It is actually kind of wild that we're simultaneously in an era of people complaining that Wayland is destroying choice and also maybe the greatest number of high-quality desktop environments aimed at different use cases the free software world has ever had

I'm pretty fascinated about why this is happening and my gut feeling is that people have simply written better abstraction layers on top of Wayland than were possible on top of X and now people can just write shit without needing to care anywhere near as much about the sharp edges that exist everywhere

And yeah there's all sorts of cases where X exposes functionality that doesn't really exist in Wayland and I understand people wanting that but also I shared my desktop with some people today on a call and got a notification icon letting me know it was happening and I didn't have to rely on the client to behave correctly for that to be something I could rely on

@mjg59 i'm not sure possibility even factors into it to be honest, it's really just just that wlroots now exists and xroots didn't.

@mjg59 I have been a happy GNOME on Wayland enjoyer since 2020 now, and am increasingly struggling to understand the complainers (with the notable exception of those that rely on accessibility features that continue to be poorly supported). I saw a post recently titled something like "is it possible to get wayland working in 2026?" and I genuinely have no idea where this sort of thing comes from

@mjg59 On the one hand, people have wildly varying needs and wants, and I'm perfectly happy to assume Wayland is actually not a good fit for some people, even many people. That's true for X11 too, of course.

But the X11/Wayland discussions I see are mostly populated by people who dismiss one or the other without much explanation, or giving reasons that don't match my perception of reality at all.

As such, I've ended up ignoring the complainers a lot.

@mjg59 I like changing desktops now and then just to relieve the monotony..I've never had any problem with #Wayland or #X11.. I suppose if you have a specific need to use X and you're really used to #plasma or #gnome it would be a bit annoying.

So: In the past the infrastructure was simpler, the underlying hardware was simpler, the number of use cases you had to satisfy was smaller. And now everything is fundamentally more complicated and you're competing with platforms that have millions of absolutely normal computer users using them. But we've also got greater avenues of sharing knowledge, collaboration, better understanding of how to build abstractions. Was the golden age 30 years ago, or is it now?

@mjg59 two words: fractional scaling

@rive You're not wrong and also it really feels like people are taking this as a moment to build their own stuff

@mjg59 yes - said from 100 years in the future

@mjg59 I'm just grumpy because I never cared about any of the things Wayland fixes, but I do care that I've had a nice, usable, accessible, working xmonad config for 11 years , and porting xmonad is, AIUI essentially impossible (if not literally, then by virtue of being an unapproachably massive task).
Old man hates change. Not much of a news story.

@tmcfarlane So keep running that there's literally infrastructure that exists to make that possible

@tmcfarlane And sorry that's kind of flippant but we're in this bizarre space where free software is supposed to meet the needs of users currently using proprietary software but also we're never supposed to change anything ever and there's really no way to manage all of this! But the beauty of free software is that if enough people care the old style can be maintained for as long as they keep caring which is something that just won't happen in the proprietary world

@mjg59 sometimes you can't beat the fear of the unknown and the idea that's "if nothing's broken don't fix it" even if it's 40 years-broke and there are more workarounds than real applications.
Also there was in the past some "adoptions" of new technologies which were not ready for prime time yet (due to them or to applications) and someone got burnt.

@mjg59 i honestly understand bugger all about the technical details but it does seem weird to me that a screensaver is apparently deep magic that cannot be comprehended by Wayland.

@sesquipedality Screen *locking* is actually pretty arcane magic and it's been very bound up in screen saving and as a result the complexity of both has become conflated

@mjg59
When do I get gnome flashback again? If I can access it on Ubuntu already please tell me how. Or at least the four workspaces. My atypical brain needs that.

@liw @mjg59 I would be happy to enjoy Wayland if it was an option for me. As there is no ability to provide prior consent, Wayland is dead on arrival and I have to rip it out of perfectly functioning freshly installed computers just to get a basic feature to work.

https://www.edu4rdshl.dev/posts/solving-the-remote-unattended-access-problem-on-wayland/

This is both a complaint and a fact. I have dozens of computers to manage for work. They are located in a building miles away from home, hidden behind televisions or twelve feet up a ladder. Remote access is required to be able to operate these screens, and Wayland simply refuses to allow it. Somebody would have to go and connect a mouse to the computer to click a stupid prompt to allow me to access the system EACH TIME.

I get the security concerns. I would enable the feature on most of my personal devices (but not all, because then how would I remote into my own desktop???). The issue is that Wayland fails to offer the user any choice in the matter. Business relies on remote access, and Wayland told them “Nah. Go elsewhere.”

@ClickyMcTicker @liw This isn't fundamental to Wayland - whatever is asking for consent could have a pre-provisioned policy.

@mjg59 jwz says he can't port xacreensaver to Wayland and I guess he'd know. It's just so Linux for a bunch of people to go around saying "but we don't need (locking) screensavers any more" which (a) misses the (lack of) point spectacularly and (b) is arguably wrong in the case of OLED screens. I have no skin in the game really and will probably use Wayland when it makes its way into Mint, but it seems like something every other windowing systern can do and the design should have allowed for.

@sesquipedality Devolving locking to the desktop environment makes it harder to plug in a screensaver in the process, but that's also true back in the X days in terms of Gnome taking over screen locking - it's not inherently a Wayland thing (but also Wayland doesn't make it practical to have a separate app take that role on)

@mjg59@nondeterministic.computer look, people want to keep using their own low-quality desktop environments that they grew accustomed to

@ignaloidas @mjg59 this, but unironically.

My "desktop" is ratpoison as a window manager arranging a bunch of xterms and an occasional Firefox. It's not supposed to be a "high-quality" environment, but a "low-noise" one.

One of my projects for 2026 is to build a ratpoison-like window manager with Wayland support. Not sure what to use as a replacement for xterm, ideally it should be something that *doesn't* integrate with a desktop environment, assumes its presence in the background and fails with weird messages to stderr if assumptions do not hold.

@mjg59 There's parts of Wayland that are significantly simpler than the X11 equivalents too.

Once you've decided that you're okay with allocating a few multiples of the framebuffer size, you can remove a lot of complexity. There's no need for all those clip path aware drawing primitives that draw directly into a shared framebuffer

Or a message for "the window that was on top of yours has moved, so please redraw the exposed portion".

@mjg59 @tmcfarlane I thought free software was supposed to meet the needs of the people who use free software

@aeva @tmcfarlane Which of the four freedoms covers that

@mjg59 @tmcfarlane which of the four freedoms says we have to chase after the whims of people who don't use free software anyway?

@aeva @tmcfarlane None of them say we shouldn't, so that's a freedom that exists within free software

@mjg59 unfortunately (as you'll know better than me), just keeping the existing stuff going is only part of the battle. We'd have to keep X11 support going in toolkits, keep key apps on supported toolkits (chrome dropping X11 is not an unimaginable situation), and that's before we get to worrying about hardware.
I suspect I'll be able to keep a desktop system going for as long as I need one, but if users dying off before the code dies is not guaranteed.

@liw @mjg59 Trying Wayland once a year since 2017 and ignoring everything else (including discussions) eventually worked well for me. I went from random lockups and crashes to a stream of annoyances to just the occasional quibble. Now with KDE6 on Wayland, everything works well enough on a daily basis. (Interestingly, I still know next to nothing about Wayland internals, though I learned quite a lot about X over the years..)

@tmcfarlane everything that works now will continue working. The future may not, and that's kind of the fundamental nature of the future - when the people who do stuff choose to do stuff differently we either get to take on responsibility for what they left behind or we follow them

@aeva@mastodon.gamedev.place @mjg59@nondeterministic.computer @tmcfarlane@toot.community meeting the needs of the people is a personal choice. It is provided with no warranties, with no guarantees of it being fit for any kind of use. Because a bunch of people are nice, it doesn't mean that anyone actually needs to be.

@aeva @mjg59 in reality Free Software has only ever really successfully met the needs of people that *write* Free software (and a good chunk of those people decided that they needed to sell software to people that didn't).
Without non-ideological users we're probably still have MetroX and commercial audio, printer drivers and PDF viewers. Mass market has boosted the stuff we care about enormously.

@mjg59 I have mostly stayed out of such arguments, but I am worried that at some point my decades-old fvwm2 setup (which is now hardwired into my brain as What My Computer Should Be Like) will no longer be usable - AFAICT there is (yet?) no Wayland version of fvwm. While distros ship both X11 and Wayland, though, I'm happy :)

@mjg59 Having only used Wayland in my own very esoteric corner, but having read a lot about how it works, I'm honestly shocked by this, because my impression of whether, let's say, VNC was capable of working would rely on which of 17 different and incompatible "extensions" your particular desktop environment happened to implement.

@mjg59@nondeterministic.computer @sesquipedality@mendeddrum.org I'm 90% certain that drawing stuff on top of everything and disappearing could be done with the existing wlr_layer_shell protocol that's implemented by many compositors, so what you need for the rest is to trigger the screensaver on idle. There's org_kde_kwin_idle which would seem to be perfectly fit for that, but only kwin has implemented it so far.

@ignaloidas @mjg59 @sesquipedality There is the ext-session-lock-v1 protocol, but it is not supported in the most widely used compositors (presumably because they implement their own screen lockers): https://wayland.app/protocols/ext-session-lock-v1#compositor-support

Unfortunately, it bundles screen locking (i.e. being responsible for deciding when to unlock the screen) together with drawing the screen saver animation. You can't leave locking decisions up to the compositor and just render the screen saver to a surface when requested.

@mjg59 everything that works will continue to work on the hardware that continues work.
As stubborn as I am, running an unmaintained browser is not something I'd do, and keeping a usable desktop browser going requires a pretty substantial base of committed developers. I'm not sure purely idealogically driven browser dev ever got much beyond lynx/links/elinks.
It's funny though that I do genuinely see a point, if things continue as they are, where the web becomes functionally useless.

@jamesh@aus.social @mjg59@nondeterministic.computer @sesquipedality@mendeddrum.org I wouldn't say that it bundles screen locking with drawing a screen saver - the locking program may do that, but doesn't have to. I believe that even with a screen locker active you could still draw a screensaver with wlr_layer_shell, though I'm not certain because the interaction isn't docu,ented afaik

@ignaloidas @mjg59 @sesquipedality The protocol I linked to definitely combines the two: the client doing screen locking decides when to lock the screen and when to unlock again.

I'm not sure wlr_layer_shell would help here: if the compositor is locking the screen, it's going to hide all the application surfaces. Saying you want your application window to be displayed on top of other application windows isn't going to change that.

@mjg59 yes, but at the same time applications that we are used to (Deskflow, Xscreensaver, workrave, …) are not especially happy.
And now we have to wait until someone implements the APIs to enable these use cases … or worse: someone writes alternatives from scratch.

@mjg59
I dunno guys but figure it the fuck out. Shit or get out of the kitchen.

This limbo twilight hellscape of "wayland is here" but "okay not for reals but soon now" to "wayland is antichrist" needs to end. I'm on mint - wayland disabled by default. Screen control, screen locking, remote desktop etc all work, but buggy as hell *without* wayland. And they say wayland is less stable???

FFS a computer needs to do screens - THIS is why it is not the year of linux taking anything over.

@mjg59 i like it on kde (which i use on live medium, 'cause i have a potato for a pc and have to use debian+xfce to have a non-frustrating experience), and will probably get on it when xfce does so..

@mjg59

It's definitely better than it's ever been. Not perfect, but better.

@wouter@pleroma.debian.social @GyrosGeier@hachyderm.io @mjg59@nondeterministic.computer there's some ideas about doing something like that with Wayback (which is essentially utilizing XWayland to provide a X server that isn't using the most crufty and least maintained parts of Xorg)

@wouter@pleroma.debian.social @mjg59@nondeterministic.computer @GyrosGeier@hachyderm.io I thought there were some discussions about also providing a way for the XWayland to see wayland clients as well, but apparently not and they link to a couple of projects that do what you want in the FAQ https://wayback.freedesktop.org/docs/faq/

@jamesh@aus.social @mjg59@nondeterministic.computer @sesquipedality@mendeddrum.org unless screen locking for you implies sceeen savers (for me it doesn't) I don't see how it would? It's a protocol to have a lockscreen, that handles locking and unlocking. Whether there's something else shown during on idle isn't a concern - the lock screen could do it, but it doesn't have to, if the compositor wants it can show something on top of it

@ignaloidas @mjg59 @sesquipedality Perhaps take a look at swaylock, which makes use of this protocol extension.

The other lock screen stuff it is doing is presenting a password prompt and doing PAM authentication to check that the password typed in is correct.

You could implement an animated screensaver with this protocol extension (since you could draw whatever you want on the lock screen surface you provide). But it would be nice to be able to do so without being required to do auth or make policy decisions about when to engage the lock.

@mjg59 systemd, pulseaudio, wayland. I can draw a line.

I assume this is a generational thing, and the previous point on the line was approximately "linux".

@mjg59 It took a long time for Wayland to get to this stage; 5(?) years ago people were already telling people to switch to it, when it was still missing stuff that lots of people needed and when there weren't many DEs that worked. That put an awful lot of people off; waiting until it actually works before telling people it's time to move would have helped.
What surprises me, was that for a fresh design, how the common functionality was missing at the point people were told it was ready.

@mjg59 I still think the golden age was the OS/2 WPS. ;-)

@demoographics @mjg59 I hope I have inadvertently solved this problem by happening to use the same window manager as Simon Tatham, which I expect means something will be done about it...

(I don't want a desktop environment, I just want a window manager!)

@denisbloodnok @demoographics @mjg59 I am _considering_ writing a Wayland compositor with a similar UX to sawfish, including programmability. I certainly haven't seen anything else remotely like it. It looks like a hell of a lot of effort, but on the other hand, it may still turn out to be my best option!

When I hear this "great variety of desktop environments" line I always suspect that the great variety is in some other dimension that isn't the one I care about. As far as I could see the last time I looked, Wayland compositors consisted of the GNOME and KDE ones, plus a bajillion hobby projects on Github that had the air of "Baby's First Compositor" – make something as simple as possible, which usually means it's a tiling one, and stop developing as soon as it can basically show windows.

I didn't find anything remotely like sawfish, by which I mean
• conventional overlapping, draggable, resizable windows, none of this tiling business
• user-configurable choice of window furniture – not so much the looks as the functionality, e.g. I want a Send To Back (But Don't Minimise) button
• user-configurable treatment of particular applications which I have standard places on the screen I want to make them appear
• combine the two, so I can configure a UI action that does a different thing depending on which application I click it on
• focus-follows-mouse
• ideally, prevent applications from thinking they know better, e.g. Zoom wants to re-grab the input focus whenever someone starts or stops a screen share, and I want it not to be able to do that. (Sawfish can't do that AFAIK but Wayland's architecture looks better suited to it.)

@wouter@pleroma.debian.social @mjg59@nondeterministic.computer @GyrosGeier@hachyderm.io wayback + one of those wayland to x11 basically does this, no? I can't think of a way of exposing only the window management bits of x11

@wouter @ignaloidas ICCCM compatibility layer for Wayland? :>

@jamesh@aus.social @mjg59@nondeterministic.computer @sesquipedality@mendeddrum.org I do know how it's used, but I just see it purely as a part of authentication policy, rather than any kind of screensaver stuff. Showing images instead of the desktop is for security first, with aesthetics taking a backseat. Screensavers meanwhile are primarily an aesthetic device, and have very little to do with the security side of things in my mind. ext-session-lock is very much a security focused extension - it's not the right place to use for aesthetics minded stuff.

@ignaloidas @mjg59 @sesquipedality What I was trying to get across is that parts of the protocol looks almost the right shape for a pluggable screensaver protocol.

The client implementing the animation would need to provide surfaces for each output where it will draw the animation, which is basically what the get_lock_surface() method and ext_session_lock_surface_v1 objects are doing.

What would be different is having the server notify the screensaver client that the screen lock is activating and it should create its surfaces. The server should probably also not send any input to the screen saver's surfaces, and could also composite any unlock UI over the top of that surface.

@wouter @ignaloidas the main problem I see is that window manager decorations are meant to be drawn on demand.

Not a problem for non-reparenting window managers, but anything that creates an extra window for the decorations gets a massive bonus VRAM allocation.

@mjg59 @ClickyMcTicker @liw Indeed, Wayland explicitly isn't concerned with this problem (this is up to the compositor to define). The portal system (which is independent of Wayland) is where this is usually delegated and there's no universal method to handle this, but some backends offer it.

For example, @kde has this feature: https://develop.kde.org/docs/administration/portal-permissions/

@palin @mjg59 I do think a lot of it boils down to "stop changing things, I just want to use what I know" -- which, you know, is not an entirely unreasonable position.

Few people tend to express it that way, though - and it tends to be coupled with "don't change things *except* the things I perceive as being better" where users want the new shiny features X, Y, and Z, but want them piled on top of a 40-year-old foundation that doesn't pass inspection. "What do you mean I have to fix my roof before I can get solar panels?"

@jzb @mjg59 this is one of the reasons, yes. But I think it's not the only one, I mean it's not the same for everyone.