4 ms·

Someone explain why the push to wayland when there is no discernable performance increase [0] and lots of broken applications. [0] https://www.phoronix.com/sca

by verroq 6y ago

Someone explain why the push to wayland when there is no discernable performance increase [0] and lots of broken applications. [0] https://www.phoronix.com/scan.php?page=news_item&px=GNOME-Xorg-Wayland-AMD-Renoir https://www.phoronix.com/scan.php?page=news_item&px=GNOME-Xo...

roenxi 6y ago
Most of X.Org is horrible and the entire system is unmaintainable. It is hard to even get maintenance releases for it released, as the article alludes to. Development hasn't stalled because people think X is good. Development has stalled because the fundamental design is so misaligned from the modern graphics stack that improvements are not worth attempting.
- p_l 6y ago
  The thing is, X.Org server is not X11 the protocol. X11 could be easily implemented on new driver framework supporting new graphic stack. Instead we got a piece of sh*t that is fitting for a custom embedded device or some closed environment, but not an X11 replacement.
  - roenxi 6y ago
    Nobody in particular wants to use the X11 protocol. It is complicated and supports a bunch of weird functionality that is rather niche. The trend seems to be to use OpenGL and call that done. I haven't seen many serious complaints that Wayland doesn't support the X11 protocol since people can run an X server directly using XWayland.
    - p_l 6y ago
      As I mentioned elsewhere, XWayland isn't actually compatible if you want to support modern applications, for example ones that might expect a systray. ICCM is, iirc, broken (I haven't spent time checking after finding out some apps critical to me didn't work at all). X11 could be updated a lot. But some of the "weird functionality" is stuff that is slowly becoming available for normal people that was thrown out with the bathwater by GTK3, like ability to use more than 8bit per colour channel.
      - roenxi 6y ago
        > As I mentioned elsewhere, XWayland isn't actually compatible if you want to support modern applications Yeah, Wayland is a bad protocol. It isn't flexible enough to do what X11 does. But if it was a good protocol, and capable of implementing X11, then it would get X11 for free through XWayland. They don't need to reimplement X11. They can use X.org or whatever for that.
        p_l 6y ago
        My point is that XWayland, for various reasons, does not replicate X11 fully. So yeah, I can open something like Xnest/Xephyr. At that point, the utility value of Wayland drops ridiculously heavy from my point of view.
  - sergeykish 6y ago
    It is theoretical until you've done it. People who worked on XOrg decided to abandon it and create Wayland. X11 is not good enough for them but is good for you...
- verroq 6y ago
  So where are these Wayland “improvements”? A Wayland stack is “better” but has same perf as X11.
madmulita 6y ago
RedHat
boudin 6y ago
Wayland is not just focused on performances but on addressing security flaws of X11. Those benchmarks were benchmarks of gnome 3.36, it doesn't represent "wayland" but gnome 3.36 implementation of it. If your concern is performance, it's still pretty much on of the point of focus. Gnome 3.38 already brought some good improvement there and there's still a few things to come in the area. In term of broken application, what do you mean?

by verroq 6y ago

roenxi 6y ago

Most of X.Org is horrible and the entire system is unmaintainable. It is hard to even get maintenance releases for it released, as the article alludes to. Development hasn't stalled because people think X is good. Development has stalled because the fundamental design is so misaligned from the modern graphics stack that improvements are not worth attempting.

p_l 6y ago
The thing is, X.Org server is not X11 the protocol. X11 could be easily implemented on new driver framework supporting new graphic stack. Instead we got a piece of sh*t that is fitting for a custom embedded device or some closed environment, but not an X11 replacement.
- roenxi 6y ago
  Nobody in particular wants to use the X11 protocol. It is complicated and supports a bunch of weird functionality that is rather niche. The trend seems to be to use OpenGL and call that done. I haven't seen many serious complaints that Wayland doesn't support the X11 protocol since people can run an X server directly using XWayland.
  - p_l 6y ago
    As I mentioned elsewhere, XWayland isn't actually compatible if you want to support modern applications, for example ones that might expect a systray. ICCM is, iirc, broken (I haven't spent time checking after finding out some apps critical to me didn't work at all). X11 could be updated a lot. But some of the "weird functionality" is stuff that is slowly becoming available for normal people that was thrown out with the bathwater by GTK3, like ability to use more than 8bit per colour channel.
    - roenxi 6y ago
      > As I mentioned elsewhere, XWayland isn't actually compatible if you want to support modern applications Yeah, Wayland is a bad protocol. It isn't flexible enough to do what X11 does. But if it was a good protocol, and capable of implementing X11, then it would get X11 for free through XWayland. They don't need to reimplement X11. They can use X.org or whatever for that.
      - p_l 6y ago
        My point is that XWayland, for various reasons, does not replicate X11 fully. So yeah, I can open something like Xnest/Xephyr. At that point, the utility value of Wayland drops ridiculously heavy from my point of view.
- sergeykish 6y ago
  It is theoretical until you've done it. People who worked on XOrg decided to abandon it and create Wayland. X11 is not good enough for them but is good for you...
verroq 6y ago
So where are these Wayland “improvements”? A Wayland stack is “better” but has same perf as X11.

madmulita 6y ago

RedHat

boudin 6y ago

Wayland is not just focused on performances but on addressing security flaws of X11. Those benchmarks were benchmarks of gnome 3.36, it doesn't represent "wayland" but gnome 3.36 implementation of it. If your concern is performance, it's still pretty much on of the point of focus. Gnome 3.38 already brought some good improvement there and there's still a few things to come in the area. In term of broken application, what do you mean?