Here’s a few scenarios off the top of my head:
one is a regrettable and avoidable chore
Typically for the novice user, there will be more than a single install from a USB stick, it’s pretty useful to be able to create your own sticks to help yourself get out of trouble when you inevitably get there as part of learning.
Also, those aren’t unrelated skills. That’s a bit of a stretch.
The vast majority of NixOS configuration doesn’t need to be any more complex than the equivalent JSON file
That depends on the users needs, and my experience (and the experience of anyone that has provided support on the matrix channels) tends to show that the users need are for more than the equivalent of a JSON file.
I’m not saying not to do USB sticks.
How would we know without even trying it?
You’re implying it’s impossible to verify the contents of a USB stick. I expect we have the technology to do that just as easily as verifying the contents of an ISO.
The support on the Matrix channels is needed exactly because people are going past simple configuration. People with a really simple configuration simply won’t need to ask. The vast majority of beginners probably only ever need to configure simple things like which programs to install and which users should exist.
I do think that the concept of NixOS has at least the potential for that, but even very basic setups quickly run into edge cases.
Just have an nvidia GPU, which is a solid 75% of users according to at leat the steam hardware survey (which is skewed, but so is the base of potentially interested users), or try to use electron with modern GNOME or KDE, which will be practically anyone who might have an otherwise simple setup. Anyone who is not a programmer likes to use software that doesn’t just sit in nixpkgs IME, too, and often ends up needing to escape the linker (at best, let’s not talk about wine), which is always a can of worms.
On top of this, the nix ecosystem is so full of footguns that I doubt any users who are at that level have a good long-term experience, and the community is largely not very welcoming of complete beginner questions either, as much as some of us try - ultimately they’re frustratingly hard to help without giving a programming 101, which this community just isn’t the right place for; understanding JSON is a much higher bar than it seems.
Sooo… I think your heart is in the right place, and I agree that NixOS could serve as a basis for a kick-ass newbie distro. Even I have to acknowledge that NixOS is not that distro today, though, and unlikely to get there without a good decade of dedicated effort (see snowflakeos for some experimentation in that direction) and a massive shift in general Linux desktop use, and even desktop software vendors. We’ll probably see complete android/iOS domination long before NixOS becomes relevant to a user unwilling/unable to burn a USB with rufus.
Giving branded USBs away at conferences is a cool idea, though, I imagine giving people something physical to remind them to try that fancy distro is a good marketing gimmick, and it’d probably filter for users who are at least receptive to learning if nothing else. Selling such a USB fairly cheaply would let people giving speeches and doing other grassroots efforts get their hands on them to give away, too. Hell, maybe we just need a branded usb and a neat script to let you flash them in bulk (though, security of branded installers remains a concern here, hard to make this safe).
Looking at it from that angle seems like a better perspective for the distro NixOS is today.
You are the one that needs to convince others it’s a good idea. Go find the data to base your theory on. This entire thread feels like you demanding to solve non-existent problem.
Can you think of 5 real people that wanted to use NixOS but didn’t end up doing so because they didn’t have an installer USB stick?
The accessibility (that is, actually getting to the point of accessing it) of Linux is a very real problem which NixOS is (IMO) in a really good position to remedy, with the simple configuration (yes, really, compared to all other Linux distros).
I can’t think of five real people who want to learn Linux, but that doesn’t mean they don’t exist. Why would my knowledge of five randos convince anybody of anything?
Because then you’re just wasting our time…
You claim there is a problem, yet you can’t think of a single person that’s encountering the problem… How do you not see an issue with your behavior?
Also it’s incredibly dishonest to say “people who want to learn Linux” when we’re specifically talking about people who want to use NixOS but can’t because they can’t flash their own USB stick with the installer. And if we’re talking about people that want to learn Linux, learning how to create your own bootable ISO is one of the most important things you can learn initially, as it helps a lot with recovering a system from a broken state.
The conversation seems to diverge on its premise.
On the one hand, would NixOS be easier to use with easier access to pre-flashed USBs? Maybe!
On the other hand, is this a worthwhile pursuit of the existing NixOS community as a way to drive engagement? Probably not!
When I first installed Linux in 1997, you would order CDs from linuxpusher.dk. Mainly because I didn’t have a CD burner or a high-speed internet connection. It significantly lowered the barrier to entry.
I would like to believe that handing someone a graphical installer USB still does lower the barrier. Even though making a USB installer is easy in any OS and isn’t “the hard part”, each step in the process has a drop-off rate; giving up because of difficulty.
Now, for whom is it worthwhile to optimise on this step of onboarding? When I started my last job, there was a Nix component to it. My colleague gave me a USB key with a “nix” label on it. A year later, it’s still this USB key I use when installing anew.
I’d recommend giving new colleagues a similar boost. For the greater public, perhaps DistroWatch or similar organisation might be able to lift this task. Distributing Linux USBs is such a niche market you’d better be doing something similar already.
I would like a NixOS branded USB stick. I don’t really care if it’s preflashed.
It’s nice to have an easy to recognize USB stick that has a dedicated use of installing OSes, so I don’t put any other files on it which I then have to contemplate and decide whether to back them up or delete them.
Preflashing them it is imho rather futile with how short the NixOS release cycle is. It’s better to tell novices to flash an up-to-date image.
As others already said, I think that the subset of NixOS users (potential or experienced) that find
curl https://channels.nixos.org/nixos-24.11/latest-nixos-minimal-x86_64-linux.iso -o nixos.iso
sudo cp nixos.iso /dev/sdc
too complicated is close to the empty set. The utility of such an installer would also be pretty limited considering it would be 100% stale by the time the USB stick is shipped to me.
That said, a USB stick with NixOS logo would be a cool gadget to own by itself: there’s no need to justify this project with usability.
@Kranzes you make some fair points but you strike me as needlessly hostile and unwelcoming in how you present them.
So there’s kind of three things going on here:
I agree with many people in this thread that (3) is a pretty small usability benefit, and certainly not where I’d put effort first if I was trying to make installing or using NixOS easier.
About 2, I’d say that if it makes sense to sell a NixOS USB for this, surely it makes more sense for the more popular distros to sell such things, so if they don’t, that’s some evidence that there wasn’t / isn’t a market for it.
Based on e.g. live usb - Where can I get an Ubuntu Installation CD/DVD/USB? - Ask Ubuntu it seems like Canonical used to sell an Ubuntu USB drive, and they don’t anymore. (Third parties still do, and indeed third parties could sell NixOS USB sticks if they wanted to, though they’d presumably not be able to use the official branding.)
Debian doesn’t sell them itself, but has a list of third-party vendors who do: Vendors of Debian installation media
I couldn’t find any official materials for Arch Linux based on a quick Google.
I don’t know quite what to conclude based on this, but I think the discontinued Canonical product is evidence that this isn’t a very important thing to be doing.
That’s a fair argument, but Canonical - and thereby Ubuntu - are more profit oriented than the average distro community. Given the nix consultancies work in and around the community rather than, well, being it, I think debian is a closer benchmark to NixOS, and getting in touch with their vendors may not be out of place.
That said, I don’t think many distro communities outside of the big three have ever directly sold physical media, and clearly fewer still do. Arch certainly never has, I imagine mainly because of point 3.
Regarding point 2, I think once you take into account the ever increasing download speeds (we live in an age of cheap, omnipresent satellite internet), and compare them to the comparatively minor incease in image sizes, pretty much nobody will be better off waiting for a USB to ship than to just run a download overnight. The resulting lack of demand probably means that only debian, who really care about meeting even the remotest accessibility concern, actually still care about providing support for 2.
Yeah I think a great next step for someone excited about advancing this proposal further would be to talk to existing vendors to try to get a sense of how much interest they receive.
Agreed with your list and the discussion of the merits of those ideas, but there’s another aspect as well, and that’s the part where an initial request / suggestion morphs into a direction of how people should spend their volunteer time.
That’s the part that has been met with the strongest reaction, and I agree that it can seem hostile, but want to be clear where that reaction comes from by noting it explicitly. Neither aspect is helping the situation.