Funny, I was literally just looking at this issue today. I switched to wayland a few weeks ago, just because it seems to save a lot of battery power. But so much basic functionality is still broken it's infuriating. And then so many things seem to follow the "unix functionality" and as a result they're half baked tools that aren't really working properly IMHO.
A lot of these issues have fixes or workarounds directly in KDE.
For reference I'm using sway. swayidle is an idle daemon, where the maintainer doesn't think that it matters whether its connected to power or not. You're supposed to create your own scripts around it that handles ac connect and disconnects.
There's a tool to do flux like color warmth setting. One of them doesn't allow you to toggle, so you have create your own toggle script that kills or restarts it. The other one is controllable, but doesn't actually account for time or timezone.
XWayland has had 2 or 3 patches to handle hidpi when the main wayland screen has fractional scaling, none of them are merged and they seem hardly active. KDE works around that by allowing you to turn it off for xwayland clients. Sway just passes it down and blurs everything.
When I exit a wayland session and then restart it the screen locks up. This doesn't happen with normal X.
And then there is electron. Slack is not the only app that ignores electron settings and doesn't run with wayland support. In chrome and electron it's supposedly supported but you have to toggle it yourself? What is this madness?
These things seem like basic functionality for me. I don't really get it. Sure, maybe I shouldn't expect a proper experience for random sway tools, so that makes the first two points irrelevant. But the fact that years in they still haven't found a proper solution on passing down hidpi for xwayland? That's incomprehensible for me.
>For reference I'm using sway. swayidle is an idle daemon, where the maintainer doesn't think that it matters whether its connected to power or not. You're supposed to create your own scripts around it that handles ac connect and disconnects.
>One of them doesn't allow you to toggle, so you have create your own toggle script that kills or restarts it.
If you're not a fan of making scripts then sway (and wlsunset I assume) is not for you. My sway install is held together by a myriad of scripts - keybindings are handled by a script, starting programs via bemenu goes through a script, sway itself is started by a script, etc. That's just how this compositor and its ecosystem are designed - they're for people who want the software itself to do the smallest possible thing with the least possible assumptions so that the user can do whatever they want to fill in the rest.
The problem isn’t that it’s held together by individual scripts and apps that one can replace or hack at. The problem is that none of these pieces ship in place.
Ideally you’d have some sort of offering that gives you a reasonable starting place with all these composable pieces in place, and then you can hack/replace/add/remove as needed.
Everyone starting from scratch doesn’t make sense. It’s like shipping a Unix system with no app doing it’s own paging because the devs rightly expect you to use some other tool to page output for you, but there’s no less/more installed (and no back buffer in the terminal or terminal emulator to boot). Then you complain and the answer is that you’re supposed to write your own (or research, download, compile, and install something someone else might have hacked up).
> The problem is that none of these pieces ship in place.
Sway is not a desktop environment. What you are describing is a desktop environment. I too would like a sway-based desktop environment. Let's stop complaining that sway is not what it is or that somehow this is related to wayland. This whole thing is just a consequence that writing cohesive graphical/desktop software is tedious and that tinkerers cannot rely on the myriad of Xorg desktop environment, in particular the modular ones like lxde/xfce to mix and match components.
> Then you complain and the answer is that you’re supposed to write your own
Yes, obviously. No-one is saying this would be a bad project. And if you blatantly misunderstand their project's scope it's no wonder you're getting dry responses.
I think gp and gggp are arguing that the diffusion of responsibility means that nothing takes responsibility for baseline functionality, and as such there are huge gaps.
With X all these things are in X or the single shared X compositor itself. Yes, it's bloated and sometimes lacks flexibility in configuration or doesn't work perfectly, but there's no gaps.
I don't think the author is saying these individual projects should do these things, but if they don't, who does? It means that everyone using wayland effectively is using dwm, and dwm isn't for everyone (or even most people?). It feels like you're saying users should suck it up and learn to love the dwm philosophy, C and all, which doesn't seem quite right to me.
In practice, Linux distributions have been the system integrators that pull these things together and handle the gaps. That's really the value of a distribution - pulls a set of components together in a way that works.
In this case, I use Fedora (Gnome). Most things are on Wayland and it works pretty well.
I am not saying sway isn’t allowed to do what they did. But the normal approach is for the 1st party dev to also provide reference implementations of supporting/complementing utilities to smooth adoption.
No one is saying they have to be shipped as part of the compositor, obviously.
Says who? You're getting software for free, from someone who is not beholden to you for anything. Either you accept it as it is, or you fix it; that's the nature of open source. Whining about it on the internet is rude and disrespectful.
If you can't be bothered, and want all this stuff included, or at the very least recommended or installed automatically on the side, switch to something else that does that for you.
The alternative to taking a piece of open source as-is isn't "complain to the maintainer until they fulfill your wishes". That kind of thing is why open source maintainers burn out. You are not entitled to anything from them; the alternative is that the software doesn't exist at all.
It is not the normal approach for a project author to write a whole ecosystem.
If you want the Desktop Environment experience, pick a Desktop Environment. This can be based on sway (e.g. Manjaro spins), but sway is not more a desktop Environment than the Linux kernel is an operating system.
I don't see how being "just a compositor" has to do anything to "has to bundle complementing utilities", without mucking the whole thing and turning it into a desktop environment.
Thought I agree that for most people, a desktop environment, "everything integrated, just change this specific thing" instead of "let me just DIY my entire environment from those separated parts" makes more sense.
afaik there are various distros that will give you a preconfigured out of the box swaywm based desktop env. Garuda linux comes to mind.
The sway wiki is pretty detailed and iirc has a helpful page full of options for various "not the compositor's job" tasks like notification daemon, app launcher, etc.
The reason why Chrome and Electron still need a flag is simple: Chrome still has a decent chunk of work to do before Wayland is supported to the same degree as X11. They had a lot of false starts: it took years before Wayland support actually got upstreamed in any regard.
Likewise for Electron. It does generally work to the point of daily-driver usability today with stable Chrome and Electron, but some apps are shipping old and out-of-support versions of Electron, so unfortunately it has become a bit of a waiting game.
Anyways, you can't actually do proper DPI virtualization for X11 apps because afaik there is no protocol for apps to communicate the scale that they are actually rendering at, so there will never be full support for mixed or fractional DPI in XWayland to the same degree that it is supported in Wayland. I'm not 100% sure there would be a point to making such a protocol at this point, since blurry-but-correct surfaces seems like an OK trade-off to keep things working in the long term. If you run real X11 with mixed DPI or fractional DPI, you get quite a range of potential failure cases depending on what apps you run.
At this point most of my screen is filled with native Wayland apps, and Electron apps that use reasonably up-to-date Electron (like VS Code), so it's rare to see a blurry window.
ChromeOS uses Wayland as a bridge between Android and Chrome because it's a politically neutral technology neither of them control, and thus can both agree on implementing. It's not about integrating better with other Wayland compositors or the wider Wayland/Linux ecosystem, it's about using Wayland as a piece of technology to sling buffers back and forth between two different camps who are otherwise on bad terms.
ChromeOS implements a Wayland compositor to support Android apps running as Wayland clients. Chrome being a well-behaved Wayland client is something completely different.
I'm not familiar with the situation, but my guess would be that ChromeOS has created its own Wayland compositor with a bunch of private protocol extensions that Chrome on ChromeOS uses to get stuff done. But for some reason or another, those protocol extensions are not suitable for submitting as new public extensions for other compositors to implement. Or they are suitable for submission, but the official extension roadmap already has different extensions in the works (but not yet complete) that will do the same thing.
Electron has different issues it's been working on, though. I believe it has some issues with window border rendering in some cases as well as issues with cursor size on fractional scale.
If you want full desktop integration why are you using sway. Most of the functionality you're talking about requires writing scripts if you used i3 or awesome WM on X11 as well so has nothing to do with wayland.
Regarding exiting and restarting a wayland session, how do you do it? GDM, the console? It obviously works for most other people so it seems you're encountering a bug related to your hardware or setup so you could file a bug report with the appropriate project (again doesn't really seem like a wayland issue though).
> For reference I'm using sway. swayidle is an idle daemon, where the maintainer doesn't think that it matters whether its connected to power or not. You're supposed to create your own scripts around it that handles ac connect and disconnects.
> There's a tool to do flux like color warmth setting. One of them doesn't allow you to toggle, so you have create your own toggle script that kills or restarts it. The other one is controllable, but doesn't actually account for time or timezone.
I used to write little tools like his and my ~/bin is still littered with scripts from 10 years ago
Nowadays I want this functionality to work by default and I won't settle for a less usable system
> And then so many things seem to follow the "unix functionality" and as a result they're half baked tools that aren't really working properly IMHO.
I'm going to disagree pretty strongly with that. In fact, I think most of its problems are that it threw the Unix philosophy out the window. While it's true that it suffers from having made everything a implementation detail that can theoretically be plugged in with a different component, the problem is that they forced everything important to go through the compositor, which bottlenecks composiblilty and doesn't allow you to actually replace individual components without support from the (usually incomplete) monolith in the middle.
Yes, actually; the X server was awful, but it was a single contained point of awfulness that allowed everything else to be trivially interoperable. Breaking that apart could have been good except that the way they actually did it mixed far too many concerns into the graphical server (compositor), so while we have more of them, in practice you can't actually mix and match the rest of the system anymore.
Not sure what you mean; wlroots still assumes that the window manager and compositor is in the same process. Wayland itself assumes that those things are also in the same process as the display protocol/server.
With X11, if someone wanted to write a window manager, they could do that and only that, without having to write a compositor or display server. Wlroots certainly simplifies those latter two bits, but there's still a lot of work to do to use it.
Beyond that, Wayland's mess of extension protocols are generally incomplete. I've been working on "porting" parts of Xfce to Wayland, and there are a significant number of things that just cannot be done at all right now without integrating it directly into the compositor process. The extension protocols necessary to do even a fraction of this table-stakes stuff are often experimental or non-standard, and even when they do exist, are often missing critical pieces.
As an example: say you want to write a dock-type app that only shows the list of toplevel windows present on the current workspace (and when you switch workspaces, it'll show the windows on the new workspace). Nope, you can't do it. The workspace protocol itself has been languishing in Freedesktop GitLab[0] for two and a half years, and even if that was standardized, it doesn't interact with the Foreign Toplevel protocol at all, so you have no idea what windows are on which workspaces.
So currently the only solution is: your dock app has to live in the same process as the compositor, where it has access to all the compositor-specific data around workspaces and windows. (Sure, you could build a d-bus interface, or something like that, but that would be specific to your one compositor... and a d-bus interface is the moral equivalent of a Wayland protocol, which is what you should be using anyway.)
> It is a display protocol, responsible for displaying things. A compositor is a must for that
Er, are we using different meanings? Obviously something has to display pixels, but I literally do run X without a compositor.
> a window manager, and there is no reason why that can’t be done with this model (and is pretty much what wlroots does).
No, you can't (excepting rootfull xwayland, obviously), Wayland forces the display server, compositor, and window manager to be a single program. All wlroots does is make it easier to write that compositor, it hardly changes the model.
The price is tearing. Which may not be too noticeable at 1920@60fps, but playing a video on a bigger screen at higher refresh rate will very very seriously tear. It is not an accident that every other OS is compositor-only.
> Wayland forces the display server, compositor, and window manager to be a single program
Only the first two. You can allow a third-party program to control your windows’ state via any IPC mechanism you choose.
Sure, there are trade offs in skipping the compositor. But the claim was:
> It is a display protocol, responsible for displaying things. A compositor is a must for that, it is not additional complexity at all.
Which is just false; a compositor isn't a must unless you move the goalposts beyond "displaying things".
> You can allow a third-party program to control your windows’ state via any IPC mechanism you choose.
I guess you could, by adding a lot of API surface; has literally anyone ever done this, or is it just a hypothetical solution? Because I'm not aware of any existing compositor that lets you plug in a window manager.
Wayland compositors have provisions for skipping the compositing part when a single client is running full screen. So, by your argument, Wayland is also "not a compositor".
I don't think that follows. You can run a fully-functional X session, with an arbitrary number of windows, without any compositing. It's not some special case reluctantly added after people pointed out that the initial "you must fully double-buffer every single pixel every single time" wasn't actually universally a good idea. But even at face value... okay, so you'd like to extend the argument to say that not even Wayland actually forces you to composite? Because that's agreeing with my argument, just more strongly.
Except that nearly everything that makes the usual years-out-of-date Wayland gripe lists here are ALL specified protocols now, it's just a matter of time until they're implemented widely enough to put this to rest.
Fractional scaling, hotkeys, gamma control, screen capture, etc, all have protocols.
I don't believe that this will happen in practice in any event. First, implementations have... let's say personality conflicts? That is, I would be shocked if GNOME replaces its screenshot protocol with the "standard" any time soon. Second, even when it's true it requires every single compositor to actually connect all the bits up. For instance, I don't use QWERTY if I can help it. So a little while back, I tried a new compositor that was based on wlroots, which supports changing keyboard layouts. Guess what I couldn't do, because that specific compositor hadn't gotten to plumbing that specific feature through? Guess what works on every single X11 window manager, no matter how barebones, because X11 can manage separation of concerns?
It's just early. It's fast to make little scripts and glue it together yourself, and until folks really enumerate use cases and build to meet them, it's going to feel unfinished. It takes years to build out a big project like this.
If you want "just works", stick to X and a mature desktop environment like KDE or Gnome for now (especially if you do any gaming).
This stuff “just worked” in the late 90’s (and probably earlier, assuming access to unix workstation hardware) with X11, when it was roughly the age Wayland is now.
I have two displays on my computer, one is 4k and the other is 1080p. One has HDR. One is 400 nits brightness, the other is not. Games are built across several engines with an amazing collection of rendering technologies. We have apps that embedded web browsers.
I'm sure if you restricted yourself to strictly first party native applications and a single display within one or two resolutions and no DPI scaling, Wayland probably meets most of your needs now.
XWayland doesn't really help you with things that are part of the desktop environment. You can't run a desktop background/icons manager or a dock/panel in XWayland, as it won't have access to the kind of information it needs to implement all its features.
> For reference I'm using sway. swayidle is an idle daemon, where the maintainer doesn't think that it matters whether its connected to power or not. You're supposed to create your own scripts around it that handles ac connect and disconnects.
Yeah, the existing tools are pretty low level. For now you need to write your own "scripts" for matching different states from different places (e.g.: idle + battery charging state). Hopefully someone will build higher level tools on top of the existing tools. The APIs are all there (i.e.: there are no technical limitations, it's just that nobody has done it yet).
> There's a tool to do flux like color warmth setting. One of them doesn't allow you to toggle, so you have create your own toggle script that kills or restarts it. The other one is controllable, but doesn't actually account for time or timezone.
I've been using GNOME on Wayland for over a year and have literally none of these issues other than X fractional scaling not being great. But even then fractional scaling is good enough that I generally don't notice it. (But all of my daily apps are Wayland native)
> swayidle is an idle daemon, where the maintainer doesn't think that it matters whether its connected to power or not
Sounds like a problem with the tool, not Wayland.
> color warmth setting
This is a general issue on Wayland but GNOME has this built in. I'm surprised that KDE doesn't.
> In chrome and electron it's supposedly supported but you have to toggle it yourself?
I guess Chromium's Wayland support isn't considered production ready? I don't use Chromium often but running on XWayland hasn't bothered me much. I guess this would be more annoying if it was your primary browser.
scripting around hobby projects to fit your needs is the hacker ethos/mindset/what makes using linux/unix fun. But give the ecosystem time and Sway and its ecosystem will be just as good as X and i3 is.
The other one is controllable, but doesn't actually account for time or timezone.
Alright first off: file a bug! How am I supposed to know there is a user with a problem, if they don't tell me!
That said not sure what you mean by this exactly... gammastep 100% does account for the local time. But it does choose the time of day to transition based on the location. Are you in a different location than your timezone is set to?
XFree86 (and later, x.org) have always had better support for hidpi rendering than MacOS X and Windows (by my definition, which is based on the quality of tear-free vector, font and bitmap rendering at non-standard DPIs while using mainstream hardware).
The wayland developers apparently couldn’t figure out
how to read the X11 manual. Why expect them to reimplement it in a superior way?
You might want to research who the first wayland developers were and are now. You might be surprised. Let's just spoiler, that they not only read the X11 manual.
IIRC, a big chunk of that was the trend that led to wayland causing "effective deprecation" of some more complex aspects of X11, like replacing multihead with windows into unified frame buffer (which now required single DPI for all screens while multihead allowed completely different settings)
Xorg does support having multiple monitors with different DPI, however applications need to use the RandR extension to obtain that per-monitor information as an older API that provided per-screen information (which doesn't make sense for Xinerama anyway) is hardcoded for backwards compatibility reasons (an attempt was made a couple of Xorg releases ago to remove the hardcoded value and present a more appropriate value but it broke a bunch of programs so it was reversed).
Intent is one thing, whenever userland GUI developers actually uses the API the intended way is another. It's not a bug, it's just a undocumented feature.
If everyone uses the API the "wrong way", now you're stuck supporting the "wrong way" or everything breaks and people WILL complain.
So your issues are basically exclusively with X, X clients and most of all, proprietary X clients that are using major-revs-old versions of Electron. Yeah, that sounds like a real drag, glad I use Wayland and only have Ledger Live that I'm forced to use twice a year.
> But the fact that years in they still haven't found a proper solution on passing down hidpi for xwayland? That's incomprehensible for me.
It's crazy how this is incomprehensible despite it being explained quite clearly every time it comes up. There might be patches, but they are incomplete and they are kludgy and the experts that could push it through ARE WORKING ON IMPROVING WAYLAND. If it remains incomprehensible, go read through the fractional scaling protocol merge and see how much discussion there was for designing that protocol, not even considering the X interop scenario you want.
You can: (1) try really hard to convince volunteers to spend huge amounts of time supporting something they deemed dead years ago, or (2) try to get shitty proprietary companies to, you know, fund a single developer part time to do basic maintenance, (3) use less crappy tools and/or demand that you aren't forced to use shitty tools.
Why is it ALWAYS that people reach for (1)? I swear to god I hope it's folks that don't do OSS development themselves. The entitlement and ignorance makes me head spin. But hey, random OSS devs are much easier to rage at than your CTO, or some random "chat startup" that doesn't give a fuck. So rock on with your ire against OSS devs not working on your explicitly deprecated scenario.
A lot of these issues have fixes or workarounds directly in KDE.
For reference I'm using sway. swayidle is an idle daemon, where the maintainer doesn't think that it matters whether its connected to power or not. You're supposed to create your own scripts around it that handles ac connect and disconnects.
There's a tool to do flux like color warmth setting. One of them doesn't allow you to toggle, so you have create your own toggle script that kills or restarts it. The other one is controllable, but doesn't actually account for time or timezone.
XWayland has had 2 or 3 patches to handle hidpi when the main wayland screen has fractional scaling, none of them are merged and they seem hardly active. KDE works around that by allowing you to turn it off for xwayland clients. Sway just passes it down and blurs everything.
When I exit a wayland session and then restart it the screen locks up. This doesn't happen with normal X.
And then there is electron. Slack is not the only app that ignores electron settings and doesn't run with wayland support. In chrome and electron it's supposedly supported but you have to toggle it yourself? What is this madness?
These things seem like basic functionality for me. I don't really get it. Sure, maybe I shouldn't expect a proper experience for random sway tools, so that makes the first two points irrelevant. But the fact that years in they still haven't found a proper solution on passing down hidpi for xwayland? That's incomprehensible for me.