New X.Org Server Release While Maintaining Separate XWayland Being Discussed

[oiaohm]

However look at it this way, if BSD didn't use X11 but used Boogaloo as its display system. How would Linux's choice of preferred display system change the relevance of the BSD display system? It wouldn't. So if Linux does lose access to X11, this would not really have much of an impact on FreeBSD's use of Xorg or OpenBSD's use of Xenocara.

This has made a very critical error. The core graphical kernel drivers for all the BSD based distributions covering over 80% of PC systems in existence come straight from the Linux kernel tree ported to BSD systems normally over 12 months behind. So if Linux is no longer testing their drivers with a solution like what freebsd or openbsd uses by the time freebsd or openbsd or any of the bsd locate the problem the programmer with the knowledge to fix it could have moved to another company.

So there will be quite a big impact on BSD long term as Wayland comes more used in the Linux world. Xenocare is OpenBSD own port of X.org X11 server. Reality up until the start of Wayland FreeBSD/OpenBSD/other BSD and Linux were all on something x.org based so meaning defect found by one most likely effected all. This is changing with Linux moving to Wayland first.

[kpedersen]

This has made a very critical error. The core graphical kernel drivers for all the BSD based distributions covering over 80% of PC systems in existence come straight from the Linux kernel tree ported to BSD systems

It is true that FreeBSD does have a Linux compat layer for Linux DRM drivers. However the OpenBSD approach is quite different and ports the drivers in a more meticulous (but better integrated manner). This is less affected by the Linux display system than one might imagine.

Another example of this is, Haiku which follows a similar approach and yet doesn't use Xorg at all. Their graphics stack is actually pretty good with a fraction of developers (albeit without 3D acceleration but that is from Mesa rather than display server). Arguably because Wayland is simpler than Xorg, it will actually be easier to untangle and port drivers from.

[Azrael5]

In the near future, also PLASMA is going to be based on Wayland. Xorg is a dead project because it is no longer maintained nowadays. Gnome is based on Wayland. Ubuntu and Fedora applies Wayland apis as default. Nvidia, the major responsible of this delay, is going to release a new driver which implements GBM in order to provide Wayland compliance. When I read about certain comments supporting a legacy graphical stack such as Xorg I understand that people miss in objectivity acting as stupid fanboys. Wayland is ready, unlike many graphical environment of many Oses which must be adapted to integrate Wayland as well as many other applications. By 2/3 years, the main Linux Oses will be Wayland based. Wayland is much more efficient and make Linux better when it is well implemented.

[Mez']

I recommend to start searching for developers who will spend their time working on this dead project - number of commits says it all (exlude XWayland of course). Maybe you'll find some under a bridge.

There you go: "dead".

[oiaohm]

It is true that FreeBSD does have a Linux compat layer for Linux DRM drivers. However the OpenBSD approach is quite different and ports the drivers in a more meticulous (but better integrated manner). This is less affected by the Linux display system than one might imagine.

Another example of this is, Haiku which follows a similar approach and yet doesn't use Xorg at all. Their graphics stack is actually pretty good with a fraction of developers (albeit without 3D acceleration but that is from Mesa rather than display server). Arguably because Wayland is simpler than Xorg, it will actually be easier to untangle and port drivers from.

Haiku is stuck of software rendering every year with Haiku there is the question how they are going to change out of this and every year they fail. There is a lot of complexity in doing hardware 3d acceleration including many different hardware quirks. 2D support is fairly simple to pull off 3d support is a real problem.
https://discuss.haiku-os.org/t/plans...eleration/7272
Yes this one of the many debates about how in heck they are going todo it with Haiku. The reality is like it or not the existing Haiku graphics stack is not up to the job and one of the considered options is to rip it completely out and copy what freeBSD has done and just port in Linux drivers.
https://discuss.haiku-os.org/t/is-ha...racts/10427/11
Yes they have voted on going gut the current solution and just use Linux drivers a few times is now if they get the man power to-do it.

OpenBSD approach results in them being even more behind than FreeBSD in hardware support as compared to Linux.

Reality the complexity of 3d drivers really does rule out reinventing the wheel as you will just not have enough developer hours heck taking a existing 3d driver from another platform to your own also still risks not having enough developer hours.

[kpedersen]

Yes they have voted on going gut the current solution and just use Linux drivers a few times is now if they get the man power to-do it.

This will be very interesting to see what they succeed with here. To be honest it could also have far reaching consequences with what the BSDs do if Wayland ever starts to take a foothold.

What I can't quite see Haiku doing however (almost by definition) is dropping their existing user-land stack and going with a Wayland compositor.

OpenBSD approach results in them being even more behind than FreeBSD in hardware support as compared to Linux.

Offtopic: Very true. It is a shame because it has ended up with a really clean result. You can even see a difference just booting it up in the framebuffer console, FreeBSD gets the resolutions all wrong, has conflicts with the weird logos at the top, flickers around and has strange spaces in the console just as it kicks in. Basically with FreeBSD, it works but it feels ratty as hell. With OpenBSD it is completely seemless. Luckily with Intel and AMD, the card support is still very reasonable.

By 2/3 years, the main Linux Oses will be Wayland based..

It has taken 12 years to get 10% user share. I think 2/3 years is optimistic

Xorg is a dead project because it is no longer maintained nowadays.

Half of useful open-source software is no longer maintained. It only takes one guy to pick it up again once things start to break. Then suddenly it is maintained again. A lot of commercial software is unmaintained, you just don't hear about it because it is behind closed doors.

When I read about certain comments supporting a legacy graphical stack such as Xorg I understand that people miss in objectivity acting as stupid fanboys

Like Autodesk and Valve right?

[tomas]

To be honest it could also have far reaching consequences with what the BSDs do if Wayland ever starts to take a foothold.

This is happening as we speak. As I said, Ubuntu 22.04 is going to have Wayland by default and the next Nvidia driver will finally fix Wayland support. Those two together will have a huge impact on the adoption of Wayland.

It has taken 12 years to get 10% user share. I think 2/3 years is optimistic

Can you please stop citing this misleading and frankly false statement? The specification for the wayland *protocol* was published 12 years ago. A specification, you know like text? Not actually any implementation, you know like code?
Actual implementations of said protocol started to appear about 6 years ago in the form of support in toolkits (gtk and qt) and then in window compositors like mutter, kwin/plasma and sway.
[QUOTE]

Like... Valve right?

Yes, what about Valve?

https://www.phoronix.com/scan.php?pa...escope-XDC2020

Do you think Valve is somehow opposed to Wayland for some reason? They will update Steam to run with Wayland natively when the time is right, and that time is fast approaching with the next release of Ubuntu 22.04.

[oiaohm]

It has taken 12 years to get 10% user share. I think 2/3 years is optimistic

https://linux-hardware.org/?view=os_display_server

The speed here is not constant over that 12 years. There is a big speed up at 2019-2020 this lines up with release of RHEL8 that default to Wayland outbox. Each distribution to default to wayland out the box will speed up Migration.

Half of useful open-source software is no longer maintained. It only takes one guy to pick it up again once things start to break. Then suddenly it is maintained again. A lot of commercial software is unmaintained, you just don't hear about it because it is behind closed doors.

That the thing we get to see behind closed doors with open source so know when stuff it unmaintained and the future trouble coming.

Like Autodesk and Valve right?

Autodesk Linux products due to be certified fro RHEL8 and higher are in fact certified to work with XWayland. Remember 2019 RHEL8 released with Wayland default that has seen large number of the commercial closed source either have Wayland support or at least have test support on XWayland so their software don't need bare metal X11 server any more. Yes starting 2019 RHEL8 out box does not install bare metal X11 server it only installed XWayland for X11 support. Bare metal X11 is install optional packages with RHEL8 that are not marked for long term maintainer-ship.

Valve with gamescope is very interested in Wayland.

Nvidia dragging feet on Wayland support has been a problem.

[kpedersen]

This is happening as we speak. As I said, Ubuntu 22.04 is going to have Wayland by default and the next Nvidia driver will finally fix Wayland support.

No, at that point we were discussing non-Linux platforms. Of course Ubuntu (as a Linux distro) is going to (finally) use a Linux-centric display system. It is what Haiku will do that is interesting.

Autodesk Linux products due to be certified fro RHEL8 and higher are in fact certified to work with XWayland.

Well yeah but that is kind of admitting defeat. Isn't the whole point of a migration to actually migrate from X11?
If popular wayland compositors ever dropped support for an Xserver layer, Autodesk simply would not support them at this point.

As with your discussion on RHEL, they are still committing to X11 (albeit with a different backend rather than Xorg). This certainly isn't an expression to migrate.

They will update Steam to run with Wayland natively when the time is right

When the time is right will be a long time. Then that is just the start, many of the games within the catalog are reliant on X11. They don't all base on SDL2. It will simply never happen. You will either be stuck with X11 (via Xwayland), possibly longer than the lifespan of Wayland (i.e Xnext) or you will simply be happy for those titles to be unplayable / removed from your collection.

[oiaohm]

Well yeah but that is kind of admitting defeat. Isn't the whole point of a migration to actually migrate from X11?
If popular wayland compositors ever dropped support for an Xserver layer, Autodesk simply would not support them at this point.

As with your discussion on RHEL, they are still committing to X11 (albeit with a different backend rather than Xorg). This certainly isn't an expression to migrate.

This does depend what you goals are. XWayland allows X11 server instance per application. You do need to look at gamescope here. This is a hard reality Wayland compositors are stack-able. So if your bare metal wayland compositor does not support XWayland you can run gamescope on top of bare metal wayland compostiro and have Xwayland support because gamescope is loaded. So even if popular wayland compositors decided not to support XWayland item does not break.

XWayland failure is if graphics drivers does not support it and valve has been working on a plan to counter that. The reality here XWayland is a official part of RHEL8 from 2019 that means has to be maintained until 2029 by RHEL 8 for the 10 years of official support. Yes each major RHEL release is resetting this. Autodesk choice not to support Wayland directly is at least 10 years into the future.

Major reason for XWayland over bare metal X11 is the security faults fixed by not having a global bare metal X11 server. Like being able to enter a password without any random running application being able to keylog it. And other security fixes like that.

When the time is right will be a long time. Then that is just the start, many of the games within the catalog are reliant on X11. They don't all base on SDL2. It will simply never happen. You will either be stuck with X11 (via Xwayland), possibly longer than the lifespan of Wayland (i.e Xnext) or you will simply be happy for those titles to be unplayable / removed from your collection.

Please remember valve supports dos titles.

https://store.steampowered.com/curator/35328265-Powered-by-DOSBox/

Yep 221 dos based games on sale in the steam store. Some of the games on sale in the steam store were on sale before the first copy of Windows sold.

The reality is gamescope is setup as the DOSBOX equal for X11 applications by valve. Right down that old X11 applications will not support modern screen res or colour depths so will need a scaler and a colour depth conversion. So valve has really planned ahead here right out to the point compositors start dropping support XWayland. Yes work on zink by valve is to cover when platforms start only shipping with Vulkan with no Opengl also was going to help of Nvidia was going to keep on not providing support for XWayland.

kpedersen I have tried to be clear that its X11 bare metal that has been unmaintained and is going away unless something changes. When it comes to Xwayland there are interest parties in maintaining Xwayland between RHEL and Valve and others there is quite a huge budget to get it right. X11 support for running legacy applications with Valve involvement I expect to be dead before that support ends. X11 support for running X11 windows managers/DE I do suspect that could be gone inside a few years remember Xwayland has cut away most of the x11 protocol that X11 windows managers need to function. Xwayland is a very cut back X11 protocol.

X11 applications will require some changes to support Xwayland like if they do screen capture or the like that things will have to work differently. Some of these changes could result is us getting the reverse where some applications work on Xwayland but will not work on bare metal X11 server any more.