After more than forty years, everyone knows that it’s time to retire the X Window System – X11 for short – on account of it being old and decrepit. Or at least that’s what t…
And X11 will never be ready for most modern users. They have different goals. But that’s the thing with open source. As long as someone somewhere needs it. Even if 90% of us don’t need X11 for legacy software. It will still be here.
Even if 90% of us don’t need X11 for legacy software. It will still be here.
I most agree with you. The Xlibre project may become popular and do something to make X11 popular again. Who knows?
And I just argued on a forum yesterday that Xorg will keep working for 20 years at least. But a lot of smart people claimed I was wrong about it being able to support new hardware. But I think Xorg is likely to build and run for decades yet.
But the X server implementation that is likely to last the longest is Xwayland. And with Wayback, the “stand-alone” X server that many distros will bundle will be Xwayland running on Wayback (Wayland) and not Xorg.
As I have said elsewhere though, few people will be daily driving an X server (Xorg, Xlibre, or Wayback) simply because many desirable applications will require Wayland.
And what will be the x11 only applications that will make people run an X server to use them? Xeyes? Xfig?
I think even running Xwayland will be pretty niche. X11 is going to be a software preservation project. You can boot up OpenLook, CDE, Trinity, or i3 for the memories (and then go back to Wayland for the apps you need).
I could be wrong. Time will tell. Within a couple of years after the release of GTK5 at the latest, we will know. By 2030 maybe.
I spent 4 years with and external monitor on my desk that I couldn’t use because it was absolutely painful to find a consistent way to make the 2 different DPIs of the screens work in a way that made sense. Only now with proper Wayland can I enjoy and use it. Yeah there’s hacks, but I’d rather let it be dead in a corner than try to work around it. It was a bunch of black screen, inconsitencies between the order I’d plug the external screen, when i did it (before or after logging in), etc… I can’t even imagine all the other pain points about hdr, variable or high refresh rates, etc.
Wayland is great.
Had to wait a bunch of time and tried many times before and it wasn’t ready for my needs, but now it is and I’m happy. God knows how many rants I’ve done on fedi about it not working for a lot of time on plasma and weird bugs everwhere.
As a user of 3 monitors with different resolutions, different refresh rates, some HDR, different UI scaling, who games and wants to use VRR - Wayland is literally why I was able to effectively switch to Linux as my daily driver.
a consistent way to make the 2 different DPIs of the screens work in a way that made sense
What do you mean? I used multidisplay setups for 15 years, I never checked what’s the DPI of my monitors is and never had issues. I just plug in any external monitors I have around and it works. I did it on desktop machines and many different laptops. I’m always baffled when people say their monitors don’t work because of sync rates or DPI. What are they trying to do and what’s not working?
I am not sure about that. They have different designs for sure. Mostly because one was designed 25 years later. I guess you mean they have different goals because Xorg did not incorporate some goals in its design (like security). But is it a goal of Xorg to be insecure? That feels like a stretch.
There are design goals in X11 that are not included in Wayland. Take asking the display server to draw primitive shapes for you as an example. But modern X11 apps do not do that either. That is not how things like Qt and GTK work. So, more of a “25 years later” thing than a true difference in goals. The “compositor” approach. The DDX layer. These are more of a reflection of “how things work today” on both systems than they are differences in goals.
Perhaps you mean things like “network transparency” as I hear that one a lot. Wayland’s design is to have a simple core that can be extended. But the same capabilities exist for Wayland. For example:
What goal does Xorg have that Wayland does not? Again, other than poor security (not a goal).
The lack of security in Xorg makes many things easier. Wayland apps run in a sandbox which makes some things harder. Many complaints I see ultimately boil down to this difference. Flatpaks are also sandboxed and a lot of the solutions on Wayland are similar (eg. XDG desktop portal). But again, am not sure crap security was really a “goal” of Xorg. It is simply convenient.
Because of security, things have to be explicitly supported on Wayland while X11 apps can just do them. There is no official way to capture a screenshot on X11 even after 40 years. But any X11 app can do it pretty easily as all apps have access to the entire display (even contents of other windows). On Wayland, there is a protocol for screen capture. There has to be, or it would not be possible. The same is true for many other features. And, I fully admit, some protocols for Wayland to do things done by some x11 apps do not exist yet (or are not yet widely supported by compositors or apps).
But again, I do not really see “poor security” as an x11 design goal. It was simply born in an era where that did not matter as much. Projects that want to modernize X11, like Xlibre, will have to break things on X too. Time will tell what that looks like.
X11 is a display server. Wayland is a presentation layer. Different goals. I have run graphical multi-seat systems using x11. Something like that will never be possible in the same way for Wayland because it is out of the design scope
Xorg is a display server. In Wayland, your compositor is the display server.
“I have run graphical multi-seat systems using x11. Something like that will never be possible in the same way for Wayland”
I have to give you this one. Wayland is not designed to be multi-seat. I do not know about “never” but you are right that multi-seat is a design difference.
I need it to run like 3 things via its original use case of “log in to remote computer, run it on linux, see it on your local machine”. still works like a charm.
Yep absolutely. It’s been years since I’ve done that myself. But there’s lots of Legacy software out there. Especially on Legacy systems that are not being developed for at all anymore. That will continue to require X11. One of the other more Niche uses which Wayland doesn’t support I believe are multi graphical users on a single system. Again probably something I don’t think I’ve messed around much with in the last decade. But it was a fun feature. Wayland is much more focused on a single session.
The biggest problem is for new users. Once the dust has settled and Wayland is the default for everything (and there’s plenty of searchable threads for how to fix X problem) then it will be great. But currently if you’re a noob and you install a distro you don’t know what either is. If you have this problem do you fix it with X or Y? Choice is great for enthusiasts, but just another hurdle for new users.
We’ll get there. Honestly I think in the long run Wayland will be easier to troubleshoot and maintain. But then that may just be memories of troubleshooting XFree86 back in the 90s. I still have flashbacks.
Most new users won’t even know that there is a choice until they’re presented with it, and most will just stick with the default option anyway. (which most distros have/are switching to wayland)
Currently, X11 is not really being developed, just maintained, which is the real issue. In this piece they are questioning whether Wayland was a good choice or not. I am using Wayland, have for some time, and I do acknowledge it is still a work in progress, validating the articles list of ‘issues’ yet to be addressed, but unless you are running a really old system, I am guessing the complications affect a very minimal group of users. There are also workarounds, for example on KDE, the gtk apps don’t adhere to those using the global menu. However, there is a fix to get around it.
In reference to using a completely different solution, isn’t it a little late in the game (16 years in development?) I think we are stuck with Wayland, no?
X11 would have needed almost a complete rewrite. Wayland made sense. Eject the technical debt and focus on your use case. We aren’t time sharing on a large central mini computer/mainframe anymore. And even then they generally are full single user systems run in parallel under a hypervisor these days. As wasteful as that might be.
But there’s still occasions when you need to run a legacy application on old AIX, Irix, etc, or vax Hardware. And need a workstation. Which right now Wayland simply can’t do without x.
And X11 will never be ready for most modern users. They have different goals. But that’s the thing with open source. As long as someone somewhere needs it. Even if 90% of us don’t need X11 for legacy software. It will still be here.
I most agree with you. The Xlibre project may become popular and do something to make X11 popular again. Who knows?
And I just argued on a forum yesterday that Xorg will keep working for 20 years at least. But a lot of smart people claimed I was wrong about it being able to support new hardware. But I think Xorg is likely to build and run for decades yet.
But the X server implementation that is likely to last the longest is Xwayland. And with Wayback, the “stand-alone” X server that many distros will bundle will be Xwayland running on Wayback (Wayland) and not Xorg.
As I have said elsewhere though, few people will be daily driving an X server (Xorg, Xlibre, or Wayback) simply because many desirable applications will require Wayland.
And what will be the x11 only applications that will make people run an X server to use them? Xeyes? Xfig?
I think even running Xwayland will be pretty niche. X11 is going to be a software preservation project. You can boot up OpenLook, CDE, Trinity, or i3 for the memories (and then go back to Wayland for the apps you need).
I could be wrong. Time will tell. Within a couple of years after the release of GTK5 at the latest, we will know. By 2030 maybe.
I spent 4 years with and external monitor on my desk that I couldn’t use because it was absolutely painful to find a consistent way to make the 2 different DPIs of the screens work in a way that made sense. Only now with proper Wayland can I enjoy and use it. Yeah there’s hacks, but I’d rather let it be dead in a corner than try to work around it. It was a bunch of black screen, inconsitencies between the order I’d plug the external screen, when i did it (before or after logging in), etc… I can’t even imagine all the other pain points about hdr, variable or high refresh rates, etc.
Wayland is great.
Had to wait a bunch of time and tried many times before and it wasn’t ready for my needs, but now it is and I’m happy. God knows how many rants I’ve done on fedi about it not working for a lot of time on plasma and weird bugs everwhere.
As a user of 3 monitors with different resolutions, different refresh rates, some HDR, different UI scaling, who games and wants to use VRR - Wayland is literally why I was able to effectively switch to Linux as my daily driver.
What do you mean? I used multidisplay setups for 15 years, I never checked what’s the DPI of my monitors is and never had issues. I just plug in any external monitors I have around and it works. I did it on desktop machines and many different laptops. I’m always baffled when people say their monitors don’t work because of sync rates or DPI. What are they trying to do and what’s not working?
I am not sure about that. They have different designs for sure. Mostly because one was designed 25 years later. I guess you mean they have different goals because Xorg did not incorporate some goals in its design (like security). But is it a goal of Xorg to be insecure? That feels like a stretch.
There are design goals in X11 that are not included in Wayland. Take asking the display server to draw primitive shapes for you as an example. But modern X11 apps do not do that either. That is not how things like Qt and GTK work. So, more of a “25 years later” thing than a true difference in goals. The “compositor” approach. The DDX layer. These are more of a reflection of “how things work today” on both systems than they are differences in goals.
Perhaps you mean things like “network transparency” as I hear that one a lot. Wayland’s design is to have a simple core that can be extended. But the same capabilities exist for Wayland. For example:
https://www.mankier.com/1/waypipe
or even better:
https://github.com/wayland-transpositor/wprs
What goal does Xorg have that Wayland does not? Again, other than poor security (not a goal).
The lack of security in Xorg makes many things easier. Wayland apps run in a sandbox which makes some things harder. Many complaints I see ultimately boil down to this difference. Flatpaks are also sandboxed and a lot of the solutions on Wayland are similar (eg. XDG desktop portal). But again, am not sure crap security was really a “goal” of Xorg. It is simply convenient.
Because of security, things have to be explicitly supported on Wayland while X11 apps can just do them. There is no official way to capture a screenshot on X11 even after 40 years. But any X11 app can do it pretty easily as all apps have access to the entire display (even contents of other windows). On Wayland, there is a protocol for screen capture. There has to be, or it would not be possible. The same is true for many other features. And, I fully admit, some protocols for Wayland to do things done by some x11 apps do not exist yet (or are not yet widely supported by compositors or apps).
But again, I do not really see “poor security” as an x11 design goal. It was simply born in an era where that did not matter as much. Projects that want to modernize X11, like Xlibre, will have to break things on X too. Time will tell what that looks like.
X11 is a display server. Wayland is a presentation layer. Different goals. I have run graphical multi-seat systems using x11. Something like that will never be possible in the same way for Wayland because it is out of the design scope
X11 and Wayland are both protocols.
Xorg is a display server. In Wayland, your compositor is the display server.
“I have run graphical multi-seat systems using x11. Something like that will never be possible in the same way for Wayland”
I have to give you this one. Wayland is not designed to be multi-seat. I do not know about “never” but you are right that multi-seat is a design difference.
My mind goes to this project again: https://github.com/wayland-transpositor/wprs
But wprs only runs one compositor so it does not inherently address multi-seat. Support for that would need to be added.
I need it to run like 3 things via its original use case of “log in to remote computer, run it on linux, see it on your local machine”. still works like a charm.
Yep absolutely. It’s been years since I’ve done that myself. But there’s lots of Legacy software out there. Especially on Legacy systems that are not being developed for at all anymore. That will continue to require X11. One of the other more Niche uses which Wayland doesn’t support I believe are multi graphical users on a single system. Again probably something I don’t think I’ve messed around much with in the last decade. But it was a fun feature. Wayland is much more focused on a single session.
The biggest problem is for new users. Once the dust has settled and Wayland is the default for everything (and there’s plenty of searchable threads for how to fix X problem) then it will be great. But currently if you’re a noob and you install a distro you don’t know what either is. If you have this problem do you fix it with X or Y? Choice is great for enthusiasts, but just another hurdle for new users.
We’ll get there. Honestly I think in the long run Wayland will be easier to troubleshoot and maintain. But then that may just be memories of troubleshooting XFree86 back in the 90s. I still have flashbacks.
Most new users won’t even know that there is a choice until they’re presented with it, and most will just stick with the default option anyway. (which most distros have/are switching to wayland)
Currently, X11 is not really being developed, just maintained, which is the real issue. In this piece they are questioning whether Wayland was a good choice or not. I am using Wayland, have for some time, and I do acknowledge it is still a work in progress, validating the articles list of ‘issues’ yet to be addressed, but unless you are running a really old system, I am guessing the complications affect a very minimal group of users. There are also workarounds, for example on KDE, the gtk apps don’t adhere to those using the global menu. However, there is a fix to get around it.
In reference to using a completely different solution, isn’t it a little late in the game (16 years in development?) I think we are stuck with Wayland, no?
X11 would have needed almost a complete rewrite. Wayland made sense. Eject the technical debt and focus on your use case. We aren’t time sharing on a large central mini computer/mainframe anymore. And even then they generally are full single user systems run in parallel under a hypervisor these days. As wasteful as that might be.
But there’s still occasions when you need to run a legacy application on old AIX, Irix, etc, or vax Hardware. And need a workstation. Which right now Wayland simply can’t do without x.