After more than forty years, everyone knows that it’s time to retire the X Window System – X11 for short – on account of it being old and decrepit. Or at least that’s what the common narrative is, because if you dig into the chatter surrounding the ongoing transition there are some real issues that people have with the 16-year old spring chicken – called Wayland – that’s supposed to replace it.
Recently [Brodie Robertson] did some polling and soliciting commentary from the community, breaking down the results from over 1,150 comments to the YouTube community post alone.
The issues range from the expected, such as applications that haven’t been ported yet from X11 to Wayland, to compatibility issues – such as failing drag and drop – when running X11 and Wayland applications side by side. Things get worse when support for older hardware, like GeForce GT610 and GT710 GPUs, and increased resource usage by Wayland are considered.
From there it continues with the lack of global hotkeys in Wayland, graphics tablet support issues, OBS not supporting embedded browser windows, Japanese and other foreign as well as onscreen keyboard support issues that are somehow worse than on X11, no support for overscanning monitors or multiple mouse cursors, no multi-monitor fullscreen option, regressions with accessibility, inability of applications to set their (previously saved) window position, no real automation alternative for xdotool, lacking BSD support and worse input latency with gaming.
Some users also simply say that they do not care about Wayland either way as it offers no new features they want. Finally [Brodie] raises the issue of the Wayland developers not simply following standards set by the Windows and MacOS desktops, something which among other issues has been a point of hotly debated contention for years.
Even if Wayland does end up succeeding X11, the one point that many people seem to agree on is that just because X11 is pretty terrible right now, this doesn’t automatically make Wayland the better option. Maybe in hindsight Mir was the better choice we had before it pivoted to Wayland.
There’s definitely regressions that need to be fixed, but the way it is presented here is just misinformation, mixing things like project-specific bugs and misunderstandings in as Wayland problems.
*BSD is officially supported by Wayland and by several display servers (a better state than for X11 where the *BSD’s had to patch things quite extensively downstream), the graphics tablet thing is a KDE-specific bug, and global hotkeys is available in some display servers through XDG portals (albeit a bit slowly), and using multiple independent mouse cursors is very specifically a Wayland feature (wayland multi-seat). Restoring window state is also supported, it just works differently than X11, and sway at least supports global fullscreen the same way as i3.
Most Wayland protocol discussions also take extensive note of Windows, macOS, Android, iOS, and even smartTV UX behaviors and upcoming web standards to see how other platforms solved a particular issue, what restrictions they apply, and what apps are likely to both expect and need. This does not mean that one can simply copy these verbatim – some of these rely on signed entitlements handed out by a centralized entity for special behaviors, which I’m sure no one is interested in.
Nothing wrong with seeing issues with Wayland or even disagreeing with it entirely, but let’s not misrepresent the project and its issues when we discuss it.
The important missing pieces will get filled in over time. Some of the people who were holding up the works when wayland was small will start to be outvoted. Some things will never be implemented, and the world will adapt after the screaming dies down.
Eventually people will wonder what all the fuss was about.
It was first released in 2008. This isn’t something new, and it still falls short in a lot of basic ways.
And X11 took how many decades to gain all these features and stood entirely still since Wayland first existed as a tech demo too? First releases are rarely going to instantly become the industry standard when something else already exists – heck if it was that easy to replace existing industry standards with some of M$’ crap ‘upgrades’ over the last few years….
Pelrun is quite correct Wayland is this generations systemd really, lots of noise around it, but in the end it works, solves many problems, is a better framework for the modern user/system’s needs etc – it is going to become the default you have to work rather hard to avoid it seems, and in a year or two now its been in that transition to being the default for a while the majority of the problems folks have with it are going to be fixed, either in Wayland patches, in the applications that are not playing nice, or in the fact that all applications are defaulting to wayland so no compatibility layer with X11 is needed.
I don’t like it, but X11 remains important for Unix.
It’s a relationship like with GDI on Windows..
Also: Terminals. X11 graphic servers exist for about any OS, making porting easy.
On Mac OS X, even good old Tiger, it’s possible to compile/run graphical Unix applications with X11 output.
Wayland is Linux specific, it will never make it to the countless niche platforms that got X11 support in the past 50 years.
It’s comparable basic and universal like ASCII, IBM Model M keyboard layout or say, OpenGL 1.1/1.2..
Sort of a graphical IPv6, with as much promise, and rate of success.
To be fair, I remember the days of using XFree86 in the 90s.. 😟
On a more positive note, early 90’s DESQView/X (for DOS) was a neat environment.
It allowed porting Unix applications and using an humble PC (386SX+) as an X11 terminal.
So I think it’s good if X11 remains available as a fallback.
Wayland is too graphics-card centric and not really intended for unaccelerated graphics over various network connections.
For the basic stuff, X11 thus is still valuable.
Which is a bit ironic, considering how complicated and difficuot to understand X11 is by itself.
I remember using it and it either corrupting or even completely wiping the EDID from my monitors.
How? Unless there is something totally wrong with the monitor? EDID eeprom isn’t supposed to be writable
KiCad mostly sticks to X11 for now because wayland has too many unresolved issues.
https://www.kicad.org/blog/2025/06/KiCad-and-Wayland-Support/
KiCAD. Sigh. I don’t like KiCAD too much, it looks like bloatware to me. Like some Symantec or Windows printer suite software.
And the guys dropped Windows XP support way too soon at the time, IMHO.
I know, I know. XP is ancient and so it was back then. But it had a long tradition in engineering and CAD, also.
The KiCAD guys should try to get KiCAD going on more down-to-earth hardware, too, I think.
Like an old Unix-based SGI workstation, for example. IRIX GL support would be a challenge, for example.
Considering how humble the graphics of KiCAD look, they should try porting it to OS/2 or KolibriOS and OpenGL 1.x.
As a testbed, to learn how to write slick software. IMHO.
Why??
My interpretation of Joshua’s comment is that KiCAD is inefficient and it’s resource requirements are at odds with what it achieves graphically. The suggestion to target older hardware would force the developers to actually optimized the code.
If you think that’s bad, try Altium. KiCad is anorexic by comparison.
People complain about X11 but it’s really not that bad.
Came here to say this. I haven’t had problems with X11 in decades. And even way back then, it was just challenges configuring it, but modern distros are so much better at that now that I don’t even think about it anymore. X11 does everything I need it to do (give me a nice graphical display). If Wayland were obviously superior, I’d switch to it. But it’s clearly not, so why bother?
The big problem with X11 is if you are trying to use it over a WAN connection. It is actually rather inefficient. Some tools are unusable if the server and client are separated by a few hundred miles. The fact that projects such as NoMachine and Exceed were able to overcome this and make a remote session snappy from a continent away proves this. I don’t know if Wayland is better in this regard, though.
Wayland doesn’t even support remote connections. The main reason I never switched.
Most of the problems were for the developers/maintainers of Xorg, not the users. Most of the paid development has stopped, so gradually it’s going to be worse on newer hardware.
This is interesting, because one of my biggest gripes about Wayland is that it doesn’t fix some of X11’s longstanding UI security problems… but, per this article, apparently it does fix the minor problem that an application can arbitrarily plonk itself anywhere it wants on the screen… and the article thinks that’s a bug. Same for “global hotkeys”, probably.
For the record, if you want anything like security, We Hold These Truths To Be Self-Evident:
An application should have no control over its window decorations (so that it can’t impersonate other applications). Apparently Wayland does the opposite of this. And 1a: an application shouldn’t be able to go full screen on its own initiative; the user should have to take an explicit action to tell the window manager to give the application the screen, and there should be some other action the user can take that will always force the application out of full screen mode.
An application shouldn’t be able to grab the keyboard focus. In fact, it should never, never, see a keystroke unless the user has taken some action to direct the focus to that application. Popups especially shouldn’t get the focus. Nor should it be possible to keep the user from taking the focus away once granted. So no global hotkeys unless they’re provided by “the system” or by something to which the user has given special privileges. And 2a: an application shouldn’t be able to mislead the user about where the focus is. Which means, for example, that if focus follows mouse (which all sane normal people know is how it should work), the application shouldn’t be able to arbitrarily change the mouse pointer glyph.
An application should not know what’s going on outside of its own windows. Things like screenshots should again be special privileges, rarely granted. I will grudgingly allow the application to know whether its window is actually mapped at any given instant, but am not thrilled with letting it know which parts are obscured. 3a: An application should not know what the mouse pointer is doing outside of its own window. 3b: Information about whether the user is present or active should be on a need-to-know basis, and your application probably doesn’t need to know.
An application shouldn’t be able to synthesize keyboard or mouse events that will be seen by other applications.
For extra credit, also don’t give applications random information about the hardware.
Except for a few, very special cases, maybe.
Such as emulators or VMs, in which the keyboard must be exclusive to the guest in order for input to not to conflict with the host.
I don’t think this is meant. The application should never need to grab the focus by itself.
Of course a wayland compositor can give an emulator, a vm or a game exclusively the input focus.
This approach also properly allows centralized mapping specific input devices to specific applications by the compositor. instead of the application just grabbing the first device it finds or worse needing application specific logic where each app has it’s own quirks.
Your points 2 and 2a contradict each other. If focus follows mouse then any popup that happens to open under the mouse cursor will steal focus.
That’s why focus follows mouse is such a bad idea. You should have to take an intentional action to give a window focus, and simply passing over a window because you have to in order to reach the window you actually want shouldn’t count.
Being able to bring across remote GUI apps with “ssh -X”. I do it daily. My work laptop has a crap screen, and won’t drive two 4K displays.
Wayland has already succeeded X11.
Windows didn’t die when XP rendering stack was replaced in Vista. Vista got a bad rap but then windows 7 was loved universally.
Some people are simply stuck in the Vista era with Wayland and fail to find ways forward and keep blaming the tools instead of their own lack of solutions.
This article existing in 2025 is a bigger problem than any point made in the article. Time to accept the path forward and adapt you solutions.
Your comparison isn’t even remotely close. It reads like someone whose use case it works for, who’s incapable of considering other people’s cases. Wayland would be fine if people only worked on their own personal devices, sure. It’s utter garbage in a multiuser headless environment.
Counterclaim: all versions of Windows ever released have been worthless garbage.
“For various stupid reasons, we substituted something not meaningfully better for a central component of your working system, and now it’s your job to do a bunch of work to fix our screwups”.
Windows 2000 had been respected by Linux users, though.
That’s at leasthow I remember from about 20 years ago.
The common statement was lke “with Win 2k, Microsoft did it right for once”.
I remember re-starting explorer multiple times in Windows 95A, it was still capable of many more things than Win 3.11, Win 95C however worked much better, I only moved to 98 for dual-monitor support IIRC. Sometimes PEBCAK.
I’ve had enough hardware and used enough flavors of Linux and Windows to know if you stumble on an issue with your particular match-up of hardware you’ll have a bad time. Whether it’s wake/sleep bugs or your EDID or SCSI being unreadable.
The real problem sometimes lies in the UEFI, most BIOS don’t expose settings we would like to use. See above 4G encoding and turning off software RAID (I Looove slip-streaming specific RAID drivers into a system with a single NVMe drive, said no-one ever).
By and large Windows ‘just works’ for what I do. Meaning a hardware or software issue I have is generally solved in 1/10th the time, and is also 1/5 less likely to occur in the first place.
I do like SteamOS, have yet to try Bazzite. Kinda hope Windows goes to a Proton/Vulkan Compat/render backend too though. Seems like it might if Intel Arc drivers already emulate any render pipeline older than DX12/Vulkan.
Windows XP SP2 had its fans, too.
To many, it was their first real, non-DOS based Windows.
Or real OS, in general (not counting OS/2)..
It’s also notable that it had best audio support and no DRM yet.
You could capture StereoMix, have DirectMusic, DirectSound 3D, Aureal A3D, OpenAL, ASIO, a real MPU-401 interface etc.
Especially high-quality MIDI Software Synthesizer such as SYXG-100, Roland Virtual Sound Canvas (SC-55/SC-88) came as 32-Bit device drivers that ceased to work on Vista onwards, because DirectMusic became emulated.
And no VSTs are no replacement, except for a MIDI editor.
The driver-based softh synths could be used in games directly via DirectX. Like the Microsoft GM synth.
XP was also the OS of hacks and CD/DVD ripping..
Beginning with Vista, a lot of stuff nolonger worked.
2D GDI drawing was very slow, suddenly. WDDM 1.0 driver model was limited.
Windows 7 fixed these things (brought back GDI acceleration), but also looked rather dull in direct comparison.
Vista Ultimate with latest Service Packs and Platform Update was on par with Windows 7. Even got DirectX 11 (Tesselation!).
Now Vista basically got a lot of Windows 7 system files and software compatibility was about same.
Unfortunately, by that time, Windows 7 had replaced it largely.
Which is a shame, because Vista was the opposite to minimalism.
It was a generous, powerful OS for high-end PCs, with lots of animations, gadgets and such.
Unfortunately, it needed a Shader Model 2.0 capable GPU to offload graphics.
Which most users didn’t realize. They turned off Aero Glass for better performance, when in reality it was helping performance.
With Aero Glass, the desktop elements are drawn by the GPU.
So a cheap Nvidia FX 5200 would have been sufficient for an upgrade, already!
It even was mentioned in books about the RC1 (release candidate) of Vista.
Well, in principle that was helping.
Because of the limits of WDDM 1.0, the graphics memory had to be mirrored in PC RAM, too.
Which is next point. Lots of existing XP PCs were underpowered at the time Vista was out.
In fact, they were basically re-used, merely slightly upgraded Windows 98SE PCs that crawled on XP already.
It’s no wonder, that Vista experience was so bad.
1,5GB RAM would have been reasonable, but many existing XP PCs had 256 to 512MB, at best!
Some Windows 2000 PCs ran on 128MB of RAM, still! 😢
Anyway, just saying. There are indeed some similarities of Vista rendering and Wayland.
the only reason folks like windows 7 was that windows 8 was so awful.
I liked Windows XP. The versions of Windows that shipped after came with a different mindset. MS accepted fundamental architecture issues, silent fault. Windows had become unreliable by… choice.
Example, when you plug 100 different USB devices into Windows Vista, and newer, devices no longer register. You will get no error. One of the reasons factories kept running XP for years after support for it stopped.
As far as “love for win 7″… I hated the move away from task bars and borderless windows. MS followed the trend with a sledge hammer. It was either Google or the web, that introduced that UI design paradigm . “Everything must be frameless, everything is clickable, no more buttons.” Even today, on windows, “Think before you click!!!” just to drag a window. Widescreen TV displays, becoming monitors, killed the window taskbar. I got a few more pixels of window content, but have to hunt for one of the 2 spots on my browser window just to drag it around.
So yah, I like my OS with a clear UI language. Instead I get “apps”, each with their own interpretation.
“This is how it works in Netflix.” and “This is how it works in Hulu.”
It’s a mess that had been solved but “branding” needed to be creative and control the whole screen.
PS: MS tablet/windows hybrid experience, it’s a joke, look at that calculator, yikes. If it was possible Steve Jobs would have done it.
I’m just a ‘casual’ Linux desktop user, but that sounds like a mighty long and troubling list of issues with Wayland. Makes me not wanna touch it.
Luckily, 95% of these problems have nothing to do with Wayland.
I’ve no objection to Wayland in principle, I just find that in practice I encounter many problems when using it – with a fairly bog standard AMD GPU from a couple of years ago and Gnome. I regularly see lots of people posting online claiming that the Wayland experience these days is pretty good, and I tried it again earlier in the month to see if that was true – I was back on X11 within days after encountering so many graphical glitches I could barely use the thing.
I’m not sure what I’m doing wrong – I thought I had what has long been considered the perfect setup for running on Wayland. My main concern is that with all the talk around disabling X11 support for various desktops, I’ll be forced onto something that doesn’t work. Over the last few years, Linux on the desktop finally became a near perfect experience for me, but I feel like I’m about to be thrown back to the fiddling and extremely careful hardware selection of the mid-2000s, even though my needs are modest (browser, terminal, IDE).
No, I don’t feel the needs to replace X, especially if after 16 years–it still is not mature enough. I see it just like Flatpack. Apt and Debs work, yet I have already had 2 different problems with Flatpack on 2 different systesm.
There are people trying to make Linux better; there are trying to make Linux different, and there are people trying to sabotage Linux.
Sabotage Linux?! How? I’d like to know.. just asking for an, uh, friend. 😃
ask Lennart Poettering
Ah, yes.
Mr. feature creep and ego projects, people like him should never be the lead in any capacity of any projects.
“After more than forty years, everyone knows that it’s time to retire the X Window System”
Please, show me exactly where I can find the statistics.
Reading comprehension gone on holiday?
i like and use X11. xorg has improved quite a bit from when i first started using xfree86 30 years ago. for example, i appreciate how easy it is to build a xorg module — like a custom input driver — from source these days.
this article — and the comments here — paints a foggier picture of wayland than i expected.
but obviously, the biggest thing i don’t like about wayland is that i don’t know it. but i’m also concerned because when i do dig into it a little bit, it seems like there’s not only an uncomfortable relationship between the display server and the window manager, but also between the display server and the apps?? it seems like even from the perspective of an app writer who is gung ho for wayland, there are several incompatible waylands??? i assume that’s not as big a deal as it looks like but at first glance it seems like a dramatic step backwards.
but personally, i’m not worried about it, and i assume i’ll switch someday. 20 years ago, i was happily using a rootless X11 server on macos x. and today, for my remote X i have switched mostly to VNC instead of X11-over-TCP. so i don’t know what my solution will look like but i’m confident i’ll be able to find a combination of wayland features and compatibility wrappers that is comfortable enough for me. but the fact that it hasn’t happened yet is making me wonder if i won’t just use xorg until i die
“Some users also simply say that they do not care about Wayland either way as it offers no new features they want.”
It offers me less features. Either I run both Wayland and xorg or just xorg. And running both offers me nothing in advance. Last I checked, X forwarding was still not ported over directly to Wayland and until that is done, I see no point in running it as xorg has to run anyway to provide that feature. I have been using xorg for over 25 years now and never had any major issues. So yeah, I’m in the camp of ‘I don’t care’. Xorg is stable, it’s fast, it runs incredibly smooth on embedded devices, runs smooth on specialized computers running long term availability hardware. We still buy brand new computers with Intel Atom N2600 CPU’s (I blame Det Norske Veritas). You need to find ways to limit the power used by the CPU and Wayland provides nothing extra.
Exactly this. I use X forwarding extensively. It’s easier not to even mess with Wayland.
Newer, therefore better. SPIT…
That’s the motto now-a-days :) . “If newer … it’s GOT to be better.” Part of human nature I think…. Never mind that older is usually much more stable and bug free, kinks worked out, etc. See that with computer languages as well.
I’ve no axe to grind with x11 vs Wayland. As long I can get in remotely and locally (multi-user) and my GUI does what I want it to do. Golden.