I guess it's time? [LWN.net]

I guess it's time?

Posted Jun 10, 2025 20:14 UTC (Tue) by yodermk (guest, #3803)
Parent article: Ubuntu 25.10 to drop support for GNOME on Xorg

In 2018 I got a fairly high-end Dell Precision with nVidia Quadro P3200 graphics. Only in the last few months (using continually updated Arch Linux) have I been able to reliably run it on Wayland. Prior, I could try it, but would run into unbearable quirks that forced me back to X11 within a day or two. Now it seems (finally) pretty solid in Wayland, with an occasional quirk that may or may not be Wayland related, but not a big enough deal to get me to go back. So with that, and hopefully others with similar experiences, I guess it's time to finally excise X11. It really does need to go.

to post comments

I guess it's time?

Posted Jun 11, 2025 14:31 UTC (Wed) by cortana (subscriber, #24596) [Link] (3 responses)

After I've been logged into (Wayland) GNOME for a few days, my mouse pointer is noticeably jerky as it moves across the screen. Logging out and in again fixes it, for a time. This doesn't happen with Xorg (which has other problems, I really wish I could switch to Wayland permanently, but it's just not there yet). This is with an RTX 3070 Ti and a GTX 1080.

I guess it's time?

Posted Jun 11, 2025 18:10 UTC (Wed) by mathstuf (subscriber, #69389) [Link]

Wild idea: floating point precision loss in some accumulator? The test would be: is it actually time-correlated, or is it actually correlated with total "distance" traveled by the cursor over that time? Is it compositor-specific?

I guess it's time?

Posted Jun 13, 2025 13:08 UTC (Fri) by mgedmin (guest, #34497) [Link] (1 responses)

Xorg had a special server thread to watch for mouse events and update the hardware cursor position to make mouse movements feel smooth (at some cost of accuracy: the cursor shape could lag a little bit after moving to a new location if the system was heavily loaded).

I don't believe any of the Wayland compositors have anything like that implemented, although I recall hearing about plans of doing something similar at some point.

I guess it's time?

Posted Jun 20, 2025 13:36 UTC (Fri) by daenzer (subscriber, #7050) [Link]

Mutter has an input thread. It currently can’t update the cursor position as directly as Xorg‘s input thread though, largely due to limitations of the current atomic KMS UAPI. Mutter‘s cursor movement is quite snappy anyway though, the difference in terms of input –> output latency is at most a few milliseconds.

AFAIK the situation is similar with other Wayland compositors.

None of this has anything to do with the Wayland or X protocol, how cursor movement is handled is purely between the display server and the kernel. (As a side note, Xorg‘s asynchronous mouse cursor movement was first introduced as „Silken Mouse“ in XFree86 4.0, when X was of similar age as Wayland is now)