Please, please stop using Xorg with Asahi Linux
social.treehouse.systems
social.treehouse.systems
I'm not sure 'quickly' really is the right word choice there.
On what planet is this true?
update: I sit somewhat corrected. I have retried KDE with an NVidia card and /usr/bin/Xwayland and am impressed. Not 100% perfect, but very usable.
Being Asahi is not only brand new but not even near a first stable release the amount of surprise about the lack of xorg focus seems as silly as those that act like wayland reached parity 2 years ago
I do wonder if that isn't a solution. Couldn't they just ship rootful xwayland as the officially supported X server on Asahi, thereby being able to drop X.org without trashing people's setups?
Edit: This is a suggestion, not a "well why didn't they just do this".
Red Hat has announced they are depreciating Xorg already; which is a huge deal because they are almost single-handedly the only contributor keeping Xorg alive. Once they are done, work on Xorg will basically cease forever.
Considering Xorg is based on X11 which dates back to June 1984, I can’t blame them. A graphics system from 1984 is not suitable for 2023 in any capacity. Not without a ton of hacks, workarounds, and jank everywhere.
"depreciation" happens when capital investments lose value over time, often for tax purposes.
"deprecation" is when the builders of a tool want to discourage you from using the tool going forward because a better solution exists.
The latter is the one that applies to software.
So, yes, there's striving to a Wayland, post-X11 future, but it's far from the sure thing that X11 is.
Ultimately X11 works out of the box and with a quick .conf change I can easily setup my default monitors at the gdm login (or whatever Ubuntu uses these days).
I don’t have a dog in the fight on a display interpreter. But wayland just doesn’t work in my use case. And hasnt through multiple iterations (this laptop was built on 20.04 and went through a few non-lts releases before I just stuck to 22.04. )
For every FOSS project, it is up to the maintainers to decide the scope of what the project supports. Asahi Linux maintainers have every right to drop support for X11 when it increases the development burden and, according to the post, has lower performance than Wayland on the supported hardware.
Anyone who absolutely needs to stay on X11 due to a use case not yet available on Wayland-compatible software, your options are to switch to a different distro, fork Asahi Linux, or wait for that use case to be supported on Wayland-compatible software.
Based on the decisions mainstream Linux distros and desktop environments have been making, it's clear that Wayland adoption is increasing while X11 is gradually being deprecated. By the time Asahi Linux hits stable, supporting only Wayland will become an even more obvious choice than it is today. The words "don't come to us asking for help" may sound harsh, but they clearly communicate that X11 will no longer be supported on Asahi Linux.
Asahi just happens to focus on hardware where Xorg's flaws are more obvious, so X will be more broken than usual. But it reflects the future of Linux graphics.
I don't have any issue with Asahi focusing on Wayland first. But this attitude doesn't make me feel great. Is it really so difficult to have a little bit of empathy for your end users?
That’s easy to say if you’ve never worked on Xorg. The codebase is one of the worst in existence, few people fully understand it, and fixing bugs in it is a modern Sisyphean wheel.
- GNOME is Wayland by default
- Red Hat has announced Xorg is depreciated. A big deal considering they were basically keeping Xorg on life support single-handedly. Without Red Hat, there’s basically nobody working on it anymore.
- KDE Plasma 6 has announced Wayland is default now.
At some point, it becomes like a user complaining that OS/2 is no longer supported. It’s dead, X11 is older than OS/2 (June 1984!), we are all collectively exhausted, time to move on.
If xmonad doesn’t work for your system; take it up with them.
XWayland is not going away. X as a display server is more or less abandoned.
The users will move on when the replacement is unambiguously better, and/or at least has feature parity. Telling someone to throw out their working setup is a lot of why Wayland is viewed by some people with derision.
> The X protocol [...] doesn't spoil like cabbage.
It actually does. The protocol doesn't handle things we expect from modern hardware and software. And the implementation certainly rots away as the rest of the software ecosystem changes in ways which will slowly make Xorg more and more broken. Drivers will change in ways which expose Xorg bugs. GUI toolkits will introduce bugs in their X backends and introduce new features which don't work on their X backend. Nobody maintains X anymore, and something like a display server in a rapidly changing software and hardware ecosystem requires maintenance to keep functioning.
None of this is to say the Xmonad developer should spend time rewriting their project to use wlroots instead of X. But as long as it requires X, it's living om borrowed time.
Avoiding jank is also more or less impossible on X. There is no substitute for Wayland's "every frame is perfect" model.
Oh, and you can't connect a high DPI screen and a low DPI screen at the same time with X; to the degree that DPI scaling works at all, it's one global setting so you get to choose whether stuff should be absurdly big on your low DPI monitor or absurdly small on your high DPI monitor.
X: I'm not dead yet
Wayland Proponents: Well, he will be soon. He's very ill.
X: I support rootless operation and mixed DPI
Wayland Proponents: You'll be stone dead in a moment.
X: I think I'll release a new version of a window manager
Wayland Proponents: You're not fooling anyone, you know. Look. Isn't there something you can do?
X: [singing] I feel happy. I feel happy.
[whop]
In general this sentence is why the Year of Desktop Linux won't come in this millennia. Not only XOrg vs Wayland. Many such cases. Sad!
Which is why people keep using xorg… wayland is NOT a drop in replacement.
my experience with kwyboard shortcuts on sway/gnome is that they're implemented differently on a wm-by-wm basis
keyboard shortcuts is what I'm referring to
But, Android is a thing, and Linux is literally everywhere. OS has won, even if the mythical "Desktop Linux" didn't.
After installing their first distro, most people have a good experience. Then they install an update, that causes some ridiculous regression. At this time they have 2 choices: spend 2 weeks on reading all kind of arcane documentation, or go back to whatever they were using before. Or ask the maintainer, who responds "deal with it".
Just as another, fresh example, look at this: https://blogs.gnome.org/mcatanzaro/2023/05/10/gnome-core-app... - Gnome finally has thumbnails in the file picker, but they removed the music player. You can't have everything, I guess.
This kind of attitude why the Year of Desktop Linux will not be seen by our generation, not evil manufacturers.
Do you have to burn a flashdrive image, navigate through odd commands to disable secure boot, and find the magical incantation to boot from USB to install a soundcard driver?
I would love it to happen but the suggestion that there is anything the developer community can do is wrong. This isn't like the move to get friends and family to swap to Firefox in 2004, that was easy, it was in parallel with no data or app loss. With no new UX to learn.
I can get my parents to change to a different web browser. I could never get them to swap to Linux. I doubt I could get them to reinstall the os they currently use!
- Chrome browser,
- Skype, Discord or other chat applications,
- update GPU drivers regularly,
- tolerate the abusive Windows update/reboot policy,
- install multiple anti-virus software just to fight with Windows Defender,
- etc...
but not willing to install a new OS, even if it's literally a few click.
I think we kept saying this "Linux can't be mainstream desktop player because it isn't default" so long, that we can't even imagine anymore that there can be other, stronger reasons too.
Defaults are definitely not meaningless, but users are willing to change bad defaults, if there is a better alternative. Sure, not your parents, neither mine. But most of people around you (I assume) and around me (I know) would.
Sure, just anecdata, but over time I have had more than 1 colleagues telling me proudly that they started to play around Linux on their machine, only to learn a few weeks later that they have abandoned it due to some (for them) insurmountable problem they faced.
Of course if the user is technical enough they can figure out how to resize their existing partitions & set up a dual boot. But that's not the default, and that still risks corrupting the existing partition and therefore deleting all their programs & data.
And yet the average user will probably not do that, just like their eyes will glaze over if you say the incantation "soundcard driver" in their presence. It's unfamiliar and their friends don't use it, and that's it. Easy to install? Okay, but can you give a reason for them to even try?
Many people don't buy a Windows/Mac OS laptop or Android phone or iOS tablet. They buy a MacBook, Samsung S42, iPad, Steam Deck. OS? Web browser? Doesn't matter.
switching your desktop to Linux is still niche.
i think it being niche and also “the most popular it’s been” doesn’t have to be mutually exclusive.
That doesn't seem to have hurt the year of Desktop Windows too much.
For example, look how saddled MS Windows is by the burden of binary reverse-compatibility. It makes the leap to other arches like ARM nearly impossible.
Yes, some features of X11 are not present in Wayland. But the incomplete reverse-compatibility allows us to decide if those features are worth it in the first place. And if you don't like it you can patch it yourself.
User experience is ALL that matters. Apple has this ingrained in their ethos. While the OSS communities are bickering over X vs Y, they chose to pick what's usable and polish it until it delivered great UX.
Last time I checked, Wayland was broken in fundamental ways on my NixOS/KDE machine. So I went back to Xorg, that for all its ancient faults, delivers a better UX _today_. I'll keep trying Wayland every few months, and when it becomes an improvement, I'll consider switching to it. No, I won't "deal with it". Why don't YOU fix it?*
*If you tell me that it's not up to you, but to application developers to add support for Wayland, I'll drop kick your sticker-infested laptop, and curse whoever came up with this ridiculous system.
At work I have deprecated many older software components in favor of implementing newer, less painful to maintain ones. Lots of other software engineers use those components, so a big part of it is documentation or training. Rather than harp on the flaws of the old thing, I try to objectively present the lessons learned from it, with an emphasis on all the good aspects of the original that carried forward into the replacement. That generally makes the transition smoother, especially for getting buy-in from the folk that were involved with developing the original implementation. I also wouldn't tell someone not to use an existing component that works, unless I have a concrete recommendation for a better alternative.
That's great, but know that you're optimizing your QoL over the users'. If and when that new shiny thing becomes an objective improvement of the user experience, only then will users share your joy of switching to something easier to maintain. Just don't expect, or even worse, force users to switch just because it makes life easier for you.
> I also wouldn't tell someone not to use an existing component that works, unless I have a concrete recommendation for a better alternative.
Yes, this is key. And precisely what the person in TFA is failing to do. UX >>> DX, always.
Do you have any open bug reports you can point to for "Wayland" not working? Which compositor were you try to use?
This all reminds me so much of ESM Vs cjs. Tech sometimes direly needs a kick in the ass. The situation cannot hold with something old & not good & the new works for most people, but also there's 1% valid problems. The world deserves to be able to eventually move on. The world can be bold & progressive & make changes, it should, and we should expect more of each other about being timely, about doing thr right thing, about making shifts. But we become so trapped by such resistance & unadaption, by staying where we are. Progress is a spirit I think should be Integral to open society & especially open source, but people don't see themselves as part of the bigger thing, they view themselves as merely at ends.
No. Why should a user go hunting for bug reports, or make an effort in making them to begin with? Developers shouldn't expect users to report every issue they run into. Most users will see something not working as expected, and move on to something that does. Fighting with software to make it work, a.k.a. "dealing with it", is a behavior exclusive to technical users and tinkerers. And even this cohort won't be bothered to do so when they just want their machines to work.
> Which compositor were you try to use?
How should I know? All I know is that I selected "Wayland" in my login screen, and ran into a half-dozen UI issues in less than 5 minutes, which I can't be bothered to list now (I posted about it here a few weeks ago if you're interested to know the details). Switching back to Xorg fixed all of them.
With the complexity of a DE like KDE, I, a technical user, wouldn't even know where to look for the culprit. Was it KWin, XWayland or Wayland? When Firefox had UI issues, is that because of Firefox, XWayland or Wayland? Or something with my combination of hardware and drivers? ¯\_(ツ)_/¯
Developers shouldn't expect users to debug their application issues. Users are not testers.
You should know to be able to type `journalctl` which will almost certainly give you a report on what just happened.
Filing a big report feels like a pretty low barrier to me. It's stunning to me to see such a strong rebuttal, expecting developers to mind read & inuit every possible thing a user might run into. To shun openness, to remain closed & curt & expectant, feels more like a consumer relationship, where you don't have any real ability to speak or be present. But do have that ability, you can provide info that would make things better for everyone, and to me that barrier to helping feels extremely low. Not a burden.
I feel like people spend more effort trashing & talking down software in forums like this, than it would have taken to make an even mostly-constructive tiny little data point that could actually help open source, and open society. The author, a NixOS user, almost certainly had plenty of basic technical chops they could have brought to bare within mere seconds to get some first level idea what happened, which they refused. The malice/concentrated-negativity, combined with the lack of effort, to me, invalidate my willingness to take someone seriously.
There are plenty of good valid useful ways to share the view & be part of open society. Just trying to tear out the supports of the building & make it crash down- anger- is not an acceptable enduring view; it's an ok initial reaction, but eventually realer more mature cycles have to begin (unless despair & ruin is the objective)
As a passionate long time supporter, user and developer of OSS, my reaction was purely based on that "deal with it" quote from TFA.
All software, regardless of its licensing terms, should first and foremost deliver a usable product. The expectation that users of free software (in both meanings of the word "free") are somehow under moral obligation to test the software and report issues to the authors is wrong, just as OSS developers aren't under any obligation to implement specific features or provide support. These freedoms go both ways.
But when a developer takes the hostile stance of prioritizing their QoL over their users', as in TFA, that is the attitude that's hurting users, and the adoption of OSS. I appreciate the effort the Asahi team is making, but this is a toxic attitude that shouldn't be part of any project.
Most of the time, I do take the effort to diagnose issues and report them where appropriate. But as you can see detailed below[1], the issues I experienced were so obvious that I assumed they had already been reported. Given the complexity of an environment like KDE, it would have definitely taken me hours of my time to properly diagnose and report these issues. And sometimes, I just want to get work done, without "dealing with" my machine... The reality is that most users feel this way, so this response shouldn't be surprising.
User: I tried it and it didn't work well don't care whose fault it is I just care that it didn't work.
Response: Every developer that flings non-working buggy software into the void is entitled to the free labor of their users who will test said software and write bug reports so that in the fullness of time it may someday be worth using. If you by the mere act of downloading something once don't see yourself as part of that developers community then your opinion is clearly invalid because you are like unto a leper unclean and outside the nourishing light of our lord!
A proper reproduction of the bug, troubleshooting, research into existing bug reports and possible causes both in the immediate and upstream project and writing it up could easily be half an hour not seconds. Meanwhile 17 other things have claims on our time. It is perfectly acceptable to shoot the shit on a forum about how you dropped a non-working project like a chicken nugget with a human finger in it without also opening an issue.
Do you know what people do when they install your software and it crashes on first run. They delete it without comment because they are not indebted to you by being granted to privilege of non-working software and running something else is liable to be a better investment of their time. This isn't abnormal or malicious its absolutely normal. This is also why your software which has a bug that effects 10,000 people gets 5 bug reports. It's not because the other 9,995 people noticed the bug had already been filed its because none of them bothered.
Pretending this is something abhorrent and invalidating is quite frankly a gross failure of analysis committed in service of invalidating a negative perspective that seemed to stick in your craw.
But quickly their position crumbles. No advocates for their position maintain. What they were doing stops working. The user is swept under the sands of time & forgotten.
This putting all the onus on the developers is insufficient. The user doesn't have to advocate for themselves, no. Time happily can ignore their plight. We don't need them. The rest of the open society world is participating onwards.
(these are not necessarily the same issues, but at least similar-sounding ones I found by googling)
• widgets opening in wrong place: https://bugs.kde.org/show_bug.cgi?id=459188
• ignoring icon size configurations: https://gitlab.freedesktop.org/wayland/wayland-protocols/-/i..., closed as wontfix
• on touchscreen menus open wherever you left the cursor: https://bugs.kde.org/show_bug.cgi?id=406025, more info in https://github.com/PeterCxy/evdev-right-click-emulation/issu...
• cursor framerate issues: https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/1827 maybe, it's in the gnome issue tracker but described as wayland-specific, so might also be happening on KDE
• firefox using wrong screen's DPI: https://bugzilla.mozilla.org/show_bug.cgi?id=1632829, resolved by original reporter but last comment is a repro from someone else. also reported in https://github.com/flatpak/flatpak/issues/5300 but closed as "wrong bug tracker"
• fullscreening a video rotates it from portrait to landscape: https://bugzilla.mozilla.org/show_bug.cgi?id=1600962 possibly, https://bugs.launchpad.net/ubuntu/+source/mutter/+bug/190522... possibly
In general wayland seems too buggy for major software like firefox to use it by default, per https://askubuntu.com/questions/1409389/why-doesnt-firefox-r...
Impressive list to hit in only 5 minutes!
Some of those might be the bugs I experienced, but then again, some of them claim to be fixed years ago, and I was running a fully up-to-date NixOS unstable (rolling release) system.
Reporting an issue in an OSS project is a great freedom that users have, but it involves so many steps, especially in these complex environments, that it's often not worth the effort. Even if you manage to figure out which component is misbehaving, to search if there's an already open bug report, and take the time to create one if not, there's a chance of running into hostility and WONTFIX responses.
> Impressive list to hit in only 5 minutes!
All it took was logging in, launching Firefox and playing a video. If these issues were so obvious in 5 minutes, I couldn't imagine how many there would be had I used the system for longer. Switching back to Xorg was the quickest fix to my workflow. Besides, I figured that since these issues were so obvious that they were already known. I was just befuddled as to why they remain so prominent given that Wayland is already 15 years old, and there are mainstream distros offering it as an alternative, or switching altogether to it.
It came in the '90s. Not sure what people keep going on about.
No you won't be able to edit system files on Mac OSX without rebooting into the bios effectively. Deal with it.
Every platform does dumb, annoying stuff. Deal with it.
Still, X.org is terrible and can be reasonably called the Internet Explorer 6 of the linux desktop. At least Wayland will be easier to displace if anyone comes up with something better. This is a step forward.
That’s FUD. The reason around the controversy about screenshots is as simple as gnome wanting to do it right from first principles, that is to also include proper support for taking video recordings/streaming, which is much more complex, while wlroots were looking for a simple “take a picture” mode at first. They couldn’t come to agreement on a common protocol for a time, end of story.
This is due to the bazaar style of development, has no technical reason. Apple can just say they will switch to this new tech in x time, linux can’t really.
We're probably never going to be in a situation where someone can "implement Wayland" and have an acceptable desktop. The real standard to be implemented was written as part of the sway team circa 2018 (eg, https://gitlab.freedesktop.org/wlroots/wlr-protocols/-/commi...).
We have to pick where we put our energies. No one should be expecting to see much effort or caring on X.org. It's not what's happening, going forward.
By all indications, neither is Wayland. It's barely any more usable without running an X server on top today than 10 years ago.
I personally almost never need to run Xwayland. My terminals (kitty, alacrity), web browser, all work great in Wayland with good hidpi scaling. Vscode has good support now too I hear. KiCad seems close to landing good support. I use OBS & it works amazing; Pipewire+Wayland is a modern marvel & such a faster cleaner better system than the slow costly trashfires we had. https://arewewaylandyet.com/ fills most of the rest of the little needs I have.
This isn't a good argument for adopting new technology when the new contender offers nothing of note over the old and the all but unmaintained, broken in fundamental ways option continues to work the way it did a decade or two ago.
It's his playground, it's free, and he can do what he likes with it but if people are using X11 its probably not because they hadn't heard such rants before its because people know their own needs better than the distro developer.
Seriously develop whatever you want to, but telling users what to do is a whole other class of unaware.
Who really cares what this dork thinks...