Fedora Workstation has long defaulted to using GNOME’s Wayland session by default, but it has continued to install the GNOME X.Org session for fallback purposes or those opting to use it instead. But for the Fedora Workstation 41 release later in the year, there is a newly-approved plan to no longer have that GNOME X.Org session installed by default.
Michael Larabel
Expect more and more of the major distributions to abandon X.org completely. For the KDE version of Fedora, X.org will be dropped entirely in Fedora 40 already, so one release earlier.
It definitely makes me nervous. At least X will still be available in Fedora 41, just not installed by default. I’m willing to move to Wayland but I’ve run into application issues each time I’ve tried. I’ll keep trying, I assume they will sort out the issues that impact me sooner than later.
But my biggest problem is, like most people, I’m in hopelessly deep with my window manager, dwm. There is a dwl but not maintained by the same group, although based on the same principals. But I also use a ton of patches and … anyway, it’s a whole thing I’ll have to deal with when I get there.
I’ll miss Xorg once we finally say goodbye, but I will move on when it is feasible for me.
cmdrlinux,
Most users including myself don’t care as long as things work. This later point is extremely important though! It would be foolish to sever X if things are still broken on wayland since that would tend to justify some of the negative images people have of linux. As a long time linux user, I really hope everything with wayland gets fixed (*) before removing X. *Actually fixed*, not just another “it’s fixed I tell ya, wayland’s the bomb” only to find out that things are still broken and/or not available with the desktop or hardware you use. Such things are extremely important if linux + wayland are going to be a robust platform for everyone.
There’s still a lot of work to be done, I moved to Wayland as my main environment recently. MATE Desktop has a lot of things which need to be done to make it fully Wayland compatible. There’s also issues with Wayfire and GTK4, windows are being drawn without borders etc. These are bugs I’ve been reporting to the various projects which have been confirmed. I’ve moved all of my application development to gtk4 to try to be as forward compatible as possible.
I didn’t really care before, but examples like a completely garbled Lyx GUI are further proving my argument that desktop Linux isn’t production ready. Lots of such issues because desktop Linux doesn’t have enough users to test and catch them, and Ubuntu/Red Hat know their main support contracts are servers.
1 app doesn’t work properly makes Linux “not production ready” jeeeez, go into somewhere like Google where 99% of engineering is done on Linux (code is only allowed to live on Linux desktops, not mac or windows — it’s sshfs mounted). Or the countless people who use Linux for their job (like me).
one or two apps may not be working – the example you give Lyx has screenshots from Windows 7 so I think your entire argument is grounded in FUD.
Then you know it’s disingenuous to say Google is doing that because they like Linux, but rather because they’re a huge target that gets immediately attacked by 0 days.
Also, “just one app” is a screw users mentality. Doesn’t matter, what matters there is at least one app (with multiple reports) that Wayland breaks instead of being a proper replacement. (Obviously that means there are a bunch more) That’s stuff you only expect from beta products before release. Absolutely unforgivable in a production environment.
Fine, continue your engineering work, but stop lying to users who rely on GUI apps that Linux is ready, after you’ve stop deluding yourself that breaking one app like it’s still 1995 is normal.
My guy, if a single app breaking on a major OS update is the line in the sand for an OS being “not production ready” and causes you to brand every dev a delusional liar, I’ve got some news for you about literally every single operating system since mankind created them…
It’s not ONE App. It’s GIMP, Mate-Panel, Caja, Blender, OBS Studio, and a bunch of other applications. Some of them even fire up and look on the surface like they’re Wayland compatible, then you start using them and notice some weird behaviours etc. Some of them aren’t being packaged correctly by the distros e.g OBS on Debian is built for X.org, it runs on Wayland but hasn’t got the Wayland video capture plugin installed so you can’t capture videos on Wayland. The list goes on. This is very far from ready for end users.
dark2,
Not everyone has the same requirements, so the situation naturally differs from one person/org to the next. But in principal I have to agree that it’s not the number of apps that are broken that matters so much as whether those apps/use cases are critical to one’s needs. All it takes is one critical breakage to disqualify a solution as production ready.
I don’t shy away from being critical of linux when I think we can do better. But, I am well aware of your agenda in framing breakages as only a problem for linux when the truth is it happens with commercial operating systems too. Sometimes microsoft push beta quality updates to their own customers.
https://www.bleepingcomputer.com/news/microsoft/windows-10-kb4532693-update-bug-reportedly-deletes-user-files/
The team I was on at the time reported these bugs to microsoft when the update was in preview before it got published to the masses, but microsoft ignored the reports from us and others. This went on to be a production issue for many. The costs of a botching an upgrade is high enough that the best practice for many enterprises is to push windows “upgrades” as far out as possible.
I don’t say any of this to deflect from wayland breakages – we should do better. But when you say “stop deluding yourself that breaking one app like it’s still 1995 is normal”, well it comes across as rather biased since it happens on windows more than you probably care to admit.
Both Microsoft and Apple are shipping processors which break more than one app.
Have you investigated the issue personally?
My first guess: line endings or a BOM messed up something in production one time and management banned anything that wasn’t Linux for touching code. LOL
Linux security has kind of been coasting, and it’s fallen behind macOS and Windows. Not to mention the security products which are built to manage endpoints rarely support Linux as an endpoint.
For real. There is a weird issue with Emacs which keeps Spacemacs from working on one of my macOS profiles, so macOS: not ready for production.
>”I didn’t really care before, but examples like a completely garbled Lyx GUI are further proving my argument that desktop Linux isn’t production ready.”
I’ve been using it for 26 years but I agree with you that it’s still not production ready, if by “production ready” we mean that nothing ever breaks or goes wrong. But then again, by that standard Windows will never be production ready, not even close.
I think if you wanted something that was really production ready it would only work in kiosk mode with only one or two programs that the user needs, and the entire thing would be completely locked down with no ability to make any changes to the OS, the window manager or the programs, and no ability to add any additional peripheral devices. The OS and programs and drivers and firmware would all be produced in one immutable image. That kind of ultra locked down experience would be good for corporate worker drones.
Windows has been rock solid since NT 3.51 barring some rare update related screw-ups affecting a very small percentage of Windows users running a very peculiar software selection.
Windows has been production ready on the desktop for over three decades.
I know not installing Windows update is a strict no-go nowadays but in the late 90s when the Internet was metered, I knew and met a lot of companies where Windows servers and workstations ran for months if not years non-stop.
Artem S. Tashkinov,
Sure, it’s production ready, except when it’s not :-/
Breakages happen with every operating system! Anyone working for a large enterprise company knows the importance of testing upgrades before deploying, it can take weeks or even months to be sure they are *actually* production ready with no hidden gremlins. This isn’t just theoretical, some of our customers were facing problems on windows 10 due to schannel protocol changes there. We had to modify the software to keep working after upgrading to windows 10. People often ignore these facts when they want to claim that their preferred operating system just works and is automatically production ready out of the box. Unless your software is certified to work with your OS, it’s just not true that the software you need will always work no questions asked, not for windows, not for macos, not for linux.
You stick to anecdotal evidence trying to portray it as a rule. Sorry, I prefer standard logic and with that I will ignore your failed attempt of an argument. If you’re hell bent on breaking something literally nothing will survive. Not Windows, not Linux, MacOS on the other hand is near impossible to break unless you boot in special mode.
The problem is that Linux is riddled with regressions and bugs every major release of almost anything, most importantly the kernel. Unlike your surface level speculation I am working on that, and I speak from experience. Isolated Windows screw-ups affecting 0.001% of 2 billion of Windows users do not count, period.
Linux on the other hand sports regressions including data loss. I’ve not lost a single file to any of my hundreds of Windows systems in over 30 years of using it. People lost data to Linux multiple times as recently as in 2023: https://lore.kernel.org/stable/20231205122122.dfhhoaswsfscuhc3@quack3/ – yeah it affected Debian “Stable”. So much for QA/QC in Linux.
Trying to gaslight me just doesn’t work no matter how hard you try. I will not descend to your “reality” with finely selected facts. I deal with the world in general. In general we have RHEL and everything else which is a complete cluster f*ck.
I said multiple in 2023 alone, here’s the second one:
https://news.ycombinator.com/item?id=36082788
https://bugzilla.redhat.com/show_bug.cgi?id=2208553
Too lazy to continue searching.
Artem S. Tashkinov,
Your anecdotal arguments are no better than mine. Anyways it’s not just us, most of the time when we encounter upgrade problems we’re not alone…unfortunately things really do break in windows.
https://learn.microsoft.com/en-us/troubleshoot/windows-server/windows-security/ssltls-communication-problems-after-install-kb931125
Artem S. Tashkinov,
You don’t seem to realize what I said, I already know there are linux breakages. Linking to them is simply redundant with what I am saying “Breakages happen with every operating system!”
Microsoft has recently sort of “broken” Windows on several occasions, namely SChannel and the Printing subsystem but 1) it was not regression 2) it was done to enhance security.
Speaking of the article you’ve mentioned: These problems may occur if you updated your Third-party Root Certification Authorities. I have to admit it still looks like a real deal and I guess a number of people were affected but it seemingly was handled gracefully and the pertinent information and remedy were published.
You’re right, “perfect” modern complex OSes which are constantly updated don’t exist but I just see a ton more very serious regressions in Linux in a year than I’ve got in Windows in a decade or two. Are you reading LKML? Kernel bugzilla? AMD/Intel GPU drivers bug trackers? I am. You can’t imagine how depressing it all is even though if you wait long enough it may actually work for you.
Artem S. Tashkinov,
I don’t doubt your experience, but it is obviously that’s the same sort of anecdotal evidence that you criticized me for. I’ve experienced breakages in some for or other with every major version of windows. I acknowledge my experience is unique to me and I might be exposed to more problems because it’s my job to fix it when customers call in with their problems. It might only be 5% of customers who experience it, but for them it might nevertheless be a critical issue. The best way to put it may be to say “your mileage may vary” depending on your specific setup.
Um, not really.
Between Windows 95 and Windows 8 I was constantly on the phone or running around to my friends fixing their computers. With the advent of Windows 8 with built-in Windows Defender things have calmed down so much people started to forget my number.
Windows updates, automatic drivers installation via Windows updates, it’s all gotten effortless and automatic recently barring rare situations when Microsoft pushes its own WU drivers over specific local drivers and by doing so things stop working. But then again it was many years ago that I last had this issue. OEMs have realized having their own specific drivers when generic WU drivers exist is a recipe for disaster and the whole situation has improved to the point no one remembers this issue ever existed.
I don’t think you’re sincere saying that Windows has required as much maintenance in the past decade as it did before. In my experience of dealing with close to a hundred people, the number of support calls has decreased by two orders of magnitude (i.e. 100).
At the same time Linux has also gotten dramatically better recently but still, darn regressions.
Artem S. Tashkinov,
…
I don’t think you’re sincere saying that Windows has required as much maintenance in the past decade as it did before. In my experience of dealing with close to a hundred people, the number of support calls has decreased by two orders of magnitude (i.e. 100).
I am very sincere. It’s your prerogative to not believe someone else’s anecdotal experiences. But then you shouldn’t be asserting that I’m wrong based on your own anecdotal experiences. Don’t be a hypocrite.
There might be something odd going on there.
This week I ran lynx to test some websites, and I didn’t even think about it. I just checked, and yeah, lynx is still working.
ENV: Fedora 39, Gnome, Tilix, Intel + Nvidia Thinkpad.
Perhaps the issue may be that terms like “Production ready” mean what you think they do.
FWIW you’d be surprised that desktop linux is deployed at scale at some pretty big engineering organizations.
I’d like to learn about such orgs more. Germany has tried to embrace Linux on many occasions and they’ve backtracked as many times. China has said a lot about Linux (e.g. openKylin) and then you come check them and they are running Windows everywhere.
With Windows you’ve got AD an GPO, and in Linux you’ve got … devops, e.g. Chef, for that which is not something too many Windows administrators are fond of. It’s so radically different it’s mind numbing.
With Windows you’ve got a ton of extremely useful and easy to setup organization management features which is a total pain in the ass for Linux.
Even as simple as file sharing. In Linux you’ve got what? NFS? It’s absolutely terribly horrible. It’s a pain to set up, it’s finicky as hell, it’s not discoverable, it breaks easily, it may crash your PC. Linux has existed since 1993 or something and there’s still no alternative to Windows File Sharing (CIFS/SMB/whatever you call it).
And people need to collaborate. Of course we have crazy fast unmetered Internet pretty much everywhere nowadays but what if your users need to send one another terabytes of data daily? And open such gigantic files in an instant?
Artem S. Tashkinov,
Samba works on linux and many of us use it fine.
I don’t understand your point at all. It doesn’t even seem tangentially related to what Xanady Asem posted.
Xorg is not going anywhere, it’s here to stay, only now it’s relegated to Wayland output (XWayland).
The hardware Xorg server is the thing that’s being slowly fazed out.
Those pesky details.
So just like XQuartz on Mac then.