Wayland is a piece of chemo

Wayland has changed a lot within a short time, since the protocol logjam was finally freed up and more development has occurred, so I think it's time to begin a Wayland thread from a more positive perspective to talk about the protocols and programs, and related topics. A large amount of programs have been ported as well, including GIMP, so it may be time to try out a compositor and see what works. FAQ: General answers: https://wayland.freedesktop.org/faq.html >I need some programs! This list isn't too bad, although some programs may vary in quality: https://github.com/rcalixte/awesome-wayland >But what about my <input problem here>? Typically, you will end up using libinput and xkbcommon in practice. For most users, it's basically setting up XKB and libinput settings they were already using in the compositor's config file. Some compositors are poorly written or incomplete and don't expose all their options, use a different one, send in a patch, or write your own. The rest of the issues are more of an accessibility issue and not an actual input issue. Accessibility isn't really something I have to deal with, but perhaps someone else knows the situation. >Where are the protocols?! https://wayland.app/protocols/ You can check which ones are supported using Waycheck https://gitlab.freedesktop.org/serebit/waycheck or with wayland-info if you prefer the CLI. >I want bitmap fonts in my terminal! Simply put, use foot: https://codeberg.org/dnkl/foot/ >Wheres my xdotool?! This isn't well supported inherently, but the situation is slowly improving. Promising new tool, better than ydotool: https://git.sr.ht/~brocellous/wlrctl Older keyboard only tool: https://github.com/atx/wtype There's also tools for using the mouse pointer without a mouse interactively on a keyboard. >But I have legacy applications that will never be ported to Wayland, or programs that are extremely buggy as native Wayland programs! For X11 clients that don't need to control the root window you can use XWayland, as well as these variations/scripts: https://github.com/Supreeeme/xwayland-satellite https://gitlab.freedesktop.org/ofourdan/xwayland-run https://github.com/ValveSoftware/gamescope For running X11 programs that need to control the root window, you can use XWayland in rootful mode, with the caveat that not every feature may work as it's like Xnesting: https://ofourdan.blogspot.com/2023/10/xwayland-rootful-part1.html https://ofourdan.blogspot.com/2023/11/xwayland-rootful-part-2.html https://bennett.dev/rootful-xwayland/ https://github.com/YaLTeR/niri/wiki/Xwayland#directly-running-xwayland-in-rootful-mode There's also an experimental project called wayback: https://github.com/kaniini/wayback >But what about network transparency? It has to be done by a separate program, like waypipe, or a RDP or VNC server. Typically this is pretty complex stuff and not the kind of code you want running in your very privileged compositor, nor does it make any sense for your compositor to be directly connected to the Internet anymore. >I want to switch to a different compositor from a compositor (weird old X11 feature) You can run nested Wayland compositors (done it many times) or theoretically could run multiple compositors at once on different VTs, but you'll probably need a seat manager of some sort at least (seatd, elogind, systemd-logind) plus adequate system resources. I haven't tested the latter use case. >But I have a NVIDIA card! NVIDIA's support of Wayland has been nonexistent to spotty, although some newer cards may work. The only real hope for older cards is if NVIDIA suddenly hits their collective heads and decide to open up the kernel and userspace drivers for those old GPUs. This situation affects X11 as well with X.Org master and XLibre being left in the lurch lately with NVIDIA, leading to NVIDIA GPU support being broken. There's nothing really in the protocol that prevents them from having drivers that work with Wayland, they simply choose not to spend developer time on this and wish to also keep their userspace drivers proprietary so far, which makes this whole thing harder. NVIDIA is a rich company that pretty much wants to dictate terms to GNU/Linux and BSD, but don't put in as much work as Intel and AMD on the platforms. If you're really on a budget and want a working GPU on Wayland, I recommend you save up for a cheap Intel or modern enough AMD card (amdgpu driver) as they support the platform fully and have an actual stake in it, or use an iGPU alternatively. Compute can work on certain AMD cards, you have to go digging as to which will work with an env var, unfortunately AMD isn't the best at supporting compute on consumer cards in general. That being said some NVIDIA cards work apparently, however especially old ones may not work at all or may only work on GNOME. There are some caveats to using a NVIDIA card even if it "works", you may run into problems. >But my window manager isn't ported! Is it really? Check the lists of compositors, there are many analogues of old styles of window managers as well as new ideas. Sway = i3, river = dwm-like, dwl = dwm, Hyprland = kind of like bspwm but really its own thing, and so on. But if it really isn't there yet and you have the skills, you could attempt to make a new compositor. Libraries like Louvre, wlroots, Smithay, etc are out there to help you. https://www.gilesorr.com/wm/table.html https://github.com/rcalixte/awesome-wayland?tab=readme-ov-file#compositors https://wiki.archlinux.org/title/Wayland#Compositors

>just buy a new graphics card and write your own compositor! That is not a very convincing argument. I will switch to Wayland if/when Slackware switches. I am no longer convinced Wayland will succeed in replacing X11, best it will get is being the default in some distros who all offer X as well.

>Wayland has changed a lot within a short time, since the protocol logjam was finally freed up and more development has occurred, so I think it's time to begin a Wayland thread from a more positive perspective please remind me what issues it solved that couldn't be solved by making X12?

the arcan dev says wayland is terribly designed i don't really care and have been using it since my distro migrated it's been buggy as fuarrrk and still is somewhat

Do people even use it for anything other than hyprland/sway ricing? Most people using GNOME/KDE are still on Xorg from what I've seen.

>>87291 X can't move forward without breaking backwards compatibility in major ways, it is part of why we have been at version 11 for just shy of 40 years now. Only options are write more extensions for X11 or start fresh and have it just run an Xserver to get the backwards comparability as Wayland does. The world and technology has changed a fair amount since the 80s and the X protocol is showing its age.

>>87297 showing its age how exactly?

saw this (image) and my opinion dropped from neutral to its probably a mess if that kind of crazy participates in it.

>>87302 imagine open source software being driven by community drama, politics etc.

>>87291 There are existing compositors if you aren't picky. I'm assuming your GPU is just too old for Wayland support, so you could wait until your distro discontinues X.Org or yes, you could sell it and get a GPU that works with Wayland. I don't think most maintainers are salivating over maintaining and supporting a X.Org package indefinitely, most distros were quick to drop 32 bit binaries when their userbase no longer had enough of the hardware. Since NVIDIA is supporting newer cards on Wayland to some extent, all distro maintainers need to do is wait for the rest of the too old cards to EOL and then drop the X.Org package on their next release after that. >>87292 X12 is Wayland's former working name, so it solves all issues solved by making X12. >>87293 There's a difference between an ideal hobby project and something you can use now or implement programs for in a stable way right now. I'd say the biggest mistake with Wayland was not going with explicit synchronization from the start but it's been merged already. >>87296 Generally it's a combination of DEs porting their environments to Wayland and then distros discontinuing X11 support for those DEs, or just dropping X.Org altogether. GNOME is dropping X support completely and KDE is already a majority Wayland userbase, so they'll probably be next. XFCE is in the process of porting to Wayland although a lot of the work is done already. As for the usecases, it varies from automotive UIs, phone UIs, kiosks, and any sort of tiler or stacking window management. >>87302 GNOME developers have always been in a walled garden long before they moved to Wayland, it doesn't really surprise me that their political beliefs are extreme. There's nobody to moderate your beliefs when you're in that sort of echo chamber. They're not the only stakeholder in Wayland protocol development so they don't get to make all the decisions anyway.

>>87314 >They're not the only stakeholder in Wayland protocol development so they don't get to make all the decisions anyway. Any influence is sufficient to cause damage. Weakest link can kill the entire project. I stopped using a lot of software that was 99% excellent work but a critical bug made it unusable.

>>87314 >I'm assuming And you assume wrong, I was just pointing out your silly argument. I try out Wayland once or twice a year and have done so since it got added to Slackware current, bugs generally drive me back to X within a few days. "Most distros" is fairly meaningless and supporting an entire architecture is a very different thing requiring far more resources. Wayland will most likely become the standard when it comes to the linux desktop, but at this time I don't see it killing X anytime soon, Haiku will probably hit version 1.0 before that happens. Wayland was never X12, some people called it that but Xorg was never on board with Wayland being X12 and they are the ones who get to make that call.

>87321 arch was pwned already when they adopted systemd, it was a given they would also adopt wayland.

>>87321 Linux is far more than the linux desktop, all those embedded systems, industrial controls, POS terminals, etc etc etc are not going to switch to Wayland until it is time to replace those systems themselves and/or what they control. This is the entire reason for X taking backwards comparability to the extreme it did, it offers certain assurances which Wayland breaks or might break and the potential cost of seeing if Wayland breaks something is often not worth it. Baring some major setback, Haiku will hit 1.0 before then. Minimum of a decade before X will be dead and even then we will still have many legacy systems still running X. Has even a minor distro announced dropping X yet? Has any DE dropped X support yet? I know Gnome announced they were going to a couple years ago but last I heard it sounded like it was not going to happen until 50 comes out and even than with the caveat of getting the Wayland version up to spec? KDE has only said that X support will eventually be dropped and implied Plasma 7 would be Wayland only but did not state it.

>>87320 >And you assume wrong Then you shouldn't bother talking about it if it isn't your actual problem. People who have compatible NVIDIA GPUs have some sort of functionality on Wayland, the rest don't and have some GPU that is 5 years+ old. Inevitably this will no longer be an issue given enough time as they'll get too old to be relevant. >Wayland was never X12, some people called it that but Xorg was never on board with Wayland being X12 and they are the ones who get to make that call. They are the one and the same people who called Wayland X12 and maintained and developed X.Org. This isn't up for debate no matter how hard you want it to be. >>87324 X was created for a very different world of graphics and networking, when computing was assumed to not be personal and that the layout of networking would be very different than it is now. It was not created for any of those things you list, it was invented as a part of Project Athena, working with other sibling systems like the Andrew filesystem. Those days are long over, the sibling software is practically used by no one, the old weird use cases are dead, and the graphics hardware has changed considerably to the point of the X model being completely ineffective, it just gets in the way at this point. The X11 protocol will probably be around forever for legacy reasons and on legacy platforms, longer than Haiku can count to 1.0, but full on traditional X servers in the driver's seat with all the extras aren't going to be relevant going forward on anything greenfield. What's going to be in the driver's seat is Wayland. >Has even a minor distro announced dropping X yet? RHEL 10: https://www.redhat.com/en/blog/rhel-10-plans-wayland-and-xorg-server It's a gradual phase out, defaulting to Wayland sessions for some DEs, removing X.Org support for others. Nobody in the open source community has the resources to maintain both Wayland and X11 versions of big DEs indefinitely, so inevitably they'll follow suit at some point when their X user base dissipates, but that will take longer. >KDE has only said that X support will eventually be dropped and implied Plasma 7 would be Wayland only but did not state it. You're either unaware or omitting that they have a majority of users on Wayland sessions. Whenever the number goes high enough from distros defaulting to Wayland sessions they can drop it for their next major version. >I know Gnome announced they were going to a couple years ago but last I heard it sounded like it was not going to happen until 50 comes out Ubuntu wants to do it before GNOME deprecates it: https://discourse.ubuntu.com/t/ubuntu-25-10-drops-support-for-gnome-on-xorg/62538 It sounds like 49 or 50, but major distros have to shift into the Wayland only session first before that happens. Also it's confirmed the xorg-server package is moving out of main to universe, so the phaseout is beginning on Ubuntu: https://discourse.ubuntu.com/t/ubuntu-25-10-drops-support-for-gnome-on-xorg/62538/2 Then it trickles down to spinoffs, Kubuntu defaulting to Wayland session for 25.10, tentatively 26.04 for full removal of the X11 session: https://bugs.launchpad.net/ubuntu/+source/ubuntu-meta/+bug/2114015/comments/14 >>87319 They have been there for the entire time X.Org was the only option so they influenced it, in fact GNOME is from the XFree86 days so it's older than the X.Org fork. Anywhere the Linux desktop is, you're basically stuck with these guys.

>>87324 give up. you're too few and dumb to compete with the Corporation. as that paid rhel subject directly above my post explains, the phase out is inevitable. they worked long and hard to take over all major distributions. the only truly independent distros are Slackware and small ass ones like Void, which are not relevant these days. maybe gentoo too, but it 90% consists of troons and their asses belong to the Corporation, so gentoo is a very frail rallying point that will be broken easily. you have no hope, really. i strongly advice you to archive good old distros, like debian 10-12, gentoo (including their distfiles subdomain), their sources, everything. also kernel 4.x, 5.x, 6.x. the open source thing is very nearly done for, it is not worth anyone's time any longer. i remember just 7-10 years ago it was different but slowly jews won. maybe chinese guys kick in and save us but i wonder who's gonna review their software. Linux desktop meme has always, since very inception, been a propaganda term to enshittificate the ecosystem and bring it down to the lowest common denominator. now enjoy it.

>>87334 Red Hat's Wayland and SystemD are for enshittifying Linux. Technically open, but like the web standard it's controlled by the Corporation. Make it complex for no reason, citing security bullshit, so only the Corporation's implementation works. Bribe distro maintainers to use the official implementation by default. Users stay on default, that's how monopolies stay alive. Add hostile features like the default DNS that Red Hat keeps shoehorning into SystemD, add vulnerabilities like OpenSSH xz, make switching impossible by making userspace depend on Wayland and SystemD.

>>87334 they dont need to "compete", linux won in many spaces against microsoft. A small group can maintain a better linux while corpa creates a linux windows frankenstein

it is shameful that wltools, screencap protocols etc are essentially mandatory for every reasonable use case for wayland why is it better that your protocol require such a spiderweb of additional protocols and shims which must support you but are themselves unsupported, unfunded, undermaintained and insecure and man it seems like a lot of the people who maintain these necessary and essential external module components of any functional wayland system get some sort of brain damage about it like the quiet part of their wayland advocacy is "wayland is perfect because it forces you to depend on me"

>>87297 >Only options are write more extensions for X11 or start fresh How is that different from Wayland? The only way to solve it's issues is extensions and mediocre hacks through PipeWire or dbus. >>87331 >They are the one and the same people who called Wayland X12 and maintained and developed X.Org Nope. Most of Xorg developers left the project a long, long time ago. Very few of the Wayland developers were Xorg developers and even fewer still have remained in the 18 years of development it has. What you have is basically a bunch of unrelated devs claiming to be Xorg devs to give themselves a semblance of legitimacy but even that doesn't work because that was not the only X implementation.

>>87342 >why is it better that your protocol require such a spiderweb of additional protocols and shims which must support you but are themselves unsupported, unfunded, undermaintained and insecure You have accidently described X11, except the protocols are extensions, and there are a lot of shims... and the whole thing is unfunded, undermaintained, and insecure. To answer your question, if a protocol is insecure it can be replaced by a v2 protocol (has already happened), and some upstream protocols have already replaced wlr protocol behavior in practice (also happened already). It's not set in stone forever. >>87353 >Very few of the Wayland developers were Xorg developers Kristian Høgsberg and Daniel Stone were both working on X.Org, others from X.Org were supportive of the new protocol. There's not really anyone else who would be initially interested. As for today, they'd be a failure of a project if they couldn't get new developers who aren't former X.Org developers. The point is that when you exhume the name "X12", thats just Wayland's old name, and you bring up the era when it was some X.Org developers trying to make a new protocol to replace X. >but even that doesn't work because that was not the only X implementation. What one is actually actively developed that isn't just another X.Org fork? And why does it even matter if there's some other, unknown, unused X implementation bitrotting somewhere? >What you have is basically a bunch of unrelated devs claiming to be Xorg devs to give themselves a semblance of legitimacy but even that doesn't work because that was not the only X implementation. They work on a display server started by X.Org devs, and some work on a fork of X.Org meant for running nested inside a Wayland compositor, thats a very similar job to the point of this almost being semantics. You can't really expect there to be additional X.Org devs joining anymore after so much time, that is an endangered species.

>>87354 >You have accidently described X11, except the protocols are extensions, and there are a lot of shims... and the whole thing is unfunded, undermaintained, and insecure. lmao. prove >There's not really anyone else who would be initially interested. that's right, all we want is to be left alone. i wish you fuarrrked off instead of taking over every distro. but you're too greedy for backdoors and data aren't you? can't allow any independent soykaf to exist, can you?

I have been using Hyprland for a month now, and besides smooth animations, there is nothing this display server does better for me than xorg did. Xorg has much lower latency, all of my x11 applications run better and it does everything i need it to. Also games on Hyprland have weird microstuttering and input lag, so despite it supporting VRR with multiple monitors, gaming is still something I would rather do on xorg.

>>87289 Wayland development is just waylandisms and stupid design decisions being undone. Almost everything that sets wayland apart from any other modern display server protocol is retarded. They managed to got everything wrong. The only thing that wayland ever had going for it is that was not designed to be used on mainframe terminals in the 80s, which is the bare minimum that could be expected from it. I also find it curious that design decisions that Wayland fans used to defend to death are never mentioned again ones they are undone. I bet that if window placement gets added, not a single soul will ever complain about the existence of it. But until then it is actually good thing that Wayland doesn't have this.

Wayland is pretty good in 2025.

>>87379 who is the president of US in your parallel dimension?

>>87354 >You have accidently described X11, I have intentionally described X11. Now tell me why deliberately and intentionally reproducing this core unresolvable problem in wayland was good for me and was a step forward for desktop design philosophy and the right way to spend a decade and a half.

>>87383 >Now tell me why deliberately and intentionally reproducing this core unresolvable problem in wayland was good for me No, because I have no idea what your issues are and you argue from false pretenses. >and was a step forward for desktop design philosophy and the right way to spend a decade and a half. You could have practically anything modern and come out better than X11 in the long run. It really doesn't matter because the last time X was impressive technically was the early 90s. The thing is these people did this for free, gave it out for free, came with up a compatibility plan, nobody came up with anything better in the meantime, certainly not you, and now it's time to start moving people over because X.Org session is a big problem from a maintenance and security standpoint. They don't need to really convince some sort of Unix-like dinosaur like you, they just need to get the big distros to drop X.Org session. Eventually that trickles down to DEs dropping it, new proprietary software developers dropping it, and so on. >>87372 I figure window placement of some sort will finally make in it, essentially it already has in session management and relative placement, but this gives compositor implementers the flexibility to not have it. With X11 you bring in all the baggage all the time. >>87363 >lmao. prove Okay. It's full of extensions, hacks, and shims for it to even work in the modern desktop sense. For example, X used to not be capable of contiguous multi monitor support, the hack was to put it all in one addressable space (a Screen in X terminology), you may notice this if you use very old X programs, they will appear in the middle of your monitors between the bezels. It's obviously unfunded, it's obviously undermaintained (that's why you're here whining), and everyone knows it is insecure already (X snooping, has a TCP server in it, ancient code which turns into a new vulnerability at least once per year). There's nothing left to prove. >that's right, all we want is to be left alone. i wish you fuarrrked off instead of taking over every distro. but you're too greedy for backdoors and data aren't you? can't allow any independent soykaf to exist, can you? It would be trivially easy to put a backdoor in X.Org, hardly anyone understands the whole codebase, and by forking it to something like XBased they could target a subgroup if they wished, plus X has input snooping guaranteed regardless of VMs or containers/jails as a bonus. In the case of Wayland, since everyone must make their own implementation outside of some common libraries, you'd need to backdoor all of that code governed by different people or make a protocol vulnerable to something which is harder to cover up. A backdoor in X.Org could last years, if not decades. >>87368 >I have been using Hyprland for a month now, and besides smooth animations, there is nothing this display server does better for me than xorg did. Xorg has much lower latency, all of my x11 applications run better and it does everything i need it to. >Also games on Hyprland have weird microstuttering and input lag, so despite it supporting VRR with multiple monitors, gaming is still something I would rather do on xorg. Hyprland implemented tearing control which you can enable per window or generally, which basically allows immediate mode like X. Some compositors simply struggle with games running in Proton, you can use Gamescope for them. Gamescope can also work for Wayland native games with the --expose-wayland flag. Gamescope is what the Steam Deck uses for games so it's pretty much what any game running in Proton would be tested on. If you have input lag still, you may have some sort of issue with how your input is set up on Wayland versus X.Org, or an issue at a lower layer. I have no microstutters or input lag in native games or Proton here.

>>87391 >It's full of extensions, hacks, and shims for it to even work in the modern desktop sense i thought we were talking about x11

>You could have practically anything modern and come out better than X11 in the long run. Yet wayland managed to suck so hard its still worse somehow.

Ultimately what people here don't understand is that it comes down to supporting hardware. Nobody wants to work on xf86-video drivers, they might not even be legally able to if it's some proprietary slop. Newer hardware, such as on the ARM Macbooks, makes no attempt to support the X driver model or its riveting 90s ideas about graphics. Projects don't want to deal with unsolvable screen tearing tickets from users who have no built in tear free solution in their drivers or other strange issues emanating from X that don't exist on Android, iOS, MacOS, or Windows. Continuing on course with a native X session course leads to maintaining xf86-video drivers, maintaining and developing a separate compositor, X remoting that is pretty much objectively worse than its competition at this point, having a big X snooping hole in every project that uses containers or VMs in concert with a display server, and a very dated graphics model. Now it certainly hasn't been the smoothest ride for Wayland, but you're pretty much guaranteed it will turn out like the systemd transition, even the people who dislike systemd know it runs on most GNU/Linux systems now. In the end, very few people stuck with the dinosaur SysV init setup or experiments like Upstart, they moved to systemd, runit, OpenRC, s6, and so on. Ultimately I think very few people will be on native X session in 10 years as Wayland stabilizes. I think you should probably be grateful the Linux desktop is growing up, going from Pulseaudio to Pipewire/Wireplumber, X to Wayland, SysV to systemd. I don't think you can outrun it, in any case. Genuinely better outcome for the majority of users in the end. I can imagine the contrarians will ultimately be forced to use a rootful XWayland session at some point when native X session isn't supported by anyone anymore, sort of the equivalent of running OpenRC. Even major critics of Wayland are admitting it has improved: https://dudemanguy.github.io/blog/posts/2025-02-03-wayland-xorg-2/wayland-xorg-2.html Anyway, regardless of the growing pains, this is a Wayland thread, it isn't about a bunch of people who can't get past fizzbuzz trying to "save" X.Org (if you balk at coding a Wayland compositor with a huge pre-written library doing most of the heavy lifting & docs and XML protocols telling you this-is-the-thing... yeah good luck fixing X). It isn't a NVIDIA thread either. Try to stay on topic.

>>87483 >Try to stay on topic. which topic exactly? i installed linux to use linux, and 10 years ago it was indeed linux, but now it's more like macroidows. wtf? i just want a system that isn't owned by china/america please take you systemd elsewhere

Honest question, does Wayland yet support controlled sub-windows? X11 is the only display server on Linux that things like Audacity and Renoise work on. In fact, last I tried, Audacity crashes when starting through xwayland and Renoise was so unusable I had to move back to X11 for it as well. >>87391 I'm tired of this security theater. Most to all wayland compositors rely on wlroots, so there's your single point of failure. Not to mention that you can also add a backdoor to libc, or any kernel module, or the kernel itself. All things that no singular or small team would ever hope to have full understanding of, they rely on people who are experts in their specific section of the behemoth. Even without a backdoor, adding a shim to a Linux system is as easy as LD_PRELOAD, yet alone the other ways to shim a library that replaces syscalls into your LD process. Security is just not a compelling reason to leave X11 when every layer of the Linux technology stack is vulnerable if you're a moron and run untrusted code on your local machine. >>87483 Interesting to read that MPV devs update post, glad to know a lot of his issues were resolved. Still insane that we are only having a semblance of a complete display system 15 years after inception. Talk about wasted time.

>>87484 >i installed linux to use linux, and 10 years ago it was indeed linux, but now it's more like macroidows. The part you're complaining about is more of a hybrid of Solaris and GNU/Linux, systemd's main inspiration is SMF from Solaris. And the majority want these features, people go out of their way to use OpenZFS, containers and namespaces are used all over the place, and yes, they use systemd. There's no hard requirement to use systemd at this point though if you really don't like it, there's several distros with alternative init systems. I'm pretty puzzled that the anti-systemd clique has this big overlap with pro X session crowd. X is basically a huge god program that does way too much, it used to do even more before they cut out the print servers and such. Some people are even running it as root still. If you hate scope creep you should appreciate Wayland, X is basically the biggest cautionary tale of scope creep of a program and protocol that's still around. > i just want a system that isn't owned by china/america please take you systemd elsewhere X was made in America. I don't think you're ever getting away from America's legacy in computers, it's just too huge. China simply makes a lot of the hardware which has given them a lot of power, it probably isn't even possible to undo that now, even big countries trying to do it are flailing ineffectively. >>87509 >Honest question, does Wayland yet support controlled sub-windows? X11 is the only display server on Linux that things like Audacity and Renoise work on. In fact, last I tried, Audacity crashes when starting through xwayland and Renoise was so unusable I had to move back to X11 for it as well. Are you talking about MDI? It's an ongoing issue, there is an attempt to add this in, in fact the protocol is done but they need ACKs: https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/264 >Most to all wayland compositors rely on wlroots, so there's your single point of failure. These compositors do not use wlroots: GNOME, KDE, Hyprland (formerly used wlroots), anything that uses Smithay like the COSMIC desktop, Weston. As time goes on there will be more. > Not to mention that you can also add a backdoor to libc, or any kernel module, or the kernel itself. That would require the patch to get past the scrutiny of the developers reviewing patches, or a commit bit. It's basically true of any software project that accepts patches or new developers. >Even without a backdoor, adding a shim to a Linux system is as easy as LD_PRELOAD, yet alone the other ways to shim a library that replaces syscalls into your LD process. That's out of scope of any display server project, this is where people use VMs and hypervisors or containers/jails/sandboxes, or replacing ld with a binary that doesn't allow LD_PRELOAD or any such shim, or well all of the above. There's a whole industry that uses this stuff already while the single user software developers languished in horrible security practices while admonishing anyone who dared to start fixing any said dumb practice for years. It's hard to begin any of that work of introducing any sort of isolation or end user understandable permissions or any graphical use case when there's this gaping security hole in every graphical application that can't really be mitigated. Obviously an attempt has begun to bring some sort of sandboxing to desktop users outside of their one sandboxed program (their browser) via Flatpak, although usually the packages (the older ones especially) are too permissive. systemd also uses some of the Linux kernel's features to provide some level of isolation in services. There are small modern jails and application kernels like nsjail, GVisor, Sydbox 3, and lightweight VMs like Firecracker and so on for more custom use cases. > Security is just not a compelling reason to leave X11 when every layer of the Linux technology stack is vulnerable if you're a moron and run untrusted code on your local machine. Well yeah in a perfect world you shouldn't do that, but people will do it regardless of whatever you say. People who have to deal with thousands to millions of end users are not going to listen to you, though, they'll listen to whoever is offering any solution to mitigate the problem. People don't want to deal with some dinosaur security model that doesn't work if one 10 line malicious script is on their machine, they expect the security model of Android or iOS at least. Leaving the front door open by having no real alternative to X discourages any new attempt to make a secure Linux distribution at any point in the future.

>>87543 >If you hate scope creep you should appreciate Wayland Actually, I just hate non binary devs who enjoy programming but don't enjoy putting thought into their programs. I also hate people like who argue about software from ethical point of view. systemd & wayland & pipewire & similar are EEE. The issue isn't ethical. The issue is that a certain government is overtaking the last rallying point of secure software - open source. After they are done assimilating all distributions, there will be no more ways to reliably deploy a transparent and secure system. This issue isn't ethical. It's a basic concern about my security. I do not think that a government is entitled to hacking into my data without a reason, and truly I don't consider a court order a good reason.

Just found out that Wayland doesn't even support Emacs. What a piece of unfinished garbage

>>87546 You need to build emacs with pgtk for wayland support. pgtk is soykaf though thats why i use lucid and xwayland.

DON'T MIND ME, BISECTING TO FIND THE BAD WORD IN A LONG POST >>87543 >People who have to deal with thousands to millions of end users >the security model of Android or iOS >security >Android or iOS Cattle security. Made safe from the user by taking away their freedoms, while having no privacy or security. One phone call to the NSO Group and they're into any smartphone. No phone calls at all and the smartphone is scanning and uploading personal data. My $100 Lainchan soykafbox is more private and secure, don't ask me how I know. We saw how SystemD worked out, they're taking away the ability to study and modify source code by making it a thousand times more complex than necessary, just like web browsers. There's also an effort to make all software and all distros depend on it which goes against Unix modularity. The init system is no longer replaceable like it used to be. SystemD already brought a worldwide backdoor to Linux, there will be more because it's too many lines of code. They are trying to install Google DNS again. It's corporate software aligned against the user like any proprietary OS.

DON'T MIND ME, BISECTING TO FIND THE BAD WORD IN A LONG POST There are many of us that want Stallman freedoms without the security theater. Want isolation? Get a physical computer or a VM, these sandboxes never worked. I don't need isolation because that model encourages running malware. I never thought "do I need to run GIMP in a sandbox" because I know it doesn't phone home or scan my drive. >These compositors do not use wlroots: GNOME, KDE, Hyprland... >As time goes on there will be more. I hope so. Or it will get complex like browsers and init systems and there will only be one or two that work.

DON'T MIND ME, BISECTING TO FIND THE BAD WORD IN A LONG POST Maybe Wayland will be fine. Maybe it's early days SystemD hype again and they'll pull the rug later. To me it looks like they're taking over over another cornerstone of userspace and making everybody depend on their software, standards they control. They're trying very hard to sunset X11 and have been blocking updates.

DON'T MIND ME, BISECTING TO FIND THE BAD WORD IN A LONG POST >>87544 I agree and I'm blackpilled. Free software is being exterminated and we're banished to our b**mer could it be the B word that triggers the filter? reservation distros which will die with us. Let's not get into the reason behind it all or we'll get derezzed. Certain people we're not allowed to criticize.

>>87551 Seriously? Censor the word b**mer and instead of informing me, make me waste 5 minutes? I found it anyway so what's the point not displaying it in the error?

A Canonical employee and GNOME developer who just so happens to be a convicted child rapist who abused his own kid sisters signed the petition against Stallman and vandalized the XLibre Git by accusing its devs and users of being evil nazis or something. Other GNOME devs claimed that people who wish to continue using X are "fascist maggots" and they continuously antagonize X users. That's all I need to know about Wayland to reject it. Malicious people who wish harm upon me are behind it so I cannot for any reason trust that they have my best interest at heart. What even is the point of it's security model? Any app, they claim, can be a keylogger under X11. The issue is that any app that becomes a keylogger maliciously is, by the very definition of the word, malware. On GNU malware without root access can: - Modify bashrc to download and execute scripts. - Read and write ~/ - Capture the root password and attempt to escalate privileges. - And much, much more. It's security model makes no sense and the restrictions that come with it are too heavy.

>>87558 >It's security model makes no sense and the restrictions that come with it are too heavy. You don't get it at all. They plan to make linux distributions "android-like" in the sense that software will be distributed in the form of "Apps". Each one will be completely isolated from each other, that's where wayland comes into play. >makes no sense After they have converted linux userspace into a secure and reliable thing, they will withdraw their efforts. After that, linux ecosystem will collapse, since freedesktop software is not sustainable and cannot be maintained long term. Then they will offer their proprietary solutions. The End. Also don't forget that linux offers too much freedom and endangers child safety. Think about the children. PS I don't believe they have plans to make linux their go to OS, I am confident their end goal is just killing the userspace.

>>87558 I do not follow this logic. Canonical has typically been more of a thorn in Wayland/Freedesktop's side than an ally, tried and failed to make Mir into something meaningful out of little but NIH syndrome. And GNOME appears to despise humanity at large including every human artifact that interfaces with them and or GTK, GNOME devs incessantly find new lows to sink to.

Don't forget that there is Xenocara, which is also ported to GNU+Linux (https://www.hyperbola.info/packages/extra/x86_64/xenocara-server/). By the way, if you want to learn about X11 protocol, I suggest reading https://magcius.github.io/xplain/article/index.html I also liked the FOSDEM talk by the Hikari guy: https://www.youtube.com/watch?v=8E0SOWo-Gsg (rip Hikari) Be sure to also scroll through the old thread: >>63336 >>87296 >Do people even use it for anything other than hyprland/sway ricing? I have used it for months now. KDE, GNOME, Sway, CageBreak (mostly), Labwc, Wayfire and LXQT (tested with Labwc) just works. Not sure about other Wayland compositors (I'm interested in DWL and WlMaker). To be fair, you still need Xwayland. The biggest mistake that Wayland developers made was that wlroots wasn't there from the start and it caused unnecessary fragmentation. X11/Xorg is good enough to be honest. Actually, Xorg is better than Wayland is currently (like how accessibility features or input methods don't works as well on Wayland). Tearing can be fixed on Xorg if you enable the TearFree option. >>87322 But Arch doesn't have a default GUI.. >>87551 >Free software is being exterminated Stop the fearmongering/FUD. Free software is an enormous success. Nobody is going to kill all those projects (especially if they use a copyleft license).

>>87565 Mir wasn't NiH, Mir was a logical response to the garbage that is Wayland. Back when Mir was introduced Wayland was complete, utter vaporware with nothing to show for it. Mir had a better design than Wayland and what's more important: Canonical produced a working implementation of Mir in a couple of years while Wayland had nothing after 7+ years of development. Red Hat pushed a lot of FUD for Mir, the community followed and then Intel removed all support for it from their drivers. Canonical decided that the project was unsustainable so they killed it. Ironically enough Wayland still can't run properly on Nvidia hardware 18 years after it was designed but its developers simply blame Nvidia and carry on.

>>87543 >These compositors do not use wlroots Fair enough, all that means is even more fragmentation to grapple with. Different issue, and not a user one, so I digress. >It's basically true of any software project that accepts patches or new developers. Yes, it is. The point is simple: A protocol or software that harms basic functionality in the name of security is itself conducting in an act of security theater. If you run malicious software on your system, you are pwned. It doesn't matter how secured you think you are, or what "security measures" are implemented. >Flatpak Ah yes, the "sandboxing" tool where a majority of software asks for full network and filesystem access, standardizing that lazy practice as "acceptable". So much more secure indeed. >Well yeah in a perfect world you shouldn't do that, but people will do it regardless of whatever you say. Then they deserved to be pwned, simple as. The act of "streamlining" computers to such a point where people who hardly think a day in their life can use it is not in the best interests of anybody who gives a soykaf. Building opaque software layers will do nothing but normalize spyware, adware, and other malicious practices. Something that is already entirely normalized on other opaque systems like mobile phones. And do define opaque I don't just mean "non-free software". Free software has provably failed its goal and been successfully co-opted by corporations. Opaque means a lack of flexible interaction between software layers, attempting at all costs to avoid allowing one layer to talk to another outside of sanctioned interfaces. This builds a hilariously strong sense of false security that does nothing but harm functionality as well as actual security, as now it's "okay" to run anything because "the developers thought about it".

>>87544 The funny thing is this software is probably no more insecure, and in some cases probably more secure, than what was used before it. >>87548 >The init system is no longer replaceable like it used to be. There's multiple distros that don't use systemd, you can use them now. I haven't heard of any issue using alternative init systems on Linux with Wayland so this sounds like a non-issue to me and more like unrelated ranting. >Cattle security. Made safe from the user by taking away their freedoms, while having no privacy or security. One phone call to the NSO Group and they're into any smartphone. No phone calls at all and the smartphone is scanning and uploading personal data. The main difference on a smartphone is that there's a cellular modem using insecure protocols, and unfortunately until the cellular industry changes that it will always be like that. Airplane mode does mitigate this but of course you won't have cellular functionality. Other than that, probably an iPhone or Pixel is much more secure than your lainbox, a Pixel with GrapheneOS loaded up being the gold standard. Getting a rootkit to persist on that cheap box is child's play especially without secure boot. The TPMs are pretty much useless on x86 platforms so you need to use something like a FIDO/U2F key that isn't compromised yet instead if you want a second factor, and you need a high entropy password. Even then there's just so many weird vulnerabilities on PCs and OSes made for PCs that won't exist on mobile platforms. >They are trying to install Google DNS again. Do you not know how to configure resolved or use an alternative DNS program? systemd functions without resolved so this is more of a distro failing to configure its defaults sort of thing. It's extra software like networkd, you don't actually need to run it when you use systemd. >It's corporate software aligned against the user like any proprietary OS. It's open source so it can't quite be on that level. It's not even a closed development process with a libre license, it's much better than how Google develops Android for example. >>87549 There's also someone working a C++ library for compositors, so thats another wlroots alternative: https://github.com/CuarzoSoftware/Louvre >I never thought "do I need to run GIMP in a sandbox" because I know it doesn't phone home or scan my drive. Some people can have both so they're not completely compromised by one project being taken over by some rogue actor. I don't really see the problem. In any case it's out of scope of Wayland alone so you'll have pretty much what you just asked for. >these sandboxes never worked There's just no way to tell how many attacks were prevented by browser sandboxing alone, but browsers in the pre-sandboxed era were a complete and total joke security wise. >>87546 Emacs has been supporting Wayland natively for 2 major versions now, you need to build or install the pgtk version like the other lainon mentions. The Lucid one does work fine in XWayland from what I remember. >>87568 >Hikari Someone could probably fork Hikari if they have the desire for a cwm clone. I don't think it helped much popularity wise that the developer was a version control system hipster who was using darcs, but at least it wasn't some closed up VCS. >Tearing can be fixed on Xorg if you enable the TearFree option. It's true but in my experience it only solves tearing, not other artifacts. There's also some overhead that should get worse as monitors become higher resolution, and as you add more monitors. >Ironically enough Wayland still can't run properly on Nvidia hardware 18 years after it was designed but its developers simply blame Nvidia and carry on. Wayland developers can't do anything about NVIDIA drivers and hardware, it isn't even really at the same layer of the OS anymore. You're expecting the tail to wag the dog and get upset when the dog wags the tail. In any case it does run on some NVIDIA devices with new enough drivers reportedly. >>87559 I don't really understand the FUD crowd. Pipewire/Wireplumber are very expressive and allow more functionality, not less, Pipewire is also faster than Pulseaudio. XDG portals and D-Bus APIs exist as well but a lot of compositors offer IPC of their own. It just isn't locked down like Android or iOS. Fully sandboxed builds and sandboxed programs are more of a distro level thing and there's hundreds of them, maybe a lot of them will adopt a more secure setup. >>87595 >Building opaque software layers will do nothing but normalize spyware, adware, and other malicious practices. Been normal before this stuff made it to the mainstream, so it would change nothing at all. >Then they deserved to be pwned, simple as. The act of "streamlining" computers to such a point where people who hardly think a day in their life can use it is not in the best interests of anybody who gives a soykaf. That's the power of abstraction in software design. I don't see what you actually lose. People expect a different security model now, not the bad ideas from Bell Labs and MIT which lead us to being between a rock and a hard place here. Either the userland gets something like this security model, or the various libre incumbents could lose out to an alternative OS. >Ah yes, the "sandboxing" tool where a majority of software asks for full network and filesystem access, standardizing that lazy practice as "acceptable". So much more secure indeed. I'll agree that the defaults are pretty bad, they could begin fixing it by asking for permissions instead of just automatically granting broad ones, have the user grant access to only certain parts of the filesystem beyond what the program needs to run for storage if they want, etc. However Flatpak permissions are configurable at least so if you know what you're doing you can lock down a package. Either they could fix up the UX/UI or someone else could start over with a better design. Kernel-wise most of what you need to have a fairly decent permission system is already here or on its way. Anyway again, it really isn't very relevant to the display server beyond the display server enabling the possible implementation. One of the attempts being terrible doesn't mean something better is impossible.

>>87689 Only someone who is hiding cheese pizza is this schizophrenic about security. Destroying the usability of an opensource system by sand boxing everything is just dumb.

>>87689 >Been normal before this stuff made it to the mainstream It being normal outside of the niche reality of Linux userspace doesn't make it acceptable to normalize it within it now that it's gaining popularity. >...or the various libre incumbents could lose out to an alternative OS. That's fine, truthfully. Linux doesn't have to win, in fact I'd prefer it didn't. The people who don't care can live in Windows and Mac, and as those stacks enshittify they can face the reality that caring about their technology stack is worthwhile. The best thing for Linux is for it to "lose". >However Flatpak permissions are configurable at least... Believe me, I've tried to configure flatpaks to have more restrictive filesystem access, it just breaks software as it begins a wild goose chase of "oh sorry, we also use this obtuse directory". If software maintainers documented their own out-of-band interactions this wouldn't be as problematic. >Kernel-wise most of what you need to have a fairly decent permission system is already here or on its way. Wish Linux had a sane interface for it, though. Instead it's just another obtuse API that most programmers will never willingly interact with. I'm shocked there hasn't been more effort to port the ideas of pledge and unveil to Linux. >One of the attempts being terrible doesn't mean something better is impossible. Impossibility was never the problem, the reality of "worse is better" continuing to propagate throughout the Linux landscape is. Wayland is simply a terrible solution that took so long to reach a usable state that I see no reason to trust the people behind it. X11 is also quite terrible, bloated, and antiquated. But at least it can run on near anything, and degrades gracefully when faced with subpar hardware. Wayland gets the hate it does simply because the developers have consistently claimed how much better Wayland is, while simultaneously acting extremely lazy. The amount of time for Wayland to gain standards for simple features like keyboard shortcuts is a prime example of this. Wayland is a lazy replacement for X11, not an attempt to make something better. The programmers outside of Wayland itself are the only reason it has any functionality at all, and they will all be forgotten as the Wayland team "revolutionizes the Linux desktop", while being lazy bags of dogshit.

>>87689 >People expect a different security model now, not the bad ideas from Bell Labs and MIT which lead us to being between a rock and a hard place here. Either the userland gets something like this security model, or the various libre incumbents could lose out to an alternative OS. My friend, your beloved circus of insider trading monkeys doesn't have a security model. It doesn't have a threat model either. It just implements "secure" software, but what it secures me against is a question that not even God can answer.

I've been using wayland for about 8 years. With portals, it runs all the background config stuff I used to have to run gnome-settings-daemon for. Now I don't have to do that. There's no screen tearing. Cool easy-to-configure programs like waybar exist. Sway and River work great. There are lots of other compositors. bemenu works on both x11 and wayland. Honestly the only problems I ever had with wayland was using sway on a work computer and screensharing with Zoom. Zoom! That's the only thing that ever gave me problems, a piece of Chinese malware. Wayland's fine. It works well and didn't break anything I cared about. Some programs I use run in Xwayland but I don't even care about that. They just work. Early on in my usage, if something didn't work right, I'd log out of sway and log back into i3, thinking it was a wayland problem, but it never was. And honestly I like it better. It works better. There's less setup to do on a new machine.

>>87711 >There's less setup to do on a new machine. That's the most moronic bullshit I've read in a while. You're not even talking about wayland.

>>87289 Is it only me , or doesn't it feel retarded to have and embedded systems opengl API instead of the full opengl API ? I mean the coupling between X and the regular opengl API seems retarded also

>>87713 That's not very nice. Why would you assume I was lying? Wayland: I just install the window manager^W^Wcompositor and run it from the console. xorg: install my wm, xorg, xf86-video-whatever, set up .xinitrc, ensure gnome-settings-daemon starts There's less to do from scratch. Wayland WMs usually have slightly more modern configuration and include autolaunching for things like portals, where xorg wm default configs usually are a bit more conservative. I don't have a problem with people wanting to use xorg at all. Use whatever you like until it doesn't work anymore. But pretending it's wayland vs xorg, or one doesn't work, is just some weird infighting. We should spend time on the really important things, which is insulting macos users.

This might be interesting. Wayback 0.1 released! >Wayback is an X11 compatibility layer that allows for running full X11-only desktop environments using Wayland. It is essentially an X11 server backed by Wayland, leveraging wlroots and Xwayland. >Our goal is for Wayback to eventually be a completely drop-in replacement to the Xorg binary, thus reducing maintenance burden >https://wayback.freedesktop.org/news/2025/07/23/wayback-0.1-released/

>>87731 not really since it is hosted on freedesktop >>87721 sorry but you spew underage ramblings i don't even want to try and figure out what you were trying to say. if your issue is installing a package uh go install windows then

>>87693 I absolutely guarantee you Linux has been hacked before they had sandboxing available in any form. >That's fine, truthfully. Linux doesn't have to win, in fact I'd prefer it didn't. The people who don't care can live in Windows and Mac, and as those stacks enshittify they can face the reality that caring about their technology stack is worthwhile. The best thing for Linux is for it to "lose". Linux isn't really the best designed, or smallest code footprint, or most elegantly coded, or most fit for purpose or whatever. It's the one that won the Unix-like popularity contest by getting the most people working on it, so it has the most drivers, supports many filesystems, supports many architectures, etc. Once that goes away Linux is dead in the water. >Believe me, I've tried to configure flatpaks to have more restrictive filesystem access, it just breaks software as it begins a wild goose chase of "oh sorry, we also use this obtuse directory". If software maintainers documented their own out-of-band interactions this wouldn't be as problematic. They do need to document this more and there probably should be a better filesystem hierarchy standard that specifies directories for somewhat random reasons. There's ways to hunt that down though like using a tracing tool, and more programs use the XDG directories now so you can look there too, user programs tend to hide a lot in ~/.local/share/ and ~/.config/. If programmers won't document it distros that are interested in sandboxing/security could, and they could publish their results in a repo somewhere regularly. But generally a new FHS that embraces the presence of a package store (Gobo/Nix/Guix all have one, some non-Linux based OSes do it too for reference) and deprecates vestigial directories is the way to go. >I'm shocked there hasn't been more effort to port the ideas of pledge and unveil to Linux. There's an equivalent already called Landlock, although it's a bit more complex as usual. I've seen at least a couple sandbox tools that already use it. >Impossibility was never the problem, the reality of "worse is better" continuing to propagate throughout the Linux landscape is. Linux is worse is better, always has been. >X11 is also quite terrible, bloated, and antiquated. But at least it can run on near anything, and degrades gracefully when faced with subpar hardware. It used to basically fry your monitor when faced with an incorrect modeline, but that was a long time ago and nobody needs to set a modeline manually anymore. In any case, it does not run on everything natively, not anymore, and more odd devices will be essentially incompatible with it as time goes on. You're talking about something that thinks it is the 90s forever at a low level. >Wayland is simply a terrible solution that took so long to reach a usable state that I see no reason to trust the people behind it. In the time it took for it to reach that state no usable competitor showed up, nor did anyone develop at a pace faster than them. I think they could have been not so stubborn about the protocol process but they did ease up on it finally, and the floodgates are open now essentially. Whatever is missing will be filled in by a protocol or a program. You can maintain or support someone maintaining X.Org as long as you like but I think programs that already support Wayland will drop X support once it's the incumbent on the popular distros, so you're going to see a shrinking library of programs at some point. >Wayland gets the hate it does simply because the developers have consistently claimed how much better Wayland is, while simultaneously acting extremely lazy. The amount of time for Wayland to gain standards for simple features like keyboard shortcuts is a prime example of this. Wayland is a lazy replacement for X11, not an attempt to make something better. The programmers outside of Wayland itself are the only reason it has any functionality at all, and they will all be forgotten as the Wayland team "revolutionizes the Linux desktop", while being lazy bags of dogshit. It's simply how bazaar development works, the people who know how to design the core protocol do what they know, they aren't going to be the ones integrating it into your distro or working on most of the software. It's unrealistic to expect this from people pretty much working for free. Other people do the rest of the work, again most of them are working for free. >>87713 There's all sorts of redundant configuration occurring in X.Org, XResources, Xdefaults, xorg.conf, xinitrc, xserverrc, then the GTK/GNOME stuff, the Qt/KDE stuff, and then your window manager's settings. So you've got 6 redundant files for your X configuration experience, the GTK/Qt bazaar junk is inevitable but the rest of that doesn't need to be separate, plus the xrdb binary, plus the window manager might provide another binary for some IPC on the fly configuration stuff. All of that X stuff can be in one file read by the compositor and it can even be hot reloaded if the developer wants it to be, the compositor can have a binary for IPC commands. That's a big difference in the number of configuration files you MUST be on the lookout for.

>>87746 >I absolutely guarantee you Linux has been hacked before they had sandboxing available in any form. I can equally guarantee you that Linux has been hacked after they had those tools, this isn't an important point. The point I made earlier is that assuming sandboxing tools "make you safe" is going to create the perception that running unsafe code is acceptable. Sandboxing is useful regardless, but it shouldn't be used to mean "security". I recognize I hold this perception that things should be closer to some form of ideal, but that's because to accept the lack of general competence and care in this discipline I care deeply about is to accept my own death. Things should be moving to be better than presently they are, instead of moving to accommodate simpletons. >It's simply how bazaar development works, the people who know how to design the core protocol do what they know, they aren't going to be the ones integrating it into your distro or working on most of the software. I never asked them to, I asked them to develop a protocol that handles basic use cases. IPC between windows is an excessively normal task, sub-window control is a normal use case, I could keep going. I recognize D-Bus exists, and it does solve some of the limitations, but I struggle to see the justification for yet another "simple" system that I need to add onto my dependency chain. As you've pointed out, they've finally eased the protocol process, but I am still reluctant to trust the people behind Wayland itself given their track record. Adoption wouldn't have take 15 years if they had designed a protocol that was meant to be useful from the start. >You can maintain or support someone maintaining X.Org... >Linux is worse is better, always has been. It's all worse, not just Linux. OpenBSD is the only "worse is better" system with any semblance of quality. >You're talking about something that thinks it is the 90s forever at a low level. And we're all using machines that still pretend to be a machine from the 70s, so what's your point? I have never found a machine an X server doesn't support, and that includes niche architectures and organization. >Once that goes away Linux is dead in the water. Good, I can only hope to live to see that happen. The death of Unix, and its philosophy, will be a good day. Call me bitter, because to an extent that is what I am, but it's comical to me that after all these years the only viable replacement to the Frankenstein's monster of X is Wayland's shattered window and a note saying "if you put this together the right way, it's really good". I hope that with the looser process for adding features to the core protocol will make it more palatable without the need for a million secondary protocols to support it. I've appreciated our dialogue, but this will likely be my last response as I can feel myself falling onto my urge to bicker. If the core standard is ever featured enough where I can use a minimal compositor without having a collection of missing features, or implementing multiple different portal protocols, then I will switch. Until then, my LISP machines will hum along happily with X as their display system. I wish you a good day.

>>87746 >In the time it took for it to reach that state no usable competitor showed up i wonder why. must be that every one was just too happy to notice. "competition" in open source is no longer a thing. not since corpos overtook practically all distros and most of open source communities. >the floodgates are open now essentially more soykaf for the god of soykaf >It's simply how bazaar development works that's called Indian bazaar development, i wonder why you omitted the word Indian. maybe because you're one yourself? >w-we're not responsible that our godsent protocol is being implemented by monkeys! holy soykaf dude. your posts make me feel nausea

I'm not deep enough into this soykaf to care about the details but wayland is the only thing that solved screen tearing so I use it on my emulation/movie box.

I read that design of Wayland might lead to increased attack surface, since xorg-server + wm + compositor is moved into 1 binary/process aka "Wayland Compositor". Under Wayland, individual client programs might be isolated but what about various parts of the Wayland compositor itself? Any thoughts on this? >>87293 Arcan looks really interesting but sadly it's not usable, yet (AFAIK). >https://arcan-fe.com/2017/12/24/crash-resilient-wayland-compositing/ >https://arcan-fe.com/2025/01/27/sunsetting-cursed-terminal-emulation/ More? * https://arcan-fe.com/videos/ * https://www.youtube.com/watch?v=07nqZIFRDJg

>>87758 >I've appreciated our dialogue, but this will likely be my last response as I can feel myself falling onto my urge to bicker. I understand, but I'll reply to some of your points anyway to clarify some stuff. >Sandboxing is useful regardless, but it shouldn't be used to mean "security". It's one tool available to mitigate breaches. It doesn't really cost much in terms of system resources, it just means you have to get it together and eat your metaphorical vegetables, which GNU/Linux people suck at because they develop in the bazaar model. I'm not saying it's as good as a VM or an airgapped machine, and yes a zero day or misconfigured sandbox means it can be breached. >The point I made earlier is that assuming sandboxing tools "make you safe" is going to create the perception that running unsafe code is acceptable. Linux distros are one supply chain attack away from serving malware in their official repos. Those programs should ALL be run in a security sandbox ideally, and the package manager/build process should be sandboxed as well. Of course they should try other ways to mitigate the risk like vetting developers of crucial software and making more builds reproducible but there's basically no reason to not move towards sandboxing at this point as well. As for the perception.. that cat left the bag completely in the early 2000s probably. There are adults walking around who have experienced the new security model their entire life with the smartphone in their pocket. It's just not what the new generations want or expect anymore. >Good, I can only hope to live to see that happen. The death of Unix, and its philosophy, will be a good day. I don't have rose tinted glasses for Unix or Unix-likes, they have limitations and problems. The problem is there is basically nothing else that will run on this volume of real hardware, alternative OSes always run in a VM or on a few select machines. > If the core standard is ever featured enough where I can use a minimal compositor without having a collection of missing features, or implementing multiple different portal protocols, then I will switch. My guess would be within the next ten years. >And we're all using machines that still pretend to be a machine from the 70s, so what's your point? I have never found a machine an X server doesn't support, and that includes niche architectures and organization. In userspace it's much, much easier to move on from historical baggage than in a kernel. I'd say the lack of support is more of a possibility if there is no attempt to make newer devices work on X.Org/XLibre no matter what via a generic X driver or something. OEMs and developers really don't focus on 2 or 3 display servers for one platform, they will pick one if it comes down to it, and that will be whatever has the most users. They don't have to do this idiotic dance on any other OS after all. >>87768 >i wonder why. must be that every one was just too happy to notice. "competition" in open source is no longer a thing. not since corpos overtook practically all distros and most of open source communities. Anyone could have tried, you don't need to be a part of a corporation to try, and nobody is really preventing it. There's like one guy doing that and the reason why nobody uses his display server is because there is no standard, everything could change completely every time he makes an update. Canonical was free to continue developing Mir if they really wanted to, but it was not in a great state back then. People like to make a big deal about something that was going to function worse ultimately, they have never seen the code or how it runs in practice, it's just idle talk. >that's called Indian bazaar development, i wonder why you omitted the word Indian. maybe because you're one yourself? It's from a book by a pasty white dude that came before most of the Indian tech outsourcing. A little bit of advice: if you don't know what you're talking about, maybe you shouldn't say anything at all. >>87816 >I read that design of Wayland might lead to increased attack surface, since xorg-server + wm + compositor is moved into 1 binary/process aka "Wayland Compositor". Under Wayland, individual client programs might be isolated but what about various parts of the Wayland compositor itself? Any thoughts on this? There is basically no benefit to separating the compositor and the display server, and it's almost always done by the same developer(s) in a Wayland compositor so if you trust them you don't really have a problem with it. The only thing I can see is if you wanted to make a modular compositor with a sort of plugin system for different shells (closest equivalent to a WM), that would need to be reflected by the design: privilege separation and sandboxing and so on, since a shell might not be written by the people who wrote the compositor. I'm not sure how unprivileged you could make those shells in Wayland in practice. Of course a X WM was always capable of logging all input and uploading it to a server using curl or something, or making a thousand popups appear, malicious stuff is always possible if you don't audit the code. What Wayland does provide is reduced attack surface from the Wayland clients, and before someone wastes even more time arguing about LD_PRELOAD or something endlessly for no reason because they did a Linux red team for the first time, the rest of the system is up to you to secure. You care about this infinitely more than the compositor/server/shell because there's endless varieties of client programs out there.

>>88223 >Linux distros are one supply chain attack away from serving malware in their official repo Compile from source + stop being a schizo. >Those programs should ALL be run in a security sandbox ideally No they shouldn't. I want my software to just work, I don't want the bloat and unending headache of sandboxing. It prevents software from integrating with each other in a simple, manageable way (the UNIX way). >and the package manager/build process should be sandboxed as well Already is. >Anyone could have tried, you don't need to be a part of a corporation to try, and nobody is really preventing it t. 19 years old 4chan regular. Don't forget that all skills can be acquired in a few weeks on the internet. >if you don't know what you're talking about, maybe you shouldn't say anything at all. Literally follow your own advice. >There is basically no benefit to separating the compositor and the display server Yes there is. Compositing is a very bloated way of pleasing snow flakes' misaligned expectations of an OS. Windows management and compositing ARE different things and therefore it makes a LOT of sense to separate them. >What Wayland does provide is reduced attack surface from the Wayland clients In the most atrocious way possible - by default and without a "TURN THIS SHIT OFF" button.

>>87289 >this is niri! a scrollable tiling wayland compositor i can feel the lag just by looking at that screenshot. does wayland even do beam racing or what? whats the point

19 ☠️☠️ uncs it's time to retire anyone above 18 has run their race and should let it rest

Anytime I begin considering finally switching to Wayland, I remember that, what is there to switch to? The nu-linux "community" can name all of these gimmick WMs but in reality they're all wlroots trash with not even all of Wayland's potential features, much less all of X11's. The only choices are to use GNOME, KDE, and Hyprland, and fuark you if you're not satisfied with any of them. Wayland fuarking devastated the linux desktop.

>>88232 >Compile from source + stop being a schizo. No, you make as many builds reproducible as possible instead. Whether an edge device compiles something or not is not really relevant. Sandboxing adds a layer of defense in case that isn't available or if something went very wrong. >Don't forget that all skills can be acquired in a few weeks on the internet. I don't see how this sarcastic jab disproves anything I said, if anything it just reasserts that there is no other alternative than Wayland. >No they shouldn't. I want my software to just work, I don't want the bloat and unending headache of sandboxing. You use it already every time you start up a modern browser, or use some of the underlying features when you run a container. People use it when systemd starts up a service most of the time, it's one of the lowest cost options for isolation. In any case I'm not here to endlessly litigate your cluelessness about sandboxing, it's just an example of a use case that benefits from Wayland, and people have that use case even if they're not you. >It prevents software from integrating with each other in a simple, manageable way (the UNIX way). Ironically enough it's partially UNIX's legacy that is preventing it from being easy, with possibly the worst filesystem hierarchy I've seen in a family of OSes that survived past the year 2000. One of the directories simply exists because some Bell Labs employee ran out of space on a drive and needed to mount a new drive. >Compositing is a very bloated way of pleasing snow flakes' misaligned expectations of an OS. Windows management and compositing ARE different things and therefore it makes a LOT of sense to separate them. People generally expect there to not be screen tearing or weird driver specific hacks to try to prevent it that you have to go and turn on. What you're proposing just makes everything worse. In any case X session has fuarrrk all to do with reducing software bloat so I don't really see how this supports your argument. For the very few use cases that need immediate mode, you can use the tearing protocol. >Already is. I'm speaking in general, I know many distros do it already, but it's a diverse ecosystem. >Literally follow your own advice. You've offered nothing that shows you know any better, so no. >>88478 >The only choices are to use GNOME, KDE, and Hyprland, and fuark you if you're not satisfied with any of them. There are Smithay based DEs/compositors like COSMIC, Niri and so on, and hopefully we'll see more Louvre based compositors in the future as well. The other elephant in the room is Weston but it's a reference compositor, kind of a boring generic stacking window manager type deal. >>88261 With the animations turned off it's probably the same as any other Wayland compositor in responsiveness. It's not like they wrote it in Javascript or Python or something. >beam racing There's no electron beam anymore unless you actually use a CRT. What you can do is use the tearing control protocol (immediate mode) and enable adaptive sync if your monitor supports it. Most compositors also support direct scanout as well. If the compositor doesn't support tearing control and you just want to run one program (probably a game), you can switch to a different VT and run Gamescope directly and use the --immediate-flips option. Just keep in mind if your GPU drivers are buggy then they're buggy and you may see more artifacts than you expect when using immediate mode. I personally don't notice input/display lag/response time issues compared to TearFree enabled desktop usage and gaming on X11 in any Wayland compositor I've tried.

Forgive me if I'm wrong, but as I recall: Wayland requires compositors and does not dictate much besides base protocols to said compositors. The wayland compositor that I hear the most positive buzz about is hyprland, because ricing. This has created a highly dedicated, annoying fanbase, which is generally criticized for political leanings which largely involve inane interpersonal drama and girlish clique gossip (e.g., "Vaxry hates troons, you can tell because he had this tepid spat with Drew DeVault which hurt DeVault's ego and that means--" "No akshually Vaxry is pro-trans his community manager is transgender") BUT- hyprland has and had for most of its existence such severe security flaws that it has been blacklisted from many major linux distros including fuarrrking gentoo based solely on its history of high consequence technical deficiencies, which then get papered over by hyprland fans as politically motivated character assassination due to the obvious out in the open virulence of retarded girl talk about irrelevant poor attitudes so: if hyprland bad, how wayland good?

>>88792 idk. what's the state of the drama outside imageboards? anyway ibm is like kicking dead whales down the beach its hardest to get rid of x11 in *fed*oora so it means other distros are gonna follow very soon. this means we're gonna be using gayland instead of improving the x11 protocol implementation. hopefully though x11 will live for 5-10 more years and after that i don't really care because the future seems to be pretty fuarrrked up in general, so what happens 5 years from now on is something i just don't care to ruminate about

The this that annoys me most about Wayland is how triggered the anti-wayland crowd gets any time it is mentioned online. Anyway, I can't imagine not using sway in this day and age. It just works. Does anyone use a non-sway wlroots compositor? How is that?

>>88806 >The this that annoys me most about Wayland is how triggered the anti-wayland crowd gets any time it is mentioned online. dbus and all software that depends on it must die, as well as people behind it. wayland is a non-dbus part of this must-die mix. since we can't combat corporations at least we can call you names.

>>87289 I've been running XLibre for the last few weeks, and it's doing its job, so… Why bother with Wayland? Wayland lost because its developers have massive egos and see themselves above the users or even other devs. X11 being outdated and in dire need of cleanup is a valid point; rewriting literally everything and expecting everyone to adapt to changes that only Wayland developers wanted is not. Right now the only people still wanting Wayland are its developers, the companies trying to seize control over Linux through making Wayland de facto a standard, and dimwits high on political bullshit. There's no practical reason to use Wayland when XLibre exists and is slowly but steadily being improved.>>87289

https://www.dedoimedo.com/computers/plasma-6-4-performance-wayland-x11-comparison.html This person's experience does not reflect mine. I use and like XFCE, on my NVidia 10 series card, while frequently task switching between wine games (not proton) and probably several other things that have turned the wayland experience worse. So from my perspective it looks like his comparisons were pretty fair, he wasn't going around trying to do things that I would do and he isn't running unnaturally performance stacked environments such as compositorless Xinits. These benchmarks are MUCH BETTER than I thought, based on my previous personal experiences with wayland. Since it is impossible that my memory could be faulty in any way, that means that there have been tremendous leaps in wayland performance in only a few years. Since the GabeCube is going to be set out with KDE Plasma and AMD graphics I assume it will be the environment to benefit the most from wayland fine tuning in the near term, which doesn't help me much personally but does suggest we might see wayland surpass X performance threshholds in a few years, at which point it might also be usable even for holdouts like myself.

>>88807 fuarrrk you too, I guess. >>89717 My last recollection of using plasma was in 2015, it was really slow on my soykafty computer. Probably one of the main reasons I switched to a tiling window manager. Is performance really still so bad that you found it unusable? You might consider trying a wlroots-based compositor, it works very well when task switching.

>>89123 >I've been running XLibre for the last few weeks, and it's doing its job, so… Why bother with Wayland? You can probably use XLibre or X.Org for some years to come. Problem is, X support will increasingly be dropped as it's just another dependency developers don't want to maintain, so new programs or updated programs will no longer be an option at some point. >Wayland lost It's the default going forward no matter what, it does not have to do with its merits at this point, once the major distros switch it's already over. Since it became usable enough and they're switching, X as the main display server is going to become a legacy use case. It's a bit like how systemd replaced most init systems since it made distro maintainers' lives a lot easier. > There's no practical reason to use Wayland when XLibre exists and is slowly but steadily being improved. X11 is completely awful outside of the traditional desktop or the very traditional X server <-> X terminal(s) setup (nobody does the latter now, by the way). Automotive Linux and mobile Linux is pretty much all Wayland now and for good reason, X11 halted any sort of mainline mobile Linux project early on. Android explicitly avoided X11, MacOS dropped X11 years back, they want compositing and want to reduce legacy code footprint. Google and Apple love grabbing permissively licensed code that's available for free but not even they want to touch X anymore. It's not even that good at the desktop setup. Want multiple monitors with different refresh rates and VRR on all of them? Nope! Want your display server to scale UI on different displays correctly, automatically? Nope! Want something other than screens merged into one big screen for X to poke at? Nope! Want compositing? You need this weird semi-maintained unofficial compositor program that doesn't work well. Want to isolate graphical programs? Well now you must run multiple instances of X. It just gets in the way and makes the job harder. >>89717 I don't think there have been performance problems for years in most compositors that are well maintained, basically nobody even thinks about this anymore except a few developers. If the setup is identical visually I can't tell the difference between immediate mode X and Wayland except that some X window managers have non-tearing artifacts even with TearFree enabled. Most performance issues you will see are from unmaintained outdated projects or bugs that could be in Mesa, the kernel, in the compositor itself, or elsewhere, but this is not behavior specified by any protocol. If you really believe you can actually see the tiny delay from compositing, go use a compositor with tearing protocol support. The anti-Wayland crowd has a bunch of outdated information they typically complain about, the only valid complaint I can think of has to do with accessibility having gotten zero developer hours spent it on in practice. It's a good point but a more general issue with bazaar style development, accessibility was never defined well generally in the X era either, and now someone has to rush to implement it.

>>89736 xdotool? screen/window recording without pipewire? xhost? not yet? >X11 is completely awful outside of... what is the point of wayland? pipewire integration? "window isolation"? why does it exist at all? "X codebase is too fuarrrked up to fix it". are you kidding me? x11 is a protocol if X.org implementation is bad, make a good one and fix the said issues in X12. why does wayland exist at all?

>>89742 Wayland exists at all because X has always been a political and social soykafshow, beginning even before they tried and failed to relicense. When a political and social soykafshow is allowed to constantly escalate you get breakaway soykafshows. >x12 would have just been wayland with a legacy name. Same social problems, same devs telling you that you don't need things, same end-user dependence and reliance on externals, extensions, modules and shims, same monstrous impact on fracturing the entire ecosystem to the point where software developers cannot effectively target a linux desktop, and same attempt at resolving said issues through political action regarding deprecation of the predecessor/competitor stack.

>>89743 How does Wayland not have the same problems? Different compositors with different features and bikeshedding making development of new features glacial. It was unusable until Valve came along.

>>89742 >xhost This would not work, unless rootful XWayland supports it now, and you'd only see the X clients anyway. There is waypipe and some compositors are adding RDP servers, KDE and GNOME at least. Software like Rustdesk has begun to support Wayland. Pretty much RDP & VNC like every normal OS has. >xdotool? wlrctl, wl-kbptr, wtype. >screen/window recording without pipewire? Why without pipewire? But yes this has been possible for a while now through wlr-screencopy or ext-image-copy-capture primarily and a couple other protocols typically. > x11 is a protocol if X.org implementation is bad, X11 is a protocol so you cannot fix the problems it creates by improving the implementation, you need to break compatibility at some point. They've been skirting around breaking compatibility for years and decided it would be possible to slowly get rid of X11, and get rid of X.Org quicker than that of course. >make a good one and fix the said issues in X12. Wayland is X12. >>89744 >How does Wayland not have the same problems? It was designed not to. The rest are human problems that were always present no matter what choice was made. >Different compositors with different features and bikeshedding making development of new features glacial. There are libraries being reused when you make a Wayland compositor nowadays, unless you really start from scratch. The "bikeshedding" gets generalized into protocols, the process for making new protocols became less onerous than it was in the past, now there are a lot of them. Again, this is basically old criticism back from when someone from GNOME used to block most proposals, they should have kept him from NACKing everything long ago but it seems like they have enough stakeholders now other than GNOME to ACK things and ignore bizarre ideas from GNOME.

>>89745 >Wayland is X12 Wayland is X11.4(9) kek not even a server-client architecture. I think wayland will fit it nicely with rust coreutils and others. >Why without pipewire Because it's soykaf? >X11 is a protocol so you cannot fix the problems it creates Such as? >RDP Lmao >It was designed not to It was never designed. It's just a bunch of indians slapping together another modern private secure and highly optimized piece of soykaf.

>>89746 >It was never designed. Wayland is the example I point to whenever I want to make the argument that software can be overdesigned. The point of wayland was to be a protocol built according to strong and specific principles regarding what a display protocol should be and many of those principles were proven incorrect over time. Dogmatic design is bad because design does not account for eventualities and external developments. Not because it is slapdash or underdesigned or unprincipled.

>>89772 How about you study what exists and use your brain to implement something that fixes existing issues and doesn't touch upon anything else? That'd at least ensure that after a certain amount of iterations the software turned out fine. How about not beginning from scratch every 15 years?

GNOME and KDE cooks dropped X11 support. As far as I understand, more are to follow.