4 ms·

This kind of project might see more success if it was made to work on Android devices. I'm cautiously optimistic that once KDE works on Wayland and Wayland work

by anon4 12y ago

This kind of project might see more success if it was made to work on Android devices. I'm cautiously optimistic that once KDE works on Wayland and Wayland works with Android drivers (or there are good quality mainline video drivers for the various chipsets), we would be able to root most any Android tablet and install a KDE/Wayland stack on it. I think that enabling that kind of thing - running an alternative environment on already existing hardware is a better goal than making your own hardware from scratch. Better as in more realistic to gain a userbase, I mean.

sp332 12y ago
Why wait for Wayland instead of running KDE on X on Android?
- ChuckMcM 12y ago
  I take it you have run KDE on X on Android on a small memory foot print ARM system? I tried something similar on the Pandaboard (OMAP4, 1GB, VRX) and it was unusable but if you've got a system or perhaps a configuration that makes this work it would be useful for development.
  - sp332 12y ago
    Nope, just wondering why Wayland would be any less painful.
    - yellowapple 12y ago
      Wayland cuts out a lot of the legacy cruft that X11 has accumulated over the decades, and is inherently much simpler; resource overhead is expected to be much lower on Wayland-based systems. It might not be less painful in terms of getting it to work, but it would be less painful in terms of getting it to work in a resource-constrained system like an ARM-powered smartphone.
      - kefka 12y ago
        The legacy cruft is easily served by not loading the respective crufty modules. And frankly, I like Xserver support across a variety of devices.
        yellowapple 12y ago
        The problem is that the legacy cruft is intertwined in the very architecture of X11 itself. That's what Wayland is intended to correct; it's a redesign to better accommodate things like compositors while not being held back by the need to explicitly support 80's-era graphical terminals and such (though similar functionality should still be possible, since Wayland is - IIRC - network-transparent).
        sp332 12y ago
        Wayland is not network transparent. They say it's "out of scope". You'll have to use something like RDP. http://wayland.freedesktop.org/faq.html#heading_toc_j_8 http://wayland.freedesktop.org/faq.html#heading_toc_j_8
        zanny 12y ago
        The X network transparency is already hugely inefficient compared to RDP or VNC. I also really hope when Dalaa 1.0 comes out (soon, please?) someone starts an initiative to make first class networked display / video streaming a reality on it. You don't really need a new protocol, just vnc-dalaa or rdp-dalaa rather than pixel buffer updates. See how Steam / Nvidia streaming works, or Miracast, except substitute hardware accelerated h264 with opencl accelerated Dalaa for a nice open standard.
        sp332 12y ago
        The issue for me is that I'll need a working video driver and framebuffer on the remote device, whereas with X I only need to set those up on the viewing end. Theoretically, someone could make a Wayland server that didn't render locally, but no one has made one of those yet.
    - ChuckMcM 12y ago
      Not a question of pain, more a question of performance. The current X server on Android on ARM is really really really slow. The hope is that Wayland will be more performant due to its design choices.
xanderstrike 12y ago
> I think that... running an alternative environment on already existing hardware is a better goal. I agree, and we know for a fact that it's a good goal because it's pretty much been the strategy of every desktop Linux distribution for decades. You don't need to own the hardware, you just need to make it work on someone else's.
- zanny 12y ago
  Consider though that having the software only gets you so far, though. the modern Linux desktop languishes from a lack of any compelling vendor options for Linux powered devices. You don't get mindshare and take over the personal computer world if nobody can get your software. And having to manually install the OS instantly knocks out 99% of potential users. So I do think at this point desktop Linux is at a crossroads where someone really needs to throw the money in to take over the personal computer space that Chromebooks are already dominating. Except instead of ChromeOS, throw around some Fedora, Lubuntu, Ubuntu, OpenSuse, etc devices. Same general idea, just without saying "were Google, use our cloud stuff because your computer can't do anything else", instead "heres a computer that does all the things the Windows machines do, except on ARM CPUs that cost a third as much and don't compromise". I guess for that, though, you need freedreno and lima to get some love so we can finally have some first class mobile GPU Mesa drivers, because the binary blobs are hopeless wrecks even if they ran fine in a classic X or Wayland stack.

by anon4 12y ago

sp332 12y ago

Why wait for Wayland instead of running KDE on X on Android?

ChuckMcM 12y ago
I take it you have run KDE on X on Android on a small memory foot print ARM system? I tried something similar on the Pandaboard (OMAP4, 1GB, VRX) and it was unusable but if you've got a system or perhaps a configuration that makes this work it would be useful for development.
- sp332 12y ago
  Nope, just wondering why Wayland would be any less painful.
  - yellowapple 12y ago
    Wayland cuts out a lot of the legacy cruft that X11 has accumulated over the decades, and is inherently much simpler; resource overhead is expected to be much lower on Wayland-based systems. It might not be less painful in terms of getting it to work, but it would be less painful in terms of getting it to work in a resource-constrained system like an ARM-powered smartphone.
    - kefka 12y ago
      The legacy cruft is easily served by not loading the respective crufty modules. And frankly, I like Xserver support across a variety of devices.
      - yellowapple 12y ago
        The problem is that the legacy cruft is intertwined in the very architecture of X11 itself. That's what Wayland is intended to correct; it's a redesign to better accommodate things like compositors while not being held back by the need to explicitly support 80's-era graphical terminals and such (though similar functionality should still be possible, since Wayland is - IIRC - network-transparent).
        sp332 12y ago
        Wayland is not network transparent. They say it's "out of scope". You'll have to use something like RDP. http://wayland.freedesktop.org/faq.html#heading_toc_j_8 http://wayland.freedesktop.org/faq.html#heading_toc_j_8
        zanny 12y ago
        The X network transparency is already hugely inefficient compared to RDP or VNC. I also really hope when Dalaa 1.0 comes out (soon, please?) someone starts an initiative to make first class networked display / video streaming a reality on it. You don't really need a new protocol, just vnc-dalaa or rdp-dalaa rather than pixel buffer updates. See how Steam / Nvidia streaming works, or Miracast, except substitute hardware accelerated h264 with opencl accelerated Dalaa for a nice open standard.
        sp332 12y ago
        The issue for me is that I'll need a working video driver and framebuffer on the remote device, whereas with X I only need to set those up on the viewing end. Theoretically, someone could make a Wayland server that didn't render locally, but no one has made one of those yet.
  - ChuckMcM 12y ago
    Not a question of pain, more a question of performance. The current X server on Android on ARM is really really really slow. The hope is that Wayland will be more performant due to its design choices.

xanderstrike 12y ago

> I think that... running an alternative environment on already existing hardware is a better goal. I agree, and we know for a fact that it's a good goal because it's pretty much been the strategy of every desktop Linux distribution for decades. You don't need to own the hardware, you just need to make it work on someone else's.

zanny 12y ago
Consider though that having the software only gets you so far, though. the modern Linux desktop languishes from a lack of any compelling vendor options for Linux powered devices. You don't get mindshare and take over the personal computer world if nobody can get your software. And having to manually install the OS instantly knocks out 99% of potential users. So I do think at this point desktop Linux is at a crossroads where someone really needs to throw the money in to take over the personal computer space that Chromebooks are already dominating. Except instead of ChromeOS, throw around some Fedora, Lubuntu, Ubuntu, OpenSuse, etc devices. Same general idea, just without saying "were Google, use our cloud stuff because your computer can't do anything else", instead "heres a computer that does all the things the Windows machines do, except on ARM CPUs that cost a third as much and don't compromise". I guess for that, though, you need freedreno and lima to get some love so we can finally have some first class mobile GPU Mesa drivers, because the binary blobs are hopeless wrecks even if they ran fine in a classic X or Wayland stack.