I’m sure this won’t make anybody mad, and we can all have a reasonable discussion about this. A change proposal for Fedora suggests Fedora should drop the X11 GNOME session from Fedora 43, moving GNOME’s target of removing X11 in GNOME 50 to the Fedora release carrying GNOME 49. Fedora 43 will be released in roughly six months.
The reasoning behind the proposal should be no surprise.
This Change effectively implements the GNOME 50 target in GNOME 49 because there is no one to support any issues with GNOME X11 upstream. The X11 session is already quite buggy, with serious unfixed issues in Mutter (like rhbz#2179566 and glgo#GNOME/mutter#3868) and recently Phoronix could not benchmark GNOME X11 vs Wayland because “GNOME on X11 wasn’t even working due to bugs”.
Dropping the GNOME X11 session now allows us to reconcile with reality that the GNOME X11 session is simply not supported anymore (either by us or upstream GNOME).
↫ Fedora change proposal
If you’re still using legacy X11 on Fedora GNOME for some reason, upgrading to Fedora 43 would mean being moved to Wayland. If you would want to continue using X11, you’d need to install a different login manager than GDM, and a desktop environment that still supports X11 (like Cinnamon or Xfce). Since all of this is the plan anyway, what this proposal effectively does is move the removal of X11 from Fedora GNOME ahead by six months. Do note that this is only a change proposal for now, and it will have to be voted on and agreed upon before it becomes official policy.
This is just yet another nail in the coffin of X11 on Linux, as more and more distributions and desktop environments move to eliminate it entirely from their installations and stacks in favour of Wayland. Big concerns here for some X11 users are definitely accessibility, where tools are still relatively young, and to a lesser degree NVIDIA users, whose GPU drivers are a bit of a mess when it comes to Wayland. To this day, NVIDIA with Wayland can be very hit or miss.
Regardless, we all know which way the wind’s blowing.
KDE Plasma also supports x11, i don’t recomend it tho, because of lack of testing
Not for long.
Other than you Thom, I can’t think of much of anyone who uses Fedora, and certainly not Fedora Gnome. Seems to be nearly all Debian-based, Ubuntu-based, and Arch-based distros in use these days. This probably doesn’t matter that much, other than it continues to speak to Fedora’s lack of package maintainers.
I do. I’ve been Fedora user and maintainer for decades. And I’ve been using GNOME with Wayland since 2013. Personally, I have not run an X server for years.
Well, I believe that Linus Torvalds remains a long-time Fedora user for one. As Fedora has about the best Apple Silicon support, I do not imagine that this has changed.
It would be interesting to get a somewhat reliable break-down of real desktop Linux use by distro though. I would have guessed that Fedora had gained on Ubuntu. There are certainly quite a few Fedora spins-offs that “feel” popular, especially on the immutable side of things.
It would be interesting to see where Arch-based distros are in the mix these days as, again, it “feels” like they have really gone mainstream.
I’ve been using Fedora with KDE Plasma as my daily driver on all of my devices since Fedora 18. While I wouldn’t say I’m 100% happy with the transition to Wayland, it hasn’t been a disaster either. If your job happens to involve a lot of enterprise or scientific GNU/Linux deployments, you’ll find a lot more Red Hat than Debian or Arch families, and running Fedora lets you stay ahead of the changes coming in the next release of RHEL & Friends.
I also use Fedora, specifically the GNOME DE. Been using since Fedora 29 back in 2018. Before that, I’ve toyed with Yellow Dog Linux (PPC, when they still released the OS for Macs), then moved onto Red Hat before it turned into Enterprise Linux. Kept an eye on the Fedora project since the Core OS days but never made the switch until I finally got fed up with Apple and just jumped in and installed Fedora 29 on my Intel MacBook Pro and never looked back. It may not be for everyone, and that’s ok.
Yet another example of something opensource just simply couldn’t do. MacOS X changed the full GUI stack, from bottom to top… in 5 years support for old GUI was dropped. Android did similar transition in Android 3/4 (Android 3 was tabled-only if anyone remembers, phones went from Android 2.x to Android 4.x). In about 5 years old approach is history.
With X11/Wayland… saga goes for 15 years with no end in sight simply because there are no one who may declare that shift is needed and it’s happening, whether you like it or not.
The only company that is having almost as much trouble forcing the switch is, ironically enough, Microsoft… who, somehow, managed to be **both** indecisive and too audacious, simultaneously: Windows 8 attempt to bring radically new way of doing GUI was quickly rolled back while Windows Phone introduced not one, but three incompatible ways of writing apps: one with Windows Phone 7, then with Windows Phone 8, and with Windows Phone 10 another one, again…
I hear where you’re coming from but it is a bit ungenerous. The transition to Wayland really began in about 2020, when toolkit developers began supporting it in e.g. GTK 3.20 — not when the first lines of code were written back in 2012 or whenever. Back in 2012 it was not at all clear that a transition to Wayland was in the cards; Xorg was still under heavy development and there were competing technologies like Mir and Y. If you wanted to go that route you might as well say that Apple began the transition to a new display server when NeXT wrote OpenStep.
And in the meantime, GNU/Linux has recently managed at least one other orderly top-to-bottom reworking of the display stack inside a five-year timeframe: the transition from XFree86 to Xorg back in the early 2000s. You could say that those are both different implementations of the same protocol, but really Xorg made some huge architectural changes that allowed e.g. hardware-accelerated compositing, and drove coordinate changes to window managers, etc.
No, “transition to a new display server” haven’t started when NeXT wrote OpenStep. You couldn’t run OpenStep programs on MacOS, after all. And I’m not talking about development: proprietary world is full of projects that take years to develop and then lead nowhere.
I’m talking precisely about switch to a new API. I’m talking about development of a platform. That’s never a bonanza of choices: if have “old way of doing things in version X, then new way of doing things in version Y”. Even when there are apparent choice (like that five-years period when both Windows 9X and Windows NT was available) – these are provided by two distinct teams, that don’t try to support bazillion things in parallel (even if both work for the same company).
But with opensource any such switch turns into tragedy where both old choice and new choice is supported by the same underfunded and overwhelmed team with people who argue the most against switch being the ones least likely to invest any resources in support of what they need or want.
Opensource is all to ready to say “yes”, but it’s basically couldn’t even say “no”.
The fact that proprietary company may say “no” doesn’t guarantee success, of course, that’s just the necessary ingredient, not a sufficient one… but still – it’s very sad to see how long that switch to Wayland took.
zde,
IMHO the wayland project shot itself in the foot by breaking things and xwayland not really being feature complete as a compatibility layer. Honest to god my complaints about wayland had nothing at all to do with resistance to change, just everyday practical issues. Most of these problems could have been solved a decade ago if only project leaders would have listened to users up front rather than be dismissive of our needs. I love FOSS and I want it to keep improving, including wayland. But sometimes the way we go about it is asinine. We have lots of opportunities to learn from past project management mistakes, yet we still repeat them and it gives us a black eye. Oh well, I hope we can do better next time…
My desktop will likely end up running wayland once the new version of KDE reaches deb stable.
@Brainworm
I agree with you. In the proprietary world, we would not have heard about Wayland for the first decade. And when it shipped, we would have all been forced to use it ready or not. Remember Vista? And a commercial forced march would almost certainly have come with hardware obsolescence. Is all that “better”?
The ancient Macbook that I am typing on will not run Metal even though it is Apple hardware. Yet I am happily running the latest KDE on Wayland and Mesa 25 with Vulkan 1.4.311 and OpenGL 4.5 support. This is on an NVIDIA GeForce 9400M. This is what Open Source gets right.
The reason that we are seeing this story is because the Wayland transition has already happened. Both KDE and GNOME already default to Wayland. The most popular desktop distros have been shipping Wayland by default and mainstream users have not been complaining. Even ultra-slow distros like RHEL defaulted to Wayland years ago. The majority of Linux desktop users already use Wayland. With great respect for @Alfman, Debian Stable does not represent the state of desktop Linux today.
What we think about that here does not change the reality that it is true. They want to remove Xorg because only niche users want it and they do not want that to burden development overall. It no longer matters that the transition could have been better managed or that the stewards of early Wayland were unresponsive. It is done now and, at the end of the day, the transition has been quite successful.
We all understand of course that X11, the protocol, is still with us even though Xorg, the display server, is being left behind.
I wasn’t claiming to be representative. However I’m not sure redhat is that representative either. I don’t have user numbers, but going by branching activity debian & ubuntu not only seem to be very popular, but may actually be the majority.
https://distrowatch.com/dwres.php?resource=family-tree
As for wayland specifically, I found this stat from firefox 2022 telemetry data suggesting 10% of users running wayland. I’m curious if someone can find more recent information…
https://www.phoronix.com/news/Firefox-Wayland-X11-Stats
I don’t think it really matters to the debate at all though. Typical users just want things to work. Wayland devs have been focusing on Gnome, so it works better there than on other desktops. The next version of KDE is said to fix many issues, if true then wayland may be a go for me. My kid’s computer is running mint & cinnemon, which is still using X for the foreseeable future. Same with other computers where I installed xfce. Wayland would be fine, but I keep getting the impression that gnome users in particular don’t seem to realize that wayland just isn’t as mature elsewhere.
I love many legacy Window Managers and my preferred desktop is XFCE (still X11) so I am not representative either. I am talking about reality, not preference. That said, I have moved many machines to Chimera Linux now and those all run KDE on Wayland (Plasma 6.3.4) which works well.
Firefox on Wayland may have been 10% in 2022. Firefox began defaulting to Wayland at the end of 2023. Either you or your distro would have to enable it before that (even if you were running a Wayland desktop).
I really wish I could find good numbers. The best I can muster is the February 2025 survey from Gaming on Linux which shows Wayland at 45% share increasing at about 3% per month:
https://www.gamingonlinux.com/users/statistics/statid=106/
I expect gamers to reflect a lower percentage than average Wayland use, especially considering that 40% of them use NVIDIA. According to Firefox, overall NVIDIA use on Linux is closer to 13% with 66% using Intel:
https://data.firefox.com/dashboard/hardware
Normal Linux desktop uses are going to use the distro default. For any distro shipping current GNOME or KDE, that is Wayland. So Wayland is the default out-of-the-box on Ubuntu, Fedora, Manjaro, EndeavourOS, CacheyOS, and SUSE. Just this list must be two-thirds or more of Linux desktop users. Ubuntu alone is something like one-third.
Marketshare-wise, the two big outliers will be Linux Mint and maybe MX Linux (if Distrowatch is to be believed). If Cinnamon, XFCE, MATE, and LXQt do not all default to Wayland this year, they certainly will in 2026. The new COSMIC is Wayland only of course. That takes us to 90% of the Linux desktop.
My point about RHEL is not that it is representative but rather the opposite–that it is a laggard. If even RHEL is defaulting to Wayland and removing Xorg, you know the market has moved. Even Debian defaults to Wayland for GNOME and Debian will likely default to Wayland for KDE in Trixie.
I really wish I could find reliable numbers but I think it is obvious that more than half of Linux desktop users today are Wayland users. It is already the majority. It will be 75% by the end of the year and 90% or more by the end of 2026. By the end of 2027, using Xorg on Linux will be as common as using FVWM is now. Wayland will likely be the default on FreeBSD by then.
In terms of Wayland “maturity”, my experience has been that what matters is how well your environment is supported by xdg-desktop-portal. Today, the best experience is KDE with GNOME a close second. Everything else is behind.. Even Wayland native options like Hyprland and Sway still have a few gaps. XFCE is not there yet.
EDIT: I just found a video from The Linux Experiment YouTube channel that says that his audience was 66% Wayland by mid-2024:
https://www.youtube.com/watch?v=8V8uQbIFlh0
Ah, Wayland. The solution in search of a problem.