NixOS needs what is IMO the killer feature of Arch: the wiki.
Comprehensive documentation on not only the OS but the additional packages that we use is what drew me to Arch, and the thing that makes me swear in frustration whenever I have to use Ubuntu/Debian.
NixOS is an excellent OS that has the promise of being every bit as hackable as Arch, but far more stable. Problem is, configuration is very different and needs extensive documentation to reduce that friction point.
The Arch wiki is pretty distro-agnostic (barring package names and pacman specific stuff). I've been distro-hopping for past decade and I've always used it as a reference for setting things up.
It's distro-agnostic because Arch does very little to modify packages when they're put in the repos, which means they'll line up with the packages own man page & readme. The issue comes when opinionated distros modify things like command syntax, etc file locations and default behaviour.
If NixOS is similarly unopinionated, it'd only really have to document its own system layer, but my point is that Arch being guaranteed to reflect a well documented system is what drew me to it.
NixOS needs what is IMO the killer feature of Arch: the wiki.
NixOS has a killer feature which obviates a wiki for most such purposes: NixOS options. They document themselves!
You don't need to look up a wiki on how to install and enable i.e. paperless and all the other services it depends on, you simply set services.paperless.enable and NixOS configures everything for you internally.
The option tells you roughly what it does internally and the other options provide pointers for things you might want to tweak about it. The services.paperless.extraConfig option for example tells you how to configure it (pointing to upstream documentation in this case) and even gives an example on what you might want to do.
Another example is how to install Steam. In Arch, the wiki must tell you all the manual steps required to enable multilib, install the steam package, install 32bit dependencies, yada yada.
In NixOS, you simply set programs.steam.enable = true;. Off to your games.
You wanna customise the Steam package to add additional flags, pass env vars or add additional packages your weird Linux-native indie game needs? programs.steam.package tells you how to do that right in the place where you do it.
While you're looking for steam, you might also come across hardware.steam-hardware.enable which you need to set in order to make your Valve Index and Steam Controller work properly.
You wanna start Steam in a gamescope session right from the display-manager? programs.steam.gamescopeSession does it for you. No need to copy paste some snippet that you'll instantly forget about and maybe breaks in a few months. programs.steam.gamescopeSession is maintained upstream by NixOS, so if it breaks, someone will go and fix that and nobody needs to adjust any of their copy-pasta because they'll just update as they always do and it just starts working again.
None of this is perfect yet and the quality of documentation of NixOS options really varies but I think you get the idea here. I already rarely look at the NixOS wiki to configure my system because the system configuration tells me what I need to do already and this will only get better as options get refined.
the promise of being every bit as hackable as Arch
I don't think it makes that promise and I don't think it's true.
NixOS is about doing things "properly"; applying software engineering to software environment management.
Whipping up a quick hack is much more complicated and time intensive on NixOS than doing so on Arch because it's way more abstract. You can't just quickly replace some binary with your own compiled one, you need to use NixOS' systems to wire in the binary and build it with Nix to begin with.
Maintaining a system (even one with terrible hacks) is much simpler in NixOS however.
You're underrepresenting the complications of NixOS and overrepresenting the complications of Arch. For example, to install Steam I would run sudo pacman -Syu steam. On a typical Arch setup that's all that's needed.
Another example is how to install Steam. In Arch, the wiki must tell you all the manual steps required to enable multilib, install the steam package, install 32bit dependencies, yada yada.
And that's why the Arch wiki is so great - it has details and links about everything that goes into making something work. If you want to learn more or if something goes wrong it's all right there.
But yes, I think you hit the nail on the head at the end there - hackability is Arch's strength, everything is exposed and flexible to tinkering. It's easy to make almost anything work, and easy to learn how it works. That's very different from NixOS's core philosophy of stability and reproducibility.
There are inherent pros and cons to both approaches - it really comes down to a mix of personal preference and using the right tool for the right job. They're apples and oranges, and the article framing NixOS as a superior successor to Arch is as silly as the reverse would be.
I installed NixOS on a laptop and tried to run a steam game and it just straight up didn't launch anything, went to the wiki to figure out the amd drivers, opengl whatever put like 10 new lines in my nix config rebuilt restarted still nothing works, after about 2 hours i just swapped back to arch and the games launched straight away, so for me it wasnt as easy as you may claim it to be. I also tried it on my desktop before and it was a better experience, but still not great. The nix config file is a bit of a mess of options that you have to dig into wiki pages and searching stuff to figure out how to get some stuff to work
Having options is not the same thing as documenting those options; well outlined documentation doesn't just dictate how to do something but also points out what you may want to do i.e. filling out unknown-unknowns.
Just because NixOS makes for an excellent DevOps template doesn't mean it can't also be an excellent platform for hacking together random crap. I understand that NixOS advertises itself as the former, but when I say "promises to be" I don't mean "makes a promise to be", but "has promise for being".
Features like: a common configuration interface, safe rollback, atomic changes, nixos-hardware all are features that enable developers to safely hack together solutions, and then have an excellent log detailing what they just did.
Note that, while the Nix package manager can technically run on OpenBSD to some capacity, that doesn't mean packages in Nixpkgs are compatible with OpenBSD.
I can't comment on the current situation from first-hand experience but I can say that there is no support guarantee as there is for Linux and macOS and that there is no binary cache either. You have to build everything yourself and I'm not even sure we can build even basic packages such as hello on BSDs yet.
Yeah true, whenever I have problems with some packages like Wayland and its alternative Xorg tools or games, its Arch wiki that helps extremely to fix or understand the situation. Its like many experiences are combined and written in a simple language everyone can easily understand.
On Ubuntu for example, everything feels like its hardcoded, not the standard and its just not even documented. And the wiki has minimal info about the packages.
Fully agreed- I experimented with it around November of last year and absolutely love the idea of it, but the documentation just isn't there. At the time I found nothing explaining flakes in a clear and concise manner so I had no idea how to use them or add them into my configs. People online kept saying to port the rest of my configuration to flakes but all of the examples online were complex and there was no simple example to build off of. I ended up settling for Universal Blue since it just uses OCI containers and I don't need a PhD to have a pseudo-declarative environment in it, but would love to revisit NixOS if the documentation ever gets better.
I'm new to NixOS, just installed it a few days ago, so I can't say much about it's pros and cons, but the installer was easy and I installed and booted into the new system very quickly. I think it might have been udpated in the past year, because I am watching a tutorial video from a year ago and he installs it via command line from the live iso.
edit: it also gave me a default configuration.nix which I've just been adding to (to get nginx with letsencrypt running, plus extra packages I wanted installed)
The article felt like it just kept going on and I was just not able to follow it very well, despite understanding the basics of NixOS. Went to the conclusion and it just feels like there isn't anything. It just feels like a pointless article with font way too small and gifs littered about.
Fun fact: I use NixOS since six years now and at least in the first two years the Arch Wiki helped me a lot to understand the NixOS configuration options.
That's the main crux with NixOS, it does a lot of stuff in the background for you that in my opinion you should know why it's being done the way it is. As such I consider Arch a good distro for a beginner who wants to learn the inner workings of Linux, while NixOS is a better-engineered distribution that takes care of the system for you. Arch's goal is to be simple for the maintainers which means it's very close to what one might consider a "standard Linux", and its wiki is mostly a documentation of exactly that.
This article reminded me of how I haven't run in a single dependency version conflict for years, I'm starting to get what debian users feel like seeing all this new distros fixing problems they never had in the first place
You can thank Flatpak for that. Dependency hell is real, especially on Debian, which ships old libraries. If you stick to default repos, you're unlikely to directly run into dependency issues, but once you install a program manually or from another repo, it's another story.
One example you may not have noticed, but which is a direct consequence of dependency hell, and a serious security issue, is for Firefox on Debian 11: it took around 6 months after it was EOL for Debian to update Firefox ESR. Twice (in other words, every single Firefox update on Debian 11).
It's almost like patience has its benefits, even if it means being forever behind the proverbial Joneses.
I tried using Arch for a year+, and spent too much time finding ways to fix things that broke with each update. Or fixing Pacman errors that made every package fail. Or filtering the update list to prevent breaking things. Or fixing the errors that using the AUR had introduced to Pacman.
On my debian PCs, I haven't even had to deal with version upgrades breaking them. I'm definitely missing out on the latest and greatest for most software I run, but I much prefer the peace of mind not worrying about updates breaking anything. I'd probably be a more powerful user if I had taken the time to learn exactly how to balance everything in Arch, but sometimes I just want to spin everything up, patch any major flaws or issues, and then get on with doing what I set out to do. An OS should be transparent when you need it to be.
Every server I run has been Ubuntu LTS since 14.04. I always read articles like this and wonder, "what are people who have this problem running?". Especially now-a-days where web services are always packaged up into Docker containers.
By 2015 I would have readily agreed to something like "It's the promised land where dependency hell and version conflicts are mere whispers of the past. Just imagine the pure bliss of a package management utopia that will make you question your life choices up until this very moment.".
Who are these people having dependancy issues now? And what are they running?
I'm assuming you mean the ability to run an AUR-helper and automagically install from PKGBUILDs; in which case, the answer is no. IMHO, lack of AUR access isn't as big of a deal as it used to be since Flatpak covers a good portion of what I have downloaded from the AUR in the past.
NixOS, package availability is comparable to Arch with the AUR. Arch has a total of about 80-90 thousand packages (That is, the regular repos + the AUR), and NixOS has over 80,000 packages in its own repos, that are actually maintained, and are more up to date than Arch (meanwhile, Arch often has a bunch of outdated and unmaintained packages in the AUR, such as all the Waybar packages that haven't been touched since April (there was a new release in June).
There is a part of me that wants to try this, but I have one question.
I believe this distribution allows you to have multiple versions of the same library to work with different programs at the same time, correct? Does this mean that each program downloads all its dependencies independently? If the answer is yes, I am staying with Arch. Too much bloat.
this distribution allows you to have multiple versions of the same library to work with different programs at the same time, correct?
That is correct. You can even have programs from entirely different releases running on the same system without conflicting with another; one with i.e. glibc 3.34 and one with 3.37 for example. Or even wilder setups with some packages using musl, other glibc and others yet being built statically.
Does this mean that each program downloads all its dependencies independently?
Each program references all of its (exact) dependencies. Nix then looks at the program and its references and builds a tree of dependencies.
Then it tries to "realise" these dependencies (make them, well, real), possibly by substituting (downloading) them from a binary cache or automatically building them on your machine if they're not available in any configured cache.
However, if some package with the same exact version already exists in the Nix store, no action will be taken. Why should it, it's already there.
For example, if you were in an empty world and built an environment with hello and coreutils in it, they'd both depend on glibc. If both came from the same revision of Nixpkgs, chances are that they depend on the exact same version of glibc.
What Nix would do here is fetch 1x glibc, 1x hello and 1x coreutils.
Note however that you don't need to manage any of this. You just say "I want hello and coreutils". Nix takes care of getting the correct dependencies in place but they won't be in your immediate environment. When you then say that your environment should no longer contain hello, it does that. At this point hello will still exist in the Nix store but it won't be in your PATH any longer, so it's not polluting any shared state; it just sits there on disk and the worst it could do is waste disk space. It's not "installed" in the same sense as what it'd mean to have an unused dependencies installed on an FHS system.
In order to reclaim disk space from unused Nix store paths, you can simply run a garbage collection. You don't need to care about that one specific hello store path here though, you just say "remove all unused store paths for me, thanks" and Nix removes it along with all other unnecessary paths. In NixOS, you can even have that ran periodically for you.
(Note that this is distinct from autoremove and the like; those "clean up" the shared state and free up disk space. In NixOS, these are separate processes and dependencies which you don't explicitly declare are never in the shared state to begin with.)
You see, while this could be seen as "bloat", it has none of the negative consequences bloat has on other systems such as more packages for you to manually manage, more binaries in your PATH or weird interactions of other programs. It's just easily managed disk space and disk space is honestly quite cheap.
Not exactly. If one package needs foo as a dependency, and another package also needs foo, it won't download a second copy.
But if another package needs, say, an older version of foo, it can download the older version of foo and you'll have two foo packages.
This is possible because Nix packages have hashes to differentiate between them, so any package can ask for the exact version of foo it needs instead of asking for foo in general and hoping the version it gets isn't incompatible.
Pretty sure this is also different from other containerized package managers, like Flatpaks or Snaps, which I believe throw all the dependencies in with their packages so each package has exactly what it needs in its container, and which is obviously going to be much more bloated than having shared dependencies. As far as I know Nix on the other hand doesn't get any more bloated than other distros (keep in mind that regular distros like Arch will still sometimes have multiple versions of a package, for example Python).
I think Nix is a really cool distro but the whole config things is really hard to learn, I wish there documentation was easier to understand as a nix noobie. I've used arch and many rolling release but nix configs are hard to learn. I really want to learn how to use nix, if anyone has any good sources for learning configs, id be much obliged
Nix and Common Lisp seem to sit in the same space -- it's spoken of extremely fondly but has difficulty escaping the lab. For some reason it's extremely technically capable, but fails to find widespread adoption.
Nix has just been removed from the university computers here. They admit it's nice, but the new (smaller) team just doesn't have the time to create the packages themselves. That's the flip side.
I have fiddled with Nix for some time (because of some Docker shenanigans) and there is some... weird stuff going on with it. Like nix-env not behaving the way it should and some flags that does not do what it is meant to do. And that is not a "Your problem" thing -- other users have reported the same kind of/similar issues.