💾 Archived View for splint.rs › arch_sanity.gmi captured on 2024-03-21 at 15:09:31. Gemini links have been rewritten to link to archived content
-=-=-=-=-=-=-
Arch Linux often does what Ubuntu tries and fails at.
I started with Arch due to a simple problem; every so often I'd read on some cool new program I wanted to check out, then feel disparaged as I read over the instructions.
# apt key add some-guy.gpg >> /etc/apt.d/unknown_repo # apt install lib-blorb3 lib-xyz gnome-desktop kitchen-sink # git clone https://github.com/rando-guy/cool-project # cd cool-project # make # sudo make install
In the early days of the command-line, I hadn't a clue what I was doing, and the cheery articles never elucidated much. It felt even worse when they explained the commands, because I didn't understand, but had to read an extra few pages just in case they contained necessary instructions.
Then, every article, without exception, had this bit:
Installation for Arch Linux:
yaourt -S cool-project
Every.damn.time.
After some errors, and reinstalls, I eventually made a bash script to install the i3-gaps desktop on Ubuntu. The script ran badly, and when Ubuntu updated, the script needed manual intervention. So in the end, it saved me no time at all.
At this point, I finally realized Arch really would be easier. Someone else should be making and maintaining all these installation scripts, not me!
So I did a bit of reading on 'what is this Arch business?', found all the Arch community comments where they told anyone who dared to ask a question rather the sacred Arch wiki to go and hang themself, and noted the 'noobs forum', for simpletons had a whole load of high-level, difficult questions. Despite this, I installed the damned OS, and indeed installations after that were easy.
And I kind of get the community ire. Clearly, the nerds on Arch go a bit overboard, but I've had someone ask me to explain something, repeatedly, while refusing to read the things I read, or even read their own error messages. It can really sap the energy, and that energy was meant to go into helping someone solve their own problems, not to baby them.
Here's the thing that perplexes me; official installation instructions don't always have any suggestions for updating. Businesses which need `kubectl` to control their k8 clusters just have to make a script to install it on this or that, or rely on `docker` for everything. Once again, this ends up recreating what's on Arch with extra steps.
Ubuntu's narrative seems to be something along the lines of 'stable OS, with well-tested packages'. ...and then people ignore the existing repositories in order to install the latest tarball from the source. And companies do this too.
This decision does not universally come from excitable engineers who want some bling. It comes because `terraform` and `docker` and whatnot rely on remote services, and having an old version can cause more problems then the recent ones. But if we're all agreed on loading the latest version of packages, why bother to install Ubuntu?
I don't see what's achieved. Do they want to make sure that `ls` and `top` aren't too bleeding edge? Certainly having an OS which was well-tested with `terraform` can't be doing too much of the work, since Ubuntu in fact updates rather a lot of the OS. And besides that, whatever tests may be performed on the latest `k8` package to ensure it's ready could equally run on Arch.
All this means of course is that I don't understand. Perhaps this is a matter of habit for old engineers, who always wanted to use Ubuntu for physical servers as it provided stability. Most likely this workflow has good reasons I'm not seeing.
And finally, to clarify, I wouldn't really demand Arch on a production server.
...but I would be curious to see it.