I don't necessarily like having to do these extra calls whenever threads are
spawned (and what happens to signals during the time signal delivery is
disabled?) but it seems to fully fix os.execute.
NOTE: Also disables signals when loading openal, as it or one of its backends
spawns threads internally, and does not disable signals itself.
--HG--
branch : minor
Sadly, it's not a complete fix for #1042, as there seems to be no good way to
deal with system() in multithreaded applications. At least os.execute can
return values other than -1 now.
- Added love.system module, including clipboard functions.
- Added custom hardware cursors (love.mouse.newCursor, love.mouse.setCursor.)
- Revamped love.joystick:
-- added Joystick objects and moved love.joystick functions which operated on individual joysticks to methods on the Joystick objects.
-- Added love.joystick.getJoystick and love.joystick.getJoystickCount.
-- Added events for when joysticks are connected and disconnected.
-- Added 'Gamepad' methods to Joystick objects: Joystick:isGamepad, Joystick:getGamepadAxis, and Joystick:isGamepadDown. They're a common abstraction for xbox controller-like joystick across operating systems.
- Added monitor choosing support and "fullscreen desktop" mode to love.window.
- Moved unicode text input events to their own callback function (love.textinput), removed the key repeat functions from love.keyboard and made the second argument of love.keypressed a boolean saying whether the keypress was a repeat.
- liblove now works with love.window and love.graphics in OS X
Alex Szpakowski
Tiago Koji Castro Shibata
Tiago Koji Castro Shibata
Bart van Strien
T-Bone
Martin Felis