The day is finally here. Xwayland now has the code for Explicit GPU Synchronization merged after waiting over a year for everything to be ready.
This merge resquest has finally been merged into XWayland, which means NVIDIA users are just one beta driver release away (May 15th) from probably having a good experience on Wayland now!
If Nvidia had worked together on Xorg, we wouldn’t have had all of those shitty Nvidia unique x11 bugs with Nvidia’s crappy proprietary drivers in the past.
Cannot confirm any Nvidia unique X11 bugs since everything works perfectly fine since a decade or so for me – but tons of Nvidia unique Wayland bugs (same driver version and same card). I am not quite sure if Nvidia are really the bad one in this specific case.
My whole invironment is full of stuff to make Wayland actually work with Nvidia.
You miss the entire point of WMs then.
You’re complaining about 7 lines of configs when the entire point of WMs is for you to literally customize everything through configs.
Shit, you probably don’t even need half of that crap. QT_QPA_PLATFORM=wayland is already handled by qt, you only need that when running apps across Waypipe from a headless server environment. By default QT infers your environment from $DISPLY which is always :0 in Wayland.
I don’t see any graphical glitches on my hardware.
Perhaps you’re still on the qt5 version?
Maybe a bug with your Nvidia driver?
Perhaps your WM doesn’t have explicit sync yet and your card & driver needs it in this particular case? Perhaps it’s running under XWayland?
Wayland#Qt : While it shouldn’t be necessary, to explicitly run a Qt application with the Wayland plugin, use -platform wayland or QT_QPA_PLATFORM=wayland environment variable.
As it says above, it should really be inferred by default. Something weird is going on here.
If Nvidia had worked together on Xorg, we wouldn’t have had all of those shitty Nvidia unique x11 bugs with Nvidia’s crappy proprietary drivers in the past.
Cannot confirm any Nvidia unique X11 bugs since everything works perfectly fine since a decade or so for me – but tons of Nvidia unique Wayland bugs (same driver version and same card). I am not quite sure if Nvidia are really the bad one in this specific case.
My whole
invironmentis full of stuff to make Wayland actually work with Nvidia.Not one single line of configuration is used for X11.
Here’s just a tiny fraction of the Nvidia specific bugs on X11 :
https://github.com/NVIDIA/open-gpu-kernel-modules/issues/380
https://bbs.archlinux.org/viewtopic.php?id=283461
https://forums.developer.nvidia.com/t/16-bit-overlays-broken-with-x11-and-4xx-series-driver/128067
https://bugs.kde.org/show_bug.cgi?id=460341
You’re using a WM bud, wtf do you expect?
I expect it to work, like Openbox works on X11.
You miss the entire point of WMs then.
You’re complaining about 7 lines of configs when the entire point of WMs is for you to literally customize everything through configs.
Shit, you probably don’t even need half of that crap.
QT_QPA_PLATFORM=waylandis already handled by qt, you only need that when running apps across Waypipe from a headless server environment. By default QT infers your environment from$DISPLYwhich is always:0in Wayland.Thank you for educating me on basic knowledge about WMs, which I use since ove a decade.
I just expect applications behave the same. I have literally not one single line of configuration for X11 in any of my configuration files.
I need this QT variable, otherwise Cura does not work properly.
Cura as in this one? :
Exactly. When the variable is unset the UI shows massive artifacts and weird rendering glitches.
I don’t see any graphical glitches on my hardware.
Perhaps you’re still on the qt5 version?
Maybe a bug with your Nvidia driver?
Perhaps your WM doesn’t have explicit sync yet and your card & driver needs it in this particular case? Perhaps it’s running under XWayland?
As it says above, it should really be inferred by default. Something weird is going on here.