HN Explorer

Ask HN: Why no one forks systemd and “fix” it?

6 points by RaitoBezarius ↗ HN
The annual Pwnie Awards at Black Hat named systemd as the lamest vendor. [1]

Though, systemd is under LGPL v2.1+ according to Wikipedia [2] and their GitHub repository [3].

Thus, it should be possible to fork systemd under LGPL v2.1 + and take the project in a different way, IMHO, a way which would listen more to its users.

But, as we have seen in the past, Debian's community has been strong enough to present a fork: Devuan [4].

Why don't we see a similar effort towards a systemd fork, given that the <wildcard>nix (even BSD?) community should be larger than the Debian's one (which is included in the <wildcard>nix community by definition.)

Is there a technical reason? Or a political reason? A social reason?

I would like to understand more about this, because it feels like to me that "open source" [5] is somewhat broken (and of course, as a developer who could lend a help, I am at fault.)

What could I do as an individual developer who could not possibly maintain an entire fork of systemd alone (with or without appropriate knowledge)?

[1] : https://www.theregister.co.uk/2017/07/28/black_hat_pwnie_awards/

[2] : https://en.wikipedia.org/wiki/Systemd

[3] : https://github.com/systemd/systemd/blob/master/LICENSE.LGPL2.1

[4] : https://en.wikipedia.org/wiki/Devuan

[5] : (F)OSS, Free software, call it like you prefer. I refer to a more eerie magic of the "open source" (e.g. RethinkDB, Python, and so on.)

7 comments

[ 2.8 ms ] story [ 25.6 ms ] thread
itwy
Because it's an amazing piece of software that works great. It's just 'cool' to shit on it.
ahoka
It keeps having serious bugs, which is not exactly amazing.

My personal encounters: - Stopped reaping child processes (which is pid 1's main job). Actually it stopped doing everything at all because of its stupid error handling - Floods syslog with useless messages (google for Time has been changed) every seconds - Keeps rescheduling random offsets in timers every few seconds, so they never really run

systemd is a most unrealiable system software in Linux I have ever seen. Typical over-engineered 'enterprise' software from a 'enterprise' company. Do you know what? Red Hat is as harmful for Linux as IBM was for Java.

wmf
There was uselessd but only one person was working on it and even he gave up. http://uselessd.darknedgy.net/

I don't think open source is broken, but it may have led some people to think that if you can imagine it someone must already be working on it. But that's not true because the set of things people can imagine is far larger than what people have time to work on. There are few people who want a "fixed" version of systemd to begin with and literally only one or two people who are willing to work on that.

digi_owl
Because it is simply too big and still growing.

End result is that unless you can get some deep pockets to fund your ongoing efforts on the fork, you are unlikely to overtake systemd any time soon.

Hell, Canonical and Debian tried for the longest time to maintain a shim that would enable systemd-dependent upstream projects to work without systemd. They simply could not keep up with the interface changes and feature creep.

rleigh
Forking implies inheriting the design, including all its flaws. You can hack around with the codebase all you like, but if you can't fix the design issues, because it would be breaking interoperability, you're a bit limited in what you can do.

Other init systems were sufficiently constained in scope that they could be swapped out with relative ease, and modified or rewritten entirely as you liked.

romanovcode
Because last thing you want is another another init system. You just don't know it.