I mainly tried GNOME on Wayland, but 8+ years my 1000Hz mouse would move inconsistently with high CPU/GPU load. I've seen
gamers praising Wayland before GNOME 42 where this was particularly bad, which has me questioning the whole thing. Fedora's very interested in getting rid of it (silently botching it F41 by not having dependencies right, straight-up disabling it right before F42 released as an undocumented changed and had to revert it); they've claimed X a maintenance burden. Yet they're providing a desktop OS to other people.
All the other Wayland-specific issues aside (that are still there today; Fedora
had to revert that X disable), the mouse cursor thing is a major issue for me. I do 6/11 no-accel UT99 and osu! and
know mouse accuracy
Xorg's been perfect with it that I've never seen it be an issue since 2016. Windows has been fine with it too forever even 11. High CPU/GPU
should not affect mouse cursor movement! And yet,
the most mainstream DE is still guessing at it 2025 as if it's an afterthought (and still doesn't sound like it solves the main issue; priorities do whatever they want when there's a lot of em, so the cursor
might be allowed to move with 7.1 HRTF 200 fps destructive physics going on in a game depending on what's more important to an algorithm, per-frame; fun prediction
)
I'm not sure what GNOME's deal is, but I have a feeling Wayland doing libinput -> evdev is slower than using evdev on Xorg anywhere. I don't like the idea of having slower input because of theoretical security benefits (I hear keylogging between apps on X used a lot, but maybe don't run malicious apps to begin with? I trust what I run
). With not being able to disable compositors certain DEs and forced Vsync/reduced tearing emphasis I'm also not sure if Wayland sessions can beat properly-tuned Xorg set-ups.
I had to figure out
MESA_LOADER_DRIVER_OVERRIDE=zink MESA_VK_WSI_PRESENT_MODE=immediate for some OpenGL games because they'd run at 40 FPS on Wayland for some reason, but translating to Vulkan and disabling its Vsync has them run at high FPS fine; meanwhile I can run them native OpenGL no problem on Xorg.
And although this might be a bit harsh for free software and volunteers (even though GNOME/RH/IBM/Canonical/SUSE pull in big bucks and fund), I feel there's something odd with taking an easier development path at the expense of worse performance from a coding standpoint; but this is mainly aimed at projects
intentionally disabling Xorg/X11 with consumer software (GNOME with GTK, particuarly Fedora, any Linux distro primarying GNOME by association). Like, can't everyone just wait until Wayland is good before trying to force it onto others? (it'll be good when it's not noticed any-app vs Xorg)
I'm still pretty sure Wayland should be marked experimental on desktop today if you're offering a mainstream experience to other people.
), accelerated video playing is perfect, suspend/resume works reliably 100% of the time and so on.
github.com
