Announcement

Collapse
No announcement yet.

Wayland Meets Some Summer Love w/ New Changes

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

  • Wayland Meets Some Summer Love w/ New Changes

    Phoronix: Wayland Meets Some Summer Love w/ New Changes

    Last week we openly asked the question if and when will X12 emerge to replace X11, which was met by a variety of responses. Some view the Wayland Display Server as being a potential successor to the current X11 / X.Org Server, but others don't give it much credit seeing as it's not too actively worked on -- well, directly, but it leverages a lot of work actively going on with the Mesa and kernel DRM. The last time the Wayland Display Server received new commits to its code-base was back in March, but that changed this weekend...

    http://www.phoronix.com/vr.php?view=ODMyNA

  • #2
    Great. Now the Wayland fanboys will be back out in force here on Phoronix. Can't wait!

    Comment


    • #3
      Originally posted by siride View Post
      Great. Now the Wayland fanboys will be back out in force here on Phoronix. Can't wait!
      Why can't wait? Can't get a life?

      Comment


      • #4
        Well well. Wasn't it supposed to be 'just a playground to try the new DRI2'?

        Comment


        • #5
          It would b great to have an alternative to X but it seems that the development is too slow to make that happen in foreseeable future.

          Comment


          • #6
            Originally posted by siride View Post
            Great. Now the Wayland fanboys will be back out in force here on Phoronix. Can't wait!
            'Fanboys' wanting to push their opinion versus you wanting to push yours.

            If you would not want to push yours than you would simply blindly ignore them.

            But I still do not see why Wayland would suck. If the GTK and Qt toolkits are ported and altered to be widget network transparant and use already available SSH connections, then what the hell is the problem? Well... Xlib is, but it is not as if Xlib and other X.org source is not available. Hell if Wine can make *NIX compatible with Microsoft Windows (for crying out loud) it should be doable.

            Minimalism, simplicity and strictly modern functionality just kicks ass. What if Wayland would be runnable, GTK/Qt would be widget network transparant and ported it could already exist alongside of X.org. And in the end if backwards compatibility in form of a seperate parser comes around...

            Comment


            • #7
              Originally posted by V!NCENT View Post
              'Fanboys' wanting to push their opinion versus you wanting to push yours.
              My opinion happens to be shared by many X developers and other people who know what they are talking about, so...

              If you would not want to push yours than you would simply blindly ignore them.

              But I still do not see why Wayland would suck. If the GTK and Qt toolkits are ported and altered to be widget network transparant and use already available SSH connections, then what the hell is the problem? Well... Xlib is, but it is not as if Xlib and other X.org source is not available. Hell if Wine can make *NIX compatible with Microsoft Windows (for crying out loud) it should be doable.
              Right, but it's not that simple. The toolkits haven't even been ported to XCB yet and XCB is the same protocol, just a different (more efficient) API. Also, Wayland isn't even an API yet. It's just a playground that kind of supports some things. I'm sure it won't look as pretty and orthogonal once it has to do all the things that are needed of a mature graphics system. It's easy to throw together a compositor that doesn't use X. That's barely even exciting. I want to see a system that can actually power a full Unix desktop stack. Until then, it's just a playground for things that ought to be ported to X.

              Minimalism, simplicity and strictly modern functionality just kicks ass. What if Wayland would be runnable, GTK/Qt would be widget network transparant and ported it could already exist alongside of X.org. And in the end if backwards compatibility in form of a seperate parser comes around...
              All of these things can be gotten by fixing X. Fixing X slowly over time also allows for the currently minimal manpower to actually achieve the goal of having a nice X. All the good parts can be reused and all the cruft can be deprecated and the new things that we learn from Wayland or elsewhere can be introduced. After all, render, composite, EXA, AIGLX, etc. were introduced this way, have stabilized and are now quite successful. 6 years ago, X couldn't do any of those things. Meanwhile, Y windows, Berlin, etc. all failed. What does that say to you?

              Throwing all your money in with Wayland, which is being worked on -- occasionally -- by one guy, doesn't seem like a smart bet either.

              Also, what does "widget network transparant[sic]" mean? There's no such thing as a "widget network".

              Comment


              • #8
                6 years ago, X couldn't do any of those things. Meanwhile, Y windows, Berlin, etc. all failed. What does that say to you?
                Projects created out of X.org frustration but never materialised becuase it didn't had any other goal other than not being X.org...

                Throwing all your money in with Wayland, which is being worked on -- occasionally -- by one guy, doesn't seem like a smart bet either.
                I have no idea what a single developper has to do with any of this. If one wants to make shit possible, they make shit possible. It is an idea, a concept, an abstract and all I know is that it is:
                A) Not ready yet;
                B) Not done yet.
                So?

                Whilst X.Org might be a good foundation for todays systems, it appears to be horribly complex. Wouldn't be the first time we hear the "need more devs", while new kids on the block can't figure it out, but that's just what I heared...

                Also, what does "widget network transparant[sic]" mean? There's no such thing as a "widget network".
                From: http://en.wikipedia.org/wiki/Network_transparency
                "Network Transparency in its most general sense refers to the ability of a protocol to transmit data over the network in a manner which is transparent (invisible) to those using the applications that are using the protocol.

                The term is often applied in the context of the X Window System which is able to transmit graphical data over the network and integrate it seamlessly with applications running and displaying locally."


                So what I mean by network transparant widgets is that instead of client->server pixmap delivery, the client could send "1 button here, 2 button there, text there and this there", saving an insane amount of bandwith... And this would suck if one would have a client-server setup locally, but if you have Wayland server only, and Qt/Gtk taking care of the client role (automaticaly compatible with other instances; X.org/Wayland/XFree86), this would be eliminated.

                Comment


                • #9
                  Originally posted by V!NCENT View Post
                  Projects created out of X.org frustration but never materialised becuase it didn't had any other goal other than not being X.org...
                  The Wayland fanboys kind of fall into that same position...

                  I have no idea what a single developper has to do with any of this. If one wants to make shit possible, they make shit possible. It is an idea, a concept, an abstract and all I know is that it is:
                  A) Not ready yet;
                  B) Not done yet.
                  So?
                  Well, things don't get done if nobody is around to do them. It's already bad enough that X.org has limited manpower. How is Wayland supposed to get done or work if there is only one guy working on it? It matters a whole lot.

                  Whilst X.Org might be a good foundation for todays systems, it appears to be horribly complex. Wouldn't be the first time we hear the "need more devs", while new kids on the block can't figure it out, but that's just what I heared...
                  The Linux kernel is also very complex, but I see few cries for trashing it and starting over again, nor is there a problem with manpower. X doesn't need to start over, it needs manpower and support for companies, like the kernel has, or GNOME/KDE, etc. Any large system is going to end up seeming "horribly complex". That's the reality of mature software.

                  From: http://en.wikipedia.org/wiki/Network_transparency
                  "Network Transparency in its most general sense refers to the ability of a protocol to transmit data over the network in a manner which is transparent (invisible) to those using the applications that are using the protocol.

                  The term is often applied in the context of the X Window System which is able to transmit graphical data over the network and integrate it seamlessly with applications running and displaying locally."


                  So what I mean by network transparant widgets is that instead of client->server pixmap delivery, the client could send "1 button here, 2 button there, text there and this there", saving an insane amount of bandwith... And this would suck if one would have a client-server setup locally, but if you have Wayland server only, and Qt/Gtk taking care of the client role (automaticaly compatible with other instances; X.org/Wayland/XFree86), this would be eliminated.
                  Terrible, terrible idea. So much of what a toolkit does is very client-specific. With your system, instead of just diddling around with things on the client-side (i.e., function calls), you have to send messages to the server and wait for responses in many cases. How would you deal with subclassing? Owner draw? What about new custom widgets that don't derive from existing widgets? And how could you leverage the object-oriented hierarchy of widgets when that is all hidden on the Wayland server? It quickly becomes a mess. Plus, it doesn't buy you anything. You still have to communicate with the server. You still have to micromanage things on the server (except with your system, you have to micromanage even more). You still have to listen for and handle events.

                  The fact that all other OSes have client-side toolkits should tell you something about the way things should be done...

                  Comment


                  • #10
                    There is new git tree qt-wayland.
                    http://cgit.freedesktop.org/~jbarnes/qt-wayland/?h=4.7

                    Comment


                    • #11
                      Wayland is a nice concept, but it fundamentally depends on Linux video drivers being independent from X. Until things like Gallium and all that get up to speed, and KMS is standard across all hardware, then Wayland is going to pretty worthless.

                      So what is the point of spending all this time hacking on Wayland when almost nobody can meet the requirements to use it?

                      If X itself is barely improved by the time the new Linux video stack matures and stabilizes then I am sure that Wayland will gain a lot more attention.

                      ------------------------------------

                      Keep in mind that X and Wayland is not a either or.

                      You do not have to eliminate X Windows support in order to run Wayland.

                      As proof:
                      I can quite happily run connect to Linux box from Microsoft Windows and run a X Server on Windows to display applications from Linux. I can also do that from OS X. Neither of which requires me to hand over my entire display to some X Server.

                      Comment


                      • #12
                        Originally posted by siride View Post
                        The Wayland fanboys kind of fall into that same position...
                        So everyone that is interested in Wayland is now suddenly a fanboy?

                        Well, things don't get done if nobody is around to do them. It's already bad enough that X.org has limited manpower. How is Wayland supposed to get done or work if there is only one guy working on it? It matters a whole lot.
                        By lowering the entry bar due to simplification.

                        Any large system is going to end up seeming "horribly complex". That's the reality of mature software.
                        Realy... I want proof of this. There is no way that there is a law that dictates that mature software will always turn out ot be so horribly complex that almost no-one even dares to touch it.

                        Terrible, terrible idea. So much of what a toolkit does is very client-specific.
                        Not if one changes that. Sorry, but that should have been reeeaaaaaaalllyyyyy obvious...

                        With your system, instead of just diddling around with things on the client-side (i.e., function calls), you have to send messages to the server and wait for responses in many cases.
                        X.org doesn't send messages and waits for replies? X.org doesn't transmit input and images? Wow... I guess I don't even have to have a network. I can just log in from England to Japan and it goes through thin air.

                        How would you deal with subclassing? Owner draw? What about new custom widgets that don't derive from existing widgets? And how could you leverage the object-oriented hierarchy of widgets when that is all hidden on the Wayland server?
                        Wait... Too much BS already... First of create a colaborative FreeDesktop spec for toolkit. Wouldn't it be nice to ignore other themes so GTK will not look like ass outside of gnome? Perfect.
                        Secondly nothing is hidden inside Wayland. Toolkit would create its own windows and classes etc etc tec and just feed that to the Wayland API (which like anything else on paper doesn't exist, hence the paper in the first place).

                        You still have to communicate with the server. You still have to micromanage things on the server (except with your system, you have to micromanage even more). You still have to listen for and handle events.
                        An Android powerd HTC Legend device has more compute power that the fastest super computer back when X.org waqs created. Believe me... Thin clients would be possible, don't you think? If you can find hardware that ships today that cannot run Qt/GTK+ due to speed problems than I would be stunned.

                        The fact that all other OSes have client-side toolkits should tell you something about the way things should be done...
                        Back when Windows95 was around there wasn't any compositing desktop. That should have told you that compisiting was the most stupid idea ever, right? C'mon....

                        Comment


                        • #13
                          Originally posted by siride View Post
                          Throwing all your money in with Wayland, which is being worked on -- occasionally -- by one guy, doesn't seem like a smart bet either.
                          Lol wasn't the linux kernel built by one dude on his own time? Oh yeah that's what Linus did. Way to make a huge judgement call on somebody that just simply made the choice to write their own program. Not to mention for someone that is so critical of anyone being a 'fanboy' of this dude's super new project by showing any interest whatsoever you sure are acting like a nazi over the whole thing yourself. Why the hell is fanboy the favorite world of anyone who resides on internet forums with a strong opinion about anything?

                          Comment

                          Working...
                          X