What does it mean by removing X11 support? [LWN.net]

What does it mean by removing X11 support?

Posted Jun 11, 2025 4:37 UTC (Wed) by atai (subscriber, #10977)
Parent article: Ubuntu 25.10 to drop support for GNOME on Xorg

Do X11 programs continue to run via XWayland?

to post comments

What does it mean by removing X11 support?

Posted Jun 11, 2025 5:00 UTC (Wed) by MKesper (subscriber, #38539) [Link] (8 responses)

Sure. It's a problem for people where graphics card or some exotic hardware are not supported. They still can change to Debian.

What does it mean by removing X11 support?

Posted Jun 11, 2025 5:57 UTC (Wed) by LtWorf (subscriber, #124958) [Link] (7 responses)

My exotic hardware: "thinkpad t15 gen2". I am clearly the only person in the world to have a touchpad or a thinkpad.

What does it mean by removing X11 support?

Posted Jun 13, 2025 13:17 UTC (Fri) by mgedmin (guest, #34497) [Link] (6 responses)

As a long time ThinkPad user I'm surprised to hear claims that Wayland doesn't work on my hardware.

What does it mean by removing X11 support?

Posted Jun 13, 2025 13:39 UTC (Fri) by pizza (subscriber, #46) [Link] (4 responses)

> As a long time ThinkPad user I'm surprised to hear claims that Wayland doesn't work on my hardware.

I don't mean to dismiss LtWorf's problem, but this is yet another example of hardware vendors not caring about anything other than Windows. That said, I've dealt with plenty of laptops over the years whose touchpad was nigh-unsuable (if not completely so) under Windows without the OEM driver; it wasn't Windows' fault for not magically handling those touchpads without the vendor driver, nor is its libinput's fault for not magically handling LtWorf's.

(And from the description and the bug ticket, it does seem like said touchpad "working" under xorg's old synaptics driver was more accidental than anything else, due to bugs canceling each other out)

(Disclaimer -- I'm typing this using a ThinkPad T495 running GNOME/Wayland)

What does it mean by removing X11 support?

Posted Jun 13, 2025 14:40 UTC (Fri) by zdzichu (subscriber, #17118) [Link] (2 responses)

Actually, Lenovo was supposed to cooperate with Fedora to make their hardware working well. And Fedora being Fedora mean it should be supported by upstream, no weird patches nor custom drivers.

That's the theory, though. Shortly after above huge announcement, Lenovo released ARM Thinkpad without Fedora support whatsoever.

What does it mean by removing X11 support?

Posted Jun 13, 2025 14:57 UTC (Fri) by pizza (subscriber, #46) [Link] (1 responses)

> Lenovo was supposed to cooperate with Fedora to make their hardware working well.

That's an aspirational statement, and it only applies to "select models" that were to be *shipped with Fedora preinstalled*.

If you buy one of those supposedly supported models with Windows and install Fedora yourself (eg because Lenovo doesn't offer Linux as an option in your region), Lenovo will refuse to support you. (Ask me how I know)

> And Fedora being Fedora mean it should be supported by upstream, no weird patches nor custom drivers.

This means that Fedora only ships what upstream provides. It doesn't mean they're on the hook to preemptively figure out every quirk of the dozens (of not hundreds) that Lenovo releases every year.

> That's the theory, though. Shortly after above huge announcement, Lenovo released ARM Thinkpad without Fedora support whatsoever.

The overwhelming majority Lenovo's lineup remains Windows-only.

What does it mean by removing X11 support?

Posted Jun 13, 2025 16:34 UTC (Fri) by zdzichu (subscriber, #17118) [Link]

I meant Lenovo should be on the hook to provide any necessary quirks, Fedora devs may provide guidance how to integrate them. Or even Lenovo should fix the hardware based on input from Fedora developers.

I'm sorry about your experience with Lenovo's support. But I'm not surprised.

What does it mean by removing X11 support?

Posted Jun 13, 2025 16:02 UTC (Fri) by LtWorf (subscriber, #124958) [Link]

It works fine on linux too, just not if one uses libinput.

What does it mean by removing X11 support?

Posted Jun 13, 2025 16:01 UTC (Fri) by LtWorf (subscriber, #124958) [Link]

Thinkpads only share the brand name. 2 different models will have completely different hardware.

The fact that something might work well on one model doesn't necessarily mean it will work well on all black computers that are covered in rubber and have a red LED on the outside.

Wayland works, it's libinput that doesn't like the touchpad on my model. Unfortunately Xorg supports multiple drivers while wayland only supports libinput, so that blocks me from using wayland entirely.

I hope it's clearer now.

What does it mean by removing X11 support?

Posted Jun 11, 2025 5:28 UTC (Wed) by motk (subscriber, #51120) [Link]

XWayland does rootless stuff now, there's literally no reason to complain much any more. You'll have far more pain dealing with ssh and xauth than anything else.

What does it mean by removing X11 support?

Posted Jun 11, 2025 7:11 UTC (Wed) by epeeist_pitlochry (guest, #156764) [Link] (13 responses)

Do X11 programs continue to run via XWayland?

One thing that is not there for photographers, is colour management. It is being worked on, but it isn't there yet.

We will continue to have to work with a UI that allows calibration and colour management of displays until it is ready.

What does it mean by removing X11 support?

Posted Jun 11, 2025 8:19 UTC (Wed) by Wol (subscriber, #4433) [Link] (6 responses)

Hopefully, the removal of X11 simply means the removal of the HAL.

Aiui (and it's not been clear) the hardware management side of X11 has been deprecated for years. The user-facing end has morphed into X-Wayland and is an active part of Wayland.

So basically, the deprecation/removal is just the removal of the actual client (in X-speak) that actively manages the hardware - that has now been replaced by the Wayland drivers. Is X-Wayland a compositor? Whatever, your X programs should be able to speak to X-Wayland and not realise that Wayland is actually managing the hardware for them.

Cheers,
Wol

What does it mean by removing X11 support?

Posted Jun 11, 2025 14:37 UTC (Wed) by nix (subscriber, #2304) [Link] (5 responses)

> So basically, the deprecation/removal is just the removal of the actual client (in X-speak) that actively manages the hardware

That would be the server, not the client :) X clients continue to work just fine.

What does it mean by removing X11 support?

Posted Jun 11, 2025 15:55 UTC (Wed) by daroc (editor, #160859) [Link] (4 responses)

Unfortunately, X11 does things backward. The part that talks to the hardware is called the client, and the part that talks to other software in order to render it/compose windows/etc. is called the server.

What does it mean by removing X11 support?

Posted Jun 11, 2025 16:39 UTC (Wed) by smcv (subscriber, #53363) [Link] (3 responses)

> The part that talks to the hardware is called the client, and the part that talks to other software in order to render it/compose windows/etc. is called the server.

Are you sure? That's the opposite of the diagram in the Wikipedia article you linked to.

In 1980s-era X11, my understanding is that the client is the application, which might be xterm or xeyes or something, and does not talk to the hardware at all, only to the X server; and the X server talks to the hardware on behalf of all clients. The X server is a server in the sense that it calls listen() and accept() on a socket, and the X client application is a client in the sense that it calls connect() to that socket.

The way in which it is backwards is that if you're using remote X11 to run an application (perhaps something that does heavy number-crunching) on a large computer that you might call an application server, displaying its output on a lightweight terminal that you might call a thin client, then the X client (software) is running on the application server (hardware) and the X server (software) is running on the thin client (hardware).

In newer X11 applications and toolkits, increasingly both the client and the X server talk to the hardware, and the X server's role is more about coordination than rendering, but the connect()/listen() are still the same way round as they always were.

Wayland is also the same way round as X11: the client is the application, which calls connect(), and the server is the centralized thing linking all your applications together (in this case it's called the Wayland compositor), which calls listen().

Other infrastructure components like Pulseaudio, Pipewire and D-Bus are also the same way round as X11 and Wayland, with one centralized server that calls listen(), and lots of clients that connect() to it.

What does it mean by removing X11 support?

Posted Jun 11, 2025 16:55 UTC (Wed) by daroc (editor, #160859) [Link] (2 responses)

> Are you sure? That's the opposite of the diagram in the Wikipedia article you linked to.

Argh. Yes, you're right. I had "X11 uses the terms backward" cached in my head, and didn't invalidate it properly. I suspect Wol may have done the same.

What does it mean by removing X11 support?

Posted Jun 11, 2025 19:16 UTC (Wed) by jem (subscriber, #24231) [Link]

The confusion comes from that a "server" is usually thought of as the big iron tucked away in some machine room and back in the day you interacted with it using a simple text terminal. The X Window System changed all that: now the display server was running on the machine where your graphics hardware was. The "server" was the Sun workstation on your desk, and the client app was typically the xterm program running on a mainframe in some data center.

What does it mean by removing X11 support?

Posted Jun 12, 2025 0:34 UTC (Thu) by gerdesj (subscriber, #5446) [Link]

Yes! The Xserver does the link to hardware and Xclients send their output to the Xserver.

The miss-understanding/interpretation comes about when you are running "remote X" (export DISPLAY=....). Now you run the Xserver locally and the Xclients are running remotely but displaying locally.

You can ssh into a remote box and daisy chain ssh sessions and still get a local X display for a client, if you are careful, even if the remote box does not have a graphical terminal. It will also be bloody slow. Having said that, modern internet connections are 1+Gb/s, which nicely glosses over a protocol developed in the days when LANs with 10Mb FD was a bit of a luxury. The cool kids got the 100Mb FD connections with their swanky PCI based Cisco 595s instead of a ISA based 509. Actually, X is even older than that funky late 1990's gear.

I shouldn't bother nowadays unless as a last resort. MeshCentral does the job, rather better!

Its tools for the job. I have a screwdriver that is an antique - it was my granddad's, then my dad's and I'm 54. It still works but it is a flat 1/2" bit and most screws I deploy are PH or torx. However, if I can, I love to dig out old Betsie because I'm a human and the handle is lovely to ... handle.

I still run X sessions across my home LANs because I can and in the case of MythTV (int al) - rather necessary if the server has no local display.

I rather like the old school charm of running X clients but it is just that - old school. I grew up when records and then tape was your only choice for music and the web did not exist. I do not now own a record player. I have converted my CD collection to FLAC.

So, I shall wipe a tear from my eye for X11 and Xorg (I barely noticed the passing of XFree86). If people who know more about this stuff than me and are prepared to slap code into a repo - then that's all good by me.

I remember when Linux looked shite, really shite on X. Funnily enough some console drivers with a good config got it right and looked lovely. Nowadays I have absolutely gorgeous founts and all sorts of funky stuff going on my screen. I'm happy to wait for the WL boys and girls n that to get things thrashed out. Its a really hard project that is progressing nicely but needs more time.

Sorry about the fount affectation!

What does it mean by removing X11 support?

Posted Jun 11, 2025 11:10 UTC (Wed) by pizza (subscriber, #46) [Link] (5 responses)

> We will continue to have to work with a UI that allows calibration and colour management of displays until it is ready.

This functionality has been available since at least Fedora 25 [1], released back in *2016*.

Wayland's "Color Management" has been on par with what X11 provides for at least that long. [2] The "being worked on" features vastly outstrip anything that X11 could ever do. [3]

[1] This is when Fedora switched their GNOME session to defaulting to Wayland.
[2] Associating a display with a profile, loading a LUT/map into the display driver, providing a way for applications to query that profile, and providing a way for tools to perform calibration.
[3] per-surface colorspace/depth parameterization and automagically making everything JustWork(tm) on the current display(s).

What does it mean by removing X11 support?

Posted Jun 11, 2025 12:01 UTC (Wed) by aragilar (subscriber, #122569) [Link] (4 responses)

I guess the question is then why do users/developers of applications which care about colour management repeatedly disagree with the assertion that wayland colour management is ready? I see similar arguments about accessibility (with the series starting with https://fireborn.mataroa.blog/blog/i-want-to-love-linux-i... being only the latest example)? Is it lack of documentation, do the wayland (and associated devs) misunderstand the needs and use-cases (and so they're building the wrong thing), or is it something else?

What does it mean by removing X11 support?

Posted Jun 11, 2025 12:47 UTC (Wed) by pizza (subscriber, #46) [Link] (3 responses)

> I guess the question is then why do users/developers of applications which care about colour management repeatedly disagree with the assertion that wayland colour management is ready?

It's not fair to lump "users" and "developers" in the same bucket here. I'm not aware of any specific "developer" complaints, but with regard to users, their complaints are overwhelmingly due to them not understanding what color management features that X11 _actually_ provides [1] versus the until-recently-unrealized color management goals of Wayland.

As an example of this, the >5yr-old laptop I am using to type this reply:

1) Has _never_ been booted into a native X11 session (ie GNOME on Wayland from day 1)
2) Was calibrated using GNOME's built-in profiling tools when I first set it up
3) Is currently using said calibration profile
4) Hot-plugged displays each get their own profile and JustWork(tm)
5) Color-aware X11 applicaitons see no functional difference between running under XWayland vs a native Xorg session

> I see similar arguments about accessibility

Accessibility is a different situation; the old mechanisms heavily relied on the "everything has full access to everything else" misfeature of X11 that still didn't work well without deliberate application (ie beyond what the underlying toolkit provides) support. Given that Wayland is explicitly designed to not have that massive "everything sees everything" security hole, a different underlying mechanism for conveying this information was necessary. It is my understanding that the new stuff being worked on relies on explicit exporting of high-level data instead of requiring a full introspective deep dive (filtered by brittle heuristics) into the state of every UI element in the application. This new stuff is not unique to Wayland and there is no technical reason why it couldn't be directly utilized by X11 (or any other type) applications -- not unlike libinput in that respect.

> Do the wayland (and associated devs) misunderstand the needs and use-cases (and so they're building the wrong thing), or is it something else?

Quite the opposite; they have been working hard to understand the *actual* needs [2] and carefully designing protocols and APIs to meet those needs. Taking into account the lessons learned from existing/prior efforts (ie across all OSes).

[1] ie very little, and it all must be managed completely manually by each individually
[2] which anyone that's ever done any UI/UX work can attest is _very_ different from the *expressed* needs

What does it mean by removing X11 support?

Posted Jun 11, 2025 13:34 UTC (Wed) by aragilar (subscriber, #122569) [Link] (2 responses)

My understanding is if you search "wayland" on https://discuss.pixls.us/ you will find numerous complaints from developers there (which again could be due to multiple issues and could even be resolved with the right documentation). So at the very least there is a disconnect somewhere, and given some of the more dismissive comments I've seen around accessibility which align with the blog series I linked (where the "why would you want to do that" having an answer within that series), my feeling is while there is (hopefully) work being done to improve things, such work seems to be undermined (and possibly regressed) by those within the wider wayland developer community (for lack of a better term).

What does it mean by removing X11 support?

Posted Jun 11, 2025 13:52 UTC (Wed) by farnz (subscriber, #17727) [Link]

A big chunk of the complaints are because Wayland's support for better than X11 colour management has taken a very long time to go from proposal to something that's stable and ready to implement.

Note, though, that a heck of a lot of the difficulty here is that X11 worked by assuming that you have a fixed framebuffer that correctly represents the data sent to the RAMDAC that then drives your CRT, and required hardware to emulate that as we moved away from CRTs and 13W3 connectors, while Wayland wants to work with things like hardware composition (where the final display output is created on-the-fly in hardware from multiple framebuffers, saving energy), and displays whose EOTF can be controlled by software.

What does it mean by removing X11 support?

Posted Jun 11, 2025 13:58 UTC (Wed) by Wol (subscriber, #4433) [Link]

> (where the "why would you want to do that" having an answer within that series)

"Why do you want to do that?" is a very valid question from a developer's PoV.

Equally "Why do you want us to do that?" is a very valid question from an end-user's perspective.

Speaking as a developer, the number of users who can't differentiate between cause and effect is awful, and demanding I fix the symptom rarely leads to success, it usually breaks things further.

But speaking as an end user, the number of developers who don't appear to have functional ears is also awful - they are so fixated on their wonderful solution that they don't understand how completely impractical it is. This digital world is dreadful - more and more the people I associate with are *abandoning* technology, because change is speeding up just as their ability to cope with change is fading. The latest NHS "drive to digital, everything is now moving to the app" will be a complete disaster as the people who need it most will be those least able to use it. How on earth are you supposed to use an app when you are half blind, have poor motor control, and your forgettery now works much better than your memory? Your typical NHS patient, in fact ... (and the other government drive, enhanced privacy and "you need to take responsibility for your own affairs", is merely pouring petrol on the flames!)

Cheers,
Wol