Creating a bootable NixOS installer USB stick is one of the most technical parts of getting started with NixOS, and it requires already having a machine at hand. So being able to buy official USB sticks with a recent installer would be a great step towards making NixOS a more user-friendly distro.
I’d honestly assume that if burning an ISO to an usb was out of reach for a users current skill level, editing a NixOS configuration in the functional and very odd Nix language would also be something they couldn’t do.
At least until we had a easy to use graphical tool for configuring your system, that could at the very least enable/disable packages and options, and turn it into a correct config.
Then again, maybe just doing a graphical install, and then installing packages imperatively is a place to get a foothold, and then it actually makes a lot of sense that having sticks available could be useful… But in that case I’d probably prefer to advice just installing Nix as a package manager, since being unable to change system configuration sounds highly limiting to everyday use, and may mean people abandoning NixOS (and by extension, Nix) due to the high friction.
Maybe I’m wrong, it’s a while since I started using Nix, would love perspectives on this, specially from new users.
As a new user I can say that we rely mostly on youtube. Several youtubers explain how to use nixos. Also the scariest part wasn’t burning the usb. The scariest part was formating the disk during the graphical install
And new user I much more prefer torrenting linux isos rather then downloading them
Also selling usbs with burned linux iso is a widespread way of supporting the distro financially
Don’t worry, I’ve got ideas for making NixOS much easier to configure .
I honestly suspect that it’s not worth it at all as a source of finance, especially for an entity that sells nothing so far (you know, all the work around taxes, etc.) It might even be cheaper to give them away.
In fairness, I first used NixOS like 7 years ago or so, so I am far from a new user but I definitely would not have picked this as the greatest challenge when first installing NixOS. I thought creating a live USB for NixOS could be achieved through the same apps as one uses to do this for other Linux distros. Last time I installed it on my physical hardware it could be. Has this changed? If not, I don’t see how it’s so challenging, there’s so many apps available to do this. Personally, I just use Fedora Media Writer when I want to create a live USB for a Linux distro. Either that or the dd
command.
Good for you. It’s one of many challenges. This one in particular has almost zero value for the user to learn (because they probably won’t ever need to do it again), and is basically insurmountable for anyone who isn’t already well versed in Linux. And the number of ways to do it is irrelevant for how difficult it is to do, or how difficult it is to figure out how to do it, or even what particular jargon to search for to figure out how to do it.
They can just follow the instructions in the NixOS Manual.
Why not make it easier? I don’t understand. This is the marketing category; I thought making it easier for beginners to get started would be an obvious step towards making more people use NixOS.
I don’t wish to completely dismiss the idea, but I expect that physical handling and delivery is both harder and slower than downloading the file and copying it to bootable media. I can’t really imagine a scenario where I would order, buy, and wait for such a thing to arrive by post. I’m also rather sure I wouldn’t trust what arrived without another means of verifying signatures / hashes, at which point it’s just about as fiddly as copying it to usb in the first place.
However, the handling and delivery challenge also suggests a scenario where it might work better, because those aspects are minimised: at conferences and other similar gatherings, perhaps some university computer clubs.
Our installer images are generally kept reasonably small, to minimise the initial download and because often updates are needed soon after anyway, so much of the image is already stale for install. I suspect the smallest usb stick worth buying these days is probably something like 32G. Around the time of a release, a larger distribution image could be made, containing a much larger set of packages precached, to be distributed physically. Perhaps the image on the device could be live-bootable and update that cache before redistribution again.
I can imagine that sort of thing might be useful for some situations, I’m much less sure it’s a viable to sell as a useful revenue source for the project.
There are marketing vendors out there that handle the process end to end. I have a couple of branded drives that I got from trade shows.
BTW, as an example, Trisquel sends out a branded thumbdrive as part of membership.
Yes, but I was speaking more as an end user: If I want to install nixos now, I’m not waiting for orders and delivery. These things might work at shows and conferences, as free handouts, and perhaps in some other cases that I expect probably are rare or at least specialised.
A repeatable way of generating relevant contents, and perhaps seeding the image via torrent for a release, might be a worthwhile project.
Yes, but I was speaking more as an end user: If I want to install nixos now, I’m not waiting for orders and delivery.
Oh, absolutely. I also doubt that outside of “Internet in a box” situation there would be a lot of users that would be onboarded using a pre-burnt thumbdrive. Just because of the delay between “I want to install OS” and getting the usb key in my hands. Running cat iso > /whatever/usb-path-drive
is pretty doable.
A repeatable way of generating relevant contents, and perhaps seeding the image via torrent for a release, might be a worthwhile project.
Some discussions here.
That sounds like an excellent idea. The millions of users with slow connections would probably be very happy for the sneakernet.
Can we start being realistic and bring up some data about potential new users selling a USB stick will bring?
Side note: I’m really not interested in having to maintain nixpkgs for people who can’t even install the distro on their own, I don’t want silly issues on GitHub.
Allowing a vendor to write the USB sticks implies a lot of trust.
NixOS (or nixpkgs) simply updates too frequently for USB sticks to be useful. Unlike, say, Debian, whose stable releases are set by two years, NixOS gets a new stable release every 6 months that users are expected to upgrade to as soon as possible. That also isn’t mentioning the backported security fixes. It would be a neat novelty item, but not very useful to the user long term.
But I think the installation process for NixOS is something worth looking over again. The Calamares installer afaik is quite old and mostly serves to create auto-generated skeleton configs and partition the disk imperatively. This doesn’t teach the user how to interact with a NixOS system in modern day (see: flakes) nor what the expectations are (being able to understand how to write in the Nix language).
Not only are those unrelated skills, one is a regrettable and avoidable chore, and the other is something the user actually benefits from doing regularly. The vast majority of NixOS configuration doesn’t need to be any more complex than the equivalent JSON file, and could pretty easily be abstracted into UIs like Geospatial Nix.
Good point, and that’s another part we should improve. There’s no reason we can’t do both.
The NixOS Foundation already sells merch. But giving them away is a great idea! Another two ideas:
- Send USB sticks to donors above a certain level for free.
- Let people subscribe to get USB sticks either on a schedule or every time there’s a release.