3 ms·
> I'd rather either not run applications I don't trust or just containerize them. Guess how Flatpak works? Containers are not security isolation, I'm not sure
by dijit 4y ago
> I'd rather either not run applications I don't trust or just containerize them. Guess how Flatpak works?
Containers are not security isolation, I'm not sure why this has to be repeated so often because it's very obviously not the problem containers try to solve.
Maybe you're thinking of Jails or Zones from BSD and Solaris?
> I don't trust random unpaid volunteers to do this anyway.
But you trust them to produce your kernel, coreutils and the very software you're talking about?
> [flatpak] eliminate[s] the ability for software to conflict
I can very much think of ways they can still conflict; one would be port bindings for example. But I agree with the sentiment, you're basically arguing in favour of static binaries being shipped... except instead of static binaries it's... a tarball you run in a chroot... which has some kind of value I suppose.
> I use a computer to run software. I use a personal computer to run the software I personally want to run.
Sure, nobody stops you from downloading random binaries, or an application developer creating their own package repo and allowing you to download directly. There is history with that approach, in fact I consider rpmfusion (third party repo) to be basically essential for desktop fedora/rhel/centos.
My major point is: centralising trust to a few maintainers who you can hold accountable for your entire OS experience has value, and you can still opt out. But making flatpaks mandatory decentralises that trust back to application developers, who might not be trustable.
One need only look at how many backdoored docker containers exist on dockerhub.. since it's a platform where anyone can publish essentially anything, even if you're aware you should avoid it- it _does_ pose a concern.
There's also a minor argument about blast radius', a developer account being compromised: on docker hub this would be disastrous and opaque, but for github it would not be so bad and probably noticable, and for an apt repository it would be much less dangerous (and very noticable).
- AnIdiotOnTheNet 4y ago> Containers are not security isolation, I'm not sure why this has to be repeated so often because it's very obviously not the problem containers try to solve. They aren't a perfect security solution, that doesn't mean they are useless for it. If I were really worried about an application being nefarious I either wouldn't run it or I'd run it in a VM on a dedicated device or something. You have to trust someone, and I choose to trust fewer people (the developer) rather than more (developer plus packagers). > I can very much think of ways they can still conflict; one would be port bindings for example. But I agree with the sentiment, you're basically arguing in favour of static binaries being shipped... except instead of static binaries it's... a tarball you run in a chroot... which has some kind of value I suppose. There are good reasons very few piece of software, let alone desktop software, are distributed as static binaries. And no, it isn't because of dependency size. There are things you just can't do without dynamic linking in Linux, and glibc is especially tricky to deal with as I understand it. > My major point is: centralising trust to a few maintainers who you can hold accountable for your entire OS experience has value, and you can still opt out. But making flatpaks mandatory decentralises that trust back to application developers, who might not be trustable. FlatPak still allows you to have your own repo run by the distro. In fact, Silverblue only pre-installs Fedora's own FlatPak repo, not Flathub. But regardless, I do not see a reason that FlatPaks need be mandatory, any FOSS software can just be packaged up by a distro without FlatPak. And if it isn't FOSS software than none of the advantages you speak of apply anyway. FlatPak just makes my, and a lot of other people's, lives a lot easier and fits how we use a computer better than the traditional package manager/repo model, so we'd appreciate if people would try to understand why. > and for an apt repository it would be much less dangerous (and very noticable) Would it? Heartbleed went unnoticed for years. The debian random number bug went unnoticed for years. An actor could slip in such a bug intentionally at any time and it would likely be not only overlooked, but simply considered a mistake rather than malicious when caught. I'm not convinced of the supposed security benefits of package maintainers.
- ryukafalz 4y ago> Containers are not security isolation, I'm not sure why this has to be repeated so often because it's very obviously not the problem containers try to solve. I will refer you to: https://blog.jessfraz.com/post/containers-security-and-echo-chambers/ https://blog.jessfraz.com/post/containers-security-and-echo-...
- dijit 4y agoI quite like Jessie Frazzel, so it pains me to say this: Reality is not in line with what she says in this post. The paper she's referring to specifically calls out hardening guidelines for containers to make them more secure, but that is not how containers are run, even on GKE (a platform I consider to be the defacto gold standard for container platforms). Aside from the issue of signing (which, is a continued problem in windows-land FYI) not being the default, even seccomp is disabled by default in the overwhelming majority of cases. Even flatpak only uses seccomp to filter "the most risky syscalls" (perf, ptrace, weird network protocols like DECnet and recursive namespaces). What I'm trying to say is: ticking the box that seccomp is enabled is not useful if you're allowing bind mounts. Why would you allow bind mounts? that's insane?! Well, usability is a trade-off my friend, and telling people their fancy sandboxed application can't read their ~/Downloads folder is not going to fly. The only solution is an all-encompassing one, with a prompt system like macos has to allow only a selected number of mounts, but that's going to be handled at a completely different level than seccomp itself, and seccomp will be explicitly permitted to do this; thus protections against this one specific thing could be thwarted. This is but one of many things. Solving security for the general case is really hard. As Jess herself points out in that article.