Linux distributions
The fact that software under open source licenses is free to modify and distribute made it possible to assemble complete operating systems with kernels, system libraries, and utilities, as well as a selection of application software. Since open source licenses have no restrictions on usage, there is no need to make the user accept a license agreement for each component.
In the early days of Linux, setting up a usable Linux environment was a complicated and tedious endeavor. To make that process simpler, Linux enthusiasts started preparing the first distributions – sets of packages and scripts to automate their installation. Many of those early distributions, such as Softlanding Linux System and Yggdrasil, are now defunct, but some are still maintained – Slackware Linux is a prominent example.
Package managers and package repositories
Early distributions had a relatively humble goal, which was to provide users with a working barebones system that they could then install their application software on. However, later distributions set out to rethink the process of software installations. The number of open source software projects was growing, and CD drives and internet connections were also becoming more affordable, so it was feasible to include much more software in a distribution than ever before.
However, many applications depend on shared libraries or other applications. Traditionally, installation packages would either include all dependencies or leave that dependency management to the user. Since distribution is managed by a single group of maintainers, developers came up with the idea of sharing dependencies between all packages that need them and automatically installing all dependencies when a user requested the installation of a package. That gave rise to package managers and package repositories – collections of files in a special format, including compiled binaries and metadata such as the package version and its dependencies.
The two most popular package formats and package managers that work with them were developed in the mid-90s and are still in use today. One is the DEB format, which is used with the dpkg utility, developed by Debian. The other is Red Hat Package Manager (RPM), which is used with the rpm utility and is developed by Red Hat.
The dpkg and rpm tools are responsible for installing package files on local machines. To install a package, the user needs to retrieve the package itself and all packages it depends on. To automate that process, distributions developed high-level package managers that can automatically download packages from online repositories, check for updates, search metadata, and more. Those high-level package managers usually rely on low-level ones to manage the installation. Debian’s Advanced Packaging Tool (APT) usually works with DEB packages, although it’s technically possible to use it with RPM. High-level package managers that primarily use the RPM format are more numerous: YUM and DNF, which are maintained by Red Hat, zypper from openSUSE, and urpmi, which is developed for the now-defunct Mandrake Linux and still used by its forks.
Many of the currently existing distributions have either been actively maintained since the 90s or are forks that split off at different points in time. For example, Ubuntu Linux was forked from Debian GNU/Linux in the early 2000s, while Rocky Linux is a Red Hat Enterprise Linux derivative that started in 2021.
However, completely independent distributions also appear once in a while. Some of them are special-purpose systems that have requirements that classic general-purpose distributions cannot fulfill. For example, OpenWrt is a Linux-based system for consumer routers, originally developed for the Linksys WRT-54G device, hence the name. Such devices often have just a few megabytes of flash drive space, so operating systems for them have to be very compact, and they also have to use special filesystems such as JFFS that are designed for NAND flash drives.
Other independent distributions experiment with different package management and installation principles. For example, NixOS and GNU Guix use an approach that allows the user to revert system updates if anything goes wrong with new package versions.
In this book, we will focus on Debian/Ubuntu and Red Hat-based systems because they have been the most popular distributions for a long time and remain popular.
Differences between distributions
The differences between distributions do not stop at package managers. Configuration file locations may differ, and default configurations for the same packages may also differ dramatically. For example, the configuration file directory for the Apache HTTP server is /etc/httpd on Red Hat Linux derivatives, but /etc/apache2 on Debian derivatives.
Some distributions also use high-level configuration tools and you may take them into account.
The choice of software and its ease of installation may also differ. Debian, Fedora, and many other distributions leave the choice of a desktop environment to the user and make it easy to install multiple different desktop environments on the same system so that you can switch between GNOME3, KDE, MATE, or anything else for different login sessions. In contrast, the Ubuntu family of distributions includes multiple flavors for different desktop environments and expects that if you don’t like its default choice (the Unity desktop environment), you should use Kubuntu for KDE, for example, rather than the default Ubuntu. Finally, some distributions come with a custom desktop environment and don’t support anything else, such as elementary OS.
However, experienced Linux users can usually find their way around any distribution.