Interesting, thank you for that link! I was also thinking about bootc as a candidate but that seemed unintuitive to me so I went with ostree for now. The current implementation needs some refining but definitely is a better start that I imagined, quite happy how it turned out.

Regarding the quotes, I’m brainstorming if there’s a way to, I wanna say “re-route” the pacman commands to edit the package list so you use the pacman commands like usual but they get applied as if you use the gui. It’s definitely possible but I’m thinking about how to get that to actually work in a clean way.

My intention behind that statement is that you aren’t forced to use say flatpaks only and under the hood the update mechanism is still pacman even tho you don’t call it directly, does that make sense?

I appreciate it! I’d be especially interested in knowing if there are any locations/directories that the user expects to be carried through deployments or to be writeable that get deleted/overwritten during updates or are read-only in the first place. Finding that would perhaps also fix the issues with some AUR packages not wanting to install. Ideally you can e-mail me your findings to the address on my Codeberg profile and I could write on the code during the vacation and actually test in a VM on my main PC once I get back.

Regarding the platform to discuss on, someone else asked a similar thing today so I’m just pasting my reply here “Hmm, I’m undecided. Me personally, I couldn’t handle running a forum due to my autism, that would burn me out instantly. Having one entirely community run with me occasionally lurking in like I do here would probably be the only realistic option.”

That’s nice to hear! My personal feedback and the feedback from people that have tried it so far seems way better than I anticipated, noone seems to have any big issues. A little more choice during the install, a little less weird messages during updates (nothing critical, just looks odd) and some AUR packages that don’t install in a way that conform to the Arch guidelines (which is more on them than on the OS) and that’s about all the negatives.

I’ll be on vacation for a week and only take my underpowered Thinkpad with me so I’ll be working on some little qol changes here and there to bring a more polished version in a few short weeks. If noone finds anything big that needs fixing I might even be able to justify bumping the version number of the next release to 0.1.0 instead of 0.0.2 since it’s way more stable than I dared to expect.

I was just about to go to bed so apologies for the short answer, I can elaborate tomorrow if you wish. The home folder, including your settings/configs will survive, there’s no difference in the way you organize files to a regular Arch install. The partitioning is preset and non-customizable (for now at least). So no need to worry, you simply install it with the hardcoded presets and use it like you would use your current install! I will likely set up a kind of wiki page to make this more sorted and readable than the current bloated drop-down menu so I can cover this in more detail in the future.

Edit: aah, I overread the first time that you have a home folder that you keep. In your case I would back it up on another drive/cloud first and then try one of the installer options that fit you the most. I should get the option for manual partitioning up and running, but if you don’t do it it exactly like the installer expects, it will fail, which is why I only defined three pre-set options.

Yeah, Linux distros using something like Discord always seemed very odd to me. I mean okay, if you cater to gamers then I guess it makes sense, but you won’t catch me doing that. Lemmy is the only social media I have anymore, the rest is just braindead and bot-infested.

No, you are pretty cool!

Yes, it does come preinstalled with KDE as implementing a proper desktop selection is planned but would take even more time. This is Arch however, you can uninstall everything you want. Throw KDE out and install whatever desktop enviornment you want, you still keep the kernel and the ostree update model, that’s what’s so cool about it!

Yeah, you are right, I’m not the best at explaining things, I’m sorry. I look at it like literally Arch but the updates are atomic. You can choose to install or uninstall whatever you want, but it gets written with an A/B system during updates so you can always have a fallback. It does come preconfigured from the start, because that was easier to develop for now, but you can treat this like any other Arch install, just with safer updates.

Thank you! I was wondering why noone has tried that yet and now I know why, but hey, it works now. I hope you will have fun with it!

Well, it were four 80 hour weeks and I “only” reverse-engineered the way Fedora makes their distro atomic with ostree but adapted to Arch. I did try to stay true to the Arch philosophy of keeping it minimal otherwise so it wasn’t that big of a project!

Edit: Sometimes I had to spin up Qwen 3.8 27B because the error codes the installer threw out were literally humanly unreadable so I had no clue what even went wrong but other than that it was by hand lmao

To put it in simple terms: when you update your system, you will have two deployments. One deployment stays exactly the same, the copy will be updated. Until you restart you stay on the original. When you restart, the system automatically attempts to boot the updated copy. If it fails, it automatically loads the original again.

That way the updates are atomic in the sense that you either boot into a fully updated system or simply back into the untouched original, nothing in between.

Edit: if you look up how Fedora Silverblue works, it’s pretty much the same, just applied to Arch!

I’m excited to hear your feedback! Keep in mind, some things might be buggy, but please notify me and they will be taken care of.

Hell yeah, let me know how it goes!

Thank you! In a perfect world I would have loved to name it “Arch Atomic”, but I’m not affiliated with them so I went for a wordplay on atomic with my favorite isotope!

There are a lot of benchmarks for the CachyOS kernel and it’s mostly single digit performance increases for “normal” tasks but a noticeably more snappy feeling. The mayor increases come in compiling or rendering workloads, there it can easily get into double digits. For example on my 270K, compiling this kernel takes 23 min with the stock linux kernel, 20 minutes with the linux-cachyos kernel in -v3.

I’m making my own distro rn, announced it a week ago here and will likely release today or tomorrow! https://pu-239.org/ https://lemmy.ml/post/52307735

Nice comprehensive overview of why this piece of slop called a “distro” and its backers should be boycotted, thank you! Gives me even more motivation to finish up mine today.