GNOME & Intel Developers Plan The Wayland Future

[RahulSundaram]

You do know the url for the fdo specifications is, right? http://standards.freedesktop.org/

You should read the content rather than just the url. Let me help you a bit

http://www.freedesktop.org/wiki/

"freedesktop.org hosts any "on-topic" software projects"

"freedesktop.org is not a formal standards organization"
"Unlike a standards organization, freedesktop.org is a "collaboration zone" where ideas and code are tossed around, and de facto specifications are encouraged. The primary entry to these discussions is the xdg mailing list."

Again, there is no such thing as a freedesktop.org standard. There are some specifications that are widely adopted and some that are not. Just because you host a specification doesn't make it mandatory for everyone to adopt it. It is purely consensus driven. "defacto" and not "dejure"

Any software can be hosted in freedesktop.org Even experimental projects used by noone has been hosted there. Example http://razor.freedesktop.org/

[jrch2k8]

So, as an end user, will I get better performance from Wayland? Will the transitions/animations be smoother in Gnome Shell? Will dragging a window be smoother? Also, will Wayland have any effects on OpenGL performance?

the short answer is yes.

the long answer is wayland is just a link of the chain not the chain, to extend a bit more the idea lets go back 8+ years when xfree86 was king!! graphic drivers were basically UMS with minimal acceleration and opengl 1.2 was the only thing semi supported in linux, in those days xfree86 was basically an OS on top a linux kernel designed to run on state of the art 80's PC's with amber display. !!it was horrible!!!

So around that time the community finally realized the situation was an never ending ever growing evil mess and needed to be fixed if something like a "Desktop" were ever to exist, so measure were taken and basically the answer was "rewrite everything and kill the old code with fire" but this has to be done step by step because wasn't any easy to tackle that horrible monolithic monster with millions of LOC that basically handled everything from printers to fonts without forget demonic things like motif, called Xfree86. so the work begun which is basically this resumed list:

* Murder motif with fire and toolkits like Gtk/qt/EFL were born
* kill Xprint/lpr and friends and give a shot to that small project called CUPS
* kill the horrid Xft code instead freetype/fontconfig took the job
* Modularize the beast -- current Xorg
* Kill UMS with fire and implement a proper memory manager tech in the kernel, today know as KMS/DRM
* remove PCI handling"ism" from X -- libpciaccess + kernel code
* Dance around the XAA corpse laughing -- Exa/UXa/glamor/SNA were born
* have a proper driver archicture foundation -- Gallium/dri2/mesa see the light
* Give a nice goodbye to GLX -- EGL tadaaa
* Proper Open Drivers -- currently WIP
* Proper Opengl support -- currently WIP
* Proper input handling -- XInput2 WIP
* Video accel in X? jajajaja --- vdpau/VA enter the theather
* Multimonitor? crazy talk those monitor weight too much !burn the heretic Xinerama! -- kindly Xrandr
* Composite? acceralation? i do ambar fast enough -- Xcomposite/Xdamage/indirect rendering/etc <-- hackish
* many other thing less important chopped from X

Today mostly everything important is already chopped out of X and improved as outside projects[or replace by better one] but to have nice desktops as you do today you have to basically bypass 99% of the functions of X and use things like opengl directly at the cost of performance due to round trips since X don't support it, that made guys like christian or keith to realize that is actually far better control the GPU directly with proper APIs than keep relying in this old half-deaf-half-blind police to be in the middle and that way Wayland is born as the final link of the chain of events that for years are happening to finally give a proper funeral to XFree86 days <-- thanks for 30 years of service but is time to go florida and enjoy retirement

about the performance it is very noticeable since with wayland everything talk directly to the GPU [Apps, Compositors,etc] using just a protocol to keep things coherent, perf advantages

* round trips X went to florida remember
* memcopy from X to GPU to compositor and back are past
* wayland footprint is stupidly minimal and goes inside the compositor or your apps is not a central process
* Exa/Glamor?? thing of the past baby, now is direct to the GPU using proper tooling
* proper Overlay support !!OMG!! finally
* fliker? out of sync frames? jitter? dafuq is that?
* Color accurate rendering
* I did mentioned it can run the milisecond KMS kicks in?
* etc, read wayland page lots of more info

so, yes user will notice awesomeness and fluidity when they see it

[bkor]

You do know the url for the fdo specifications is, right? http://standards.freedesktop.org/

You have no clue how freedesktop.org works. It is a place where people from various desktop environments work together. Just putting something on there doesn't imply anything. Nobody will automatically use it.

[bkor]

I guess you have never heard of gstreamer, for example.

What the fuck do you mean? Really, if you have a point, make it. Up to now I only see empty stuff from you. Spend a little time to prove why GNOME is "NIH". All these empty things from you maybe is nice for your Phoronix post count, but suggest to actually already prove the NIH.

Now: Poppler. AFAIK mostly maintained by KDE people. Not that it matters who maintains it, but apparently quite important to you. Anyway, another example of something which we use. So even in case GNOME is totally NIH except Poppler, the one example totally proves that we're not NIH. Or at least by the reasoning you've been demonstrating.

Already hinted to raise the bar of everyones argumentation a bit as I actually know what I am talking about.

[bkor]

Which is only needed because GTK refuses to implement compatibility with Qt themes. Qt is compatible with GTK themes, so this sort of thing isn't needed on Qt (or, by extension KDE).

While in real life, gtk makes this possible by providing x settings and standardizing on this. Things we talked about between KDE/Qt and Gtk+ maintainers. Yet another "let's see if this will stick" argument, similar to the other dude. I've seen 10-15 attempts already like this. Wtf.

[sarmad]

the short answer is yes.

the long answer is wayland is just a link of the chain not the chain, to extend a bit more the idea lets go back 8+ years when xfree86 was king!! graphic drivers were basically UMS with minimal acceleration and opengl 1.2 was the only thing semi supported in linux, in those days xfree86 was basically an OS on top a linux kernel designed to run on state of the art 80's PC's with amber display. !!it was horrible!!!

So around that time the community finally realized the situation was an never ending ever growing evil mess and needed to be fixed if something like a "Desktop" were ever to exist, so measure were taken and basically the answer was "rewrite everything and kill the old code with fire" but this has to be done step by step because wasn't any easy to tackle that horrible monolithic monster with millions of LOC that basically handled everything from printers to fonts without forget demonic things like motif, called Xfree86. so the work begun which is basically this resumed list:

* Murder motif with fire and toolkits like Gtk/qt/EFL were born
* kill Xprint/lpr and friends and give a shot to that small project called CUPS
* kill the horrid Xft code instead freetype/fontconfig took the job
* Modularize the beast -- current Xorg
* Kill UMS with fire and implement a proper memory manager tech in the kernel, today know as KMS/DRM
* remove PCI handling"ism" from X -- libpciaccess + kernel code
* Dance around the XAA corpse laughing -- Exa/UXa/glamor/SNA were born
* have a proper driver archicture foundation -- Gallium/dri2/mesa see the light
* Give a nice goodbye to GLX -- EGL tadaaa
* Proper Open Drivers -- currently WIP
* Proper Opengl support -- currently WIP
* Proper input handling -- XInput2 WIP
* Video accel in X? jajajaja --- vdpau/VA enter the theather
* Multimonitor? crazy talk those monitor weight too much !burn the heretic Xinerama! -- kindly Xrandr
* Composite? acceralation? i do ambar fast enough -- Xcomposite/Xdamage/indirect rendering/etc <-- hackish
* many other thing less important chopped from X

Today mostly everything important is already chopped out of X and improved as outside projects[or replace by better one] but to have nice desktops as you do today you have to basically bypass 99% of the functions of X and use things like opengl directly at the cost of performance due to round trips since X don't support it, that made guys like christian or keith to realize that is actually far better control the GPU directly with proper APIs than keep relying in this old half-deaf-half-blind police to be in the middle and that way Wayland is born as the final link of the chain of events that for years are happening to finally give a proper funeral to XFree86 days <-- thanks for 30 years of service but is time to go florida and enjoy retirement

about the performance it is very noticeable since with wayland everything talk directly to the GPU [Apps, Compositors,etc] using just a protocol to keep things coherent, perf advantages

* round trips X went to florida remember
* memcopy from X to GPU to compositor and back are past
* wayland footprint is stupidly minimal and goes inside the compositor or your apps is not a central process
* Exa/Glamor?? thing of the past baby, now is direct to the GPU using proper tooling
* proper Overlay support !!OMG!! finally
* fliker? out of sync frames? jitter? dafuq is that?
* Color accurate rendering
* I did mentioned it can run the milisecond KMS kicks in?
* etc, read wayland page lots of more info

so, yes user will notice awesomeness and fluidity when they see it

Thanks for the thorough reply. I'm glad to hear this. Responsiveness combined with the cool usability of Gnome Shell = awesomeness.

[Awesomeness]

Dantti has been working on colord support on KDE for a quite awhile.

Yeah? LOL? Daniel Nicoletti has ADS or something like that. He has a thousand parallel projects and he can't finish a single one. He already got distracted from colord-kde by Red Hat's new NetworkManager Plasma applet. Or was it his new e-mail client that's his current project? No idea. He's so jumpy.
Wake me when his stupid Apper thing is usable before bothering me with another vaporware project of him?

The KWin color management is not tied to Oyranos.

For now it is. Checking the color correction checkbox in KWin's KCM makes currently talk KWin only to Oyranos.

[Awesomeness]

http://www.freedesktop.org/software/...q.html#oyranos

Notice that colord has several important features like being async and thread safe and is far more widely adopted by major projects including CUPS, Wayland and yes, KDE. So in this case, accusing them of NIH just shows bias.

Thanks for the link as it proves that Oyranos is 6 years older than colord and therefore colord is NIH syndrome in action.

Citing projects which Red Hat made use its own colord project is not disproving NIH.

[RahulSundaram]

Thanks for the link as it proves that Oyranos is 6 years older than colord and therefore colord is NIH syndrome in action.

Citing projects which Red Hat made use its own colord project is not disproving NIH.

Since when is say CUPS a Red Hat Project? By your silly definition, Wayland is NIH.

[Awesomeness]

Qt dependency is not NIH.

Yes, it is.
It's perfectly possible to write Gnome HIG-compliant applications in Qt. If Mono with all its duplicated features could become a Gnome dependency, so could Qt.The difference is that Mono was written by Gnome's founder, while hostility towards Qt 1's licensing is the reason Gnome exists at all.

So did KDE work with others?

Of course and I already proved that.