The NixOS project and its community are celebrated for technical excellence and innovation. Yet, our primary information hub - nixos.org - doesn’t fully reflect these strengths. Many users, both new and experienced, struggle to find what they need. There is a history of anecdotal feedback that consistently highlights several pain points.
Overwhelming or sparse information:
Some sections are packed with technical details, while others lack the practical guidance users are searching for. Newcomers often don’t know where to start. Even seasoned contributors can have trouble finding up-to-date resources, as information is sometimes buried or simply missing.
What we think should be improved
Clear entry points:
We need distinct paths for newcomers, advanced users, and contributors, so everyone can get to the right place faster.
Persistent navigation:
A unified, well-structured menu will make it much easier to jump between topics and find your way around.
Less clutter, more focus:
We should reduce information overload by breaking up large pages and surfacing the most relevant content for each audience.
This redesign is for the community, by the community. Your feedback is essential: suggest improvements to our information architecture, highlight missing or unclear content, and help us prioritize what matters most.
How to take part
Leave a comment under this post with your suggestions, feedback, or concerns.
Let’s work together to make nixos.org as pure, functional, and deterministic as NixOS itself. Join us in building a better home for NixOS online!
We welcome your ideas and contributions - let’s make this a resource we can all be proud of.
A lot of drama from the past two years has given us a bad name (justified or not); we should be centering both the teams/friendly nature of the community and the values of the community. Thus, we should keep (or promote) the teams and values pages as first-class things.
The foundation always needs money. thus, we should keep the donate page as a first-class citizen.
We should see what our traffic patterns are (mobile/web traffic splits, etc.) and maybe consider streamlining experience accordingly.
We should respect users’ time and not present them with everything we can possibly think of. Less is more.
My particular gripes:
Teams should be on its own page, divorced from the Community page.
The “Explore” page, while delightful in the pictures, is too vague and should go away.
The asciienma things are cute, but can probably be stricken to no ill effect.
The “Download” page should just be divided by platform, and then inside you should have nix (the package manager), nixos (the operating system), nix (the flamethrower), whatever. Basically, just compact and streamline it down.
The “Learn” page can lose the search bars, can keep the links to the three manuals but lose the section and subsection links to parts of those manuals (resulting in less maintenance).
The “Values” page is fine, we just need to follow them as a community more reliably .
The “Community” page should have the team section broken into its own page. The RFC stuff should probably be moved to a manual or somewhere else. The online spaces and meetups stuff is good, the Nixcon stuff probably doesn’t need to be there as a first-class section (just link out to somewhere else, probably from the calendar section). The foundation stuff should be either at the bottom of the main landing page, or its own page available from the site footer. Like, there’s just waaay too much stuff happening on this page that isn’t directly “I want to go make friends with Nix people”.
The “Blog” page is fine to keep as a first-class thing but the weird filter checkboxes should all get chucked.
The “Security” stuff should be moved to the navbar or somewhere more prominent than the footer, since that’s a selling point of nix.
We should make the “commercial support” thing a first-class page (instead of a footnote in the community page) and link it on the navbar (using the space freed up by axing “explore”).
The big things I’m concerned about:
We don’t need a huge site redesign, and we’re spread thin for contributors to the homepage as-is. A big rewrite is probably not the best use of resources right now.
The problem we have right now is too much content, organized weirdly. We should take a third to a half of the content, thank it for its time, and Marie Kondo it to /dev/null.
We still have a large (but tractable!) backlog. I’d like to see us declare moral product bankruptcy and close anything older than, say, 2024 out of hand–if it’s important it’ll come back, and if it’s still open after four years maybe it’s not that important. But, we need to clean that product debt up.
Internationalization would be a bunch of extra work right now (and the one that would be worth supporting is Mandarin Chinese) and I think while it’s a worthwhile goal I’d caution against tackling it when we’ve got so much other low-hanging fruit.
Once all of this is done, maybe then we should look at a larger redesign/reworking…but as far as I see it, the main thing we need is less and organized, not more and new.
One little suggestion: Avoid marketing Nix as “reproducible” or “declarative” put more emphasis on business value like:
onboarding developers in under an hour
Better Ansible alternative
No configuration drift compared to Ansible
Basically unbreakable
Easy upgrades
Sidenote: I think it would go a long way for Nix marketing if there was a project that sets up WSL with the required apps easily. I would imagine devs nix run something and it installs IDEs and stuff which can be opened as native windows apps (x11 passthrough thing). Is anyone interested in picking such project up?
NixOS-WSL does exist, though I can personally say I’ve had some uncommon but persistent issues using it. Admittedly I wouldn’t be able to tell you if that was due to WSL itself or due to NixOS-WSL.
I think conceptually the idea is pretty great too, I do worry that people might start desiring more first class support for the platform if we actively encourage it though. I’m not intrinsically aware of what kind of burden that would throw onto Nixpkgs, and that might be something to consider.
There’s a big, big problem that many of us simpy got used to: nixos.org is used as a website for several things.
I am certain that this has been discussed and I’m willing to bet that @fricklerhandwerk knows everything about this.
To put it simply, nixos.org should be for NixOS and nothing else.
We have another problem, and that is the foundation is NixOS, as well. That should also be changed, if you ask me, to something like The Nix Software Foundation or similar.
I think with the exception of the first item (which is arguable depending on who you think “developers” are) I think they’re all pretty NixOS-centric. There definitely is a term misuse problem on the existing site, though.
Yeah, you are right. On other hand - most of the business value in Nix ecosystem is NixOS and Devshells. I guess one may argue packaging/OCI but yeah…
Easy cross-compilation is one of the things that I missed
Anyways, my whole point is that the site’s focus should be on business value. It should show teams how they can deliver faster and easier, and that it is worth the investment
In fact, it would be cool to ask some companies to speak up and share their success stories or create few mini official video tutorials that help people to get onboarded
For me the target audience should be massively enhanced. A good way would be to show usecases where NixOS would be a perfect operating system.
System administrators, e.g. in education: Its good for computing pools, but also excellent for making exactly comparable and restricted computers for taking exams (in this usecase you need up to hundreds of personal computers to be exactly the same and to be wiped after an exam).
Linux enthusiasts with more than one computer and maybe a homelab with some VMs.
Or just normal linux users who like to use office and other programs - maybe like this: explore office - HedgeDoc
I use it, I like it. In most situations, I end up using NixOS because I prefer isolation. My experimental tendencies get me in less trouble using it. NixOS, NixOS-WSL, Nix-darwin - I use them all.
I’ve been exploring nix for a while now have have set up multiple servers with home-manager as well as nixos. This is the first time I’m realising that nixos.org is not just about the OS.
To be very honest the documentation is hell and very hard to parse. If I didn’t have Claude to help me create flakes and explain things to me, I’d have given up on nix a while ago.
I think this would be a great first step that also aligns with the first item “Clear entry points”. Does the marketing team think this is feasible? @idabzo
Here’s my 2c which aligns with what some people have mentioned already. I approach this from the viewpoint of someone that:
Uses NixOS as a daily driver on multiple systems
Uses NixOS in a professional setting
Has introduced NixOS to other people (with varying levels of success)
This ties in with (3), but the tagline I use to get people excited for NixOS is “NixOS could be the Debian of the next 20 years”
Been using Ansible (and Puppet to a lesser degree) for ~12 years professionally
You may have noticed I only mention NixOS, there’s no mention of Nix anywhere. That’s because what I care about is the operating system and I believe more focus should be given on that. Particularly, I believe nixpkgs should be front and center as it’s the project that so many people have spent hours upon hours to build and the thing that defines NixOS.
The fact that Download | Nix & NixOS has instructions for downloading Nix before NixOS confused both me and other people I know. For me, using Nix or Lix or Tvix or … is just an implementation detail, the power of NixOS comes from all the packages and mainly the NixOS modules that people have written. The ideas behind Nix, the store, etc are obviously important as well, but they’re a way to compose a better OS than what we’ve doing all these years.
I understand this isn’t everyone’s viewpoint, other people care more about the package manager side of things or even only care for nix-darwin or using Nix as part of a build system. All of these things are valid as well! I just think we need a better representation, concrete examples and more clarity for Nixpkgs and NixOS on the website.