Submitted by The slow rise of Wayland hasn’t really been slow anymore for years now, and today another major part of the Linux ecosystem is making the jump from X to Wayland.
So we made the decision to switch. For most of this year, we have been working on porting labwc to the Raspberry Pi Desktop. This has very much been a collaborative process with the developers of both labwc and wlroots: both have helped us immensely with their support as we contribute features and optimisations needed for our desktop.
After much optimisation for our hardware, we have reached the point where labwc desktops run just as fast as X on older Raspberry Pi models. Today, we make the switch with our latest desktop image: Raspberry Pi Desktop now runs Wayland by default across all models.
↫ Simon Long
Raspberry Pi Desktop already used Wayland on some of the newer models, through the use of Wayfire. However, it turned out Wayfire wasn’t a good fit for the older Pi models, and Wayfire’x development direction would move it even further away from that goal, which is obviously important to the Raspberry Pi Foundation. They eventually settled on using labwc instead, which can also be used on older Pi models. As such, all Pi models will now switch to using Wayland with the latest update to the operating system.
This new update also brings vastly improved touchscreen support, a rewritten panel application that won’t keep removed plugins in memory, a new display configuration utility, and more.
I have long (>2,5 years) been using Wayland on all of my graphics-enabled Raspberries, with no glitches. In fact, I switched my main environment over to Wayland precisely because it vastly outperformed X on my ARM hardware (the raspberries, but most important, my Lenovo C630 laptop). Not a glitch… Never looked back.
As someone who’s experienced the wayland problems, I’m not opposed to this switch as long as things are working. All of my gripes with wayland were pragmatic in nature. Reports suggest the latest version of KDE should solve the last of my wayland issues, and if so that’s great! I’m just waiting for it to come down the debian stable pipeline. I also have some Cinnamon and xfce desktops, those might take more time before being ready. I really wish screen sharing would have been a global wayland feature that just worked everywhere like in X. Alas, it is what it is.
They do always seem to gloss over the legacy issues and focus on look we fixed the latest and greatest. I understand that, but it’s not going to help me or get more of my users on Linux. I must have a bullet proof OOB RDC solution that doesn’t need a PhD to implement.
FWIW, I’ve been using Wayland on both a Pi 4 & 5 for quite a while now, I’ll have to test the update on the older model and see if the claims stack up.
So as promised I tried several Raspberry PI 4 and 5 devices all operating headless needing remote terminal and RDC. I tested with the native Pi client as well as several commonly used 3rd party solutions, not a single combination worked reliably and not one functioned for more than 3 hrs without failing.
Furthermore, switching to LabWC broke many previous OS features that had been working flawlessly for many months on end, and on some older but very common hardware 3rd party drivers failed rendering various expansion like cases and hats with M.2 or PoE disabled.
Sucked in again!
So I’ll spend the next day hopefully reverting to X via raspi-config without the need to re-image boot devices.
cpcf,
I haven’t tried this on rpi, but this process is familiar to me on my computers. I genuinely hope the problems are sorted, but I’ve stopped proactively checking claims. After all my systems are working fine with X and there’s no pressing reason to switch. The next attempt for me will probably be when debian stable includes the next version of KDE.
Aside, I have a question for you: does RPI5 performs well enough to be useable as a real desktop?
I’ve been impressed with just how much linux software works on ARM, but the performance was never really on par.
@Alfman
I should have known better.
Regarding the RPi 5 as a desktop, I’d say no, in that a ten year old Core i3 or i5 outperforms the RPi 5 comfortably, I’ve even got a 2008 Core 2 Duo I keep here for reference and I’d say that is still more performant than the RPi 5 as well. Of course they all suck down energy at a vastly greater rate.
My main reason for using the RPi is as a test platform for industrial sensor / control applications, we put them in place and lock them down serving up their own data analysis. As an aside, we’ll be moving to the 2350 due to built in security features.
cpcf,
That’s my experience with RPI4, but I didn’t know if it was still the case for RPI5. There are a lot of alternatives to RPI, with better hardware and enclosures but then you take on the risk of bad support. Overall I’d say the ARM ecosystem gets a failing grade here. This is why, even though I do like the efficiency of ARM, I still end up buying x86 minipcs instead.
Yes, this is exactly where ARM SBCs really shine. I bought several rpi zero-w for this purpose. Much faster than a microcontroller with the benefit of a genuine linux stack.
This is the kind of progress I like to see with Wayland! It’s good to know that labwc is performant on older/slower hardware as well.
Now if only the Wayland devs can start caring about essential protocols like RDC, that’s the only real blocker for me to switch to using it daily. Right now all RDC clients I’ve tried under Wayland fail to render the remote desktop.
@Morgan
Yes let’s hope so, I’m keen to give it a try but because I’ve heard this all before(old hardware performance improvements) I’ll reserve my opinion. Until RDC works reliably it’s a no go for me on anything but the test beds.
So 16 years after initial release yet another of the GNU/Linux players decided to make a switch to Wayland. On where each and every project needs to do that on their own. Great, isn’t it? Perfect strategy to hold GNU/Linux on desktop back? Pure perfection.
Well, not every project needs to implement Wayland… The main distributions need to. Other than that, only if yours implements a graphical environment completely of its own, you will have any issues adapting.
Debian has long shipped Wayland by default on some desktop environments (IIRC, for GNOME it has been the default since Debian 11, or maybe even 10). OK, maybe at the beginning it was a bit bumpy, as some use cases were no longer supported (i.e. remote desktop, which I happen to use). But things have smoothed out.
Raspberry Pi OS is a Debian derivative. They do modify their graphical environment, although not fundamentally; it was logical for them to change, as it’s no longer sensible for them to keep delaying the step.
MATE is getting closer, I’m running it fully in Wayland mode now. There’s a few rough edges to smooth out but it’s already very usable.
Any chance of a news covering on why Wayland failed to deliver? As we still seem to get articles written in denial, on how it’s a best thing after sliced bread and just about any time now it will take off. Some in depth article on why it failed from technical, governance, idea point of view, on why it isn’t used in lets say Android … To get some idea on what is wrong with Wayland, to hopefully improve on that and for being able to move forward. Wasting another decade before doing that is in my opinion not a good thing to do.
Just open your text editor and write something better.
So i guess anybody can be a journalist and writer these days? I mean i could do that but i would much rather prefer if it would be done with people having much deeper knowledge into this subject. For example i can only assume on why Wayland was never being considered for Android, some people out there know, for a fact, on why. Then i read things like we don’t want for Wayland to become another Xorg, that is one solution for all and instead each project should do their own implementation. It seems this strategy is failing so people that actually want to keep it like it is, one would need to interview them, as a basis for such article? On why. This is a job for some more prominent journalist, isn’t it? If there is to be some weight behind it. As the goal would be to try to improve in the future, as this can’t be it? 16 years later and nobody rally happy with it. Dangling. Or is this it, as good as it gets and that is that, as it was imagined? We should be happy?
Geck,
X11 is old and bloated, so I didn’t really have an issue replacing it. But wayland shot themselves in the foot with bad execution. They should have been listening to user needs up front and reevaluated breaking critical features. The covid stay at home mandates required millions of employees to have access to remote desktop and wayland desktops had objectively failed us. At least now those gaps are being filled, but better management could have solved a lot of user problems a lot sooner. Oh well, it will all be forgotten and we’ll move on.
Write something better than Wayland. Get coding and stop complaining on the internet.
That’s not how it works. If someone paints graffiti on your house, you don’t tell the owner “if you don’t like it paint something better”.
It is time developers realised the have responsibility to the community and they can cause considerable damage.
Gnome3 sent Linux desktop back a decade (mostly by abandoning the Linux desktop of the time – Gnome 2). Wayland is on track of doing the same and fragmenting Linux desktos into multiple incompatible OSes. A profit funded company would have killed a project within a year.
I wonder if Gnome or Wayland teams are getting any funding from MS or Google. Correction: Google does sponsor Gnome Foundation. I guess that’s peanuts for keeping Linux desktop divided.
Initial Wayland deployments were either Wayland demos showing off fading transitions or round window corners (BTW, all Wayland features could have been implemented with a few X11 extensions) or walled-garden DEs written with disregard to desktop usability conventions and interoperability.
An actual progress is happening now, driven by people who want to build usable DEs, or simply port existing ones. It is an ugly process with a lot of infighting between thinly wailed hostile factions. I expect it to take a better part of a decade and things will get a lot worse before getting better.
We will likely end up with hybrid X11/Wayland desktops, with applications/toolkits supporting both and features being gradually unified. This is happening already – XWayland is becoming tightly integrated with compositors and there is a good reason for it: a compositor is acting as a display server to XWayland and at the same time it is being its X11 client (a window manager). Ideally both would be in the same executable, using the same mainloop.