Mark Shuttleworth Declares Mir A Performance Win

[jrch2k8]

Please, enlighten me as to my shortcomings. I don't believe I've made a single statement in this forum today that would indicate that I favor Mir or Wayland or Canonical or any particular group, and no one has actually proven any of my statements to be false. Maybe you should be the one downvoted?

I do not care whether Mir or Wayland end up winning. It seriously does not matter either.

1.) you understand how Mir/Wayland and Xmir/Xwayland works? how they process things internally in the GPU? how much stress is put into the API of the backend? you realize is a very very simple layer and the backend don't do nothing a simple SDL script + hacks won't achieve?
2.) you understand the guy that wrote XWayland and later ripped off as XMir stated is not good enough and need drm/kms side fixes, to be a real solution?
3.) you know Xmir and Xwayland are desgined for rootless apps? aka not running an XServer session in fullscreen
4.) don't you think wayland devs[Xorg/intel/amd/community/Vmware/kernel.org/redhat], that have many brilliant devs working for it would annouced XWayland if it was good enough?
5.) you realize wayland is finished except some corner cases that will be tackled in 1.2 release?
6.) explain to me why Redhat or anyone responsable will force you to test a layer that impact minimaly the features of wayland[or mir] that its own developer among many other brilliant ppl stated is not good enough yet?
7.) why you assume we wanna test XWayland?
8.) we are months away of fully native toolkits and DE for wayland upstream, meaning every distro out there will have it out-of-the-box just needeing a switch or option test to switch back and forth between wayland and X11
9.) you realize wayland is backed by the Xorg foundation and freedesktop.org and FSF? you know how much that weight? you seen its members?

my point is, if you don't have at least some of those answer how can you state what is good or bad? you tell your doctor what to do if you don't even know anatomy or medicine?

to put your post in an analogy "canonical is giving me a old VW beetle with a exterior made of paper that look like a ferrari so i can see it now, while wayland is an actual ferrari but i don't know to reach italy to see one, so it sucks"

[a user]

Maybe I'm wrong on this, but XMir runs on top of Mir, so Mir is still being used

you are wrong on this.

, and on Ubuntu 13.10 you'll have the option of testing Unity8, which will be running purely on Mir...

and here again. unity on 13.10 is running on XMir, meaning it is running on the X emulation and not on mir. the first unity running on mir was announced for 14.04 but it depends how far they will be then.

[Kayden]

And yet Wayland cannot display simple OpenGL apps, wheter running in XWayland or not. Mir can.

That's simply false; Wayland runs OpenGL apps just fine. As a demo, we even had Team Fortress 2 running natively under Wayland at one point.

[Bestia]

ok lets get the hard to miss facts that if you actually read the code would OVBIOUS before answer your post

...

if you need evidence find someone who can understand C and C++ and ask him to read both codebases

ok, now to your post

...

3.) Wayland and everybody else[Mplayer/Gtk/Qt5/EFL/SDL/KDE/GNOME/e17/etc] except canonical wonderboys are focused full steam ahead in native ports instead of make PR with some technically useless demo or push alpha software to users
4.) Qt5, GTK , SDL, EFL apps already runs very good in wayland natively and Gnome/e17 should work natively in wayland this year tops and KDE prolly will in Q1 2014, like i said before wayland ppl don't make PR they port
...

When you mention all those toolkits and their support for Wayland you should probably back those claims with this link.

Looking at that page doesn't seem to indicate that SDL has very good support in Wayland.

SDL

Benjamin Franzke is working on a port of SDL to Wayland, it's available in his sdl-wayland repo on freedesktop.org.

Last commit in this repository dates back to February 2012. Searching for more information about Wayland support in SDL doesn't bring better results.

SDL support is rough. Latest sdl with wayland support is soreau sdl. Before running any applications, please:
export SDL_VIDEODRIVER=wayland

Last commit in this repository was made 5 months ago and it's about X11 not Wayland.

You also won't find any trace of Wayland support in official SDL Mercurial repository. Look at the /src/video there is definitely X11 but no Wayland or Mir.

[coder543]

1.) you understand how Mir/Wayland and Xmir/Xwayland works? how they process things internally in the GPU? how much stress is put into the API of the backend? you realize is a very very simple layer and the backend don't do nothing a simple SDL script + hacks won't achieve?
2.) you understand the guy that wrote XWayland and later ripped off as XMir stated is not good enough and need drm/kms side fixes, to be a real solution?
3.) you know Xmir and Xwayland are desgined for rootless apps? aka not running an XServer session in fullscreen
4.) don't you think wayland devs[Xorg/intel/amd/community/Vmware/kernel.org/redhat], that have many brilliant devs working for it would annouced XWayland if it was good enough?
5.) you realize wayland is finished except some corner cases that will be tackled in 1.2 release?
6.) explain to me why Redhat or anyone responsable will force you to test a layer that impact minimaly the features of wayland[or mir] that its own developer among many other brilliant ppl stated is not good enough yet?
7.) why you assume we wanna test XWayland?
8.) we are months away of fully native toolkits and DE for wayland upstream, meaning every distro out there will have it out-of-the-box just needeing a switch or option test to switch back and forth between wayland and X11
9.) you realize wayland is backed by the Xorg foundation and freedesktop.org and FSF? you know how much that weight? you seen its members?

my point is, if you don't have at least some of those answer how can you state what is good or bad? you tell your doctor what to do if you don't even know anatomy or medicine?

to put your post in an analogy "canonical is giving me a old VW beetle with a exterior made of paper that look like a ferrari so i can see it now, while wayland is an actual ferrari but i don't know to reach italy to see one, so it sucks"

I'm aware of the technical aspects, believe it or not. But, I have neither a reason nor the time to prove it to you.

[jrch2k8]

When you mention all those toolkits and their support for Wayland you should probably back those claims with this link.

Looking at that page doesn't seem to indicate that SDL has very good support in Wayland.


Last commit in this repository dates back to February 2012. Searching for more information about Wayland support in SDL doesn't bring better results.


Last commit in this repository was made 5 months ago and it's about X11 not Wayland.

You also won't find any trace of Wayland support in official SDL Mercurial repository. Look at the /src/video there is definitely X11 but no Wayland or Mir.

1.) i did the first 200 times, now i just imply ppl know so but sure is a good advice
2.) well repo activity does not mean support, it just means is either ready waiting for upstream or waiting for fix to reach upstream.
3.) well SDL have some requests in mailing list among them i think multiprocess render for 1 surface is one of those and i think some landed in 1.1 while others are wating for 1.2
4.) looking at the code is not so far away but maybe asking in IRC which is the specific issue blocking upstream to the main developer

[coder543]

you are wrong on this.


and here again. unity on 13.10 is running on XMir, meaning it is running on the X emulation and not on mir. the first unity running on mir was announced for 14.04 but it depends how far they will be then.

See. Now I don't trust your knowledge here. Unity8 is the experimental version that runs natively on Mir. Unity7 will be running on XMir. So, I'll wait on someone else to tell me that XMir doesn't use Mir.

[jrch2k8]

I'm aware of the technical aspects, believe it or not. But, I have neither a reason nor the time to prove it to you.

well i actually don't care if you do to be honest but at least i can say you don't understand how XMir works otherwise you won't post that assumption or try to stretch the ball to redhat park that way[especially since redhat is less related to wayland than Intel but still a good supporter and contributer]

conclusion if you don't understand ask nicely but don't go agresivelly assume things out of thin air and expect not to pass as a troll

[UraniumDeer]

Lemme just add a little FUD to the fire :P

Mir is currently Wayland + libhybris (iirc).
If Mir comes out on top on all this, will Wayland die? Will the future development of the display server then be in the hands of a group of people that knows very little about the inner working of X and the creation of a displayserver?
Is so, I wouldn't really consider it a win for Linux.

[F i L]

Maybe I'm wrong on this, but XMir runs on top of Mir, so Mir is still being used, and on Ubuntu 13.10 you'll have the option of testing Unity8, which will be running purely on Mir... but the real beta test for this release is with Mir utilizing XMir. It's still a significant technical accomplishment, and more than just buffer swapping. I wouldn't be too quick to discount it. But, yes, Wayland is further along. It has been in development for many years... I would expect no less of it.

Mir and Wayland are display servers, meaning they're the API which applications (directly or indirectly through Qt/Gtk) call to receive input, request resolutions, etc.. (sorry if you know this, i'm just being thorough). XMir/XWayland is a X11 extension (or patch or both, idk) which passes on X11 events to a Mir/Wayland and creates a rootless X11 environment housable under Mir/Wayland.

Much of the Mir API useful for applications isn't complete, so with Ubuntu 13.10 Mir has basically three tasks (that i'm aware of): Display management (multiple screens, resolutions, color, etc), OpenGL ES screen presentation (i think it's using OpenGL ES), and Input handling (wrapping up linux udev events). Since Mir in 13.10 is just running a full screen XMir app, I doubt the GLES code will need to be very complicated (basically just buffer swapping) and the input management should be straight forward as well (though i believe this is an area where Mir did not copy Wayland's design. could be wrong). My guess is that most of the work is being put into Display Management right now, and I've heard it's largely a copy of Waylands design.

All the Mir window management stuff is yet to be done in Unity 8, so until that is Ubuntu's default desktop you can't really claim "Mir" is receiving a ton of beta testing where Wayland is not.. especially considering Wayland is actually testing the window management features with Weston (and now Gnome/KDE) whereas Mir is still playing catch-up on the other stuff.