Who's Afraid of Systemd?

Wednesday Jun 17th 2015 by Bruce Byfield

Can you believe that some Linux users are still opposing systemd?

Last year, the free software community was full of debates about systemd, the system manager that replaces init, the process that boots a Linux system. Now that systemd is uneventfully running the latest releases of major distributions like Debian, Fedora, and Ubuntu, you might imagine that opposition to it is melting away -- but you'd be wrong.

Instead, casual references on social media show that the rumors are as common as ever. And while you don't hear much recently about Devuan, the anti-systemd fork of Debian, it is still trudging towards a release while making the same arguments as ever.

The situation is not unique. Some free software circles have always seemed to require an enemy. For instance, in the first decade of the millennium, it was Mono, an adaptation for Linux of Microsoft's .Net. Hundreds of thousands of words were written denouncing Mono, yet today it attracts no attention, although it is still available in repositories.

Perhaps, too, free software users are becoming conservative as they age, as indicated by the user revolts against GNOME and KDE. Yet no precedent comes close to the viciousness of attacks on systemd, or had so little foundation, either.

Running Out of Arguments

Part of the opposition to systemd is a reaction to Lennart Poettering, its founder. Before systemd, Poettering was best known for PulseAudio, a sound system that caused immense problems for many users before it stabilized. In addition, Poettering lacks diplomacy, and sometimes appears arrogant. As a result, more attention has sometimes been paid to Poettering than to his code.

However, more relevant opposition is equally misguided. For example, in describing itself as free from bloat, Devuan implies that systemd is bloated. Yet it provides no figures to prove this implication, and, for all anyone knows, systemd might actually be more compact that the parts of the Linux system that it replaces.

Another common argument is that Red Hat, Poettering's employer, is using systemd to seize control of the major distributions. This argument seems to have begun during systemd testing, when the focus was on GNOME, Red Hat's desktop environment of choice. But, how it would work in practice is obscure, considering that major distributions offer multiple desktop environments.

Anyway, systemd is licensed under the second version of the Lesser GNU General License, so any Red Hat takeover attempt could be easily thwarted by forking systemd.

Other arguments against systemd are no more supportable. Devuan's home page asks, "Have you tried to opt-out of the systemd change in Debian and stay with sysvinit, or whatever other init you prefer? You will quickly notice that is not a matter of choosing packages and in fact Debian offers no choice."

Yet a search quickly unearths instructions for making an install image without systemd and for removing systemd from your system. True, neither set of instructions is as quick or as easy as installing a package, but anyone with enough experience to care about what Devuan calls "Init Freedom" should be able to follow them without much difficulty.

Nor does the claim that systemd violates the Unix design principle of having one program that has one function stand up under scrutiny. For one thing, a design principle is a preference, not an unalterable rule. Many desktop applications, including LibreOffice and Firefox, violated this principle long before systemd.

More importantly, systemd is actually a general name for a series of related, similarly structured commands. From this perspective, systemd conforms the principle of one program doing a single function in much the same way as the Linux kernel or a command line shell does. It is a suite of programs, not a single monolithic one.

Some technical arguments about systemd could be made. Linus Torvalds, for example, queried why systemd logs could not be in plain text. Yet, for the most part, these are not the arguments that are being made.

In fact, not only are the most common anti-systemd arguments easily discounted, but they are surrounded by a vagueness that raises suspicions. Although Devuan's initial announcement claimed the support of several Debian maintainers, only Roger Leigh, who maintains systemvinit and other relevant packages, has actually revealed his name. Similarly, Devuan argues that the adoption of systemd will have "serious consequences" and claims to have "many letters from concerned professionals" yet none of these dire phrases has ever been substantiated.

The result is an air of secrecy and danger that, however appealing and reminiscent of freedom-fighting that it might be to Devuan's supporters, does nothing to justify the anti-systemd rhetoric or make it plausible. If LibreOffice could fork from Oracle's OpenOffice.org without any secrecy, you have to wonder why Devuan cannot be equally forthcoming.

A Losing Battle

Devuan may eventually produce a release, or become an option for a few distributions like Puppy Linux. However, without any arguments in its favor, it is likely to become even more of a niche distribution than TDE, the continuation of the third KDE release series. Its mailing list mostly shows the same dozen or so posters, and in the last ten months it has raised only 7934 Euros, plus 2.7 Bitcoins, all of which suggests limited support.

Devuan supporters are free to do as they like, of course. That is what free software is all about.

Possibly, too, now that systemd is in general deployment, it may reveal problems that will make an alternative welcome.

For now, though, Devuan supporters sound as though they are doing more fearmongering than raising constructive doubts. Systemd may not be ideal, but most users are unlikely to make a fuss so long as their systems continue to boot and function the way they are supposed to.

Mobile Site | Full Site
Copyright 2017 © QuinStreet Inc. All Rights Reserved