Announcement

**rohcQaH** · 12 September 2009, 06:32 PM

pabloski: Gallium3D is a framework for graphic drivers. It's only purpose it to draw stuff (no matter if it's lines, textures, triangles, video or whatever).
Xorg has other components, like input devices, the X protocol, window management (with the help of a WM) and a lot of glue in between.

G3D is important as it allows many features the classic mesa architecture didn't permit, but as bridgman said, it's not meant to replace anything but the gfx drivers.

**elanthis** · 12 September 2009, 06:57 PM

There's a lot of confusion over what a gallium state tracker is. Hardware works by setting various renderi g states (eg color, shaders, matrices, whatever) and commands (eg render the last three vertices as a triangle using the current state). Different graphics Apis work using a different set of states and commands. Ms of OpenGL is still built around thefixed function pipeline function pipeline for example. X and some other protocols don't use state at all similarly to the hardware - some don't use state at all but instead pass all rendering parameters in the command itself. The basic purpose of a gallium state tracker is to convert API state into hardware state, which includes tracking both current hardware state (so you don't tell the hardware to load a shader it's already using for example) as well as API state. The X state tracker is the code to map the X protocol rendering calls to hardware equivalents. You still need the display server for applications to connec to, to interpret X requests into the gallium state tracker calls, to manage window placement and stacking, to cache font glyphs, to set indirect rendering targets for compositing, to detect monitors and control multihead layout, process input events and direct them to the proper applications.

**MuPuF** · 12 September 2009, 07:06 PM

Here is what I thought I read:

Xorg is old and doesn't really fit the need of modern desktops (ie compositing). At the moment, just a little part of the X protocol is actually used.
It is also a real pain to make a tear-less experience for videos and with compiz.
At the moment, Xorg is not really suited for embedded Linux, the memory footprint is really important and the loading time is really long.

So, when you hear about a minimal window server that would be light and would fit the modern desktop needs, you can expect a real lot of improvement

As a wonderful improvement, we can say Wayland is represents 5000 lines while X-server is at least 100 times bigger. Wayland is much more maintainable and easy to understand

There would be no copy of graphic buffers, buffers would be passed from the application to Wayland just by giving away the GEM pixmap handle.
Also, as it is really minimal, we can expect a fast boot time

At the moment, Wayland supports applications coded using clutter. The GTK support is on the way (client-side rendering branch has been merged to the new GTK). There is no news for the Qt support.
So, at the moment, you would be able to use Wayland to prompt users for their passwords and then launch a root-less Xorg with the user's right and, when the graphic toolkit would be well supported, we may see gnome and KDE running over Wayland

I remember someone writing the first feeling he had while testing Wayland was the smoothness when moving the windows and the mouse.

So, here are the advantages of Wayland (according to me):
- Lightness (fits embedded Linux needs)
- Simpler to maintain (Will be released on time ?)
- Faster and tear-less
- Made for modern desktops needs

**drag** · 12 September 2009, 07:11 PM

Originally posted by ioannis View Post

I'm sure the Phoronix guys had good intentions, but what we see here is a classic example of the detrimental effects of forking applied to a discussion.

There are some constructive comments in here that it would be good for the person that initiated this discussion, along with others that participated in the original thread, to read. Instead we ended up with 2 threads.

I can easily see the benefits of things like Gallium3D and talk about it, even if I have only basic knowledge on it's inner workings. However, even though I do understand the desire for a 'lighter' X alternative proposed with Wayland, I'd need to know a lot more about Wayland to have a meaningful opinion about it.

I find this comment by Lowe interesting:
"X is technically Free Software, but it is not a community project. "

The lack of developers in Xorg is a direct result of its complexity, becoming a barrier for anyone attempting to 'enter' it. However, as people said before me, X is getting simplified by stripping away parts that better fit in other places, such as the kernel, leaving a more manageable code base.

I like fresh ideas, who doesn't, but I'm always afraid of the detrimental effects of forking/diversifying.

There is a confusion of issues going on here.

There are issues with people thinking that X is inefficient and huge. And there is issues with stability and driver releases.

There are multiple issues people are having with X.org and Wayland DOES NOT SOLVE ANY OF IT.

Get it?

People want better drivers and lighter desktop. Wayland is not the answer they are looking for.

Gallium3D and moving X off the hardware are the very things that will improve X and seperate drivers from your X Server.

Gallium3D and moving X off the hardware is what is required to get Wayland to work.

Get it?

In order to get Wayland working as a replacement for X the driver model for video Linux needs to be changed.

This video model change will fix most of the issues with X!

Let me state it again:

THE WORK NEEDED TO GET WAYLAND WORKING IS HTE SAME EXACT WORK IT TAKES TO FIX PEOPLE'S PROBLEMS WITH X.ORG.

Kernel mode setting, Gallium state trackers, etc etc. These things are prerequisite for Wayland. These are the things that is needed to fix people's issues with X.org. That is why there is no major work undergoing for Wayland, right now. THERE IS NO FREAKING POINT TO CONTINUE IT AT THIS TIME. Until they fix the GPU driver model for Linux then a usable wayland solution is not going to be forthcoming.

Gallium3D and state trackers means that X is no longer your source for drivers. It is just another graphics application like any other and thus you will not be dependent for X for your future driver upgrades. As long as the Gallium 2D state trackers are compatible with your X Server then you can upgrade everything all the time.

And it is true that X is full of a bunch of legacy crap. But this legacy crap is good because it allows backwards compatibility. EVEN IF YOU SWITCH TO WAYLAND YOU WILL STILL NEED X.ORG TO RUN MOST OF YOUR APPLICATIONS.

Modern applications use X extensions which are about as modern as anything else.

Do not think that Wayland is a magic bullet. There remains a massive amount of work to do with updating the Linux video driver model before people can even consider using it at this time. And you will still need Xorg as a off-screen X server to run the majority of your applications. And the work that needs to go into Linux video driver stack will vastly improve and solve most of the problems people have with X.

X running hardware drivers is BAD. X running on top of Gallium as just another graphical application is GOOD and is not much different then what you'd get with Wayland anyways.

**BlackStar** · 12 September 2009, 07:15 PM

Originally posted by pabloski View Post

no one knows what google is cooking, but they have a graphics stack for android, so it is reasonable to use the same for chromeos

afterall they can coerce nvidia and the others into making drivers for their graphical stack

The thing is, they've said that they consider ChromeOS a completely distinct project from Android. They target completely different hardware (phones vs netbooks) and it's probably much easier to build ChromeOS on the X stack than not.

Time will tell, I guess. :-)

**MuPuF** · 12 September 2009, 07:24 PM