So I’ve recently taken an interest in these three distros:
All of these offer something very interesting:
Access to (basically) all Linux-capable software, no matter from what repo.
Both NixOS and blendOS are based on config files, from which your system is basically derived from, and Vanilla OS uses a package manager apx to install from any given repo, regardless of distribution.
While I’ve looked into Fedora Silverblue, that distro is limited to only install Flatpaks (edit: no, not really), which is fine for “apps”, but seems to be more of a problem with managing system- and CLI tools.
I haven’t distro hopped yet, as I’m still on Manjaro GNOME on my devices.
What are your thoughts on the three distros mentioned above?
Which ones are the most interesting, and for what reasons?
Personally, I’m mostly interested in NixOS & blendOS, as I believe they may have more advantages compared to Arch;
What do you think?
I think NixOS is awesome, but it certainly doesn’t offer “access to (basically) all Linux-capable software, no matter from what repo.” - at least not natively. You can do that through containers, but you can do that with containers on any distro. Where it shines is declaring the complete system configuration (including installed programs and their configuration) in its config file (on file-based configuration, I wouldn’t really consider blendos a viable competitor).
I don’t quite agree with this. In NixOS you can write custom expressions that fetch software from any source, and stitch them into your configuration as first-class packages. So you do get access to all Linux-capable software natively, but not necessarily easily. (There is a learning curve to packaging stuff yourself.)
I use this process to bring nightly releases of neovim and nushell into my reproducible config. Ok, I do use flakes that other people published for building those projects, which is a bit like installing from a community PPA. But when I wanted to install Niri, a very new window manager I wrote the package and NixOS module expressions all by myself!
Another NixOS user (and minor package maintainer, if it matters) here. Essentially, NixOS is actually rather simple to write a configuration file for a particular program once you get the knack for the nix language and learn how to workaround the sandboxing. I would actually consider it substantially less involved as compared to (for instance) creating your own Debian package.
However, getting to this point will take a bit of effort, and this step is more or less obligatory to use software on NixOS, whereas it generally isn’t (but still is a good idea) on other distributions.
expired
“Learning curve” is an understatement. Nix is one of the most poorly documented projects I’ve seen, next to openstack. Coming into it with no background in functional programming didn’t help.
Maybe I shouldn’t have tried to package openstack on nix.
But I’ve tried to package other stuff, like quarto, and that was a nightmare. Nixpkgs didn’t have an updated pandoc and I spent an eternity asking around for help, to try to package it. An updated version just got pushed to unstable a few days ago. The same matrix channels I joined to ask for help have been discussing this since then. Props on them for getting it working, but anyone who says that you can easily package anything, is capping. You need to have an understanding of the nix language, nix packaging (both of which are poorly documented), and a rudimentary packaging ecosystem of what you are trying to package.
Don’t even get me started on flakes vs nonflakes.
I still use nix-shell for all my development environments, because it’s the best way for reproducible environments I can share I’ve found.
how well is it suited as a daily driver for dev work & playing games?
The single killer feature that convinced me to move to NixOS is the ability to very easily keep separate development environments separate. For instance, if you’re working on multiple dev projects that have different minimum requirements, and you want to ensure that (for instance) you don’t accidentally use features from after boost 1.61 for project A, because that’s the stated requirement, but you need features from boost 1.75 in project B.
In a normal distribution, in order to set up an environment that has the proper version for project A you’d need to set up a chroot, a virtual machine, a complicated set of environment variables in a bespoke script with custom installation paths that you need to set up manually and remember to source, or just install a newer version of boost and rely on continuous integration to catch it if you screw up.
In NixOS, you can set up different shells which all reference the exact correct version of the libraries required for every project, you can have them installed simultaneously and without conflicts, and there’s even a shell hooking program that will automatically load and unload this configuration when you change directories into and out of the project folder. It makes managing many different projects much easier. It’s like a better version of venv, but for everything.
Well, for playing games I use the flatpak version of steam and it works OK.
For dev work, it’s great overall. Especially its ability to create separate reproducible environments with whatever dependencies you need for every project. However, there are some tools (rare, but they exist) that don’t work well with it, and if your dev work happens to need them, it can becomes a problem.
For day to day (i.e. web browsing), it works the same as anything, with one disadvantage: there is a disadvantage here: it downloads a lot more than other distros on update, and uses more disk space. The biggest difference between NixOS, and say Arch, is not how it behaves once it’s up and running, but in how you configure it. Specifically, you have to invest a lot of time to learn how, and set up your system initially. But then reinstalls, and (some of) the maintenance, become easier.
why is that?
In regular FHS distros, an upgrade to libxyz can be done without an update to its dependants a, b and c. The libxyz.so is updated in-place and newly run processes of a, b and c will use the new shared object code.
In Nix’ model, changing a dependency in any way changes all of its dependants too. The package a that depends on libxyz 1.0.0 is treated as entirely different from the otherwise same package a that depends on libxyz 1.0.1 or libxyz 1.0.0 with a patch applied/new dependency/patch applied to the compiler/anything.
Nix encodes everything that could in any way influence a package’s content into that package’s “version”. That’s the hash in every Nix store path (i.e.
/nix/store/5jlfqjgr34crcljr8r93kwg2rk5psj9a-bash-interactive-5.2-p15/bin/bash). The version number in the end is just there to inform humans of a path’s contents; as far as Nix is concerned, it’s just an arbitrary name string.Therefore, any update to “core” dependencies requires a rebuild of all dependants. For very central core packages such as glibc, that means almost all packages in existence. Because those packages are “different” from the packages on your system without the update, you must download them all again and, because they have different hashes, they will be in separate paths in your Nix store.
This is what allows Nix to have parallel “installation” of any version of any package and roll back your entire config to a previous state because your entire system is treated as a “package” with the same semantics as described above.
Unless you have harsh data caps, extremely slow connections or are extremely tight on disk space, this isn’t much of a concern though.
Additionally, you can always “garbage collect” old paths that are no longer referenced and Nix can deduplicate whole files that are 1:1 the same across the whole Nix store.
The way nix deals with packages is very different from most distros. If you install a newer version of a package, the older version just gets hidden, not removed. This makes it very easy to rollback or recover from errors, but it does mean you tend to use more space.
You can always configure the garbage collection to reduce disk space usage, or manually run it.
It’s the probably the best distro for dev work imo. Nix in general is really nice for development. Games work fine — you can just install steam or putrid or whatever, and you can run normal binaries with steam-run.
And you can create package configs, but you can also do that for nearly every distro. So, yeah, that confuses me too… I’m not sure what OP was trying to say there.
Sorry for my ignorance,
but why is blendOS not a viable competitor to NixOS?
To clarify, I was referring specifically to its ability to specify the full system configuration in its config file - not overall. But I haven’t used blendos, and my impression is mostly from a quick look at their documentation. They have a snippet with sample configuration. There, they have a “Modules” section, but I couldn’t find what modules are available, what options they have, how to configure them if we want to do something more complex than the available options.
Then containers are clearer: they have a list of installed apps, and then commands to bring them to the desired state (somewhat similar to a dockerfile). But even then, i imagine that if you have a more complex configuration, that’s going to get clunkier.
Thanks, that makes sense.
Do you think the use of OCI containers/images is a mistake/bad choice from blendOS?
How is NixOS different?
No. It’s probably the best way to run packages from Arch, Debian. Ubuntu, Fedora, and others, all on the same system.
NixOS simply doesn’t tackle that problem, so it doesn’t come with containers out of the box. If you want to run packages from other distros on NixOS, you’d probably need to manually configure the containers.
I feel like you’re under the impression that the three distros, NixSO, blendos, and Vanilla OS, have similar goals. I don’t know about Vanilla OS, but the main similarity between the other two is that they’re both non-standard in some way.
But they’re actually solving completely different problems: BlendOS wants to be a blend of different OSes, NixOS wants to have a reproducible, declarative configuration (declarative here means, you don’t list a bunch of steps to reach your system state, but instead declare what that state is).
You can steam-run any Linux executable, so for those cases it’s fine.
All the major software is already in Nixpkgs, there’s just some holdouts still shipping .deb and .rpm files, though
I would disagree. I feel like nixpkgs has pretty much everything, more so than any other distro in my experience. The differences in how NixOS work can make it a little weird to run something off the cuff, but steam-run has your back in those situations.