Please not, we had the discussion and the change request got withdrawn:
Aoife Moloney amoloney Change Wrangler
This change proposal has been withdrawn by the change owner for Fedora Linux 43, and the discussion topic will now be closed.
To find out more information about the Fedora Changes process, please visit our docs page.
That is the answer I found in ask fedora … i will move this topic into the The Water Cooler tech-talk
2 Likes
Xorg Officially Abandons “Master” Branch for “Main”, Throws Away 2 Years of Code
I did read the headline. If you want to have it in Fedora you will have to make a change request and justify how and why you need that.
The discussion can also be held here in the water-cooler section.
2 Likes
If and when it can replace xorg then sure why not? Wayland is already pretty good though but there are some wm and desktop that don’t support it yet so there is definitely still a reason to use X11 and also for some older programs. Though I didn’t really have a lot of problems with XWayland so far. I don’t really think that when a project stops being updated that it’s necessarily a bad thing. I think old games would be a great example of that. There are many good old programs too that I have used that still works well. I think Xorg is pretty complete now for what it does.
What happend is that there are a lot of buggy patches from one developer that messed up master and added regressions into Xorg.
Main has taken all the patches from the last 2 years that are trusted and of high quality.
Nothing of quality was thrown away.
5 Likes
I personally wouldn’t want Fedora to use it until it’s proven.
1 Like
I can prove it on probably the most bleeding-edge stack 
I dropped-in replaced it from Xorg, my xorg.conf snippets for evdev and synaptics worked still as-is, and I visibly don’t notice a difference 
(works as-well if not better than Xorg and doesn’t expose differences with games/etc)
2 Likes
I used x11 for a long time because of some weird issues with Nvidia. I use wayland now with no issue.
What i’m basically saying is I’m fine with Wayland.
Seeing that green background and the Start button makes me feel younger 
2 Likes
Ignoring whatever acrimony exists between the primary developer of XLibre and X.Org, Freedesktop, and Red Hat more broadly: there is still value in allowing users the option of X11. Fedora can go in full throttle on the Wayland-first ecosystem, and I’m perfectly able to accept that if I have my option within a community spin to use X11, be it Xorg or XLibre.
Wayland works well enough; no one is contesting this. The biggest problem is that there are longstanding functionality gaps within Wayland that cannot be fundamentally reconciled with the scope of the project. Global hotkeys are valuable, but Wayland the protocol does not implement it as it’s outside the scope of the project. Whatever falls outside the scope of the project ultimately gets outsourced to the compositor, which leads to ecosystem fragmentation and conflicting reports of what Wayland does and doesn’t do vs. what the compositor’s on the hook for vs. what neither the compositor nor Wayland itself will touch. If there’s no properly specified means of outlining which monitor is my primary one under Wayland, and this state of affairs existed for at least the last 10 years, something is clearly amiss here.
Hotkeys are the domain of your Wayland compositor (re: Mutter, KWin, Sway, Hyprland, etc). VNC is considered obsoleted by RDP not because RDP is superior but because Wayland itself fundamentally cannot facilitate VNC, and RDP emerged as a “natural” alternative. Wayland setups work great on single monitors, but dual monitors are where the experience becomes variable depending on your compositor. No singular compositor is “flawless” with dual monitors from Mutter, to KWin, to Sway, to Hyprland, basically anything else along those lines. Yet Mutter and Hyprland easily outperform KWin and Sway (at least in my personal testing across multiple computers with AMD hardware) by sheer virtue of not having as many artifacts on screen when moving windows from one monitor to another, killing unresponsive windows, etc.
Also, just putting this out there: Wayland’s more than adequate for gaming if Valve’s willing to go Wayland-first. I get that. But Wayland can only go so far, and the same’s doubly true for Proton regardless of which version you run (re: Valve Proton, GE-Proton). Anything from DX10-onward runs a-okay under a Wayland-first system; I’ve personally tested Dark Souls PTDE, Dark Souls III, Hollow Knight, Persona 4 Golden, the Nocturne HD Remaster, Final Fantasy III (both the 3D remake and the Pixel Remaster), Yu-Gi-Oh! Master Duel, even Warframe. All of that stuff works perfectly, as you’d expect.
But no one with a Steam or GOG library ever sticks with those types of games. There’s a potpourri of games made with DX9 and prior that are popular, that are beloved, that do go on sale regularly, and don’t run perfectly under Wayland.
- The Witcher: Enhanced Edition and The Witcher 2: Assassin of Kings have a launcher that allows you to tweak graphical settings, read the digital manual, and other stuff to that effect. It fires up under Wayland, but does not allow you to physically click “Play.” You need to specifically override the launcher to have it start from
witcher.exeorwitcher2.exe, but then you miss out on changing settings unless you wish to muck about with the INI files. Under X11 (Xorg and XLibre), the launcher works as expected and allows you to fire up the game that way. - Deus Ex GOTY, no matter if it’s the Steam version or the GOG version, if I unplug my second monitor and only use the first one, if I disable FreeSync and have my remaining monitor set at 60Hz, the game plays poorly. It runs. I can definitely run around and fire the GEP Gun into the Subway where NSF members are holding hostages at my leisure, but the framerate is hideously variable. This is a game that runs at 60FPS 1080p on OEM laptops with Intel GPUs; I shouldn’t be experiencing variable FPS on a game from 2001. Under X11 (Xorg or XLibre), the game runs as expected.
- The PC ports of Prince of Persia: Sands of Time trilogy are, admittedly, poorly optimised. Yet workarounds do exist to get around Ubisoft’s treachery. The problem is that this is, again, a DX9 game with strange quirks that exist under Wayland but not X11. Under Wayland, all of these games have this weird sepia haze that artifacts like crazy when an in-game cutscene (not an FMV) happens, not to mention that fog effects basically make half the map disappear. Under X11, the games all work as expected: fire it up, wibbly wobbley timey wimey stuff happens, you die and rewind because you mistimed that jump.
- Devil May Cry HD Collection is an anomaly because it doesn’t run perfectly under Wayland or X11, but Wayland is far and away more problematic because of artifacting during in-game cutscenes and microstutter during intense bouts of combat (insofar as DMC1HD is concerned, anyway). Under X11, DMC HD Collection doesn’t have the microstutter or the artifacting, but some cutscenes are basically skipped wholesale (i.e. the opening cinematic when you fire up DMC1HD and load a save file). I think it’s the convergence of weird Wayland and X11 quirks and the way Valve and GE-Proton handle the sandboxing? Either way, Wayland’s the worse experience between the two.
I’ll go out of my way to admit that it might be a skill issue on my part, hardware configuration, or the perennial assumption that my use case is invalid. To that end? I’m using dual 1080p monitors that barely crack 75Hz, my GPUs are an RX Vega 64 and a 9070 XT that, by all counts, should be first-class citizens on Linux, I’m using the most recent AMDGPU and Mesa stacks, and this is Fedora: kernel updates happen in real time (more or less). This is not hardware that should struggle against games that came out as recently as 10 years ago, and Wayland is very much a complicating factor in all cases. Even the ones I didn’t outline in detail like Star Wars Battlefront, Battlefront II, KOTOR, KOTOR2, or The Force Unleashed Ultimate Sith Edition.
The Xorg team very clearly is not interested in providing feature updates considering how Xorg 7.8 never once materialised. I can live with a Xorg that’s in “maintenance mode” and given only the bare minimum necessary to prevent catastrophic failures; that’s what I’ve been doing for basically the entire time I’ve been using Linux. The issue I personally have is that there’s a self-evident desire among the bulk of Xorg developers to wipe their hands clean of X11 and properly commit to Wayland. I can respect that decision and ultimately live with provided that there’s a meaningful alternative. If there is no more Xorg in the tentative future, XLibre must necessarily exist to offset that absence.
If you want to go all-in on Wayland, fine. If you want to keep Xorg around instead of XLibre to placate people like me, sweet. If you want to be bold and forsake Xorg altogether and allow willing community spin teams to maintain XLibre for the MATE+Compiz, Xfce, Cinnamon, and LXQt spins, my heart would leap with joy. Just don’t take away the option for me to use X11 over Wayland.
We get it: decision paralysis in Linux is a real thing but like… c’mon man, this is the one choice on the decision tree that shouldn’t be pruned.
1 Like
For this to be realistic developers need to step up maintain not only Xorg but all the GUI tool kits and desktop environments X11 features. Do these developers exist?
Yes, yes there are. Even before the Wayland/X11 split, there were community projects that emerged specifically to preserve GUI toolkits and desktop environments.
- Trinity desktop still exists, still remains in continuous development, and they still actively maintain the Qt3 toolkit, if memory serves.
- Sonic-DE exists precisely to preserve X11 functionality within Plasma 6, following the announcement that KDE will drop the X11 session entirely in Plasma 6.8.
- MATE specifically forked GNOME 2.32 to preserve both the desktop environment and the GTK2 toolkit. Nowadays, they migrated to GTK3 wholesale which broke compatibility with themes and the like, but that still indicates a vested interest to adapt GNOME 2.32 to GTK3 instead of doing the Trinity Desktop thing and maintain a hideously outdated toolkit.
- Cinnamon was originally a soft fork, later turned hard fork, of GNOME 3 meant after the original Mint Gnome Shell Extensions proved themselves untenable.
- Xfce4 is roughly two decades old, and they’ve somehow managed to survive GTK2, GTK3, and the proliferation of Wayland throughout the broader ecosystem.
- This one isn’t GUI/X11 related (at least not directly), but the Linux Mint project has gone out of their way to preserve and modernise various APT tools long after Ubuntu and Debian gave up on them (re: gdebi, aptdaemon, synaptic, etc). Lest we forget, Cinnamon itself is the brainchild of the Linux Mint project. Maintaining an entire desktop environment + assorted tooling is one substantial endeavour, but maintaining and modernising package management infrastructure that upstream deprecated is a horse of an entirely different colour. This change was implemented in 2024 following the release of Mint 22.1, JFYI.
This is a non-exhaustive list, but these are the major projects that I can immediately think of that probably won’t fizzle out, stop maintaining their toolkits and desktop environments, and “upgrade” to Wayland+insert such and such toolkit here. I can understand being skeptical of the community’s wherewithal to maintain massive projects with broad scopes. Yet we clearly have precedent for the community alternative to not only survive, but thrive in the face of institutional funding and development.
Tying this all back to Fedora: people like me who want X11 and find Wayland legitimately inadequate to suit their needs and use cases do tangibly exist. They may not be a broad majority, or a slim majority, let alone a majority of any kind, but enough of us exist specifically to justify the existence of Fedora Spins like MATE+Compiz, Xfce4, Cinnamon, and LXQt. All of these environments are not Wayland-first. At best, they’ve had experimental sessions since like 2015. This doesn’t mean that these environments are hostile to Wayland; Clem even said that stepping up development of the Wayland session for Cinnamon and MATE is on the project roadmap. Yet he said that in the same breath as saying that they want to give users the option to have an X11 session.
I’m not unsympathetic to the burden placed upon software developers, but I sincerely think we’re throwing the proverbial baby out with the bathwater by forsaking X11 entirely in favour of Wayland despite its decade plus long streak of issues that cannot be resolved (i.e. NOTABUG, WONTFIX, outside project scope, the finger pointing between Wayland to the compositor to the application developer with no side wanting to compromise). If there is a substantial, vested interest within the community to maintain these older technologies, I would think that it would behoove the Fedora Project to seriously consider either keeping Xorg as a legacy option officially or allowing the community spin teams to utilise XLibre.
This just shows that the gamer-community is expecting to much from the Linux community. If they really want to be downwards compatible, then there has to come more effort and also money from their side.
About your wish of a community based Spin which still lets you use x11 on a Fedora Image, feel free to create a WG or a SIG to create one. Or as alternative supporting the gaming WG which tries to keep such things alive.
Time is passing and new technologies getting on the market every day. The focus away from x11 is not just the fault of Fedora pushing Wayland. It is simple a consequence of time passing and Devs moving away from x11 to continue feeding their family. As this get harder everyday too, a Dev not want’s to code in his free time just to keep running some old hardware alias software.
When WinXp/Win7/Win10 gets retired and looses support we have to accept this. So why we not finally also accept that software which has been created to run on such systems, has to be retired to?
The fact that Open Source Software, inkl. older code base, still is available not means that the “free work” on this code has to continue being “free”.
About having Xlibre in Fedora is almost the same, as asking Fedora to forget the COC and do as it never would have existed. The mindset of the “main Dev of Xlibre” is as in-compatible with Fedora, as Wayland is with Win95.
Erm… the “gamer-community” is not a singular, homogeneous entity that demands, demands, and demands without ever giving back. Unix emerged in 1969 because Ken Thompson and Dennis Ritchie wanted to port a Multics game over to a PDP-7. Emulators are the domain that best performs within a FOSS ecosystem, and especially Linux (for decades even). Games are a fundamental part of the broader FOSS ecosystem, and saying that the “gamer-community is expecting to much” is kinda incoherent. We have projects of substantial quality like SuperTuxCart (an original, actively developed FOSS game with years of history), Azahar, melonDS, PPSSPP, DuckStation, mGBA, Snes9x (although technically it’s not FOSS because of the noncommercial distribution clause), need I go on? Why shouldn’t our expectations be high, if we already have precedent for high quality, actively maintained software?
Gaming on Linux didn’t emerge in a vacuum; there were active attempts to do Linux-native gaming as soon as NVIDIA and ATI released their binary drivers way back in the Fedora Core days. It only gained substantial traction thanks to Valve and Codeweavers collaborating extensively on Wine and Proton, making Windows gaming legitimately feasible under Linux. Yet even in those days, the assumption was X-first because that’s what all the OEMs and developers were targeting. Valve and Codeweavers alone can’t do it all, nor should we expect them to. Yet even the work they’ve done to this point is substantial enough to not dismiss wholesale.
Thank you kindly for the suggestion; I must admit that I’m fairly new to participating in the Fedora community more broadly. I’ve always been more of a silent user than an active participant and I’m earnestly trying to change that.
You’re not wrong that new technologies will naturally emerge, and that developers inevitably change their focus for whatever reason. What I take exception to is this zero-sum mentality regarding developer wherewithal. If the current team of developers behind Xorg genuinely don’t want to keep doing it anymore, it’s their decision to abandon the project wholesale and move on with their lives. That is 100% true. Yet that doesn’t preclude any possibility of the community picking up the slack.
That’s precisely what happened with MATE (started by some Arch Linux user in Argentina back when GNOME 2.32 was depricated wholesale) and Cinnamon (started by Linux Mint after MGSE proved itself untenable long-term). Both projects are still, ostensibly, within the domain of the community. We have strong precedent for sustained community stewardship of paradigms that paid developers cannot or will not sustain.
At the risk of repeating myself here: Linux Mint’s still a community-run project and they’re actively maintaining huge chunks of APT code that Ubuntu and Debian abandoned. They’re not just maintaining and contributing to desktop environments; they’re also actively maintaining infrastructure that upstream abandoned. This is a strong, promising indicator of the community’s willingness to maintain infrastructure that upstream abandons. Yes, it’s not a universal indicator because some stuff is more difficult to maintain and modernise than others. But the precedent still exists, and that definitely counts for something.
Forgive me for being so forward, but you’re bringing up a bad example here. This is neither the time, nor the place, to discuss Microsoft’s broader failings within the Windows ecosystem over time. Yet Windows NT is not the example you wanna bring up for “software which has been created to run on such systems, has to be retired to.”
Even if Windows XP, Windows 7, and Windows 10 don’t exist anymore, Windows NT’s own API and ABI is robust beyond compare to anything we have in Linux. On Windows 11, I’m more than capable of downloading Witcher 2 (a game released in 2012 between the transition of Windows 7 and Windows 8) off Steam or GOG, installing it, and then playing it without issue. If there’s some glitchiness afoot, I can just right-click the witcher2.exe file, hit “properties,” and then set it to run in XP, Vista, 7, or 8 compatibility modes.
The only “hard break” that Windows ever experienced was DOS compatibility, and that was specifically due to the transition from Windows 9x to Windows NT that occurred during 2000/XP. Even then, Windows NT existed parallel to Windows 9x long before Windows 2000 and Windows XP. Their legacy application support is huge thanks to Win32/Win64’s extensive scope, even if the operating system itself goes EOL. That’s why you can play Deus Ex GOTY (or basically any Unreal Engine 1 game) on Windows 11 without fail.
To wit: Witcher 2 has a Linux-native version that I physically cannot play, even on Linux Mint MATE, because it explicitly requires GTK2 which has long since been deprecated. Yet the Windows version runs perfectly fine on Linux (Mint and Fedora). I’m personally of the belief that the broader Linux ecosystem has a heinously poor capacity to maintain long-term application binary support because it’s just taken for granted that a developer will move onto the next thing that makes their life easier but leaves the user holding the bag. But that’s just my opinion.
You’re 100% correct on that front, which is why all the projects I rattled off (re: MATE, Cinnamon, Xfce, LXQt, even Trinity) all hinge on community sponsorships, grants, and small dollar donations. This is fundamentally true for FOSS across the board (re: Thunderbird actively soliciting donations on first run after a fresh install). What I take exception to is the notion that the situation is a stark binary that it’s between maximally inefficient developer input while the software remains gratis vs maximally inefficient corporate funding while charging a premium.
Commercial funding is obviously ideal, but it’s never consistent because marketplace demand and institutional dynamics will always affect the company in question’s ability to keep participating. Even so, the community’s willingness to step in, coupled with active sponsorships from vested interest groups, is meaningful in itself. Linux Mint has no major commercial sponsors, yet the project’s maintained serious viability precisely because of the grants and small dollar donations they receive, plus whatever niche sponsorships they can muster. This is enough to actively fund the infrastructure necessary to host Linux Mint itself, the forums, ISO servers, the broader tooling to facilitate active maintenance of Cinnamon, MATE, and Xfce spins, and contributing upstream to the development of MATE and Cinnamon. And again: they have room to spare for maintaining deprecated APT tooling that upstream abandoned.
Allow me to reframe the proposition: if Xorg must be removed, there must necessarily be a community-stewarded alternative to Xorg. Regardless of whatever acrimony exists between XLibre’s primary developer and the broader Fedora/Red Hat ecosystem, they’re the ones who have a community-maintained X Window System. IF, by chance, there’s an alternative Xorg implementation that’s also community-maintained but isn’t run by someone like XLibre’s primary developer, then by all means, use that. Unfortunately, it doesn’t exist. The easier solution would be to keep Xorg around if XLibre is a non-starter, but that really depends on how badly upstream wants to get rid of Xorg and the community’s desire for X11 in one form or the other.
It’s a dead project.
Hosting is provided by the partners.
Then there are corporate sponsors
Then there are user contributions.
There are also unlisted bitcoin and other methods.
There is a lot of work going on to support old games on Linux.
For example there are implementations of old Microsoft DX3D in vulkan being actively worked.
I only see this support getting better and better in the near to mid future.
There are stories on phronix.com about this all the time. A random pick is https://www.phoronix.com/news/Vulkan-1.4.344-Released
1 Like
I wouldn’t have used Linux without gaming, and don’t see a real-world benefit to using a years-long worse experience just because it sounds better on-paper.
I can use Linux where the two-top DEs dictate X’s removal, use FreeBSD that has unbiased support for Xorg, XLibre, and Wayland, or use Windows that already does the same stuff as *nix with optimal input/graphics handling. AMD used to use Never Settle for gaming marketing; why settle for Wayland?
Assuming it’s possible (not sure how with libinput and glamor), I want technical reasoning that has Wayland at lower-latency and higher performance than X (Intel DDX, evdev input, Intel tearfree/vsync/swapbuffers disabled, cairo GTK).
Even though Beat Saber PCVR does fine for a quick test on GNOME/Wayland, I know it’d be technically better on X and Windows.
Isn’t that one of the advantages of the Wayland architecture?
Assumeing that compositor steps out of the way for full screen gaming:
For X11 it’s APP → Xserver → Kernel
For Wayland it’s APP → Kernel
That trip to and from the Xserver is a latency issue.
Even when a compositor is involved the effiecient transfer of Windows content the Compositor in Wayland (only a handle is share, the data is in the GPU memory) is also a latency win.