nixos-infect VS rfcs

There is one solution I've seen being used to solve this issue.

It is to overwrite the current Linux OS with the one you want. I came across this idea here [0]. I researched and got Alpine Linux running on Hetzner (even they don't support custom images) using a similar method [1].

This seems to be the guide to do the same with Arch Linux [2], I'm not sure though.

Once you do create a successful Arch image on OVH, take a snapshot of the machine before installing anything else, in case you want to start from a fresh Arch image in the future.

[0] https://github.com/elitak/nixos-infect

[1] https://wiki.alpinelinux.org/wiki/Replacing_non-Alpine_Linux...

[2] https://wiki.archlinux.org/title/Install_Arch_Linux_from_exi...

Related:

https://www.reddit.com/r/NixOS/comments/u5gi6e/whats_a_good_...

https://github.com/elitak/nixos-infect

In my experience, the initial learning curve is "way steep" due to incomplete learning resources, the community being split on flakes (I'm not jumping on them yet because I'm sticking with the stable branch), and this one documentation gap in particular.

FWIW, I recently found a VPS offering for $1.41/month (!) @ 1.5GB RAM & 30GB HDD via https://lowendbox.com/, at https://my.racknerd.com/index.php?rp=/store/black-friday-202... (please note I have no idea how reliable it is though!). I managed to deploy NixOS there through nixos-infect (https://github.com/elitak/nixos-infect), and then further configure it with NixOps. That said, using NixOps does currently require a Linux (or Mac, probably) box as the managing one, and some Nix-fu, which is definitely non-trivial. A draft (WIP) writeup on that, if you're interested: https://github.com/akavel/scribbles/blob/main/_drafts/202308...

Those Linode instructions are about installing an OS from an installation ISO from Linode's rescue mode. -- I'd consider that more a 'plus' to Linode that you get to be able to install whatever unsupported Linux on Linode's VMs. But, of course it's not as smooth as the officially supported images.

Sibling comment mentions that NixOS provides ways to build VM images (e.g. https://github.com/nix-community/nixos-generators has a generator specifically for Linode).

Another option is to use nixos-infect, which will replace a Linux distribution with a NixOS distribution. https://github.com/elitak/nixos-infect

> At that point, I feel like all of the repeatability gains are gone. If I want to spin up a fresh server, I have to read a guide and set stuff up by hand?

Just as with ansible, "git pull to deploy". You'd keep a copy of your NixOS configuration.nix somewhere else, and would be able to apply it later.

> it really does let me just get a new VPS and deploy to it very fast

Using a tool you're familiar with to get the job done is going to be faster than learning to use a tool you're unfamiliar with.

I believe much of the use of NixOS is for hobby stuff, and for personal machines. (c.f. "In what environments do you use Nix", Development (1242) vs Home Server (845) vs Production (386) https://discourse.nixos.org/t/2022-nix-survey-results/18983).

for NixOS

Now that you both mention it though, I probably could've skipped the "Boot into the installer" step. I know nixos-infect can install in-place. I havent checked how they do it, but a lighter version that just adds nixos should be pretty easy using ZFS. It probably would've been possible for me to do replace the installer steps with something along the lines of:

You can execute nixos-rebuild on other distros but the NixOS modules and activation scripts expect it to run on NixOS system so it probably will not work (e.g. NixOS would want to control systemd, which would conflict with the distro). And if it did, you would no longer have the other distro but NixOS (see also nixos-infect).

You may consider this view biased, but we have this: https://srid.ca/nixos-mod

* September 2023: The "Nix Community Survey 2023" is looking for gender data, and the mods don't like that most contributors are men.

* November 2023: The moderation team tries to institute a Code of Conduct https://github.com/NixOS/rfcs/pull/114 ... and they get their way

* November 2023: Some are not happy about it: https://discourse.nixos.org/t/moderation-team-accountability... -- the moderators talk about their "authority" and of course lock and hide the thread. It's "disruptive" and "off-topic", you see.

* This sort of activity continues -- moderators consolidating and increasing their power, citing how they need the power to control "concern trolls" and such -- and now in April 2024, we get https://save-nix-together.org/

The "anonymous contributors" want to drive out the NixOS founder entirely, so that _they_ are in charge. They want "to hold people accountable for bad behaviour at all levels" and lament having "responsibility without authority" - in other words, they want power, power, power. They want power over everyone. Their justification is that they believe they have the moral high ground, and they deserve to lord it over everyone else.

Hold onto that hard power, Eelco, and tell this lot to fork the project. Let's see how they enjoy moderating noxious.org instead of nixos.org

https://old.reddit.com/r/NixOS/comments/1ceiz36/thoughts_on_...

And the RFC to improve the situation:

https://github.com/NixOS/rfcs/pull/175

> (after eelco ignored the PR for quite a while, also!)

Clicking that link takes us to a PR that was opened on 2024-02-02. The initial response from the Nix author comes 7 minutes later. Puck has multiple back and forths with other members Github, but her next interaction with the Nix author comes the next day on 2024-02-03. This is also the first time in the conversation where she "reminds him ... to even read her PR message". There's a second interaction later that same day during which she does similar, but it's worth noting this is pointing to a different message and appears to be less a "reminder to read" and more re-iterating what they feel is their argument against the Nix author's own arguments. Puck then continues to have back and forth with other commenters but as of today, there has been no further comments from the Nix author after 2024-02-03, and no further comments from Puck after 2024-02-08.

This hardly to my mind qualifies either as "having to remind him multiple times to even read her PR message at all" or "after eelco ignored the PR for quite a while, also!" So as I said it's a fairly weak claim, and feels more like a "bastard eating crackers" reaction to the PR than an actual showing of poor behavior.

As for the "Meson example", I didn't ignore it. As I stated in my comment, I had at that point read two of the referenced discussions in detail, and thus commented on them. I didn't comment in the "Meson example" for the simple reason that I hadn't read it.

I have read it now, and equally find it confusing.

1) The claim in the letter is that the proposal has "passed RFC, for five years", yet the RFC itself only appears to have been opened 2022-08-24. It's been a while since grade school for me, and I'll admit COVID has warped all our sense of time, but I'm pretty sure 2022 is not 5 years ago.

2) The first completed working implementation of the change doesn't appear to have been done until 2023-01-18 (https://github.com/NixOS/rfcs/pull/132#issuecomment-13874661...). Again this is much less than 5 years old.

3) On 2023-03-20, the author of the PR for this change states:

> the RFC has made it past most of the early stages and the current goal is to achieve parity with the current buildsystem before replacing it.

(https://github.com/NixOS/rfcs/pull/132#issuecomment-14768433...)

Again, this doesn't seem to fit at all with the claim that the proposal has "passed RFC, for five years"

4) On 2023-11-01, the Nix author themselves asks for updates on the RFC implementation, an action which doesn't seem congruent with someone who is willy nilly single handedly blocking things and being a disruption to the process. And the author of the PR states:

>the main block is actually a lack of free time for the main devs!

(https://github.com/NixOS/rfcs/pull/132#issuecomment-17890770...)

This doesn't seem to point to evidence that the Nix author is single handedly holding up this process.

5) On 2024-03-21 the PR author notes:

> currently working on adding support to build nix-perl, waiting for assistance

(https://github.com/NixOS/rfcs/pull/132#issuecomment-20135356...)

Not to sound like a broken record, but if the issue isn't finished as of a few weeks ago, it can hardly be considered to be held up by the Nix author for 5 years.

I agree that one of the links in the open letter is to a comment on a PR from 2019, which is indeed 5 year ago, and does indeed contain the Nix author commenting that they are skeptical of the change because "he doesn't know meson but knows his own build system". But given that there's an entire wealth of history on the topic since then, including progress on the feature that appears completely unobstructed by the Nix author and an open PR that is a mere 3 weeks old for a current implementation, I find myself again unconvinced of this rampant bad behavior on the part of the Nix author. And I reiterate again that these complaints are very weak and don't do much to support the open letter at best, and act as contrary evidence at worst.

Again there might be other context to be had that is missing, but if one is going to write a massive "open letter" complaining about bad behavior, I expect the links in that letter to point to actual bad behavior, and or provide the relevant context necessary to show how what appears to be normal dissent is a passive aggressive continuation of obstruction. I have to assume the links one provides in an open letter is their strongest evidence, and if this is all the authors have... I am unconvinced.

Nix with dynamic derivations (RFC92) could potentially beat this curse.

https://github.com/NixOS/rfcs/blob/master/rfcs/0092-plan-dyn...

See: A plan to stabilize the new CLI and Flakes incrementally https://github.com/NixOS/rfcs/pull/136

NixOS recently introduced "problem" infrastructure to deal with such problems more gracefully and explicitly:

https://github.com/NixOS/rfcs/blob/master/rfcs/0127-issues-w...

For some more context: Flawed as they are, Flakes solve a large number of problems Nix experiences without them. This is why I, and presumably many others, use them even in their current experimental state.

An RFC was recently accepted to commit to forming a plan towards stabilization of Flakes: https://github.com/NixOS/rfcs/pull/136

Personally, I don't believe there won't be any breaking changes, but I also believe that the stabilization of Flakes is still a ways away and hope that there will be a reasonable migration path.