hl.dsp.focus({ workspace = ... }) won't focus an empty workspace on a different output

[kolayne]

Hyprland version

v0.56.1 at commit 5c9377c clean (version: bump to 0.56.1).

Is this a regression?

Don't know

What's the bug?

When attempting to focus an empty workspace on a different output, the focus leaves and immediately returns to the previously focused workspace. Seems to only have effect with cursor:no_warps = true

Reproduction steps

1. Run hyprctl eval 'hl.config({ cursor = { no_warps = true }, input = { follow_mouse = 1 } })
2. On output A open workspace 1 and open a terminal
3. On output B open workspace 10 and do not open any windows
4. In the terminal run hyprctl dispatch 'hl.dsp.focus({ workspace = 10 })'

Expected behavior: workspace 10 is focused and remains focused.

Actual behavior: workspace 10 is focused for a short moment and then focus removes to workspace 1.

Attachments

No response

I acknowledge that:

• I have searched all (open and closed) discussions and issues and my problem is not there.

• I have read the Code of Conduct and agree to follow it.

• I wrote this request with my own hands, my own keyboard, and in my own words.

[kolayne]

The reason seems to be as follows. When doing hl.dsp.focus({ workspace = ...}), the control gets to Actions::changeWorkspace, which then calls CInputManager::simulateMouseMovement:

Hyprland/src/config/shared/actions/ConfigActions.cpp

Lines 998 to 1003 in 60695e1

	if (!g_pInputManager->m_lastFocusOnLS) {
	if (Desktop::focusState()->surface())
	g_pInputManager->sendMotionEventsToFocused();
	else
	g_pInputManager->simulateMouseMovement();
	}

It calls CInputManager::mouseMoveUnified:

Hyprland/src/managers/input/InputManager.cpp

Lines 187 to 190 in 60695e1

	void CInputManager::simulateMouseMovement() {
	m_lastCursorPosFloored = m_lastCursorPosFloored - Vector2D(1, 1); // hack: force the mouseMoveUnified to report without making this a refocus.
	mouseMoveUnified(Time::millis(Time::steadyNow()));
	}

It detects that the focused monitor is different from the monitor under cursor and forces focus back on the old monitor (wrong):

Hyprland/src/managers/input/InputManager.cpp

Lines 348 to 349 in 60695e1

	if (PMONITOR != Desktop::focusState()->monitor() && (*PMOUSEFOCUSMON || refocus) && m_forcedFocus.expired())
	Desktop::focusState()->rawMonitorFocus(PMONITOR);

But I'm not sure where to fix this. It seems like the whole call chain is some nasty work around something: CInputManager::simulateMouseMovement is trying to exploit and repurpose CInputManager::mouseMoveUnified, moreover, I don't even get why it is called in the first place.

@vaxerski, your suggestions are welcome.

[kolayne]

Also, should it happen at all? I just noticed this is an unconditional thing, and it's the case even when input:follow_mouse = 0. Imagine I have a workspace 1 with window A and workspace 2 with window B. Leaving my cursor over A, with my keyboard I switch to workspace 2 and work with window B. If after this I move my cursor, the workspace 1 will get focused. But the window focus remains at B! Which means that now my active window is outside of my active workspace!

That's not right, is it? What is the expected behavior?

Upd: this bit is fixed in #16060. The question from the comment above remains.

[vaxerski]

the reason it's there is so that when you focus e.g. ws 2 and it has a window, we send enter events

[kolayne]

I believe, this case is already covered by adjacent code. Could you please describe the reproduction steps that I can verify this with or add a test for this behavior?

[vaxerski]

open window on workspace 1
move to workspace 2 (new workspace, empty) on the same monitor
do not move the mouse
scroll / keyboard / etc events should not reach the old window regardless of follow_mouse settings