I believe XWayland is there to stay for retro compatibility. That seems fine to me, retro-compatibility seems like a good thing. That would not be unlike XQuartz on Mac, which doesn't make us question Mac's own display server.
> I would argue nearly "there". But not yet.
I fully believe there are use case that are not fully covered indeed.
Gaming would be one (I don't know whether gaming on Linux, outside the Steam Deck, is actually niche). But doesn't Wine work well with XWayland? (until they fully move to Wayland, which I think I've read they are in the process of doing)
I would guess NVidia would concern mainly gamers and people working with ML (maybe those people don't actually need NVidia to display stuff though, and are actually fully covered by the iGPU - I don't know).
It seems to me running an actual X11 session is now needed for niche things, but I might be wrong.
Yes but if more than 50% of the GUI softwares you run fallback to XWayland (including terminals), it makes no sense.
> But doesn't Wine work well with XWayland? (until they fully move to Wayland, which I think I've read they are in the process of doing).
Wine is already a compatibility layer, so adding XWayland on top of it introduces overheads and additional lags and visual artifacts in some games. There is an experimental version of wine 9.22+ that should theoretically run natively on Wayland, but it still crashes 100% of the time for me. Nvidia drivers were also notorious with Wayland, but they are getting better progressively.
I think the meme is nearly here: this might be the year of the Linux gaming desktop.
That just seems wrong. Default terminal apps like gnome-terminal and konsole are probably the first apps that have worked (well) on Wayland (as soon as select paste started working, which it has for years). If you are using a terminal that doesn't support Wayland, that's a choice you are making.
Nothing uses XWayland on my setup except IntelliJ (and they are working on a native Wayland support, it might actually be working well right now but I also happen to run it in a local container though ssh and I haven't yet looked into the Wayland alternative of ssh -X). And that was not even a criterion to choose my apps.
That's just not 50% for most people I think. And even if it were, running the DE in Wayland and all the apps in XWayland already brings benefits in comparison to running a full X11 session, at least for setups with multi screens with different DPIs. So yes, it would still make sense.
> Wine is already a compatibility layer
.. That doesn't introduce any lag because it's not an emulator, it's a reimplementation of Windows APIs, with native speeds, so it doesn't matter at all that it's a compatibility layer in this discussion. Okay, there is dxvk, wined3d and things like this, but apparently they can be better than the original under certain conditions [2].
> adding XWayland on top of it introduces overheads
It's not supposed to, your game renders directly anyway [1]:
> What about the overhead of running X on wayland?
> Most modern desktops already use an external compositing manager: when the X server decides it needs to update content, it informs this external process (usually your window manager), and allows it to control the rendering entirely. Using Xwayland is just the same as this, but more efficient because the compositing manager doesn't have to go back through the X server to display the content it rendered.
I'm eager to believe that nvidia has bugs with Wayland (they've been given the middle finger for a reason), and that there are Wayland-specific bugs in some apps, especially games.
> I think the meme is nearly here: this might be the year of the Linux gaming desktop.
Aren't we already there actually? Isn't it like 90%+ of games that just work now? Isn't the Steam Deck a success?
I have friends that were stuck on Windows because of games until recently but finally switched to Linux last year or the year before.
[1] https://wayland.freedesktop.org/faq.html
[2] https://www.reddit.com/r/linux_gaming/comments/nyuqou/what_m...
This has been my experience. The only games that give me grief are either just a badly supported on Windows or use some ring0 anticheat bullshit. Thankfully, I don't even like most of the sorts of games that rely on ring0 anticheat, so nothing of value was lost for me.
The only remaining games that have papercuts are caused by garden-variety Wine/Proton support that usually get patched in a matter of weeks. I honestly haven't seen a Wayland-caused game issue since the release of KDE 6.0.1, and even the general Nvidia driver issues have mostly evaporated since then.
Konsole: 24.12.1
KDE Frameworks: 6.10.0
Qt: Using 6.8.1 and built against 6.8.1
Arch Linux (Wayland)
Build ABI: x86_64-little_endian-lp64
Kernel: linux 6.12.10-arch1-1
For an Arch user, that 50% may be valid since the distribution encourages the use of an eclectic mix of software. On the other hand, most Linux users will be using distributions based upon Gnome or KDE. In those cases The majority of software will run under Wayland.
It's also kind of weird to dismiss Xwayland due to it adding overhead. Traditionally, X applications could be displayed upon a completely different machine from what they were running on. This could be a dedicated X terminal, and even a Windows or Macintosh computer. I would suggest the only reason why X is thought of as a native graphics stack is due to the rise of Linux (and, to a lesser degree, BSD) on inexpensive personal computers.
Things like XMonad or StumpWM are relatively popular among Arch users so that is why X is still featured prominently in the Wiki. I use those two, but on NixOS, and I find it annoying there is no simple migration route. Waymonad never took off. StumpWM/Mahogany seems to be getting there, but AFAIK there is no release in the horizon yet.
Of course if something you need doesn't work under Wayland, you will need to use X and you can for as long as it still works, and that's actually about the only good reason I see to pick X. But make sure issues are reported!
(Now, I'm curious, what breaks CJK under Wayland?)
As far as I understand, wayland doesn't have any common provisions for input methods whatsoever. (There is something recent and experimental, but I'm not sure.)
So far, every toolkit would implement it's own way of inputting CJK, independently. So input would work in kde and gtk programs, but not in foot or alacritty.
But it is dying. Its design doesn't correspond to the way things work anymore. It is a nightmare to maintain because its (old) code is incredibly complex, because it is full of features, and these features need to be kept for backward compatibility but aren't used much anymore. Modern toolkits actually bypass its features when they can: instead of drawing using X directives, they stream huge pictures they draw themselves. X wasn't designed like this, but it was getting in the way so this was shoehorned into it. It's now common to have two screens with different dpis, which X doesn't handle. With X, drivers are provided with X, but it is now normal to have the drivers in the kernel instead.
The X developers went away to design and develop Wayland to start from a clean slate, with all the insight gathered all these years. Drivers and apps are now developed against Wayland and are less and less tested on X. Things are going to startup breaking, I suspect librairies will startup removing X support at dôme point, etc.
X is not an alternative to Wayland. It is its past. I don't wish X to die at all. I have no feelings here. But that's what's happening.
I don't mind someone using X knowing all this. But we are commenting on a guide directed towards newcomers. Because all these reasons, making them use X is bad advice.
If you have a working install using X, there's no urge to switch. Don't fix it if it ain't broken. It might even be clever to follow the upgrade path of your distro if you didn't customize anything.
Apart from that, I don't see any reason today to use X instead of Wayland unless specifically something doesn't work. And if it is the case, I hope you reported the issue, because one day you might be forced to switch and if the bug is still there, you will have to live with it.
We are not in a situation where we have two competing alternatives between which you can freely chose. Both involve the same developers, and one of them is considered obsolete and isn't developed anymore.
It's not that we are Wayland fans (at least I'm not). It's just that Wayland is all we have going forward.
How long until running software designed in 1987 is reserved for retro-computing? The graphics stack has changed way too much since then.
(damn, the X protocol v11 is even older than I thought, but I guess Wikipedia's entry is not considering minor revisions to X11)
---
What graphics stack pushing forward is needed to write documents, spreadsheets, view charts, write code, edit video/images, do data analysis, etc?
Writing code and doing spreadsheet was already possible on the Atari ST but you wouldn't use its software stack today.
Hardware evolves and software should either grab its magic ball and plan for that, or adapt to changes to avoid dragging hardware evolution back.
(Besides stability) it offers more options. The PineNote is an e-ink stylus tablet that only supports Wayland, and I've been looking for a dead-simple mouse-driven compositor (because tablet). X had bucketloads of WMs, most of them long-since stabilized and featured complete. Wayland has nothing. Nothing in the Debian-stable (which comes shipped on the PineNote and is stable) repo, at least.
Is this somewhat niche? Sure! Is it a huge fucking pain in the ass? Also sure!
I'm not against Wayland, in premise - it supports process isolation (potentially useful for KDE Activities in some amazing ways) in a way that X just can't and never will. X is not the future. But frankly, neither is Wayland. It's barely even the present (ask wacom/Krita users, or anyone with accessibility needs, what they think of Wayland devs), and actually loses some features (multi-window stuff is impossible-by-design on Wayland, which means entire swathes of software will just literally never switch to Wayland), and while Wayland does gain features and possibilities, they're niche features just like the X features that Wayland broke.
The "X is a clusterfuck" argument becomes increasingly weak, since Wayland is more and more clearly a clusterfuck itself.
At this point, I think Arcan is the future (www.arcan-fe.com), although please note the main dev is specifically avoiding the limelight before v1.0 and thus writes in a deliberately dense and unapproachable fashion. It's super elegant and a genuinely groundbreaking project.
Is it? The article was written end of last year. I believe the most widespread DEs on the most widespread distros now use Wayland by default, don't they?
There's now stuff that just works better on Wayland today. I've kept X11 for a very long time, probably until 2023, because of little annoying details, but now it just works. I think the only remaming thing that annoys me is some apps like Kate not coming to the to the foreground when invoked.
Things are slowly going to stop working on X11. Perhaps not today, but we are talking about setting up a new Linux install.
Especially something like Arch that has all the recent versions.