Why I am Leaving Linux

I have used Linux and maintained Linux systems for decades. However, my experience is that there has been an increasing tendency for things to break on Linux and not get fixed. I have been using the Arch Linux distribution, which is bleeding edge, but my experience is that problems that show up on the bleeding edge eventually work their way into other distributions. At the same time, Linux has become increasingly hard to maintain due to increased complexity and constent churning of the software. The introduction of systemd is one example of this. I speculate that these problems are at least partially driven by the increasing corporate takeover of Linux, which has some benefits, but also seems to have a serious downside. The developers of Linux are no longer the users, so things important to users don’t get fixed.

Here are some examples of things that have gone wrong that either haven’t been fixed or have taken an inordinately long time to fix:

• I frequently use the openbox window manager, which associates mouse buttons with particular actions. However, certain other applications such as various file managers gratuitously step on this arrangement. This problem appeared a few years ago but has never been fixed.
• Several years ago, /dev disk identifiers no longer could be trusted to be associated with a particular disk from boot to boot. The problem was fixed by the introduction of the UUID identification mechanism. This arrangement was awkward, but seemed to work. However, recently I have found situations where it does not. In particular, adding a new disk should not change the reference to an existing boot disk. This promise was suddenly broken (at least with syslinux), with the result that the system would no longer boot when a new disk was added. The use of disk labels, an alternative to UUIDs was advertised, but never seemed to work.
• Around June 2019, the new 5.2 kernel had a bug that prevented the mounting of USB disks. A patch for this problem was quickly developed, but was only scheduled to appear in the kernel in late November of 2019 in 5.4. Really???

That was the last straw. I have moved to OpenBSD. So far the experience has been mostly positive. It installs rapidly and seems to have fewer problems with UEFI systems. The package management system is as good as that of Arch Linux (a very good system) and is better in some ways. There is a broad selection of packages, and though the move from Linux to BSD causes problems with some packages, there is a very active and helpful user community with OpenBSD. OpenBSD has not lost the spirit of minimal but elegant solutions to problems characteristic of the early days of UNIX, in contrast to the increasingly baroque character of Linux. With only minor exception, I can do what I need to do on OpenBSD. I have no reason to turn back. (This was written using LaTeX on a Lenovo X1 Carbon laptop running OpenBSD.)

This document was translated from L^AT_EX by H^EV^EA.