Slashdot Mirror
Ubuntu Dumps X For Unity On Wayland
An anonymous reader writes "Canonical and Ubuntu founder Mark Shuttleworth has announced that Ubuntu will move away from the traditional X.org display environment to Wayland — a more modern alternative. The move means there is now little reason for GNOME developers to recommend Ubuntu as an operating system. Shuttleworth said, 'We're confident we’ll be able to retain the ability to run X applications in a compatibility mode, so this is not a transition that needs to reset the world of desktop free software. Nor is it a transition everyone needs to make at the same time: for the same reason we'll keep investing in the 2D experience on Ubuntu despite also believing that Unity, with all its GL dependencies, is the best interface for the desktop. We'll help GNOME and KDE with the transition, there's no reason for them not to be there on day one either.'"
...but I still know a LOT of people who forward X over SSH, and there are still a lot of professors who are advising their students (at least in the engineering schools I have seen) to do the same. I guess this is one of those times that just saying, "I use Linux!" will not convey what people think.
Palm trees and 8
This article reads less like a news story than an emotional, personal rant by someone who's puckering with contempt because he got his feelings hurt.
Tech companies make crappy decisions all the time. Ubuntu probably thought it would have more time to become the king of the desktop before realizing that soon the desktop would be irrelevant and that *nix alternatives had already beaten it to the punch for being the kings of mobile.
At this point, he should probably start thinking further down the road to gesture and voice computing. My kinect tells me that it's almost time to stop touching devices at all, and I believe it.
I remember a discussion a year or two ago here on Slashdot how X was badly in need of replacing. Sounds to me like Canonical have the right idea, and the impetus to make it happen.
which is totally what she said
There have been other projects over the years that have tried to improve on X (Fresco/Berlin and picogui readily come to mind) but I don't believe any of them have demonstrated results that seriously threatened the revitalized Xorg project.
I hadn't heard of Wayland, but I must admit since Xorg got going I haven't kept a close eye on that level of the graphics stack. Mark's blog post makes it sound like they're willing to ditch network transparency for better graphics effects, which makes me a little leery. Undoubtedly for most users that's the "right" approach, but if they do lose network transparency it's going to make Ubuntu an impossible choice in a lot of business environments where running apps from a server is part of day-to-day business.
Also, the amount of work to port all the requisite software/toolkits to a non-X platform is going to be... impressive. Haiku faces this problem, as do a fair number of older applications when looking at running native on Windows and OSX - it ain't easy. Plus, we're talking an entirely new backend in Wayland, one that's going to require (from the sound of things) rock solid OpenGL support.
Ubuntu has shown they can deliver in the past, and perhaps they can do it now, but I can't help but wonder if they realize the magnitude of what they're undertaking here.
"I object to doing things that computers can do." -- Olin Shivers, lispers.org
I'd suggest it's mostly to do with the move to compositing window managers, which X was not designed for.
The most specifically obvious example - video.
Video is these days presented by slapping a video texture on a window. If the render of the video frames to the texture isn't done simultaneously with the render of the video frames to the screen, you see "tearing", or a line where one frame ends and another frame begins, something you shouldn't see at all.
The same phenomena happens with composited windows, which are also rendered textures ; drag a window around at speed on a poorly configured X / Compiz desktop and you will see the tearing where two frames of the desktop texture are being rendered out of sync with the screen.
Many users now have more than one screen. X cannot render a single desktop to two screens and keep both synced. While it's possible to get one screen synced, it's not always easy, and getting both screens synced seems to be impossible.
Now, this clearly isn't a hardware limitation, because Windows can and does do this right - I have never seen application windows or video tear on the Aero desktop. This is on the same nVidia card, both operating systems running the official nVidia drivers.
This is one of the few things about Linux that annoys me when I compare it to Windows. The other is PulseAudio - but I have workarounds for my PulseAudio problems. It just looks sloppy to have great big tears in your otherwise very pretty composite display, and if you want to enjoy a movie, you are either stuck with using only one of your screens.
Backwards compatibility is achieved by (optionally) running X(server) as a sub-process of Wayland.
Isn't this inherently inefficient?
X is designed as a network-transparent windowing protocol first, with optimizations to improve performance on a local display added-on.
If you start with a display system that's optimized for local display, and then implement, on top of that, a network-transparent display system, there's no reason the implementation of the latter should be inherently less efficient than its direct implementation - unless the display server or the compatibility layer are implemented badly, or there's some level of incompatibility in the basic concepts of the two systems that makes a compatibility layer difficult to achieve.
I know very little about Wayland - but if it's largely based on DRI and OpenGL, then implementing X on top of that shouldn't have a significant negative impact on X performance.
Personally I'm not sure how I feel about moving away from a network-transparent rendering system. It's something I've grown used to in my years using X (about 1996-present). That alone is enough to make me uncomfortable with the change. I don't relish the idea of moving to a system where some apps will support remote display via X and others won't - or where I might have to choose between an X version of an app and a Wayland version... It reminds me of the situation on my Windows machine at work: choosing between Win32-native, and Cygwin/X versions of packages...
Though, on the other hand, how frequently do I actually use this feature? I use it for Emacs and a few other things, and that's about it. I never attempt running Firefox or Blender or GIMP or VLC remotely via X, I always just run those on the local machine. If my experience really is typical, then the network-transparency feature of X is being underused, typically, to the extent that it's not worth making it a design priority. (And, actually, I think people tend not to design Linux GUI apps with remote display in mind. I think they're more commonly developed for a local display, with the result that their behavior might be a bit too network-intensive or latency-sensitive to work with a remote display...) It might really be better to optimize for local display and then have remote display via a special layer: VLC or whatever else.
The Wayland FAQ was kind of interesting to read. It's interesting what they have to say about X's legacy baggage, for instance. Of course, I've heard a lot of this stuff before... I remember "Berlin" and GGI as a previous attempt at roughly the same thing. Maybe Wayland will yield a better result in the end? I don't know.
Bow-ties are cool.