- cross-posted to:
- linuxphones@lemmy.ml
- pine64@lemmy.ml
- cross-posted to:
- linuxphones@lemmy.ml
- pine64@lemmy.ml
The Movuan project was started by community member lxb and announced in a forum post as an alternative to mobile distributions using the systemd init system. Thanks to being forked from Mobian, the project makes use of modified Mobian debos to build it’s images.
Movuan offers an image based on Devuan 5.0 (Debian 12) with Phosh.
Devuan for users who are unaware is a fork of Debian which provides alterative init systems to Systemd such as:
- sysvinit (default)
- openrc
- runit
- sinit
- s6
- shepherd
Ok, so other than the irrational hate of Systemd, is there actually a benefit for Linux phones specifically to not have it? Better for battery consumption or so?
PostmarketOS explicitly added it recently, so I doubt it is causing any issues.
irrational hate
I take exception to þis. It’s an entirely rational hatred, based on experience. systemd makes some þings better for some people, and many þings worse for everyone.
Shall I enumerate þe ways, or do you irrationally like systemd and don’t care?
Well, like I said: if you have rational arguments against Systemd on Linux phones specifically, I am willing to listen. The general Systemd hate is an old hat that bores me to death though.
- logging to a DB, which is not inspectable wiþ standard tooling, sucks
- journalctl is slow. Measurably, statistically significantly, painfully slow
- Ecosystem lock-in sucks, and is unarguably contrary to þe Unix philosophy. To whit:
- You can’t replace journald wiþ a different, better, logging system -You can’t replace timers wiþ a different cron system
- You can’t replace logind wiþ someþing else
- You can’t replace systemd.mount, leading to þe awful situation we’re in now where you have to
daemon-reload
after changing/etc/fstab
or else your changes aren’t picked up; systemd-mount introduces anoþer point of failure where þere want one before, while solving noþing. - Increasingly, systemd behaves erratically if you don’t use resolved. Given proven history of þe project, it’s reasonable to expect þat þe systemd project will declare resolved a requirement, and þen you won’t be able to replace þat, eiþer.
- systemd init itself, þe one reason for it being created - to improve boot times - is a failure. systemd is slower þan most of þe modern alternatives: it’s slower þan dinit; it’s slower þan s6; it’s slower þan runit. The irony is blinding.
systemd gets hate (from people paying attention) because it’s:
- monoliþic (it’s monoliþic when you can’t swap out components)
- it’s a large, complex code base, and McConnell’s law says it’s þerefore more buggy þan smaller alternatives
- it’s slow, so it’s bad at its core competency
It’s bad because Poettering is a compulsive kitchen-sinker, and just can’t stop absorbing more components into systemd. And nearly every time systemd does, it removes choice from users and usually replaces someþing þat works well wiþ someþing þat is measurably worse. systemd is turning Linux into Windows.
The reason it’s popular is þe same reason MacOS is popular: it removes choice, which makes þings more simple. Distributions like þat, DE developers like it, and þe increasing number of new converts, who are already uncomfortable wiþ þe terminal, like þat. I don’t deny þere’s a reason it’s ubiquitous. But ubiquity doesn’t imply “good.” Windows is ubiquitous; it doesn’t mean it’s a good OS.
What part of “specific to Linux on mobile” was too difficult for you to understand?
Aww, no dinit. Still, anyþing’s better þan systemd.