Wayland and Weston 1.0

Tue Oct 23 00:21:29 PDT 2012

On Mon, 22 Oct 2012 16:46:52 -0700
Jesse Barnes <jbarnes at virtuousgeek.org> wrote:

> On Mon, 22 Oct 2012 19:26:46 -0400
> Kristian Høgsberg <hoegsberg at gmail.com> wrote:
> 
> > Wayland and Weston 1.0.0 have been released!  
> > 
> > Thanks and congrats to everybody involved for making this happen.
> > We're entering a new, exciting and somewhat scary phase for Wayland.
> > As of this 1.0.0 release, we're changing the development model and
> > committing to the the protocol and client side API we have now.
> 
> Woohoo!  I'm really excited by this.  My hope is that with a "stable"
> release number out there, people will really start evaluating Wayland
> and Weston for their needs, and hopefully see the same sorts of big
> power and memory footprint savings I've seen when compared with other
> solutions.
> 
> I really hope one of the desktop environments jumps on board now.  With
> a modern toolkit the porting effort shouldn't be huge (mostly in the
> window management component), and the power savings on a typical laptop
> have the potential to be huge.  Given the existing toolkit support, it's
> also possible someone could start a simple, new environment with Weston
> at its core.

Indeed!

Though, I think it is worth mentioning, that Wayland 1.0 is about the
core protocol. Desktop environments are going to need a lot more, for
example:
- clients want to restack their surfaces
- server to ask a client to minimize/maximize/fullscreen or go
  normally windowed
- client wants to minimize itself
- server wants a client to not draw window decorations, shadows, title
  bar, on all or some sides (would be especially useful for tiling
  window managers, I guess)
- to be able to recall window position

I'm not aware of any of the above available through the wl_shell
protocol yet, so if you want to really implement a real DE, you will be
facing some protocol designing, too. Currently the wl_shell protocol is
just enough to be able to work with windows in Weston.

I believe this is mostly because there has not been time to concentrate
on desktop shell protocols yet, just get something that one can test the
core with, but also to avoid dictating a single desktop protocol design
without getting some real experimentation and discussion. Maturing a
generic desktop protocol will certainly take time, and more people than
there were designing the core protocol.

For anyone wanting to port or write their own window manager to Wayland:

Most likely you have a desktop window manager. A quick way to get
started, is to fork Weston's desktop-shell plugin and start hacking it.
Qt could be another good choice, but I am not familiar with it.

You also need to understand some concepts. I'm repeating things I wrote
to the wayland-devel list earlier [1], a little rephrased.

We need to distinguish three different things here (towards Wayland
clients):

- compositors (servers)
	All Wayland compositors are indistinguishable by definition,
	since they are Wayland compositors. They only differ in the
	global interfaces they advertise, and for general purpose
	compositors, we should aim to support the same minimum set of
	globals everywhere. For instance, all desktop compositors
	should implement wl_shell. In X, this component corresponds to
	the X server with a built-in compositing manager.

- shells
	This is a new concept compared to an X stack. A shell defines
	how a user and applications interact. The most familiar is a
	desktop (environment). If KDE, Gnome, and XFCE are desktop
	environments, they all fall under the *same* shell: the desktop
	shell. You can have applications in windows, several visible at
	the same time, you have keyboards and mice, etc.

	An example of something that is not a desktop shell
	could be a TV user interface. TV is profoundly different:
	usually no mouse, no keyboard, but you have a remote control
	with some buttons. Freely floating windows probably do not make
	sense. You may have picture-in-picture, but usually not several
	applications showing at once. Most importantly, trying to run
	desktop applications here does not work due to the
	incompatible application and user interface paradigms.

	On protocol level, a shell is the public shell interface(s),
	currently for desktop it is the wl_shell.

- "window managers"
	The X Window Managers correspond to different wl_shell
	implementations, not different shells, since they pratically
	all deal with a desktop environment. You also want all desktop
	applications to work with all window managers, so you need to
	implement wl_shell anyway.

I understand there could be special purpose X Window Managers, that
would better correspond to their own shells. These window managers
might not implement e.g. EWMH by the spec.

When you implement your own window manager, you want to keep the public
desktop shell interface (wl_shell). You can offer new public
interfaces, too, but keep in mind, that someone needs to make
applications use them.

In Weston, a shell implementation has two parts: a weston plugin, and a
special client. For desktop shell (wl_shell) these are src/shell.c and
clients/desktop-shell.c. The is also a private protocol extension that
these two can explicitly communicate with.

The plugin does window management, and the client does most of user
interaction: draw backgrounds, panels, buttons, lock screen dialog,
basically everything that is on screen.

Phew... this came out a lot longer than I intended, and I really need
to stop for now. This is my view of the Wayland world, hoping to
clarify concepts, and raise discussion where (if?) I got it wrong. :-)

Thanks,
pq

[1] http://lists.freedesktop.org/archives/wayland-devel/2012-August/004796.html