dinckel

Just to clarify for everyone else:

• Nixpkgs is the equivalent of a core distro repository for NixOS. Instead of Core + Extra + Multilib + AUR on Arch, we just have Nixpkgs that has everything
• A Nix “module” is essentially an app with pre-packaged declarable options. So rather than just installing something as is, you can use its options to declare how you want it installed. I can get firefox, with a package definition for firefox-nightly, and also tell it which addons I want bundled, for example. 99% of the time, this is the preferred way to do things.
• Home-Manager is a third-party Nix module, that lets you declare stuff in ~/. Very convenient for shells, browsers, and whatever else you want in there. Let’s say you want to have a specific shell, with a specific ssh signing key, and whatever prompt you like. The home-manager daemon would build it for you with a new system evaluation.
• Stylix is another third-party Nix module. For supported things, it will style your apps with the preferences you’ve defined. Handy if you want a uniform look and feel for the things you use

This is a good example of what people consistently overlook/misunderstand, when it comes to Nix.

Obviously you can remount a /home, or just pull the dotfiles from a personal repo, but the strength of Nix is also in that I can re-create my entire config exactly how it is defined. If i were to setup a machine completely from scratch, with a mature enough config, it will get me from 0 to my exact desktop completely unattended.

But there are also many more advantages to it, at least in my eyes. Let’s take trying/tweaking new packages as an example. Yesterday I pulled an old repo for an Outer Wilds mod. The thing needs a dev environment, and a mod manager for the actual game. A nix shell got me both, I finished my work, and when I exit out of fish, both are gone, just as I wanted them to be.

Another good example would be partial os updates. I’ve used Arch for almost 9 years before switching to Nix, and pretty much a top3 Arch rule is not doing partial updates, or partial rollbacks. In case of a breakage, I would have to manually redownload an older version of a tarball, pacman -U the package, and then hope i’m not cooked. In the case of gcc incompatibilities, it can quickly become a massive pain in the ass. My nix flake would never experience this problem, because I already have two different scenarios available - either i build based on an older lockfile from my git repo, or I create an overlay for a specific input I need, so that it still pulls what it needs, and doesn’t interfere with the rest of my system

Correct. Atomic distros don’t apply the update, unless it is ready to be applied successfully all together, usually with an option to restore the previous state, without the need of something like btrfs snapshots.

With Nix(-OS) as an example - your bootloader entry is just a reference a giant list of what you need to get out of the Nix store, to achieve the config you want. Many of those can coexist in the same system as a result, including different versions of the same package

This setup won’t really teach you anything different in relation to containers though.

One time I’ve lost around 200gb of data, by accidentally removing a folder, instead of its symlink. Didn’t have backups either, but it wasn’t anything I couldn’t get again

Mass layoffs suck. Yet again it’s corporate mismanagement, at best.

But in all seriousness, what was even their role there? It’s a glorified social media platform with an occasional job application. They cant possibly need that many people

Sarasa Gothic + Iosevka for just about everything

I love OSS but I won’t sacrifice my experience just to go fully libre. Sometimes it just doesn’t make sense. I’m glad it’s an option for people who do want that though

For games you’re playing through Steam, no additional steps are required, apart from enabling Steam Input support in the controller settings. Your pad should be detected automatically

That makes a lot more sense now. Love me a project with a fun story behind it

That’s the good part. There’s plenty of choice, and it’s easy to swap

What is happening to GNOME is truly one of the biggest fumbles in OSS. They could have just continued improving things, but instead choose the path of most resistance, refused to commit to any logical strategies for further improvement, and are now stuck in a loop of nothing getting done

That’s kind of the point though. One of the foundational pillars of a good distribution is mature package management, and that includes not relying on self-updaters that will pollute your system with untracked files

I have mine on a rPi 3b+, in a docker container. Works great

The repost is a week late, but yeah

I feel like it’s just a matter of time, until they pull the rug from under lifetime subs.

But in any case, this is probably it for me. I’m not completely happy with jellyfin performance on my server, but the price hike puts me outside of what i’m willing to spend for this service. I already host it myself, and i can tunnel it myself too, if i ever decide to run it outside of my home network

Are you trying to say that error 43 isn’t the only thing you’d ever need to know about it?

The Microsoft support forums are on a whole level of their own, when it comes to being useless.

sfc /scannow and the troubleshooting button in the settings do fuck all, and compared to the usual systemd journal, the event log rarely gives you any useful information whatsoever

If you know how to use VB, using Virtmanager wouldn’t be too different. You’ll have to find how these are packaged for your distribution, but the instructions for everything else on the wiki will be helpful regardless of the distro

You can keep your existing partition, and just mount the disk in a libvirt/qemu on kvm setup. Here is a good place to start

about:blank when you open a new tab