Slashdot Mirror

← Back to Stories (view on slashdot.org)

Ubuntu 18.04 LTS Will Default To The X.Org Stack, Not Wayland (phoronix.com)

Posted by msmash on from the tough-calls dept.

An anonymous reader writes: Five years after their original goal to ship Ubuntu with Wayland, Ubuntu 17.10 transitioned to using the Wayland display system by default as part of their transition to GNOME Shell as the default desktop. But with the upcoming Ubuntu 18.04 LTS release, Canonical has decided to transition back to the X.Org Server. Their reasoning for moving to an X.Org Server by default is better support for screen sharing, remote desktop, and better recovery from crashes. But for those interested the Wayland session will still be available as a log-in option.

35 of 194 comments (clear)

  1. See Saw Cycles of Adoption and Abandonment by CRB9000 · · Score: 2, Interesting

    Really, these things swing back and forth and really is a non-news item. The headline should be changed to read: Defaults to X.org Allows Choice of Wayland. Which is not really newsworthy.

    In other words, "Meh."

    1. Re:See Saw Cycles of Adoption and Abandonment by Viol8 · · Score: 5, Insightful

      "Which is not really newsworthy."

      Well it is actually. Various vested interests have been plying the X Windows is dead, Wayland is the way forward line for a few years now. For Ubuntu - a distro not exactly known for its conservatism and aversion to releasing bleeding edge sofware - to return to X as the default graphics system is a pretty obvious statement that Wayland is a long way from being ready for prime time.

    2. Re:See Saw Cycles of Adoption and Abandonment by jouassou · · Score: 2

      It's not necessarily that it's a long way from prime time. But it's definitely not stable enough at the moment to be the default in an LTS version, which has to run stable for the next 3 years without relying on any major upgrades.

    3. Re:See Saw Cycles of Adoption and Abandonment by Junta · · Score: 5, Insightful

      Well, it's a bit more than that. The statement has been in various circles 'wayland is good enough *today*, you don't need xorg anymore'

      This is acceptance that people do have things they can't do in Wayland, and it needs to be opt-in rather than opt-opt to avoid bad user experience.

      It's not 'wayland will *never* be better', but it is a statement that it has a ways to go, and some of the limitations are design choices that will require interesting conversations, particularly about security with regards to screen sharing.

      --
      XML is like violence. If it doesn't solve the problem, use more.
    4. Re:See Saw Cycles of Adoption and Abandonment by chill · · Score: 3, Interesting

      This is called the Tyranny of the Default, and is a real thing.

      --
      Learning HOW to think is more important than learning WHAT to think.
    5. Re:See Saw Cycles of Adoption and Abandonment by Fly+Swatter · · Score: 4, Interesting

      A vendor in it for money chooses to backtrack and default away from the new and shiny. I don't think that is about choice and a hell of a lot more about what they think of the current state of Wayland.

    6. Re:See Saw Cycles of Adoption and Abandonment by Anonymous Coward · · Score: 5, Insightful

      At some point that 40 year old code based designed around a 386 when the software had to draw even the primitives just isn't going to cut it anymore.

      You got the reality exactly backwards. A codebase designed around being efficient enough to run on a 386, and field-tested on every thing imaginable for 40 years.
      Anything that today's wonder boys can generate, is destined to be a steaming pile in comparison - just because they never saw a reason to learn what efficiency or portability even is.
      Barring a major miracle, X will stay with us for quite some time yet.

    7. Re:See Saw Cycles of Adoption and Abandonment by Spacelord · · Score: 3, Insightful

      > that 40 year old code based designed around a 386

      It's less than 30 years, and you can say exactly the same about Linux.

    8. Re:See Saw Cycles of Adoption and Abandonment by Viol8 · · Score: 2

      Yeah, its so easy to hack thats why it's hackers favourite attack vector into *nix systems. Oh, wait...

    9. Re:See Saw Cycles of Adoption and Abandonment by Lord+Kano · · Score: 2

      I disagree. If they were totally abandoning Wayland, then you'd have a point. They simply saying, "Hey! Look! Choice! Yay us."

      Like they did with systemd, right?

      LK

      --
      "Hi. This is my friend, Jack Shit, and you don't know him." - Lord Kano
    10. Re:See Saw Cycles of Adoption and Abandonment by Uecker · · Score: 3, Informative

      The code is actually not bad in my opinion and due to its age, a lot of problems have already been fixed a long time ago. Ilja von Sprundel (just featured in another story on slashdot) did some auditing a couple of years ago and fixed many bugs. He gave a talk about it on the CCC conference and he seemed actually quite fond of X from a security perspective (quotes: "the developer involved actually amazing" and about the core protocol "this code is actually pretty cool (from a security perspective) you can see where the code got patched over (e.g. integer overflow checks)") . In fact, he seriously complained about clients and in particular about Qt/KDE in this talk. This is a much newer code base...

    11. Re:See Saw Cycles of Adoption and Abandonment by Anne+Thwacks · · Score: 2
      is a pretty obvious statement that Wayland is a long way from being ready for prime time.

      Au contraire, mon frere

      Ubuntu has a strong track record of deliberately choosing things that are not ready for prime time - the move back to X is probably an indication that Wayland is finally stable.

      --
      Sent from my ASR33 using ASCII
    12. Re:See Saw Cycles of Adoption and Abandonment by Cyberax · · Score: 4, Informative

      I worked on X codebase and I know it's shit. It's a patched-over shit, but still. Integer overflows, memory corruption, it has everything.

      But even setting this aside, X.org is insecure by design. Any application can just send any events to any other application, so there's no point in trying to make it secure. If you have access to an X connection then you already have full access to the user's data. For example, you simply can inject "ctrl-t" into the shell to launch a terminal and then inject any commands you want into it.

      And about "todays wonder boys" - Wayland is designed and written mostly by the same developers who are working on X.org

  2. But but .... by Viol8 · · Score: 5, Funny

    ... the Wayland devs kept telling us that no one cares about remoting with X which is why they hardly bothered to work on that side of it. Were they wrong?? Say it ain't so!

    1. Re:But but .... by KiloByte · · Score: 2

      I guess that's just a hard to discover easter egg, like mouse paste.

      --
      The creatures outside looked from Alt-Right to Antifa; but already it was impossible to say which was which.
    2. Re:But but .... by jbernardo · · Score: 4, Informative

      Or like being incapable of running GUI apps as root - which breaks among others gparted, and won't ever be fixed for native Wayland apps, but you need to "think of the children" - https://bugzilla.redhat.com/sh...

    3. Re:But but .... by Junta · · Score: 2

      Actually, for executing a remote application, Wayland can accomodate with Xwayland.

      Here the thing is sharing your screen, like in a teleconference situation or accessing your whole screen remotely rather than X forwarding which Wayland can't accommodate, in part due to intentional design decisions to mitigate security risks.

      --
      XML is like violence. If it doesn't solve the problem, use more.
    4. Re:But but .... by Anonymous Coward · · Score: 3, Insightful

      "gparted should not run its UI as root. It should run its UI as a regular user and use PolicyKit or something else similar to gain elevated privileges only when necessary to query or modify devices"

      I'm all about being angry, but this makes sense. root only when needed, we don't need root for the UI.

    5. Re:But but .... by jbernardo · · Score: 3, Interesting

      I know of that workaround, but it is all it is, a workaround. The fact that you need it in Wayland is a consequence of its lack of support for running GUI apps as root. And it won't work for native apps,only for X apps.

    6. Re:But but .... by thegarbz · · Score: 2, Informative

      ... the Wayland devs kept telling us that no one cares about remoting with X which is why they hardly bothered to work on that side of it. Were they wrong?? Say it ain't so!

      Not at all. Remote desktop was always important to a subset of users that were not at all targeted by Wayland. Those same users happen to also be the ones who would use LTS releases of Ubuntu.

      If you read the original post they will use Wayland as defaults on all other releases and specifically rushed Wayland to 17.10 to gauge if it will be a default in 18.04LTS. But there are some features that need to be worked on before it will be suitable for LTS release.

      But by all means smug on.

    7. Re:But but .... by jbernardo · · Score: 3, Insightful

      "gparted should not run its UI as root. It should run its UI as a regular user and use PolicyKit or something else similar to gain elevated privileges only when necessary to query or modify devices"

      Why? Because Wayland devs decided the UI should not run as root? Because breaking functionality in the name of a misguided sense of security is fashionable?

    8. Re:But but .... by thegarbz · · Score: 2

      but you need to "think of the children"

      Actually I think it's more like "think of the security hole". I can't say I disagree with the fact that old software should use the modern way of interacting with a modern system if it wants to work for users.

      Expecting full backwards compatibility for everything for ever just invites any script kiddy to take over your computer. It's amazing how we're progressive about protocol adoption and depreciation in Linux, ... until it comes to some GUI application that hasn't gotten its act together in the past several years.

    9. Re:But but .... by serviscope_minor · · Score: 3, Interesting

      x host si:localuser:root

      was all I meeded

      So all you needed to get Wayland to work properly was... X.

      --
      SJW n. One who posts facts.
  3. Why switch to Wayland in the first place? by halivar · · Score: 3, Interesting

    Despite it's touted simplicity, Wayland lags behind X functionality in both network awareness and driver support, as well as still a slight lag in performance despite its purported closeness to the hardware compared to X. Am I misunderstanding something?

    1. Re:Why switch to Wayland in the first place? by serviscope_minor · · Score: 2

      In the meantime there's a laundry list of reasons X.org is a horrible system to use on a desktop.

      OK, I'll bite.

      Like what?

      So far about 80% of them are "I refuse to acknowledge the existence of any new API calls past about 1987". A few others centre around the kernel being unreasonably slow, and there's one or two decent points.

      --
      SJW n. One who posts facts.
    2. Re:Why switch to Wayland in the first place? by serviscope_minor · · Score: 5, Informative

      Right indeedio I will bite!

      Forget the speed,

      Yes indeed, forget that because there's no evidence X is slow. Sure it was slow on a Sun 3/60 and people were maybe right to whine then. It's been a very VERY long time since I've run a 20MHz desktop CPU.

      the issues around hardware access

      Which are?

      the issues on on exclusive access that creates a huge security issue for lockscreens

      Oh you mean the issue that doesn't happen at all if you use a compositor like 99% of modern desktops. Note: if the desktpos don't implement that, it's their fault not an inherent limitation in X.

      the inability to use most of the buttons on a laptop while a locked session is in progress (but yes having to open up and login to a device just to hit the volume down key is totally what we expect from modern 2018 systems).

      Ah yes, the thing that isn't an issue with a modern compositor architecture. Note: bugs in gnome aren't bugs in X11.

      Look: the modern X architecture routes EVERYTHING through the compositor just like Wayland. So the security tradeoffs are identical. It's impossible for an actave grab to interpose.

      If gnome have done a shit job of actually using the features that X has had for the last decade or more, blame gnome, not X.

      I'm not going to repeat them all here (because who has the time), this horse has been beaten to death so often it is now a goo of red puree mushed into the carpets

      No it hasn't. It's been beaten by ill-informed people repeating poorly understood talking points.

      --
      SJW n. One who posts facts.
  4. Honestly... I'm sure why... by ckatko · · Score: 5, Insightful

    ...we even have Wayland/Mir.

    The X Server stack was fast enough back in the days of the FOUR-EIGHTY-SIX.

    And almost all the implementations of the new system lack features that we already expect to work on x server without thinking about it.

    1. Re:Honestly... I'm sure why... by jouassou · · Score: 5, Informative

      Most of the Wayland maintainers have also been working on X.org for a long time, and I trust the developers to know better than the users when a rewrite is due. From what I've read, in addition to the issue of maintainability, X.org is inherently insecure (any app is allowed to draw over / screencapture / keylog any other app), contains a lot of code that is never used anymore (e.g. the builtin font rendering and GUI toolkit in X), while modern developments such as DRI and compositing were bolted on as ugly extensions. So if the X.org maintainers say it's cleaner to rewrite it than to keep bolting on new features on top, then I believe them.

      If you're genuinely interested in why people are developing Wayland, I recommend looking at this talk :).

    2. Re:Honestly... I'm sure why... by serviscope_minor · · Score: 5, Informative

      Because back in the days of the "FOUR-EIGHTY-SIX" your cpu was generally faster than your graphics card. Nowadays your graphics card is so much faster than your cpu the X stack leaves your graphics rendering waiting on the CPU.

      No it isn't: X has supported hardware acceleration from the earliest days and continues to do so. See, for example GlamourGL, in which Xorg uses OpenGL shaders to do all the 2D drawing operations.

      So Wayland is an attempt (successful or not is an entirely different discussion) to get the cpu out of the way for your graphics card to work more efficiently.

      No, that's utter crap. Wayland doesn't do that AT ALL. Wayland is basically a system for sending bitmapts to a compositor and have the compositor send back input. Wayland provides very little else and certainly no rendering. Applications are expected to render to their own buffers using something like DRI, which is PRECISELY the same as they use under X11 too if running locally.

      --
      SJW n. One who posts facts.
    3. Re:Honestly... I'm sure why... by halivar · · Score: 3, Insightful

      I trust the developers to know better than the users when a rewrite is due

      As a developer who has been on way too many misguided rewrite projects, I do not share this credulousness.

      "Ugh, this code is so messy!"

      "I know! Let's rewrite it!"

      And thus we enter purgatory.

    4. Re:Honestly... I'm sure why... by LiENUS · · Score: 2

      No it isn't: X has supported hardware acceleration from the earliest days and continues to do so. See, for example GlamourGL, in which Xorg uses OpenGL shaders to do all the 2D drawing operations.

      There's a reason that any serious graphics rendering under Xorg uses DRI. It's because DRI bypasses most of the X stack.

      No, that's utter crap. Wayland doesn't do that AT ALL. Wayland is basically a system for sending bitmapts to a compositor and have the compositor send back input. Wayland provides very little else and certainly no rendering.

      So what you're saying is wayland gets the cpu out of the way so applications can render on the graphics card more efficiently? wow why didn't I think to say that?

      Applications are expected to render to their own buffers using something like DRI, which is PRECISELY the same as they use under X11 too if running locally.

      So what you're saying is all that Xorg nonsense just bogs it down and anything rendering 3d any serious graphics is just going to do direct rendering like wayland prefers anyway? holy shit I wish I had thought to say that... oh wait I did.

  5. Common sense like this should be applauded. by blind+biker · · Score: 5, Insightful

    This is the first time in a long while that a company steps back from what looked like suicidal commitment to a bad idea, and actually went back to what works.

    I wish Lenovo did the same with the 7-row keyboardes on the ThinkPad. Also I wish Linux companies (except RedHat, of course) would ditch SystemD.

    --
    "The agriculture ministry is not in charge of Gundam" - Japanese ministry official.
  6. LTS by erapert · · Score: 2

    It's an LTS right? So isn't it the right move to keep it in stable territory for now?

  7. gnome-shell wayland disaster by Anonymous Coward · · Score: 3, Interesting

    The shit is a total clusterfuck with Wayland. Something happens to gnome-shell--REBOOT. Some random gnome-shell plugin acts bad, no way to unload it--REBOOT. Mouse stops working--REBOOT. Come back from screen lock and clicks don't work--REBOOT.

    All of this shit is possible to completely fix non-destructively when gnome-shell runs under X by Alt-F2 'r', or lacking input, Ctrl-Alt-F1 'killall -HUP gnome-shell'.

    Now Alt-F2 'r' is disabled, and every other previously working solution causes EVERYTHING to be killed, because now gnome-shell is the parent of the entire session. The gnome-shell developers have basically said tough, this is intended operation, and you shouldn't need to restart the shell ever. Fuck them. I leave my workstation powered up for months on end, yet I have to restart gnome-shell it seems every week or two sometimes.

  8. Good job Canonical by Anonymous Coward · · Score: 4, Insightful

    I think they should be congratulated on responding to user sentiment. I wish more companies would admit they acted prematurely and roll back changes that didn't work out. I can think of one or two very large Linux features that I could live without, but which are foisted on all of us.