Karna
LinuxAdoptions are a mistake anyways. Remove unmaintained packages and block the name for several months.
Adoption of unmaintained packages to maintain them is not a mistake. The problem is the current implementation, not the idea behind it. It's like saying the AUR is a mistake, because some people do malicious stuff.
They should find a better solution, like adoption shouldn't be granted to everyone without question, especially new accounts who didn't maintain anything before. Mass adoption shouldn't be granted automatically (limit rate), in example 1 package adoption per day and if someone wants more, admins or moderators need to approve. And updates of newly adopted packages should wait a day.
Also the AUR helpers should do a better job. Always ask if a new adopted package should be updated and give a warning the maintainer changed.
The problem is the current implementation
Yes, exactly this! I am not surprised it happens. I’m surprised it didn’t happen before.
especially new accounts
Weren’t there “sleeper accounts” registered years ago that became active in the current wave?
Also the AUR helpers should do a better job.
Even experienced people will just update as if nothing could happen. Adopted packages should have to use a different name and the current name being blocked so it WILL get attention when some tries to update their system.
Weren’t there “sleeper accounts” registered years ago that became active in the current wave?
Which does not invalidate my point about new accounts, but yes. Its important for new accounts, so once a sleeper account is banned, its not that easy to just create thousands of new accounts while everyone is focusing on the current active ones. And if, as I suggested, mass adoption per account is not possible, then the attacker has less to attack.
Edit: They need to make sure that sudden editing many packages in short time, with probably the same or similar lines should be automatically reported. They need some automated checks in place, at the very least. This would be a very suspicious behavior if many accounts are not active, and then suddenly all of them do something.
Even experienced people will just update as if nothing could happen.
Then you can't fault the system, if you are this reckless.
Adopted packages should have to use a different name and the current name being blocked so it WILL get attention when some tries to update their system.
This would break all dependencies of this package. The point of adoption is to keep it working and stable. I'm highly against force changing the name, that is not a solution (at least not one I am happy about).
[Changing a package’s name] would break all dependencies of this package.
Yes, that is correct. But that should not be a big deal for packages that are actively maintained. The maintainer can simply change the dependency to the new name after making sure the new package is legit.
OK, that's a good point. It would prevent auto updating. However any package that is NOT updated, should stay with same name in my opinion. So that everything (with the old secure code) stays intact and working. The name change should be part of the the update process. So it only breaks if you want to update, which would ensure compatibility if you choose not to (as it is safe). Something along the likes like this. I agree with your suggestion now, because that seems to be sensible idea.
Enforced commit signing and making it obvious when the signature changes would help make adoption much safer. I really hope they implement it.
Just saw DeepSeek TUI in there, that's probably going to be a whopper that hit quite a few users.
You mean https://aur.archlinux.org/packages/deepseek-tui-git from the list https://lists.archlinux.org/archives/list/aur-general@lists.archlinux.org/thread/P4WIRHTFNH2YZWQHGBAKQWX5YOAFIDLY/ ? It has 0 votes and Popularity is at 0.000000, since it was uploaded 2 months ago. And there is a non Git version that is not listed (probably not affected) at https://aur.archlinux.org/packages/deepseek-tui . On top of it, on their website https://deepseek-tui.org/#install the main recommended way to install it is with npm and there is even a way to build it with Rust it seems (I'm a bit unsure).
So I don't think it hit that many users, as only a couple of people since Mai would want to install it, and they have to have updated it to build from source in recent days. That is a very narrow number of people, just by my gut feeling.
Tumbleweed stay winning
No way to prevent this, says only repo where this regularly happens
Isn't this similar to the reason a lot of people hate snaps? Or am I misunderstanding something? I'm not an Arch user (btw) so I'm not super familiar with AUR.
You're right on malware finding its way onto the Snap Store. I find it hilarious to see that the employed tactics are basically identical 😜.
However, FYI, the hate on Snaps is a lot more broad than that.
Besides all the other non infected ways to install the software, there is a way to prevent this: Just read the AUR package before install and don't trust blindly any new maintainer.
It's metaphysical approach to security. Enshrined rules that can't be enforced don't define user's behavior.
