Wayland screenshots

Which desktop has Wayland in Ubuntu? I have MATE desktop installed on Ubuntu 20.04:
Code:
root@ubuntu20:~# uname -a
Linux ubuntu20 5.11.0-27-generic #29~20.04.1-Ubuntu SMP Wed Aug 11 15:58:17 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux
root@ubuntu20:~# lsb_release -a
No LSB modules are available.
Distributor ID:    Ubuntu
Description:    Ubuntu 20.04.3 LTS
Release:    20.04
Codename:    focal
root@ubuntu20:~# top -n 1 | grep -i wayland
root@ubuntu20:~# top -n 1 | grep -i xorg
   1211 root      20   0 1203380 113864  77744 S   6.2   1.6   0:07.04 Xorg

Edited to add: Answered my own question:

"Ubuntu ships Wayland as default in Ubuntu 17.10 (Artful Aardvark).[80] Ubuntu reverted to X.Org for Ubuntu 18.04 LTS, as Wayland still has issues with screen sharing and remote desktop applications, and does not recover as well from window manager crashes.[81][82] Ubuntu ships Wayland by default in 21.04.[83]"

Reference: https://en.wikipedia.org/wiki/Wayland_(display_server_protocol)#Desktop_Linux_distributions
Yep. And I used the default DE, which is kinda GNOME3... or is it Unity? I don't know, I like it anyway. Mean, prefer it over KDE.
But the point is, the underlying thing is Wayland.
But there's one more thing. For most apps, I guess, they use Xorg-translation layer or something. Because I also used FVWM2 there, and it won't run on Wayland, you understand. But with that layer it runs all right.
EDIT: just checked in my bhyve Ubuntu (Live-ISO). It's Xorg, no Wayland.
Am I surprised? Can't say I am :D :D :D
 
Yep. And I used the default DE, which is kinda GNOME3... or is it Unity? I don't know, I like it anyway. Mean, prefer it over KDE.
But the point is, the underlying thing is Wayland.
But there's one more thing. For most apps, I guess, they use Xorg-translation layer or something. Because I also used FVWM2 there, and it won't run on Wayland, you understand. But with that layer it runs all right.
Thanks for bring this to my attention. I'm curious enough now to attempt the upgrade to 21.04 even though it is not yet an LTS version. I've never had much luck with Wayland before, and mainly just play around with Ubuntu out of curiosity anyway.
 
Thanks for bring this to my attention. I'm curious enough now to attempt the upgrade to 21.04 even though it is not yet an LTS version. I've never had much luck with Wayland before, and mainly just play around with Ubuntu out of curiosity anyway.
Here you go:
For developers and innovators, Ubuntu 21.04 delivers Wayland and Flutter for smoother graphics and clean, beautiful, design-led cross-platform development.
 
We don't yet have an Ubuntu port, do we? ;) ;) ;)
Anyway, one can make this more of a FreeBSD thing by running it in bhyve...
Which is, actually, the most proper way of running any non-BSD OS on a modern computer...
 
We don't yet have an Ubuntu port, do we? ;) ;) ;)
Anyway, one can make this more of a FreeBSD thing by running it in bhyve...
Which is, actually, the most proper way of running any non-BSD OS on a modern computer...
Shan't be spending much time on this. My interest in byhve is presently non-existent, interest in Wayland and Ubuntu, only peripheral. FreeBSD has long been my preferred OS for most things I do. I'm retired but have some old web application software I still support, on bare-metal installations only, of FreeBSD, Debian, Ubuntu, and Linux Mint, mainly just for my own amusement. It's the primary focus of my continued interest in computers. Well, that plus youtube videos and a few other media streams I enjoy. :D
 
Ok, Ubuntu 21.04 Live CD does NOT run Wayland. Perhaps, during the installation you're offered...
Not running Live CD, rather, upgrading my existing 20.04 LTS test install, which I can easily redo later at my leisure.

"... Ubuntu 21.04 with native Microsoft Active Directory integration, Wayland graphics by default..." ~ https://ubuntu.com/blog/ubuntu-21-04-is-here

Edited to add: ha well the upgrade finished but it's still using Xorg. Might try again later with a fresh install, but starting to lose interest, and may very likely wait until it goes to an LTS version. Honestly don't have much confidence in the future of either Wayland or Ubuntu, and, as Linuxes go, tend to prefer Debian, or even Mint, which strangely outperforms Ubuntu in most areas, in my estimation, and this in spite of being a derivative thereof. But then again, even Debian has been slipping over the past few years-- so-- long live FreeBSD...

ETA 2: Just realized, duh, that "Wayland graphics by default" implies the Gnome3 desktop-- which is the Ubuntu default DE-- duh--and that's a deal-killer as far as I'm concerned. I'll be sticking with MATE for the foreseeable future, and leaving the doublespeak to their PR departments, where it, too, is the default...
 
For me Wayland is just another of these things with the big promise to "change the world for better", which never gets finished, is stuck in development like forever (~13 years by now) and therefore never gets relevant traction. Just like Btrfs in the Linux kernel as "ZFS killer." Some people might still believe that, the rest just uses ZFS instead since ages.

At best it's a nice proof of concept, but that's then already all about it.
 
Well out of curiosity I've watched some talks about Wayland, and it gives me still some mixed feelings. On the positive side it tries to resolve some problems which are really there. Most and foremost having to rely still on X11, and that X11 is not exactly the best environment to really draw nice windows without having some nasty artifacts now and then. Also X11 is quite the heavy beast, and well not the most efficient things, baroque, ugly, whatever.

Personally I found this talk about libinput, which has some fundamental diagrams of X11 and Wayland in it quite enlightening:

View: https://www.youtube.com/watch?v=HllUoT_WE7Y


On the other side so under Wayland there's now only two things left: the Compositor, which is basically doing all screen drawings plus window management, and its applications. So something like window manager plus X11 server in one package.

Since every compositor is different this means that stuff, which are doable in Compositor A might not be doable in B and vice versa. This is why for input devices libinput was created, to handle all input devices. But still this can be something which might bite some programmers now and then.

The other problem are dedicated GPUs. AMD and Intel do use GBM as technology to talk to the GPUs, while Nvidia uses EGL. So if somebody wants to use the proprietary drivers of AMD or Nvidia he's got to make sure first the compositor of choice uses that technology. If not supported most people will probably then switch to another compositor with the required support.
 
Since every compositor is different this means that stuff, which are doable in Compositor A might not be doable in B and vice versa. This is why for input devices libinput was created, to handle all input devices. But still this can be something which might bite some programmers now and then.
I think from the outcome of the talks, your view on it is pretty much spot on. This part in particular I feel there could be a good solution.
  • A single compositor that everyone uses. Anyone doing their own should be seen as odd or niche / hobbyist (like writing your own Xserver).

  • This compositor is pretty much blank in terms of looks and user-functionality but provides every feature (Xwayland, waypipe, EGL hookup, etc).

  • A proper shim library (perhaps looking like libX11 or libXCB would be convenent for backwards compatibility but not 100% necessary) that everyone uses in a standardised and common way to provide the user-facing functionality (so basically the smaller window manager part ontop of the standard compositor base)
At the moment we are leaning towards this. Most projects use wl_roots (which is almost like that standard compositor). However underlying toolkits are not using it (it doesn't provide client functionality so much). Gtk, Qt and SDL2 as the main ones are very ad-hoc, thus non-standard, duplicated work etc.

My worry is that as terrible as it sounds, I don't think modern developers (including interests, funding, ability) have the discipline to pull it off. They are a scattered bunch and something like this probably needs to be developed by a commercial entity. Commercial entities these days are also not quite what they once were. They only care about cloud and cheap monetization ventures. Everyone is in too much of a rush to do something actually impressive and useful
 
I think from the outcome of the talks, your view on it is pretty much spot on. This part in particular I feel there could be a good solution.
  • A single compositor that everyone uses. Anyone doing their own should be seen as odd or niche / hobbyist (like writing your own Xserver).

  • This compositor is pretty much blank in terms of looks and user-functionality but provides every feature (Xwayland, waypipe, EGL hookup, etc).

  • A proper shim library (perhaps looking like libX11 or libXCB would be convenent for backwards compatibility but not 100% necessary) that everyone uses in a standardised and common way to provide the user-facing functionality (so basically the smaller window manager part ontop of the standard compositor base)
At the moment we are leaning towards this. Most projects use wl_roots (which is almost like that standard compositor). However underlying toolkits are not using it (it doesn't provide client functionality so much). Gtk, Qt and SDL2 as the main ones are very ad-hoc, thus non-standard, duplicated work etc.

My worry is that as terrible as it sounds, I don't think modern developers (including interests, funding, ability) have the discipline to pull it off. They are a scattered bunch and something like this probably needs to be developed by a commercial entity. Commercial entities these days are also not quite what they once were. They only care about cloud and cheap monetization ventures. Everyone is in too much of a rush to do something actually impressive and useful
FWIW, Xorg started life as X Window System back in 1984 (v. X1 was released back then), and it was developed at MIT, before being spun off as XFree and then as Xorg. X11R6.5.1 was only publicly released in 2000. Wayland seems to be taking just as long to gain traction.
 
FWIW, Xorg started life as X Window System back in 1984 (v. X1 was released back then), and it was developed at MIT, before being spun off as XFree and then as Xorg. X11R6.5.1 was only publicly released in 2000. Wayland seems to be taking just as long to gain traction.
Indeed, it came from academia where they could take a step back and approach it from a potentially more creative standpoint rather than the majorities usecase (these days is playing games, reading email, web browsing). They also considered integration and how it could be utilised by existing institutions.

However, the very early days is kind of where Wayland is now. Where X11 really took off was (pre-Xorg) when it was adopted by a number of companies and standardised together. Check out the initial copyright comment of this file:

https://sourceforge.net/p/cdesktopenv/code/ci/master/tree/cde/lib/DtTerm/Term/Term.c

You can see the kinds of stakeholders that X11 technology had (not just CDE, these are all part of the OpenGroup that pushed Motif, Xt and things).

I don't think IT companies are able to work with one another these days. They are too interested in pushing their own agendas and locking in users with the least amount of actual effort. Open-source is possibly the only solution we have left but it is also simply too unorganised for what is needed.
 
On a recent Ubuntu install, when you log in, you can choose Gnome on Wayland. (The opposite of Fedora which runs Gnome on Wayland by default but you can choose Gnome on X).
 
First time I bothered to look at this thread. Isn't this like asking for a screenshot of sh in the console as if sh is different looking than zsh in the console?
I disagree, the screenshots of Wayland are a proof that it's working, somebody did get it to work, that it is on the way to becoming a drop-in replacement for Xorg. Evidence of working Wayland instances are like a way for Wayland movement to gather steam.
 
I disagree, the screenshots of Wayland are a proof that it's working, somebody did get it to work, that it is on the way to becoming a drop-in replacement for Xorg. Evidence of working Wayland instances are like a way for Wayland movement to gather steam.
Or you can just say so. I'd believe them.
 
I see 2 actual screenshots out of 43 posts so far. Are these "pure" Wayland or Xwayland?
How can I know the difference ?

Note there are at least 3 functional window managers : sway,river,labwc. I find sway nested within labwc interesting.
Note : labwc has no bar.
 
there are at least 3 functional window managers [...] river, labwc
Possibly don't use it for anything important. From its website:

"Note: river is currently early in development. Expect breaking changes and missing features
[...]
To enable experimental Xwayland support pass the -Dxwayland option as well."


And it seems labwc doesn't support fullscreen yet. It is basically on par with TWM :)

Also note that they are *not* window managers. They are entire compositors which is equivalent to 1/5 the work of an Xserver. They are very large project scopes for individual developers so I can't confidently say I have much trust in them to succeed. But good luck to them. i3 and emulations like Sway are fairly poor tiling environments (even the tiny DWM has more heuristics) so it will be good to have alternatives.
 
Back
Top