LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   General (http://www.linuxquestions.org/questions/general-10/)
-   -   X vs Wayland (http://www.linuxquestions.org/questions/general-10/x-vs-wayland-4175468850/)

linuxCode 07-08-2013 05:53 AM

X vs Wayland
 
Hi

I was listening to the linux action show about X vs Wayland. The debate starts at 42 minutes into the video.

X vs Wayland article

Should X still continue to be the defactor of linux or do we move forward to a future of wayland?

TobiSGD 07-08-2013 07:05 AM

Your poll is flawed. The first option, "X should stay and those limitations/bugs should be fixed" is not possible without breaking backwards compatibility, which is one of the main reasons why the Xorg developers decided to do a new start with Wayland. When breaking backwards compatibility there is point left why not just start a new design, without the flaws in the old design (which prevent fixing of those bugs). The other reason is that the X design (created more than 30 years ago) does mostly not apply anymore to how modern desktops work.

linuxCode 07-08-2013 08:07 AM

Quote:

Originally Posted by TobiSGD (Post 4986132)
Your poll is flawed. The first option, "X should stay and those limitations/bugs should be fixed" is not possible without breaking backwards compatibility,

I did not know that. I was making a point as a user and not as a developer. Since you are a mod, can you edit the poll and remove the and those limitations/bugs should be fixed part. I would do it myself but I think I can't edit it

/dev/random 07-08-2013 08:24 AM

Quote:

Originally Posted by TobiSGD (Post 4986132)
Your poll is flawed. The first option, "X should stay and those limitations/bugs should be fixed" is not possible without breaking backwards compatibility, which is one of the main reasons why the Xorg developers decided to do a new start with Wayland. When breaking backwards compatibility there is point left why not just start a new design, without the flaws in the old design (which prevent fixing of those bugs). The other reason is that the X design (created more than 30 years ago) does mostly not apply anymore to how modern desktops work.

Sorta, I have read a lot of about this and there is a lot of misinformation going around about this, to muck up the waters a bit, You are correct to say you can't fix the current extensions without breaking backwards compatibility, but notice that everyone that is pushing for Wayland has never told anyone that it's possible to come up with new extensions that fix older problems, still making X.org backwards compatible, just include both newer and older extensions.

The push for Wayland is running on pure BS for all the wrong reasons. X.org (X for short) is not slow! In fact Valve gets faster frame rates under X then in Windows with worse preforming drivers then the windows counter parts. If X were slow this would not be true.

I have a problem with Wayland it's called breaking compatibility with everything else, (this isn't the windows world, you don't need to rewrite the API's every 5 years in order to profit). Call me when Wayland is 100% cross compatible with all the BSD world, all the Solaris world and the rest of the UNIX world, if it isn't then it's not needed.

I also don't buy the whole speed thing, why? Because as of late Linux is getting bloated with more and more never needed abstraction layers and programs that are written in stupid languages that don't make sense to write these programs in in the first place... Perl, Java, Java Script and Ruby are NOT exactly low level languages. They are High Level Languages or (HLLS) so if you cry about speed, stop using useless abstractions layers and programs that are written in languages they shouldn't be written in in the first place.

Not saying this is you attacking X but if you use udev, pulseaudio, DBUS, and policy kit or any other slow honking turd you have no right to complain about how slow X is as you are using much worse and slower programs and you don't bother to attack those so I don't see why you would attack X.

TobiSGD 07-08-2013 10:09 AM

If the Xorg developers themselves claim that they can't solve those bugs without breaking the design I have to believe them, since they are far more proficient in that topic than I am. Have a look here, where a Wayland developer (and former Xorg developer) explains the problems: http://www.youtube.com/watch?v=RIctzAQOe44
If you know a way around these problems I would recommend to contact him, so that you can work that out with him. I for sure can't, so I have to believe what he tells me.

TobiSGD 07-08-2013 10:10 AM

Quote:

Originally Posted by linuxCode (Post 4986167)
I did not know that. I was making a point as a user and not as a developer. Since you are a mod, can you edit the poll and remove the and those limitations/bugs should be fixed part. I would do it myself but I think I can't edit it

AFAIK, polls are unalterable, except for Jeremy, so just leave it in or contact Jeremy about it.

H_TeXMeX_H 07-08-2013 12:53 PM

This poll cannot possibly be more biased, except if you changed the title to "X sucks, Wayland rocks".

Both links you link to are severely biased, i.e. let me summarize the sections in the phoronix article (also summarized in the proposed title change):
The Failings of X
The Fixings Of Wayland
Some Misconceptions about X and Wayland (this is also biased, as all of them are against X and for Wayland)
A Few Generic Advantages Of Wayland

How about provide a balanced article that logically takes the pros and cons of each, NOT the cons of one and the pros of the other. Provide better poll options that are not biased.

My choice is:
I do care, and I will wait and see if Wayland fails, because that's the only one in danger of failing, not X.

jamison20000e 07-08-2013 10:49 PM

...i want to vote all 3 + like :D
improvements here there, yes\thx...

jamison20000e 07-09-2013 01:45 AM

ok done editing ;) lol

always
__________________

read
__________________

cynwulf 07-09-2013 06:10 AM

Quote:

Originally Posted by /dev/random (Post 4986176)
X.org (X for short) is not slow! In fact Valve gets faster frame rates under X then in Windows with worse preforming drivers then the windows counter parts. If X were slow this would not be true.

I assume you're referring to the "Steam" software - and thus games?

Nowadays any games which uses mesa/gallium, use DRI, which bypasses the X server. Without DRI (DRI2) mesa/gallium programs would be extremely slow.

T.Boyle 07-09-2013 07:13 AM

X is a dinosaur, time to hit the road jack

/dev/random 07-09-2013 09:52 AM

Quote:

Originally Posted by cynwulf (Post 4986868)
I assume you're referring to the "Steam" software - and thus games?

Nowadays any games which uses mesa/gallium, use DRI, which bypasses the X server. Without DRI (DRI2) mesa/gallium programs would be extremely slow.

I am pretty sure Nvidia still uses GLX and not DRI, but that doesn't matter, because without X these drivers do not work at all, that's why there are no 3D programs that run without X, so it can't totally bypass X, it may not use some of X's draw functions and use its own. but without X your 3D card it worthless, not even OpenCL runs (currently, there are plans to make it work but as of now it still requires X) so with all this, how does it exactly bypass X if X is required?

It doesn't, it simply uses its own proprietary extensions to make it work. That said, with the way the Wayland team wants to do things your going to need a Intel i50 with 70 hecta cores all at 70 Terra Hertz to make this as fast as X in the long run.

Wayland is just a striped down version of X with none if its functionality. All it does it display stuff, if doesn't handle input, it doesn't give you network transparency and it doesn't give you the options X does... they want more abstraction layers and if they get what they want.. I'll be going to BSD because I like to run a modern UNIX based OS on hardware we have today with the same speed of faster that X already provides, I don't want a mess of abstractions layers that won't even run on hardware we have today. Not hardware that hasn't been invented yet. This isn't Windows

H_TeXMeX_H 07-09-2013 10:21 AM

Quote:

Originally Posted by /dev/random (Post 4987000)
Wayland is just a striped down version of X with none if its functionality. All it does it display stuff, if doesn't handle input, it doesn't give you network transparency and it doesn't give you the options X does... they want more abstraction layers and if they get what they want.. I'll be going to BSD because I like to run a modern UNIX based OS on hardware we have today with the same speed of faster that X already provides, I don't want a mess of abstractions layers that won't even run on hardware we have today. Not hardware that hasn't been invented yet. This isn't Windows

I agree. Wayland seems to currently be just a stripped down version of X. It seems Wayland's argument is that they remove X's problems ... yes, but they do this by removing its functionality. Is that the right solution ... I'm thinking not, but I will wait and see.

P.S.
What I really don't like is when someone takes a project, removes the majority of the code because "it has problems", removes the functionality along with it, and then markets the program as something better than the original. See mplayer2, mupen64, libav, etc.

cynwulf 07-09-2013 11:07 AM

Quote:

Originally Posted by /dev/random (Post 4987000)
I am pretty sure Nvidia still uses GLX and not DRI, but that doesn't matter, because without X these drivers do not work at all, that's why there are no 3D programs that run without X, so it can't totally bypass X, it may not use some of X's draw functions and use its own. but without X your 3D card it worthless, not even OpenCL runs (currently, there are plans to make it work but as of now it still requires X) so with all this, how does it exactly bypass X if X is required?

To understand how this works you would need to read up on DRM/DRI.

GLX is separate from DRI. Nvidia use their own (closed source) DRI implementation.

/dev/random 07-09-2013 12:17 PM

Quote:

Originally Posted by cynwulf (Post 4987062)
To understand how this works you would need to read up on DRM/DRI.

GLX is separate from DRI. Nvidia use their own (closed source) DRI implementation.

I get how it works, it doesn't send any data through the X server, but on screen data is not all that X.org does, DRI/DRM still needs a userland module and since it is Xorg that handles the input/ hardware and other parts of the desktop interface, you still need X. Without X how do you tell the program what card to use? (Im assuming you have multi cards here) what screen resolutions are available...etc that is where X comes in. So I still don't see how X is slow... compared to what? Windows? Apple? That's the problem with the Wayland argument it states its slow but doesn't compare it against anything... well if you don't compare it, how do you know its slow? Also how much slower is it? are we also talking about the input and other driver foot prints or just raw rending speed? Because if its just raw rendering speed, why not write an extensions that just uses DRI/DRM to create a compsiting desktop that doesn't use Xorg functions at all...

The Wayland argument seems to have no solid valid base for trying to replace X, accept for 'Wayland is better because I said it is'. That is all i am hearing, there are no solid reasons.

I can make a similar argument about anything...
I mean if we are going to replace things because they are old, lets get rid of Linux.. its monolithic its 23 years old.. must suck... lets rewrite a completely new kernel... but this time lets just make the core.. we'll let some abstraction layer deal with drivers and other useful functions. We will only make it boot.. oh and its faster!


All times are GMT -5. The time now is 02:00 AM.