Announcement

Collapse
No announcement yet.

Wayland Becomes A FreeDesktop.org Project

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

  • #31
    Wayland is pretty much FTW. X.org (X*) was necessary because of at the time that's just how things were, however, now with all sorts of acceleration a much smaller server is optimal. I might go back to using Linux if and when Wayland goes mainstream (and I can replace X11 with it/use it on my NVIDIA GPU).

    Wayland is the kind of advances the FOSS world needs to make. Innovations. Rather than trying to do old things better do new things that haven't been done because the old things don't work.

    Comment


    • #32
      It's more like do something. Let it sit until it's outdated. Argue that it's not outdated. Update it through hacky workarounds while something better is being done to take it's place. Rinse and repeat.

      Comment


      • #33
        If you want to move the window management into Wayland (as opposed to delegating it to external apps), then you will have to re-implement all the functionality from pretty much every window-manager out there, because somebody out there is using these features.

        At the very least, it will have to be good at window placement, tiling, common 3d effects, focus stealing prevention, implement different focus policies, programmable buttons and titlebars, excellent and programmable keyboard shortcuts, horizontal/vertical maximisation, and all the other things that make a Unix desktop so productive. And then it will not be all that minimal or lightweight anymore.

        Currently, I see Wayland as an interesting prototype. I don't expect it to seriously threaten X in the next 2-3 years. I'm willing to be convinced. But remote application access and advanced window management are the killer features or a Unix desktop. The weaknesses of the Unix desktop are elsewhere. If you mess with the things Unix does well, you run the danger that only Win/Mac switchers will use it.

        Seriously, switching from something like Afterstep or KWin or Compiz to the window management comparable to OSX or Windows is simply not going to happen for many long-time Linux users. And any serious development work will likely stay on X, which I use at work all the time, and it is transparent and extremely useful to run things remotely and get graphical updates on my main workstation. Mucking around with VPNs and whatnot to do the equivalent of "ssh -X" is not attractive. What do we gain in return? Faster resizing? Cute spinning things?

        I'll wait it out and weigh the pros and cons once it actually works.

        Comment


        • #34
          Originally posted by Decatf View Post
          It's more like do something. Let it sit until it's outdated. Argue that it's not outdated. Update it through hacky workarounds while something better is being done to take it's place. Rinse and repeat.
          How do you know? I mean this thread is pretty much "yay it's new, so it must be better"

          I've read many comments about X being outdated and unmaintainable. But apart from my own experience(the C-Api!) I have yet to see anyone coming up with evidence and or actual knowledge.

          So let's play a game:
          1. Think about a problem you have with X, anything.
          2. Is it an actual X issue or is it a problem with the Kernel, Scheduling or DRM?
          3. Is the issue really X or is it another part of the graphic stack, e.g. Mesa, Gallium, the drivers?
          4. Is it really a problem within X(the infrastructure, Apis, fundamentals) or is it a simple bug/unfinished feature?

          If you still have anything, you may post it here. If not, please stop talking about Wayland as the future

          Comment


          • #35
            Originally posted by pingufunkybeat View Post
            Seriously, switching from something like Afterstep or KWin or Compiz to the window management comparable to OSX or Windows is simply not going to happen for many long-time Linux users. And any serious development work will likely stay on X, which I use at work all the time, and it is transparent and extremely useful to run things remotely and get graphical updates on my main workstation. Mucking around with VPNs and whatnot to do the equivalent of "ssh -X" is not attractive. What do we gain in return? Faster resizing? Cute spinning things?

            Well.. I have some news for you:

            People have been doing remote desktops and remote applications on Windows and other platforms for a very long time now. And for the most part performance tends to be better.

            One of the fun things people are starting to be able to do with remote applications is that animations are being transcoded into lossy video formats real-time to be transmitted over the network to a remote desktop. That way you can do things like have HD video from a remote desktop.

            Basically; X Windows network was a very cool feature, but the world has moved on to bigger and better things.

            Not to mention also I can quite easily do X Windows on both Windows and OS X, and I don't see why it would not be any different with Wayland. There is nothing mutually exclusive between X networking and Wayland.


            As far as applications and window management.... the Widgets and other important desktop applications have abstracted themselves and isolated themselves so much from low-level X11 (which is mostly useless for modern applications) that porting effort is probably not nearly as significant as you may expect.

            For example...

            People have told me that Xterm would be faster then Gnome-terminal, yet time and time again I've proven that Gnome-terminal is much faster at being a remote X11 client precisely because the huge amount of work that the Gnome/GTK folks have had to put in to work around X's crappy behavior.


            I would be very surprised if Wayland developers did not have remote desktop capabilities in mind, even if it's not a primary goal.

            Comment


            • #36
              Originally posted by drag View Post
              Well.. I have some news for you:

              People have been doing remote desktops and remote applications on Windows and other platforms for a very long time now. And for the most part performance tends to be better.
              I've seen it, I've used it, and I've hated it.

              At work, I regularly run parallel processor-intensive processes on up to 10 different network computers, with the updates being transparently rendered on my workstation.

              If there is an easy and efficient way to do this on Windows (which does not involve Cygwin or MinGW or such), I'm not aware of it. But I do agree that I avoid MS Windows whenever I can, and that I'm biased against it.

              Sure, X on Wayland is a possible fallback for such applications, but I still wonder what we gain by doing this.

              Comment


              • #37
                Originally posted by fabiank22 View Post
                How do you know? I mean this thread is pretty much "yay it's new, so it must be better"

                I've read many comments about X being outdated and unmaintainable. But apart from my own experience(the C-Api!) I have yet to see anyone coming up with evidence and or actual knowledge.

                So let's play a game:
                1. Think about a problem you have with X, anything.
                2. Is it an actual X issue or is it a problem with the Kernel, Scheduling or DRM?
                3. Is the issue really X or is it another part of the graphic stack, e.g. Mesa, Gallium, the drivers?
                4. Is it really a problem within X(the infrastructure, Apis, fundamentals) or is it a simple bug/unfinished feature?

                If you still have anything, you may post it here. If not, please stop talking about Wayland as the future
                I never said Wayland was the future.

                Comment


                • #38
                  Originally posted by Decatf View Post
                  I never said Wayland was the future.
                  Argue that it's not outdated. Update it through hacky workarounds while something better is being done to take it's place
                  Was were you talking about then? Hurd? What makes Wayland better than X?

                  Comment


                  • #39
                    Originally posted by ethana2 View Post
                    I want to see what kind of power consumption can be managed playing 3d games with less context switching and strict render-on-page-flipping. Why render 230 FPS with tearing when you can render 1/2x or 1x refresh rate with no tearing and lower power usage?
                    How will Wayland, which requires at least one context switch (between the application and the Wayland server to hand off each frame's buffer) be better than the current architecture, which requires zero context switches (application talks directly to the GPU)?

                    Try running (non-redirected) glxgears on NVIDIA's driver and then sending SIGSTOP to the X server from an SSH session sometime.

                    Comment


                    • #40
                      Originally posted by nerdopolis View Post
                      There are many more Disadvantages of having the apps draw their own borders

                      http://blog.martin-graesslin.com/blo...w-decorations/


                      Have you ever opened up a maximized window in Windows, and then have it open up a progress window for a long operation, and then try to minimize or move the maximized window, only to have it throw a "ding" sound at you, and flash the titlebar of the progress window?
                      That has more to do with the way event-handling is managed in Windows, probably a holdover from the single-threaded 16-bit days. User32 could just as well respond to messages on the non-client area, even when there is something else going on. But the Windows devs must have decided that it would be dangerous to have User32 do its own thing without the applicaiton's knowledge, so they didn't make that change.

                      Comment


                      • #41
                        Originally posted by pingufunkybeat View Post
                        Having window management in a separate process is a FEATURE, and not a bug.

                        It's the best thing about X, period. I hate unminimiseable and uncloseable zombie windows on that other operating system.

                        Window management the X way comes at a cost, which is evident in issues such as rubber-banding when resizing, etc. But it's still very useful.
                        Actually, putting the window manager in a separate process complicates things considerably, in the same way that adding multi-threading complicates a program. Now for doing something simple like resizing a window, you have three programs that must coordinate their activity instead of just two. It's a much more complex problem and is part of the reason why resizing apps in X is still slow and laggy even with relatively fast hardware.

                        One mistake people keep making with this client-side vs. wm-side window management stuff is differentiating between problems due to that architectural choice, and problems due to other design decisions in the system. Having window managers as separate programs doesn't, for example, solve the issue of programs having different window borders any more than having window management functionality inside the client-side library creates it. Windows could just as easily forbid modifications to the non-client area of a window and that would be that, no more inconsistency.

                        Comment


                        • #42
                          Originally posted by fabiank22 View Post
                          Was were you talking about then? Hurd? What makes Wayland better than X?

                          For me the exciting thing is just that it is a viable alternative to X and I think that a "monoculture" always is a bad thing no manner how good that dominant technology is. X and GNU are clearly very dominant technologies among free *nix-like operating systems. That is why I find the LLVM/Clang, Wayland and Sta.li projects very interesting.

                          Comment


                          • #43
                            Originally posted by fabiank22 View Post
                            How do you know? I mean this thread is pretty much "yay it's new, so it must be better"

                            I've read many comments about X being outdated and unmaintainable. But apart from my own experience(the C-Api!) I have yet to see anyone coming up with evidence and or actual knowledge.

                            So let's play a game:
                            1. Think about a problem you have with X, anything.
                            2. Is it an actual X issue or is it a problem with the Kernel, Scheduling or DRM?
                            3. Is the issue really X or is it another part of the graphic stack, e.g. Mesa, Gallium, the drivers?
                            4. Is it really a problem within X(the infrastructure, Apis, fundamentals) or is it a simple bug/unfinished feature?

                            If you still have anything, you may post it here
                            Okay, what about this http://bugs.freedesktop.org/show_bug.cgi?id=11227 and other http://www.x.org/wiki/Development/X12 problems of X?

                            Comment


                            • #44
                              Also can I switch GPU on-the-fly in Linux? No, and it's X problem too.

                              Comment


                              • #45
                                Why do you think it is an X problem ?

                                Comment

                                Working...
                                X