I've painstakingly analyzed the existing web of calls to set the
force-feedback callback, and replaced it with a chain of functions to
set a controller pointer in ForceFeedbackHandler. EvdevController
implements functions to do force-feedback using the kernel drivers for
the input device.

rhn authored Sep 23, 2014 and Cebtenzzre committed Aug 04, 2022

Accidentally also added a message queue in evdev.

Yakov Markovitch authored Nov 14, 2017 and Cebtenzzre committed Aug 04, 2022

This receiver is not Microsoft original, though "mimics" MS USB Vendor ID;
still perfectly compatible.

David Seifert authored Dec 16, 2016 and Cebtenzzre committed Aug 04, 2022

See https://bugs.gentoo.org/594674.

Carles Pastor Badosa authored Mar 19, 2017 and Cebtenzzre committed Aug 04, 2022

btn2string returns "DPAD_UP" when passed XBOX_DPAD_UP and gives similar
responses for the other directions (DPAD_LEFT, DPAD_RIGHT and DPAD_DOWN).
However, string2btn didn't accept those as input, rendering the output of
--help-button confusing.

James Le Cuirot authored Jul 14, 2016 and Cebtenzzre committed Aug 04, 2022

There are similar Thrustmaster products in existence but I don't know
whether they have exactly the same USB identifier or protocol.

Although all the buttons, including the triggers, have a very digital
feel, almost all of them also provide analog values. Sadly the Xbox
360 controller only has analog triggers so this is mostly wasted.

Lars Kellogg-Stedman authored Sep 23, 2016 and Cebtenzzre committed Aug 04, 2022

This corrects OPTION_CALIBRARIOTION -> OPTION_CALIBRATION (which I
only noticed because I was looking for the implementation of the
calibration option and was having trouble finding it because of the
spelling).

James Le Cuirot authored Dec 28, 2016 and Cebtenzzre committed Aug 04, 2022

This was triggering an exception, causing the daemon to exit
prematurely.

Lucas Clemente Vella authored Feb 21, 2016 and Cebtenzzre committed Aug 04, 2022

After a USB transfer error, either retry or properly disconnect if the
device is no longer found.

till busch authored Nov 02, 2016 and Cebtenzzre committed Aug 04, 2022

use `libusb_handle_events_timeout_completed()` instead of `libusb_handle_events()`
should fix #144

James Le Cuirot authored Dec 27, 2016 and Cebtenzzre committed Aug 04, 2022

Now that the destructor crash is fixed, it hangs instead. The crash
has probably always masked this issue. I don't understand libusb much
as it isn't well documented but it seems that by the time we check
that all transfers have been cancelled, new ones have already been
submitted. Setting m_is_disconnected to true beforehand to indicate we
are destructing appears to remedy this.

James Le Cuirot authored Dec 27, 2016 and Cebtenzzre committed Aug 04, 2022

parse() is indirectly called by the destructor, which is not allowed
when it is pure virtual, causing a "pure virtual function called"
crash.

We don't care about parsing when the controller instance is about to
be destroyed so define a dummy method in the base case.

Fixes #252

Fixing typo sensitivty -> sensitivity