Announcement

Collapse
No announcement yet.

Ubuntu 21.10 "Impish Indri" Development Begins

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #41
    GNOME on wayland works. Want something else? Brace for impact. Alternative implementations are trouble.

    Every month there’s some poor kid on Github who decides the world needs yet another mediocre compositor implementation. Same kind of kids that did new distributions in the 2010s and plagiarized music player apps in the 2000s. Nothing changed really.

    Comment


    • #42
      Originally posted by 144Hz View Post
      GNOME on wayland works. Want something else? Brace for impact. Alternative implementations are trouble.

      Every month there’s some poor kid on Github who decides the world needs yet another mediocre compositor implementation. Same kind of kids that did new distributions in the 2010s and plagiarized music player apps in the 2000s. Nothing changed really.
      My experience tells me otherwise.

      I've been using Wayfire (which is a minimal Wayland compositor) with a mix of GNOME and Xfce applications full time for a few months now. No "trouble", very stable and reliable.
      I also used Sway quite a bit, and found it very stable as well.
      And I often read that KDE's Wayland is advancing, but I'm not a KDE fan anyway and I have never tried its Wayland session, so I can't speak for that.
      So you have at least 2 very reliable Wayland options other than GNOME.

      I'm not a GNOME hater by any means. Quite the contrary. I do like the design and workflow of GNOME, I always have it installed on my system, and I do use it from time to time.
      But the fact remains that it is more complex and relatively heavier on resources. That's why I don't use it if I'm doing something serious, where I need the desktop to be as light and responsive as possible (and being very stable of course), and Wayfire provides just that.

      TLDR: Yes, GNOME has the most advanced Wayland implementation, but there are also a few other reliable options.
      Last edited by idash; 30 April 2021, 12:02 AM.

      Comment


      • #43
        Originally posted by Alexmitter View Post

        Wayland, as a production read protocol exists since 2015, thats 6 years, not a decade.
        2016 the first desktops switched to it.



        Ah, the back to the past crowd. Yes sure, lets go back to shitty X-like protocols like Arcan or directly X11.
        i've never said to go back to X. just get rid of this wayland.

        Comment


        • #44
          Originally posted by loganj View Post

          i've never said to go back to X. just get rid of this wayland.
          And what would you propose instead?

          Comment


          • #45
            Originally posted by 144Hz View Post
            GNOME on wayland works. Want something else? Brace for impact. Alternative implementations are trouble.

            Every month there’s some poor kid on Github who decides the world needs yet another mediocre compositor implementation. Same kind of kids that did new distributions in the 2010s and plagiarized music player apps in the 2000s. Nothing changed really.
            I have to disagree. wlroots is a high quality implementation, it is also used for phosh what is largely developed by gnome contributers because mutter simply is unfit for that job on the current target devices.

            Wlroots will be the future backend for all the small desktops, there is no doubt about that and nothing to worry because it truly is excellent in its implementation.

            Comment


            • #46
              Originally posted by arQon View Post

              I'm not sure I'd even go far as to call it "pushback". As has been made (very) clear in most of the posts here already, people just want something that Actually Works. Wayland has repeatedly failed to be that for even common cases in the past, and continues to be that for some less-common cases even now, so there is a lack of *confidence* that "this time, it really does work, we promise" - which is very much understandable.

              I suspect that what you perceive as objecttions to Wayland itself are mostly just objections to the stupidity of the Wayland cheerleaders, who feel that if Wayland works for the subset of uses THEY have it must be good enough for everyone else tooo - because otherwise that might mean that they've somehow made the wrong choice, and their egos can't handle that. (Hence the distortion of "facts" pushed by them in this thread, and the backpedalling when called out on it).

              Wayland is basically in the same position as WINE. "It works 95% of the time" isn't good enough. The advantage that Wayland has for that last 5% though is XWayland: and NOBODY with a functioning brain cares whether their X is delivered "natively" via X, or if it's passing through XWayland instead. The contrast with that though is that the majority of Wayland supporters are *frothing at the mouth* against X. (Which is interesting, because that's generally behavior of the side whose position has no merit, which isn't actually the case here for once).

              So I don't think the opposition is to Wayland itself: it's just that everyone's tired of the idiot fanboys who don't even know what Wayland's limitations are (and thus why there are some scenarios where it flat-out CAN'T be used instead of X) trying to preach about how it's the Second Coming, but only being able to do so by attacking X (old / buggy / broken / omgsecurity! / etc) rather than being able to point to any actual Wayland advantages that are meaningful to anyone. (So, "tribal politics as usual", basically).
              Spot on.
              I've been expressing the same things for a couple of years now. And many others have as well.

              I specifically agree with the part in bold (something I've also shared before). It's often complemented with moronic punchlines such as "Let X die already" or the likes, for dramatic effect. And these cheerleaders are ignoring issues openly so as to maintain their agenda through fake optimism. With a bit of hindsight, it kind of makes for a good laugh... I mean a great case study for a psychology class.
              Then a couple of pragmatic people come and make a fest out of them. And the scenario repeats. Until one day maybe, hopefully, it's actually covering more than the limited set of use cases from big egos early adopters.

              Comment


              • #47
                Originally posted by Mez' View Post
                Spot on.
                I've been expressing the same things for a couple of years now. And many others have as well.

                I specifically agree with the part in bold (something I've also shared before). It's often complemented with moronic punchlines such as "Let X die already" or the likes, for dramatic effect. And these cheerleaders are ignoring issues openly so as to maintain their agenda through fake optimism. With a bit of hindsight, it kind of makes for a good laugh... I mean a great case study for a psychology class.
                Then a couple of pragmatic people come and make a fest out of them. And the scenario repeats. Until one day maybe, hopefully, it's actually covering more than the limited set of use cases from big egos early adopters.
                Interesting point of view.

                Lets talk about a quite short but severe list of things Xorg/X11 can't do:

                - Prevent a (non root) application from reading all your inputs.
                - Prevent a (non root) application for casually read your whole screen whenever it likes to.
                - Prevent tearing. You either need a compositor doing vsync for it, or buffering at the driver side. Both inducing lag, possibly stuttering and latency.

                Then there is the thing with render pipelines. In the worst (and most common case), your clients buffer gets passed on as such: Client -> Xserver -> Compositor -> Xserver -> Xserver Generic DRI driver -> Actual hardware driver via DRI. That is just horrible.
                On Wayland, this would either look like: Client -> Compositor -> Hardware Driver or with compositor bypass look like Client -> Hardware Driver

                The embedded business is all over Wayland, mobile Linux devices like Jolla's Sailfish OS use it as well as all the major platforms for the new emerging mobile linux phones.
                Simply because Xorg is too heavy, slow, wasteful, and buggy. Unacceptable for a high reliability business.

                Generally, I would be interesting in a psychological study behind the thought process of people who want to grip firmly onto a display server technology that allows any client to do whatever it likes. I know, change is hard, that's a human thing.

                Comment


                • #48
                  Originally posted by Alexmitter View Post

                  Interesting point of view.

                  Lets talk about a quite short but severe list of things Xorg/X11 can't do:

                  - Prevent a (non root) application from reading all your inputs.
                  - Prevent a (non root) application for casually read your whole screen whenever it likes to.
                  - Prevent tearing. You either need a compositor doing vsync for it, or buffering at the driver side. Both inducing lag, possibly stuttering and latency.

                  Then there is the thing with render pipelines. In the worst (and most common case), your clients buffer gets passed on as such: Client -> Xserver -> Compositor -> Xserver -> Xserver Generic DRI driver -> Actual hardware driver via DRI. That is just horrible.
                  On Wayland, this would either look like: Client -> Compositor -> Hardware Driver or with compositor bypass look like Client -> Hardware Driver

                  The embedded business is all over Wayland, mobile Linux devices like Jolla's Sailfish OS use it as well as all the major platforms for the new emerging mobile linux phones.
                  Simply because Xorg is too heavy, slow, wasteful, and buggy. Unacceptable for a high reliability business.

                  Generally, I would be interesting in a psychological study behind the thought process of people who want to grip firmly onto a display server technology that allows any client to do whatever it likes. I know, change is hard, that's a human thing.
                  Change is not hard for me on computers.
                  If I find something better (than my current workflow, an app, a DE, a distro, etc...), I'm willing to change in a snap. Always have eversince I've been using computers. But it's very different when the change is active (decided by yourself) or forced on you (only if you have a negative opinion about the change).

                  I don't care too much about security at home level. I'm an uninteresting fish in the sea for ill intended people. I value pragmatism over security when using my computers.
                  And it's not because Xorg is too heavy for embedded and mobile devices that it is for computers.

                  What is hard is launching a wayland* session and being frustrated by the bugs and things I can't do, or the weird (feeling fake) smoothness of wayland. I would even be open to an alternative way of doing the things I can't, but when there just isn't...
                  In my balance, and I believe many here have expressed a similar opinion, the negative parts of wayland outweigh the negative aspects of X. In other words, I can live with X as defective as it is, but I can't currently go on with my workflow with wayland. This is my position today, at this level of implementation, it doesn't mean I'm closing the door. I'm actually giving it a go every month or so to check on the progress.

                  If I'm not convinced, then it's not resistance to change. It might just be that the validity of the arguments is disputable to me. Or that the current state of implementation is not sufficient in my opinion.

                  * by wayland, in this context, I'm talking about Mutter as the specific compositor. I haven't tried with other compositors (hence DEs).
                  Last edited by Mez'; 30 April 2021, 11:28 AM.

                  Comment


                  • #49
                    Originally posted by Mez' View Post

                    * by wayland, in this context, I'm talking about Mutter as the specific compositor. I haven't tried with other compositors (hence DEs).
                    I actually had the same opinion as yours when I was only looking at GNOME Wayland, which unfortunately gave me a bad first impression about Wayland.
                    I also tried Sway, but tiling window managers are just not for me. I was like "if only there is something as minimal as Sway, but Openbox-like..."

                    Then I came across Wayfire, and that was it. My opinion on Wayland just flipped to the opposite.
                    Wayfire is lighter and more responsive than any minimal X window manager I've tried, that includes Openbox, IceWM, Fluxbox, etc. while also having the benefits of Wayland.

                    I'd recommend you give it a try.
                    Not necessarily saying you should switch to it, but at least as an experiment to understand that Wayland != Mutter/GNOME.

                    Comment


                    • #50
                      My initial impression of Wayland was also pretty poor. At the time Mir worked better, and that had its own problems too.

                      Developers have come a long way since then. The only thing that I’m missing is a working VNC of some kind, and I’ll make the switch

                      Comment

                      Working...
                      X