Close to nine-hundred comments¹ when Lennart posted about run0 on mastodon² a couple of months ago.

¹ https://news.ycombinator.com/item?id=40205714

² https://mastodon.social/@pid_eins/112353324518585654

Systemd, the software that recently added a feature that deletes users' home directories?

You mean a very specific command like systemd-tmpfiles that you have to manually execute and feed with the parameter --purge while omitting a config file in order for it to do so? Because of a bug that was quickly fixed in 256.1, not even 2 weeks after the 256(.0) release? [1]

[1] https://www.theregister.com/2024/06/20/systemd_2561_data_wip...

So what you're saying is that they don't thoroughly test systemd before releasing a new version?

So what you're saying is that testing can detect any bug and it's lazy developers who ship bugs?

This is not some random cgi script to generate a photo gallery that was whipped up over a weekend. This is systemd. It touches every aspect of each distribution that chooses to use systemd. It is important software. There is a minimum level of care that should go into development and release of software that does what systemd does.

They introduced functionality that deletes files and failed to run it through its paces before release. That's not merely failing a smell test, it's a big red flag that the developers of systemd should not be working on something as important as systemd.

Oh, and now the creator of systemd wants to add on functionality for privilege escalation. Knowing how they handle systemd, do you think they'll do a good job at making sure their proposed 'run0' thing will work as it says on the box, or is it going to have untested warts?

> Oh, and now the creator of systemd wants to add on functionality for privilege escalation. Knowing how they handle systemd, do you think they'll do a good job at making sure their proposed 'run0' thing will work as it says on the box, or is it going to have untested warts?

This feature isn't something new and has been present as `systemd-run` for ages...

Lenart is a MS employee. Testing at MS was always done by the user.

They are free to add whatever they want, if I'm free to not use it, to disable their stuff without break my OS. Because at the end, that is MY computer not Lennard's. I moved from Windows for that reason, avoid a OS that deprive me the authority on MY PC. If it won't be possible in future, I'm already studying OpenBSD. Anyway, my desire is a systems-free linux officially supported by major software houses.

I'm okay with this as long as they don't come for my doas.

if initd and shepherd still exist, then so will doas and sudo

Great, more lennartware unleashed upon us, more systemd nonsense to disable. Thankfully there’s s6-sudo, doas. The former solves the same problem with a daemon without messing with PID 0.

https://skarnet.org/software/s6/s6-sudo.html

Surely, just like systemd was, this will be completely without controversy.

Just switch to doas already.

Is sudo even useful ? I could argue that it has only few sane usage. Indeed, those usage are importants.

We have built massive multi-user systems with no sudo at all. Every users had the privileges required to do what was necessary.

To me, the most useless use of sudo is for admin access. When people creates nominative accounts, which are allowed to sudo everything. So people ssh to the system (via per-user account), then sudo su -, to do stuff.

Surely, it made sense before ssh keys where introduced (where there even such a time ?), because we had a single root password. A security-minded people would use auditd to link the whole process tree with a specific ssh private key, regardless of the user used.

But again, sudo is still useful in some cases.