The X Window System has provided a framework for desktop graphics in Unix and Linux for more than 30 years. X has gone through many phases since it first appeared in 1984, but it stabilized long ago. Version 11 of the X protocol (referred to as X11) has been around since 1987, and it is a fundamental part of the Linux landscape; however, computers have changed a lot since 1987, and many experts believe the X Window System needs to be replaced.

The Wayland display server protocol, developed since 2008 under the direction of Intel employee Kristian Høgsberg, is primed to take over for X11. Wayland could eventually solve a load of problems that developers have to contend with when integrating Linux applications with a graphic desktop, but is Wayland equal to the task? This article looks at the state of Wayland integration.

The Turning Point

X11 is a patchwork of code difficult to maintain and almost impossible to expand. Serious mistakes often emerge that have remained hidden in the code for years. For example, a security vulnerability in the font server dating from 1991 wasn't discovered until 2014 [1]. Rigorously practiced backward compatibility is also part of the reason X11 is not considered secure. Several of the core components carried over from the early days are no longer of any real use, but they have to be available by default.

The focus at the beginning of development was also completely different from what is expected of a modern graphics stack today. Modern graphics libraries now draw circles and rectangles or move windows, without having to resort to an X Server. The modern environment also supports shared libraries, which means graphics-capable applications don't have to carry around a ballast of graphics functions. Modern clients just expect the display server to allocate an area where they can write and display the content, which is what Wayland provides.

Detours and Wrong Turns

The person perhaps most responsible for Wayland's position as an heir to X11 now no longer supports the Wayland protocol. In 2010, Mark Shuttleworth announced that Ubuntu would run with Wayland instead of X11 from version 12.04 onward [2]. This announcement gave the Wayland project, which originally was meant to prove only that X11 could be rebuilt without too much effort, strong standing as a possible heir apparent for X11. Shuttleworth later decided to develop his own, new Mir display server rather than support Wayland.

Most Linux distributions will eventually switch to Wayland, although X11 will be around for a few more years. The Wayland developers created XWayland, a slightly modified X server that serves as a compatibility layer, to eliminate problems during the transition phase for applications that still require X11. The first major distribution that natively uses Wayland by default will probably be Fedora, which was considering making Wayland the default system in Fedora 24, although a recent blog post by developer Matthias Clasen [3] said, despite all the work put in toward that goal, Wayland was not quite ready for production distros.

Wayland/X11 Differences

Wayland differs from X11 conceptually and functionally. In the Wayland environment, the compositor is the display server. In X11, on the other hand, the compositor is an external component that requires an additional processing step. Wayland is just the protocol, and Weston is the reference implementation of a compositor (Figure 1). The differences between Wayland, X11, and XWayland become clearer if you look at how the display servers process an event such as a mouse click (Figures 2-4).

Figure 1: The display server between the kernel and hardware and the applications [4]. (CC BY-SA 3.0 [5])

Figure 2: The X server processing an event from an input device [6].

Figure 3: Wayland processing an event from an input device [6].

Figure 4: XWayland processing an event from an input device [7]. XWayland runs as a modified X Server in Wayland.