Announcement

Collapse
No announcement yet.

Sadly, Two X.Org GSoC Projects Already Failed

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

  • #11
    Originally posted by BradN View Post
    We use network forwarding every day in particular, and on relatively old hardware too. It's usable even on a crappy 100 megabit network...
    Is this some kind of joke? What kind of system could possibly fail on a 100mb network, it would have to be awful.
    Last edited by smitty3268; 27 June 2014, 03:25 PM.

    Comment


    • #12
      Why is everyone so negative about these projets? I see only one thats is for X Server (QtQuick compositor). Rest are Mesa/Wayland projects, and you want them to fail?

      Comment


      • #13
        Originally posted by magika View Post
        Why is everyone so negative about these projets? I see only one thats is for X Server (QtQuick compositor). Rest are Mesa/Wayland projects, and you want them to fail?
        Complete cluelessness and a total inability to perform basic reasoning.

        Comment


        • #14
          Originally posted by BradN View Post
          Maybe i'm in the minority but I think what X needs is aggressive refactoring and code cleanup in the name of maintainability. Backwards compatibility and network transparency are its strongest assets I think. Or at least, any X replacement should make those features first class citizens. We use network forwarding every day in particular, and on relatively old hardware too. It's usable even on a crappy 100 megabit network...
          I don't know how many times it has to be said, but modern X's components (DRI, etc) are NOT network transparent anymore! It is Network Capable, just like Wayland. If you want to use X.org's Network Transparency, you pretty much have to use it's oldest components that are still "maintained".

          As for backwards compatibility, I think there needs to be a limit. Like, you can't remove a feature added 3 major releases ago or less. With that, we'd be able to clean up old crap nobody uses in newer releases, while keeping AT LEAST 3 major versions with a feature available for those who need it.

          Comment


          • #15
            Originally posted by Daktyl198 View Post
            while keeping AT LEAST 3 major versions with a feature available for those who need it.
            So we keep features from X9, X10, X11 ?

            Comment


            • #16
              Originally posted by doom_Oo7 View Post
              So we keep features from X9, X10, X11 ?
              lol, nah. Like Wayland wouldn't remove any features in versions 1.0 - 3.0, then when Wayland hits 4.0, it can remove features from the 1.0 release, but not 2.0/3.0. Then on the 5.0 release, it can remove features from 2.0, etc etc. I figure this gives a couple years of availability for features (it's not like major version shifts happen every month, right?), but allows us to not be using legacy crap 30 effing years down the line like what we're doing with X.

              The key word here is "can". 99% Features will most likely live for more than 3 major versions, it just gives the project some breathing room when it comes to maintaining old and crappy code :P

              Comment


              • #17
                Originally posted by Ericg View Post
                Wayland. XWayland will handle backwards compatibility, Wayland can also do networking better than X.
                Wayland does no networking whatsoever; buffers are passed directly via local system handles, and communication happens over unix sockets. The point is, it doesn't have to be. Frameworks like VNC have proven that accessing a desktop over the network can be handled (almost) completely separate from the windowing system.

                Comment


                • #18
                  Originally posted by Ancurio View Post
                  Wayland does no networking whatsoever; buffers are passed directly via local system handles, and communication happens over unix sockets. The point is, it doesn't have to be. Frameworks like VNC have proven that accessing a desktop over the network can be handled (almost) completely separate from the windowing system.
                  You obviously never read my Wayland writeup haha. You're right in that Wayland has no 'built-in' concept of networking. That being said, the protocol and the extensions are efficient enough that a Networking is expected to be a very simple addition. The latest mailing list post I saw was to constantly track screen-damages and send the changed pixels over the wire (much like VNC), granted that was a few months ago so it could've changed since then. We can always do better though, perhaps Linux will have a proper RDP replacement soon that's a bit more efficient than pixel-scrapping.*

                  *Just search "limitations of vnc", you'll find a few comparisons between VNC vs RDP vs X11 Network Transparency, VNC(-style) is the most flexible but its not the most efficient.
                  All opinions are my own not those of my employer if you know who they are.

                  Comment

                  Working...
                  X