A friend of mine joked when this came to light the other day that we should start the timer to see how long it would be before some SystemD person crawled out of the woodwork to say that was expected behavior and you're an idiot for using it wrong. And we laughed to ourselves, wouldn't that would be funny, that really is the reputation they have now isn't it.
Well, jokes on us, because that's the first fucking comment on the bug.
systemd version the issue has been seen with 256 Used distribution Debian Unstable Linux kernel version used 6.8.12-amd64 CPU architectures issue was seen on x86_64 Component systemd-tmpfiles Expec...
Look, "/home is a tempfile" isn't just a footgun, or a bug.
It's an explicit ideological attack on the whole idea of what Linux is and who Linux is for.
If /home is just a tempfile to be purged when drivespace pressure hits some arbitrary threshold then Linux is a corporate-owned SAAS shim, and that's all it is. It's not just "breaking userspace", it's abandoning the idea that the people using that space and the stuff they've made there matter at all.
The whole system makes sense when your realize that SystemD's only constituency is people who want to rent out computers. Not for people who want to live under a their own roof, maybe in a community, to maybe build something together.
@empathicqubit nobody said anything about conspiracies, it's about a pattern of behaviour that harms user experience. I don't really agree with OP's "landlords" comparison but to brand this as conspiratorial thought is a poor comprehension of the problem.
It's hilarious that everyone is still ranting about this (with a questionable spin on things) while upstream already implemented and backported a restriction of the usage of the parameter in question to prevent this (deleting home) from happening and clarification for the intended use-case. Which is not at all freeing up space by deleting tempfiles. Yes the tool's name is misleading, but if you ever maintained a sizeable project you should be able to relate to not wanting to rename.
The bug got fixed after public outcry, but the working culture that allowed that bug to exist in the first place hasn't changed at all and that's the problem.
If you've ever maintained a sizeable project, you know that long term, discoverability and legibility outweigh literally everything else, and when you find a problem with supportability you can solve with grep you take that win every single time.
You may chose to believe that, the convo in the bug report and actions beg to differ. Sure the original state was bad, but it's clear there was nothing intentional or systematic about it. Like actually considering /home as temporary and wanting to purge it, as it widely insinuated. Anyway, I doubt there are any open minds in this discussions, si bit much point having them (yeah I realise that's inconsistent with me answering, I can't help it and don't claim to be fully rational (or close to) 😀 ).
@imsodin "Nothing intentional?" Guy, _read the bug_.
bluca's opening comment is quite literally "this is working as intended and you are stupid".
The second comment says explicitly "the tool itself is nothing wrong."
It's not until you're 2/3rds of the way down the page that someone says "We need to rethink how --purge works. The principle of not ever destroying user data is paramount" and the next 2 comments from Poettering and Bluca are repeating that No Actually It's Right Actually.
Yeah it is "right", as that's always in the eye of the beholder: You do want to purge user files if there user really wants to, e.g. application data when purge uninstalling that application. And it's super unlikely that anyone ever wants to do that with everything on their system, i.e. using it without a config file like the original reporter did. And now doing that is disallowed. I don't defend the original state as being good or anything, it's just ridiculous how people conflate nit being aware of/ignoring a problematic possible (mis)use of a tool with a conspiracy of systematic intention of harming userspace. Oh and yes, Luca's communication here (not dating it's the only time) is not conductive of a constructive discussion (to put it mildly) - if that was what you all were complaining about, that would make a lot more sense.
@Simon Frei @mhoye So you agree 1) the tool was improperly designed 2) the tool was improperly documented 3) the bug were improperly triaged 4) the bug improperly managed
So what is your issue with being annoyed about being annoyed that things were put in a state where you could destroy your system using a tool where the documentation implies that it does something that *isn't* destroy the system?
@silverwizard 1) not ideal maybe 2) yes 3) 4) no - there was one contributor reacting not very helpfully, but overall the reaction to the report went into the right direction and quickly so. Also I wasn't poking at "being annoyed", I was poking at all the conjecture about intent and outrage, ask the huge noise being made over this (at least many outraged and well populated toots made it into my timeline, a perfectly objective measure of scale of outrage of course).
@Simon Frei I see - so you don't understand how these things point to a core problem in the project and are not concern about the project nor its future, and think we should be less worried? I think you'll find most of the people making the noise disagree.
@mhoye Look, it's plainly in the documentation that when you press this button, the computer will shoot you in the face. It's your fault for not reading the docs.
Do not question why there exists a button that shoots you in the face without warning
@argv minus one You say that like it's a good thing. I also acknowledged the difference between / and *. My point, which remains unchanged, was that there is an effort underway to reduce the number of foot guns, and that's a good thing.
rm -rf / requires you to use --no-preserve-root. I'm not sure about using * from within the root directory and I'm not about to spin up a VM to test it, but this at least acknowledges that there's a problem.
If you don't already know what it does, it at least isn't a command that gives the impression of doing something different.
Given that they've fixed the submitted bug in 256.1, it seems that the maintainers of systemd disagree with you as well.
`rm -rf *` was the mentioned command, not `rm -rf /`.
Which kind of proves a point. Unix and its siblings are full of system-destroying footguns that can and will go off if you don't pay careful attention to what you're seeing and what you're typing. The (previous) behavior of systemd-tmpfiles is by no means unprecedented.
If I didn’t know better, I swear that systemd was an invention of Microsoft (since one of the creators is a MS employee) for the expressed purpose as a psyop against Linux users.
mhoye
in reply to mhoye • • •Look, "/home is a tempfile" isn't just a footgun, or a bug.
It's an explicit ideological attack on the whole idea of what Linux is and who Linux is for.
If /home is just a tempfile to be purged when drivespace pressure hits some arbitrary threshold then Linux is a corporate-owned SAAS shim, and that's all it is. It's not just "breaking userspace", it's abandoning the idea that the people using that space and the stuff they've made there matter at all.
mhoye
in reply to mhoye • • •SystemD is for landlords.
The whole system makes sense when your realize that SystemD's only constituency is people who want to rent out computers. Not for people who want to live under a their own roof, maybe in a community, to maybe build something together.
SystemD is just software for landlords.
Empathic Qubit (Old)
in reply to mhoye • • •@gsuberland
Graham Sutherland / Polynomial
in reply to Empathic Qubit (Old) • • •Empathic Qubit (Old)
in reply to Graham Sutherland / Polynomial • • •Okay
@mhoye
Simon Frei
in reply to mhoye • • •mhoye
in reply to Simon Frei • • •@imsodin Everything you've just said is wrong.
The bug got fixed after public outcry, but the working culture that allowed that bug to exist in the first place hasn't changed at all and that's the problem.
If you've ever maintained a sizeable project, you know that long term, discoverability and legibility outweigh literally everything else, and when you find a problem with supportability you can solve with grep you take that win every single time.
Also, this shit's not funny in the least.
Simon Frei
in reply to mhoye • • •mhoye
in reply to Simon Frei • • •@imsodin "Nothing intentional?" Guy, _read the bug_.
bluca's opening comment is quite literally "this is working as intended and you are stupid".
The second comment says explicitly "the tool itself is nothing wrong."
It's not until you're 2/3rds of the way down the page that someone says "We need to rethink how --purge works. The principle of not ever destroying user data is paramount" and the next 2 comments from Poettering and Bluca are repeating that No Actually It's Right Actually.
Simon Frei
in reply to mhoye • • •Oh and yes, Luca's communication here (not dating it's the only time) is not conductive of a constructive discussion (to put it mildly) - if that was what you all were complaining about, that would make a lot more sense.
silverwizard
in reply to Simon Frei • • •@Simon Frei @mhoye So you agree
1) the tool was improperly designed
2) the tool was improperly documented
3) the bug were improperly triaged
4) the bug improperly managed
So what is your issue with being annoyed about being annoyed that things were put in a state where you could destroy your system using a tool where the documentation implies that it does something that *isn't* destroy the system?
Simon Frei
in reply to silverwizard • • •Also I wasn't poking at "being annoyed", I was poking at all the conjecture about intent and outrage, ask the huge noise being made over this (at least many outraged and well populated toots made it into my timeline, a perfectly objective measure of scale of outrage of course).
silverwizard
in reply to Simon Frei • • •Jonathan Lamothe
in reply to mhoye • •@mhoye Look, it's plainly in the documentation that when you press this button, the computer will shoot you in the face. It's your fault for not reading the docs.
Do not question why there exists a button that shoots you in the face without warning
like this
Tim, Cainmark 🚲 🥥 🌳 🧱 🫛, samiamsam and MildDrop72 :verialt: :verialt: like this.
Cainmark 🚲 🥥 🌳 🧱 🫛 reshared this.
argv minus one
in reply to Jonathan Lamothe • • •@me
This is a Unix-like system. It has quite a few buttons that shoot you in the face without warning.
Jonathan Lamothe
in reply to argv minus one • •/and*. My point, which remains unchanged, was that there is an effort underway to reduce the number of foot guns, and that's a good thing.Jonathan Lamothe
Unknown parent • •@Krangled Failstate
This is nonsense for at least three reasons:
rm -rf /requires you to use--no-preserve-root. I'm not sure about using*from within the root directory and I'm not about to spin up a VM to test it, but this at least acknowledges that there's a problem.It was bad UX, plain and simple.
Jonathan Lamothe
Unknown parent • •Jonathan Lamothe
in reply to Jonathan Lamothe • •@Krangled Failstate As for your remaining points, watch this video:
If you still don't understand, watch it again.
Jonathan Lamothe
Unknown parent • •@Krangled Failstate
"Oh no, we found an unexploded land mine from the second world war."
"Pfft. It hasn't been a problem in all this time. It's fine."
This is basically the same argument.
Seriously. Just stop.
argv minus one
in reply to Jonathan Lamothe • • •`rm -rf *` was the mentioned command, not `rm -rf /`.
Which kind of proves a point. Unix and its siblings are full of system-destroying footguns that can and will go off if you don't pay careful attention to what you're seeing and what you're typing. The (previous) behavior of systemd-tmpfiles is by no means unprecedented.
@krangledfailstate
Will
in reply to mhoye • • •