Day 1 - The Structure and Evolution of the X Window System

📌 Key Takeaways

• The X Window System is the system that handles the graphical environment of Unix and Linux. It began at MIT in 1984.
• Server and client are reversed. The side with the screen, keyboard, and mouse is the X server, and the program is the X client.
• Because it is network-based, you can display on your own computer the screen of a program running on another computer.
• The evolution order of X server implementations is XFree86 → X.org Server → Wayland.
• The X Window System does not enforce a look or a way of operating. That is why a separate window manager is needed.

What Is the X Window System

• The official name is the X Window System. It is not written in the plural as "X Windows".
• It is not part of the operating system. It is an application separate from the Linux kernel, so the system works without it. Servers are often run without X.

Unpacking the Features One by One

A question that asks you to pick "the incorrect feature of the X Window System" flips one of the lines below into a choice.

• Even if X dies, the kernel stays alive. Being able to switch to a text console and recover when the graphical screen freezes is thanks to its separation from the kernel.
• You can see what your current session runs on with echo $XDG_SESSION_TYPE. It prints x11 or wayland.

Server and Client Are Reversed

This is an especially confusing part of the X Window System, so it makes a good exam question.

   [ The computer in front of me ]          [ A distant server ]
 
     X server                                 X client
     screen · keyboard · mouse                the actual program
     (the side with the hardware)             (xterm, gedit ...)
        │                                        │
        │◄──── request to draw on the screen ────┤
        ├───── passes keyboard/mouse input ─────►│

It makes sense if you think "the side that provides the resource (the screen) is the server". On the web, the side that holds the data is the server, but in X it is the side that holds the screen that is the server. So the computer in front of me becomes the X server.

• Thanks to this structure, the window of a program running on a remote computer can appear on my screen. This is called network transparency.
• Even when the program and the screen are on the same computer, it works in exactly the same way.

The Four Things That Pass Between the Two

   X client (xterm)                              X server (my screen)
        │                                             │
        ├── ① Request   "draw a window here" ────────►│
        │◄─ ② Reply     "that window's size is this" ─┤
        │◄─ ③ Event     "the A key was pressed" ──────┤
        │◄─ ④ Error     "there is no such window" ────┤

• Only requests are sent by the client, and the other three are sent by the server. The keyboard and mouse are attached on the server side, so input always flows from the server to the client.
• A client cannot touch the screen directly. Even plotting a single dot must be asked of the server. That is why windows from many programs coexist on one screen without overwriting each other.

Telling Who Is the Server

• There is one tip for telling — "the side with the monitor where the window is seen" is the X server. It is not the side with better performance, nor the side where the program runs, nor the machine called "the server".
• So it is common for the "server computer" to be the X client (the side running it) and "my PC" to be the X server. This reversal is the material for questions.
• Multiple clients attach to one X server at the same time. If ten windows are open, ten clients are connected.

Evolution

   1984  X Window System        started at MIT
     │
   1987  X11                    the version number still in use today
     │
   1992  XFree86                free implementation for PC Unix. The standard for a long time
     │                          split apart by a license change in 2004
     ▼
   2004  X.org Server           split off from XFree86. The current standard
     │
     ▼
   2008~ Wayland                the next-generation approach meant to replace X

Questions ask the evolution order XFree86 → X.org Server → Wayland as is. Memorize the order of the three names.

• X11 is the 11th version, released in 1987, and what is used today is still X11. That is why the notation X11 remains in many places.
• Wayland is not a new version of X but a replacement. The structure itself is different, and existing X programs run through a compatibility layer called XWayland.

Sorting the Kinds of Names

The names that appear mixed in the choices are not the same kind. First sort out what each name is a name of.

• XFree86 and X.org Server are both "programs that implement X11". It is not that X11 ended and X.org arrived, but that the program implementing the same X11 protocol changed.
• The trigger for moving to X.org Server was XFree86's license change in 2004. The new terms were incompatible with the GPL, so the major distributions moved to the X.org side.
• The body managing X's standard has changed too — starting from MIT's X Consortium, it is now handled by the X.Org Foundation.

X and Wayland — What Differs

   X11                                      Wayland
   application (X client)                   application (Wayland client)
        │                                        │
     X server  ◄──►  window manager · compositor    compositor  (display server + window management)
        │                                        │
      kernel · graphics card                    kernel · graphics card

• In X, drawing a window goes a full circle: client → X server → compositor → X server → screen. Wayland strips out the X server in the middle so that the client and compositor talk directly. That is what "stripped away the complicated structure" means.
• In Wayland, the distinction between display server and window manager disappears. The compositor does both.
• GNOME on major distributions such as Fedora and Ubuntu uses a Wayland session by default, but X.org is still used alongside it. The "current standard X server" on the exam is X.org Server.

It Does Not Enforce Policy

The X Window System provides only "how to draw" and does not decide "how it will look".

   X Window System    provides only the basic functions for drawing windows
      │
      ├── window borders · minimize button · moving?   →  handled by the window manager
      ├── login screen?                                →  handled by the display manager
      └── panel · file manager · integrated environment? →  handled by the desktop environment

• That is why Linux has several environments with different window looks. This is a big difference from Windows and macOS.
• The distinction among these three (window manager · display manager · desktop environment) is covered in detail three days from now.

Related Terms

Tomorrow we cover the layers of libraries and toolkits used to build X clients.

📖 Glossary

• X Window System: The system that provides the graphical environment of Unix and Linux. Started at MIT in 1984.
• X server: The side with the screen, keyboard, and mouse. The computer the user sits at falls under it.
• X client: An application that requests that something be drawn on the screen.
• Network transparency: The property that the screen of a program run remotely can be displayed exactly like a local one.
• XFree86: The project that was the standard X implementation for PCs for a long time.
• X.org Server: The current standard X server implementation that split off from XFree86.
• Wayland: A next-generation display server protocol meant to replace X.
• X11: The 11th version of the X Window System, released in 1987. It is the version still in use today.