SDL_GetGlobalMouseState reports weird coordinates (Linux, Xwayland)

[mgerhardy]

I hope this isn't me just not understanding some of the concepts - but I received weird values for the global mouse state

Screen 0: minimum 16 x 16, current 5120 x 1440, maximum 32767 x 32767
XWAYLAND27 connected 2560x1440+0+0 (0x24) normal (normal left inverted right x axis y axis) 600mm x 340mm
	Identifier: 0x6d3
	Timestamp:  38592
	Subpixel:   unknown
	Gamma:      1.0:1.0:1.0
	Brightness: 0.0
	Clones:    
	CRTC:       1
	CRTCs:      1
	Transform:  1.000000 0.000000 0.000000
	            0.000000 1.000000 0.000000
	            0.000000 0.000000 1.000000
	           filter: 
	non-desktop: 0 
		supported: 0, 1
  2560x1440 (0x24) 312.000MHz -HSync +VSync *current +preferred
        h: width  2560 start 2752 end 3024 total 3488 skew    0 clock  89.45KHz
        v: height 1440 start 1443 end 1448 total 1493           clock  59.91Hz
XWAYLAND28 connected primary 2560x1440+2560+0 (0x24) normal (normal left inverted right x axis y axis) 600mm x 340mm
	Identifier: 0x6d5
	Timestamp:  38592
	Subpixel:   unknown
	Gamma:      1.0:1.0:1.0
	Brightness: 0.0
	Clones:    
	CRTC:       0
	CRTCs:      0
	Transform:  1.000000 0.000000 0.000000
	            0.000000 1.000000 0.000000
	            0.000000 0.000000 1.000000
	           filter: 
	non-desktop: 0 
		supported: 0, 1
  2560x1440 (0x24) 312.000MHz -HSync +VSync *current +preferred
        h: width  2560 start 2752 end 3024 total 3488 skew    0 clock  89.45KHz
        v: height 1440 start 1443 end 1448 total 1493           clock  59.91Hz

int windowX = 0, windowY = 0;
SDL_GetWindowPosition(mouseWindow, &windowX, &windowY);
io.MousePos = ImVec2((float)(mouseXGlobal - windowX), (float)(mouseYGlobal - windowY));

the window (0:0) the mouse is hovering is on SDL display 0 (0:0:2560:1440), but the global mouse position is at 3646:769 - this looks like a bug in SDL for me. I will report it there. Weird thing is that the SDL display 1 has a x position of 2560 but is the left display. Not sure how to convert the mouse coordinates properly.

When I create the window on the other display, I also get invalid(?) values:
display 1 is reported as x-offset 2560 - the global mouse coordinate was e.g. 956 while hovering the window at 2560:0 (fullscreen)

---

the imgui issue for reference: ocornut/imgui#2117 (comment)

[flibitijibibo]

Both GetWindowPosition and global mouse state are unsupported in native Wayland, and should return -1 indicating this. This isn't the end of the world though, if I remember correctly their code already has a path to use the non-global GetMouseState while tracking the mouse enter/leave events.

Xwayland may be a different story but I would check native Wayland first, since that will be the priority in the near future anyhow.

[DanielGibson]

and should return -1 indicating this

SDL_GetGlobalMouseState() returns an Uint32 so it can't really return -1, and if it returns 0xffffffff in case of error that's not documented.
It would indeed be nice to have some way to tell if the call succeeded or not

[flibitijibibo]

Oh, right, I was thinking of Set rather than Get. I think Get falls back to the window-relative version, so you can only detect it with something else like GetWindowPosition.

[DanielGibson]

In dhewm3 I use SDL_GetGlobalMouseState() before creating the window, to create the Window on the display the cursor is currently on (with SDL_WINDOWPOS_UNDEFINED_DISPLAY(displayIndex)in SDL_CreateWindow()).
Seems to work on X11 and Windows, unsure about Mac, most probably not on Wayland..

[flibitijibibo]

On the native Wayland side I've been doing a bunch of work that makes the coordinates relatively sane for single-display setups - past that, global queries still can't be used (and multi-window is still a stretch if you try to write your own focus detection for some reason).

UE5 has been my test for the recent commits and I have to say, having a feature query for global window/mouse position does seem like a really good idea since the alternative is basically to hardcode checks for SDL_GetCurrentVideoDriver, which is a terrible idea and only a lazy person would do that.

[Kontrabant]

Is this still an issue? XWayland and the SDL Wayland backend have had a lot of work done on them over the past two years, I can't replicate the original issue on the current version XWayland, and the SDL Wayland backend manages to fake the global mouse coordinates well enough for the 99.9% of applications that just query them to manually transform the global pointer coordinates to relative (like the UE5 editor).

[slouken]

I think this is fixed, please let us know if that's not the case.