Wayland's Weston Now Handles Full-Screen X Windows

[elanthis]

I don't think they have a huge team in order to pull something like this fast. X i think has more people.

Almost any project has more people. There is a huge shortage of developers in the Linux desktop space, despite hypothetical claims of FOSS allowing millions of eyeballs and hands. Microsoft has a test team for one small part of Windows that swamps the size of the developer pools for X/Wayland/GTK/GNOME combined. Turns out most people want to get paid for their hard work.

[89c51]

Almost any project has more people. There is a huge shortage of developers in the Linux desktop space, despite hypothetical claims of FOSS allowing millions of eyeballs and hands. Microsoft has a test team for one small part of Windows that swamps the size of the developer pools for X/Wayland/GTK/GNOME combined. Turns out most people want to get paid for their hard work.

Most wayland contributors are payed devs i think. Intel pays some, colabora some, the guy working on the input thing is payed and so on. Its just that there are few and no company as a whole is interested in desktop linux much. At least in advancing it.

[phred14]

Wayland should never be adopted as it is broken by design. It will just slow down applications compared to X without compositing.

Please explain. Broken by design is a much more serious accusation than merely a poor implementation.

[smitty3268]

I won't be using it on my desktop pc untill nvidia driver supports KMS anyway. Which I'm afraid will be never.

/fail

The nvidia driver already supports kernel mode setting, they just have a proprietary implementation of it.

NVidia just has to support the buffer sharing API that Wayland uses, which should be simple.

[Ibidem]

dont make me laugh. pulseaudio has latency issues so that you have to use a different mixer if you want low latency. i hope this is not an inherent flaw, cause if it is lennard should be hung from a lamppost.

Halfway: with pulseaudio, latency is tied directly to cpu usage, but there's an old and complicated api to ask for better latency. No guarantees that you'll get it, though.
Source: http://www.freedesktop.org/wiki/Soft...actencyControl
Of course, the OSS4 documentation points out that at least for video you only need ~30-40 millisecond latencies; the OSS4 developers state that they consider these normal latencies and claim that for single-digit latencies you need special hardware...Source:http://manuals.opensound.com/develop...io_timing.html

@JS987: I prefer X to wayland, but if you're bashing wayland, present some evidence.

[pq__]

Most wayland contributors are payed devs i think. Intel pays some, colabora some, the guy working on the input thing is payed and so on. Its just that there are few and no company as a whole is interested in desktop linux much. At least in advancing it.

Yeah, there are a few paid developers, who get paid to do whatever makes the companies and their clients happy. Unfortunately, the work is usually towards some embedded thing, which does not care too much about desktop user features. At other times, the work is for such low-level features, like the sub-surfaces that I am working on, that it does not directly benefit users. "Hey, I can make videos play with a lot less CPU usage! Uhh, sorry, it doesn't really improve desktop user interfaces, but it can make toolkit developers a little happier. Isn't that cool?! No? eheh..."

Thankfully there are some contributors who work on the desktop user features, so we're not completely stalled there.

[JS987]

I prefer X to wayland, but if you're bashing wayland, present some evidence.

Wayland doesn't support server side rendering, which means 2D acceleration using graphics driver isn't possible.
Applications will have to use OpenGL for 2D acceleration.
OpenGL (snb/gl) is (very) slow for 2D graphics compared to Intel SNA (snb/sna).
http://people.freedesktop.org/~ickle...r-20120712.png
OpenGL causes higher memory usage which means less memory can be used for disk caching
which means applications run slower.
qtperf4 uses 3 MB of private memory with X11 backend (-graphicssystem native), but 25 MB with OpenGL backend (-graphicssystem opengl).

[89c51]

Yeah, there are a few paid developers, who get paid to do whatever makes the companies and their clients happy. Unfortunately, the work is usually towards some embedded thing, which does not care too much about desktop user features. At other times, the work is for such low-level features, like the sub-surfaces that I am working on, that it does not directly benefit users. "Hey, I can make videos play with a lot less CPU usage! Uhh, sorry, it doesn't really improve desktop user interfaces, but it can make toolkit developers a little happier. Isn't that cool?! No? eheh..."

Thankfully there are some contributors who work on the desktop user features, so we're not completely stalled there.

pq__ is there a timeframe for the desktop stuff??


Maybe a specific todo (and the subsequent phoronix article ) can help things a bit.

[Ericg]

Wayland doesn't support server side rendering, which means 2D acceleration using graphics driver isn't possible.
Applications will have to use OpenGL for 2D acceleration.
OpenGL (snb/gl) is (very) slow for 2D graphics compared to Intel SNA (snb/sna).
http://people.freedesktop.org/~ickle...r-20120712.png

Radeon is already doing 2D-over-OpenGL for all GCN hardware so thats a non-issue for the simple fact we are already doing it. Does glamor suck right now? Yes. Can it get better? Yes. Will it get better? Yes.

[JS987]

Radeon is already doing 2D-over-OpenGL for all GCN hardware so thats a non-issue for the simple fact we are already doing it. Does glamor suck right now? Yes. Can it get better? Yes. Will it get better? Yes.

I don't use Radeon hardware. Glamor will maybe become as fast as Intel SNA, but it can take years. Glamor will have to be supported by GTK/Cairo/Qt as Wayland does no rendering.
Memory usage won't be probably solved as it is likely limitation of OpenGL.