpressed(joystick, button)
	if button == "menu" then
		-- OUYAAAAAAAAAAAAAAAAAAAAAAAAAAA
	end
end

[slime73]

Original comment by Carlo Cabanilla (Bitbucket: clofresh, GitHub: clofresh).

---

Oh weird, glad you figured that out. So that means we won't be able to tell which joystick actually pressed the menu button huh.

@karai17: that'll only work work in love 0.8 since 0.9 passes in joystick objects instead of joystick numbers. You'll probably wanna do

function love.keypressed(key, isrepeat)
    if key == "menu" then
        love.joystickpressed(love.joystick.getJoysticks()[1], key)
    end
end

[slime73]

Original comment by Landon “Karai” Manning (Bitbucket: karai17, GitHub: karai17).

---

Heh. The point that this is mind-boggling still stands. :)

[slime73]

Original comment by Pavol Rusnak (Bitbucket: prusnak, GitHub: prusnak).

---

JFYI I created bindings for Moga Pro in Steam Big Picture mode as mentioned in SDL_gamecontrollerdb.h:

https://bugzilla.libsdl.org/show_bug.cgi?id=2357

I think you want to do the same for Ouya controller.

[slime73]

Original comment by Martin Felis (Bitbucket: MartinFelis, GitHub: MartinFelis).

---

So I guess the Ouya controller then works (somewhat mind-boggling), right? At least it does not seem to be an issue of the port, rather of SDL or the controller itself.

Feel free to re-open if I did not read correctly or you do not agree.