Announcement

Collapse
No announcement yet.

The X.Org Server's GLX Is Being Rewritten

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

  • The X.Org Server's GLX Is Being Rewritten

    Phoronix: The X.Org Server's GLX Is Being Rewritten

    While most developers are focused around new innovations for Wayland (or Mir), there's still life ahead for the X.Org Server in maintaining legacy support and other cases where the xorg-server will not die for years to come. In improving the X.Org Server, Adam Jackson at Red Hat has been working on rewriting the GLX portion of the X.Org Server...

    http://www.phoronix.com/vr.php?view=MTQ2OTY

  • #2
    Great

    Great, keep reducing the X.org codebase, make it slimmer, leaner and cleaner!

    Refactor code and remove duplicate and redundant code.
    Get rid of legacy stuff.

    libgl needs to be fixed to not depend on glx though.

    Comment


    • #3
      Why does that Mir thing have to be mentioned in every article? This has nothing to do with Mir!

      Comment


      • #4
        I find something like this to be a good opportunity to move to X12, especially if this has a probability of breaking current xorg drivers. On the other hand, there are other things done in the past few years that could have qualified for switching to X12, such as xinput2. I feel like if X12 does ever get released, it will be relatively arbitrary, much like the linux kernel hitting 3.0. Much like X11, the linux kernel had several opportunities to ditch the 2.6 series. IMO, the switch to udev would've been a good time to do so (and even then, it should have been named 2.8).

        Comment


        • #5
          Originally posted by schmidtbag View Post
          I find something like this to be a good opportunity to move to X12, especially if this has a probability of breaking current xorg drivers. On the other hand, there are other things done in the past few years that could have qualified for switching to X12, such as xinput2. I feel like if X12 does ever get released, it will be relatively arbitrary, much like the linux kernel hitting 3.0. Much like X11, the linux kernel had several opportunities to ditch the 2.6 series. IMO, the switch to udev would've been a good time to do so (and even then, it should have been named 2.8).
          Effectively X12 is Wayland.

          Comment


          • #6
            Originally posted by n3wu53r View Post
            Effectively X12 is Wayland.
            Wayland is effectively X11's replacement, but considering how architecturally different wayland is, I would not classify it as a "successor".

            Comment


            • #7
              Originally posted by uid313 View Post
              Great, keep reducing the X.org codebase, make it slimmer, leaner and cleaner!

              Refactor code and remove duplicate and redundant code.
              Get rid of legacy stuff.
              +1

              Especially when working towards making Xorg legacy to the post X11 era.

              Comment


              • #8
                Originally posted by schmidtbag View Post
                I find something like this to be a good opportunity to move to X12, especially if this has a probability of breaking current xorg drivers. On the other hand, there are other things done in the past few years that could have qualified for switching to X12, such as xinput2. I feel like if X12 does ever get released, it will be relatively arbitrary, much like the linux kernel hitting 3.0. Much like X11, the linux kernel had several opportunities to ditch the 2.6 series. IMO, the switch to udev would've been a good time to do so (and even then, it should have been named 2.8).
                I think you're conflating X11 the protocol with Xorg the implementation. Nothing that's being done here represents a change to the X11 protocol... it's just changing the implementation and some of the protocol extensions....

                Comment


                • #9
                  Originally posted by schmidtbag View Post
                  I find something like this to be a good opportunity to move to X12, especially if this has a probability of breaking current xorg drivers.
                  If we bumped the X version every time we changed Xorg driver ABI, we'd be on X25 by now.

                  Originally posted by schmidtbag View Post
                  I feel like if X12 does ever get released, it will be relatively arbitrary, much like the linux kernel hitting 3.0.
                  It would be because we changed the protocol between the X server and client applications in an incompatible way and no longer supported the old applications without some sort of compatibility shim. So far, we've gotten away with just adding extensions to add/replace functionality and not removing/breaking anything in the core protocol, though we've kept a wishlist of things to change if we ever do decide to make an X12.

                  Comment


                  • #10
                    Originally posted by alanc View Post
                    If we bumped the X version every time we changed Xorg driver ABI, we'd be on X25 by now.
                    Thank god there are still sane software people in the world. If Chrome's and FF's version inflation spread there too, we'd indeed have X25 :P

                    Comment


                    • #11
                      Originally posted by alanc View Post
                      If we bumped the X version every time we changed Xorg driver ABI, we'd be on X25 by now.
                      Considering X11 is 26 years old, that's not as unreasonable as you're making it seem. That's basically a new version every 1.6 years, which is a good timeframe for software releases. But, what separated it from X10 or any of the other previous versions? Considering X11 has been around longer than me, I don't see how any of the major changes throughout X11's history would be considered less significant than the decision behind moving from X10 to X11. I would find it hard to believe if X11 happened to be so effective that for 26 years it just needed tweaks, extensions, and driver ABI changes, especially for something as complex as a GUI.

                      I'm not going to pretend I know better, and in the end I suppose a version number doesn't seriously matter. But to me personally, sticking with 1 version for a quarter century becomes equally as meaningless of a version system as increasing the whole number once every other month like FF or Chrome.
                      Last edited by schmidtbag; 09-27-2013, 10:54 AM.

                      Comment


                      • #12
                        Originally posted by schmidtbag View Post
                        Considering X11 is 26 years old, that's not as unreasonable as you're making it seem. That's basically a new version every 1.6 years, which is a good timeframe for software releases. But, what separated it from X10 or any of the other previous versions? Considering X11 has been around longer than me, I don't see how any of the major changes throughout X11's history would be considered less significant than the decision behind moving from X10 to X11. I would find it hard to believe if X11 happened to be so effective that for 26 years it just needed tweaks, extensions, and driver ABI changes, especially for something as complex as a GUI.
                        Well, it wouldn't be true to say it was so effective to not need anything more - afterall, people have spent much of that time trying to work around deficiencies, and to design replacements for it.

                        But yes, X11 *is* the same protocol that was released in '87, and while modern apps wouldn't run on an old server (due to lack of extensions), an application written and compiled 25 years ago should still run fine when talking to the latest Xorg server...



                        Originally posted by schmidtbag View Post
                        I'm not going to pretend I know better, and in the end I suppose a version number doesn't seriously matter. But to me personally, sticking with 1 version for a quarter century becomes equally as meaningless of a version system as increasing the whole number once every other month like FF or Chrome.
                        It's the version of the *protocol*, and protocol versions aren't changed just because marketing people think that bigger numbers sound better. It's very meaningful that the version hasn't changed in such a long time, because it means compatibility with other implementations of that protocol hasn't been broken either.

                        Comment


                        • #13
                          a couple of beers later

                          Rewrites are fine if somebody has documented existing functionality.
                          MIR/Wayland != XOrg
                          That is the new idea isn't the equivalent to the old idea.

                          We're always losing functionality. Gnome 3, KDE 4, and so on, it's a new play: Functionality Lost.
                          ie. Paradise Lost.

                          Comment

                          Working...
                          X