I was looking into Hyperbola Linux, when I came across this post. In this post it is outlined why Hyperbola Linux is switching to a BSD base. A bold move, but I hope it works out for the team.
However, I am curious about the four points they outlined as reasons for switching away from the Linux kernel. Is the Linux kernel really careening down a path of instability and binary blobs, or is this just a case of the Linux kernel’s development not fitting with Hyperbola’s design goals?
People have been predicting this literally since it was released and no major issues have cropped up that I have ever heard of, except for technical nitpicks that don’t really amount to anything (ie, the hardcoded DNS issue in the optional DNS resolver). I’m still waiting to hear about a single attack that actually exploited systemd as a vector. Some bad bugs have cropped up over the years, yes. Bad bugs have cropped up in all system-critical software over the years. And usually when people point at a longstanding “bug” and go “z0mg they don’t care about security” it’s usually because they’re overlooking valid considerations about real-world use and are literally just nitpicking because it’s not abstractly perfect.
And their rationale for example disliking having a hardcoded DNS server is that that’s only valuable for embedded. So what? Now init software should only serve unixsheikh’s use cases because… no reason?
That’s because there’s nothing wrong with it and it’s the best init system and that’s why everybody uses it. Everything else is just conspiracy theorizing to be frank. If people want to displace systemd they need to write something better. Not something that has one or two features like parallel starting, like actually do everything systemd does and do it better.
A lot of the supposed “security” issues are really just nonsense. For example they cite a bug with how systemd handles invalid usernames leading to root escalation. They ignore that in order to exploit this bug you already need root privileges.
They conflate fragmentation with freedom. A fragmented ecosystem isn’t necessarily free (UNIX was non-free and horribly fragmented for a long time), and a more cohesive ecosystem isn’t necessarily unfree (almost everybody was standardized on SysVInit for a long time and there was no wailing about the death of freedom). Interoperability is valuable and not a bad thing. Standardizing certain aspects of Linux systems improves interoperability between systems.
If you didn’t admin Linux systems before the switchover I can imagine not getting how fucking direly something like systemd was needed however.