116
The cost of maintaining Xorg
(mastodon.social)
From Wikipedia, the free encyclopedia
Linux is a family of open source Unix-like operating systems based on the Linux kernel, an operating system kernel first released on September 17, 1991 by Linus Torvalds. Linux is typically packaged in a Linux distribution (or distro for short).
Distributions include the Linux kernel and supporting system software and libraries, many of which are provided by the GNU Project. Many Linux distributions use the word "Linux" in their name, but the Free Software Foundation uses the name GNU/Linux to emphasize the importance of GNU software, causing some controversy.
Community icon by Alpár-Etele Méder, licensed under CC BY 3.0
Force people to move to Wayland. Everyone that complains about Wayland breaking their setup knows how to install Xorg anyways. But most Wayland problems are software vendors not giving a shit. Make them give a shit by breaking their shit by default on most setups. 10 years was enough time to make your software work on Wayland. If your software doesn't work on Wayland by now, then your risk management is shit.
It's not only software vendors but Wayland itself lacks some crucial features. For me it's auto-type and screen magnification - both are showstoppers for me.
Autotype is already solved - ydotool, wtype and dotool exists (and possibly others as well).
Screen magnification is already present in KDE (
Meta
++
,Meta
+-
to zoom in/out). There's also a magnifier tool (KMag). There may be similar functionalities in other DEs.My issue is the lack of an overall GUI automation tool, ie, like AutoHotkey. X11 had PyAutoGUI, but there's no such AIO equivalent for Wayland yet, and the PyAutoGUI devs don't seem to be interested in Wayland support - it's neither on their road map, nor have they even answered any Wayland questions on their Github page, which is disappointing. But this isn't Wayland's fault, when other tools have shown that automating the GUI is possible, we just need someone to put together a complete package like PyAutoGUI / AHK.
These tools work by creating a virtual keyboard so they don't let you send input to a specific window. The input goes to whatever happens to be focused at the moment. This makes them less reliable than the X11 equivalents and unusable for tasks where you need to guarantee that the right window gets the input.
KMag doesn't work on Wayland.
feel free to check out map2, I'm currently working on version 2 which will do lots of the stuff you need when it's ready, but currently the API might still change and docs are active WIP
still, it can already do most stuff I need it for :)
If that's the case, then stick to Xorg for now. But that doesn't change the fact that it's in your best interest for distros to ship with Wayland out of the box.
Do you want software you use to be compatible with Wayland now or later? If your answer is later, then you have to wait for vendors to catch up, even though Wayland got ~~auto type~~ (already exists) and screen magnification by then. This is why I never understood this push against Wayland. People, your only alternative to Wayland is dead and unmaintained. If you push against Wayland as the default option, you only make your transition in the future more painful than it needs to be.
Also, I think it's still a software vendor problem. If your software can't work with the only desktop protocol with a future, then you must contribute to the protocol to create a way to make it work. If you don't do that, then shit happens, your software breaks, and you had 10 years to contribute to the protocol to fix it. Your risk management was once again exceptional at avoiding doing the necessary work to eliminate a long known risk.
Nobody's pushing "against Wayland". I don't give a shit about Wayland or Xorg. What I care about is having a full-featured, easy to use desktop stack readily available. The "dead" Xorg works perfectly with everything. That's the bar.
When I get a checkbox on the login screen saying "use Wayland" (or when the distro does it by default) I need everything to work. If everything does not work, I do not use it.
The Wayland choice of pushing complexity onto individual software projects by making them all reinvent a hundred wheels, and onto users by making them hunt down a hundred pieces of software to build a wobbly desktop stack sucks. I have no incentive to take part in this particular rat race.
Install Xorg yourself. Don't make it easily accessible to new Linux users. Software vendors will take note and postpone doing any work for as long as possible.
And you obviously care a lot about Wayland and Xorg.
No, it doesn't. And if it does, then it's still insecure by design. When I hear statements like these, I get the urge to publish PoC Linux malware code on GitHub that uses X11 specific features to show just how not fine it is.
Substitute Wayland for X11 here. Both Wayland and X11 are protocols. X11 is such a lackluster protocol that all implementations died, except that Xorg still has users.
New users will drop any distro whose default desktop doesn't work perfectly and with all the features they want. Linux already has a high enough bar competing with Windows, creating additional artificial hurdles is dumb in the extreme.
Security vs convenience has always been a give and take. There's a cutoff point that users will not cross if the software becomes too inconvenient to use, even if it means greater security. The Wayland stack is currently on the bad side of that line and needs to step over if it wants to see mass adoption.
Nobody cares, all they see is the stack, with Wayland leading the point on the bad decisions.
You are projecting. If this were any other piece of software, say, a text editor that works and does everything you need, and someone came and told you "you must use this new one, it's the way forward, but oh it doesn't have all the features you need from a text editor" you would say "thanks but I'll wait until it's ready". But you see no problem in pushing Wayland on people who can't use it?
Please understand that nobody will ever successfuly push through incomplete software. Not on Linux. There's nothing you or anybody can do to convince people that incomplete software is complete and usable when it's not.
Both Wayland and X11 are an artificial hurdle to someone, so at least pick the sane choice with a future.
No, Wayland is doing fine.
Oh no, Wayland isn't X11. It's almost as if Wayland isn't supposed to be 1:1 bug compatible with Xorg.
I don't know about what text editors you use, but my text editor doesn't allow malware to log all keystrokes, tamper with windows of other apps, steal clipboard contents without consent, inject keystrokes into other windows, escalate privileges, and install rootkits that persist OS re-installs using the escalated privileges.
People work on Wayland. Nobody works on Xorg. Alternatives don't exist.
Do you need a refresher about systemd, pulseaudio, etc.? I'm not in the systemd haters camp, but pulseaudio broke regularly for me. Yet every distro included pulseaudio.
I really wanted Wayland to work for me. I just bought a new ASUS laptop (and ASUS has a great Linux compatibility track record, mind you!), 7th Gen Ryzen+Radeon, all AMD. I figured, let's use Wayland on this one.
I installed KDE Neon, updated the kernel (some stuff is broken on the LTS kernel, no big deal, easy fix), switched to the Wayland session, everything was fine...until I opened any chromium-based app. Crashed kwin, killed the session completely, it recovered, but in a new session. Switched to X11, everything works. Maybe if I grabbed a newer mesa from a PPA it would work, but:
And I know, technically KDE could (and afaik, is) implement session management so that doesn't happen. But to my knowledge, literally 0 WMs/DEs can recover the session after a compositor crash currently, and that's a big deal.
If you still want to give Wayland a try, take a look at https://wiki.archlinux.org/title/wayland#Electron. Electron still defaults to X11, even though Electron supports Wayland. It's a bit annoying to set the command line parameters for apps that bundle Electron, but maybe it works for you.
ydotool?
Increase your scaling/decrease resolution.
Not the same as "on demand zooming", which let's one stick with a high, native resolution, but zoom in when required (e.g. websites with small text that can't be zoomed via browser's font size increase; e.g. referencing some UI stuff during UI design, without having to take a screenshot and pasting + zooming it in e.g. GIMP).
What? Strg + Mousewheel, you can even set the option to only zoom text. At least on firefox. No clue what kind of browser you are using which is not capable of that.
Yeah, that browser zoom. And I too used / use Firefox. I'm not saying these kind of sites are common, but nevertheless I've encountered them occasionally. Back then, the most pragmatic workaround was to use desktop zooming of Xfce.
My intention on the previous comment was simply to give some examples of desktop zooming that go beyond the typical accessibility viewpoint (e.g. vision impairment).
By that logic, one could answer that 15 years was enough time to make Wayland work better than it does... but that would be petty and disingenous.
Desktop stacks are very complex. X.org took 30 years to beat that complexity into a usable shape. Wayland pushed most complexity up the stack and still took 15 years to finally put together a protocol of beta quality.
It will take the rest of the stack however long it will take to build on that protocol. Most of the Linux community are volunteers, and Wayland was and still is work in progress. Nobody in their right mind rushes to write software on top of an unstable protocol.
If Wayland is truly ready I think we will see meaningful stack adoption within the next 5 years. But I don't think trying to force developers into it will achieve anything.
As for forcing users that's completely unreasonable. If you're using XFCE on Nvidia you'll have to wait for XFCE to get Wayland support and for Wayland to get Nvidia support. Very few people are willing to change their whole desktop stack or able to buy a new graphics card for the sake of... of what? Bringing about the Year of the Linux Desktop?
You know what's super ironic? I went through this exact thing with X, 25 years ago, when you had to put together a Linux desktop with spit and duct tape.
Wayland promises to be much nicer than X but the way it asks you to put together a working desktop stack yourself is straight out of the '90s.
The more things change the more they stay the same.
Step 1: Install a Wayland compositor of your choice Step 2:
[This comment has been deleted by an automated system]
Read again
You should have used a - instead of a .
This is a terrible recommendation and I hope as few people follow it as possible.
People like you are why Linux has a reputation for always being broken; as soon as we get something that works and is stable, we gotta move to the next broken thing.
The same thing will be said about Wayland in 20 years, if it ever reaches feature-parity with X.
Oh yeah?
That must explain why Xorg always crashes and burns when I don't use the correct combination of desktop environment, compositor and driver version.
Let's not ignore that Xorg doesn't and never has been working for everyone. At least default to the sane option with a future.
User error. If X is too difficult for you to use, maybe you should try Windows.
Moving to an even more broken option isn't a good solution.