Winit: Key repeat rate on Linux, Wayland, Sway is interpreted inversely

[e00E]

This is a copy of https://github.com/jwilm/alacritty/issues/2041 . I initially reported it there but was pointed to testing it with winit directly.

---

I am using Linux with Wayland with Sway.

I am setting the key repetition rate in my sway config. Per man 5 sway-input:

input \ Sets the key repeat rate in number of keypresses per second.

Alacritty is interpreting this value the wrong way.

When set to 1 I should observe one repetition per second and when set to 1000 I should observe 1000 repetitions per second. This works correctly in the browser I am typing this from (Firefox).

Alacritty seems to interpret the value as the number of milliseconds between repetitions giving me a lot of repetitions per second in the first case and 1 in the second case.

$ alacritty --version
alacritty 0.2.7

sway --version
sway version 1.0-beta.2-274-g5f45a4bb (Jan 25 2019, branch 'makepkg')

---

I can reproduce this problem with winit directly.

As explained in https://github.com/tomaka/winit/issues/776#issuecomment-457867696 I am using a https://github.com/tomaka/glutin example.

I am holding down the f key and from the console output I can see that the event appears with the wrong rate as described above.

WindowEvent { window_id: WindowId(Wayland(WindowId(94063627530576))), event: KeyboardInput { device_id: DeviceId(Wayland(DeviceId)), input: KeyboardInput { scancode: 33, state: Pressed, virtual_keycode: Some(F), modifiers: ModifiersState { shift: false, ctrl: false, alt: false, logo: false } } } }
WindowEvent { window_id: WindowId(Wayland(WindowId(94063627530576))), event: ReceivedCharacter('f') }
WindowEvent { window_id: WindowId(Wayland(WindowId(94063627530576))), event: KeyboardInput { device_id: DeviceId(Wayland(DeviceId)), input: KeyboardInput { scancode: 33, state: Pressed, virtual_keycode: Some(F), modifiers: ModifiersState { shift: false, ctrl: false, alt: false, logo: false } } } }
WindowEvent { window_id: WindowId(Wayland(WindowId(94063627530576))), event: ReceivedCharacter('f') }
Awakened
WindowEvent { window_id: WindowId(Wayland(WindowId(94063627530576))), event: KeyboardInput { device_id: DeviceId(Wayland(DeviceId)), input: KeyboardInput { scancode: 33, state: Pressed, virtual_keycode: Some(F), modifiers: ModifiersState { shift: false, ctrl: false, alt: false, logo: false } } } }
WindowEvent { window_id: WindowId(Wayland(WindowId(94063627530576))), event: ReceivedCharacter('f') }
Awakened
...

[e00E]

Found the problem in client-toolkit. Working on PR.

[trimental]

Yeah this is my bad, I read the wayland documentation wrong since the key repeat delay is in milliseconds. This should be fixed by https://github.com/Smithay/client-toolkit/pull/62

[trimental]

At the exact same time, what are the chances ahah.

[e00E]

Oh, you have already done it, great!

[e00E]

Would it be possible to update the client toolkit dependency from 0.4 to the current 0.6 which includes this bugfix?

[Osspial]

cc @vberger

[vberger]

This update is contained in the eventloop-2.0 refactor. Backporting the update to master branch would be a significant amount of work (it implies an upgrade of wayland-client, which was very largely breaking) that I am not willing to do a second time.

However, I released SCTK 0.4.5 containing this fix, cargo update should bring it to you.