Wayland Begins Porting Process To FreeBSD?

http://www.phoronix.com/scan.php?page=news_item&px=MTMwMzE

Wayland Begins Porting Process To FreeBSD
Posted by Michael Larabel on February 16, 2013

While upstream Wayland developers are just concerning themselves with Linux support and are exclusively focusing their development efforts on taking advantage of the best Linux APIs and functionality, a developer has begun proposing patches that begin to work on FreeBSD support.

A set of 12 patches were published on Friday to start off Wayland support for FreeBSD. The patches begin here on the Wayland mailing list.

The changes come down to mostly adding in support for various functions on FreeBSD to replace their Linux equivalents used by Wayland. However, even if Wayland and Weston end up working on FreeBSD, the support won't be too widespread considering the only kernel mode-setting driver ported to Wayland in the mainline kernel is the Intel KMS driver as of FreeBSD 9.1.

We'll see how long the FreeBSD Wayland work goes on, but chances are it won't see much attention or gain much ground on the FreeBSD desktop.
 
Does it really need KMS? Could it be patched so that it doesn't?

Second what really is KMS? I really haven't looked into the design since I don't run any Linux boxes personally except virtually. Isn't it just the kernel keeping track of the graphics state? Couldn't a generic wrapper be created to apply to most existing video drivers? I sure it's not that simple so I'd appreciate someone elaborating a little.
 
KMS is a driver that provides an abstraction of the concept of mode switch. KMS really belongs in the kernel because none of the userland programs, even the X server itself should know or depend on the details of the hardware. It's one of the first steps to provide proper virtualization of the graphics hardware.
 
I would be interested to know if any progress has been made on this.

Looking at the below it looks like some things have been pulled together.
https://wiki.freebsd.org/Graphics

On KMS, I found a few posts that suggest this is not dependent on KMS which I believe is the case. It can fall back to none GPU rendering with fbdev, pixman, swrast?
https://tecnocode.co.uk/2013/02/18/west ... r-backend/
http://www.phoronix.com/scan.php?page=n ... px=MTUwMzM
http://www.phoronix.com/scan.php?page=n ... px=MTY2MjQ

Either way it sounds possible. I think this would be a very cool feature to have. Also having recently discovered xrdp and X11rdp, I think it would be a great enabler for running GUI software remotely. A FreeBSD terminal server without X. I think this would be greatly used by people running servers in the Cloud. No more Xforwarding or insecure VNC sessions.
http://www.phoronix.com/scan.php?page=n ... px=MTMxNjE
 
Petz said:
I think it would be a great enabler for running GUI software remotely.
X already does this. Wayland does not.

A FreeBSD terminal server without X.
You don't need X to run a terminal.

With all the (typical) fighting and friction and non-Unixness surrounding Wayland, I don't think we want or need that over here. Linux is no longer a Unix-like system. Linux is now ... Linux.
 
Wayland is not systemd. Wayland is pretty much a rewrite of X, given all that has been learned in the meantime. There is a lot of potential.

http://www.youtube.com/watch?v=RIctzAQOe44 has a talk by Daniel Stone, one of the developers. The first ten minutes or so, he talks about his background, but after that it gets into the detail.
 
drhowarddrfine said:
With all the (typical) fighting and friction and non-Unixness surrounding Wayland, I don't think we want or need that over here. Linux is no longer a Unix-like system. Linux is now ... Linux.

Quoted for truth.
 
I have been always open towards Wayland. I've been using the X11 APIs in my small programs and the APIs feel like they have been designed in the 80s (chaotic, unstructured). Anything that comes can be potentially better than X(org), in my opinion.

I wish FreeBSD would do more to implement and fix KMS, if this is really the future. FreeBSD is a good desktop system and it should support graphics very well. If the kernel offers enough, the developers will support the it in their applications. FreeBSD needs to be one step ahead.
 
wblock@ said:
Wayland is not systemd.

Unfortunately, it's gotten to the point where I don't trust Linux developers working on new things for *nix any more than I trust Windows developers working on *nix. (I already saw the video.)
 
nakal said:
I've been using the X11 APIs in my small programs and the APIs feel like they have been designed in the 80s (chaotic, unstructured).
That's probably because it actually was designed in the 80s ;)
 
Apparently, work is underway to support Wayland. According to the wiki there is a port of Wayland, but it is untested because there is nothing yet that can use it.
It seems that the developers are focusing on resolving the dependency on udev first.

https://wiki.freebsd.org/Graphics

I haven't been able to find any other information on the state of Wayland/Weston.

I do hope we can get Wayland to work on FreeBSD because otherwise we're stuck with X (unless Apple decides to open-source their graphics stack, something I find highly unlikely).
 
If you want to open up the possibility of an Xorg alternative designed for *nix-like systems, KMS and GL will go a long way. Right now, the absolute best option for graphics acceleration are nVidia binary drivers. And those are heavily bound to X. You pretty much need them to support this.

Give developers the ability to launch a program on a terminal that switches to the native panel resolution, give them an OpenGL context to render with, and you can start to see people actually trying to make something interesting. If they then succeeded in creating GTK+ and Qt backends, things might even start to become remotely viable.

Yet as it stands, the effort to create a new display server (that isn't 1024x768 VESA) is about as daunting as creating a new operating system, and that is completely ridiculous.
 
Can anyone explain or provide a like explaining what Wayland is/does and what X does? Especially from an API/flowchart perspective. I'm sure it's more complicated than what I think. How is Wayland any different than console SDL? With graphics being moved to the kernel with KMS It seems to me Wayland would be little more than just a graphics and input sharing/cooperative API. If that's the case then it's simple and should be functional already. A little harder is rewriting all the toolkits but it's not quite that hard. Then implement an xserver display driver for the new API. Why is it harder than this?
 
Back
Top