Skip to content
HxHippy
Why My Dev Box Runs Fedora Now (and Arch Didn't Do Anything Wrong)
3 min read

Why My Dev Box Runs Fedora Now (and Arch Didn't Do Anything Wrong)

  • #Linux
  • #Fedora
  • #Arch Linux
  • #Workflow

If you've read anything on this site, you know I've been "Arch btw" for a long time. This fall, my main development machine became a Fedora box.

Arch is still in the house. My other machine runs it, and so does my airgapped system. I still like Arch. This post explains why it can't be the main dev box anymore. I'm not trading up from a worse distro.

Consistency was the requirement

A daily driver is a tool you pick for a job. The main dev box's job is to be the same machine tomorrow that it was today, so I can spend my time on the work and not on the box.

A rolling release changes under you all the time. Usually that's great. On the machine I lean on all day, it's a moving target I have to keep re-checking.

Fedora moves fast too, but it moves in releases. I know what version of the base system I'm on, I know when it changes, and I can read the change list before I take it. A machine that only changes when I choose to change it is a lot easier to trust when I'm on a deadline.

Arch can't give me that, and it shouldn't try. Its whole design is to ship upstream software as close to how upstream wrote it as possible, as soon as it's out, and let you build the system you want. That's exactly why people love it, me included.

The bonus: it lines up with RHEL

Fedora is upstream of RHEL. The package manager, the SELinux policy, systemd's defaults and the shape of /etc all line up with Red Hat-family servers. That wasn't why I moved, but when something I build has to land on a RHEL box, I'm a lot closer to knowing it'll work there.

SELinux on by default

On Arch, SELinux is something you'd build yourself, and almost nobody does. On Fedora it's enforcing out of the box with a maintained policy.

It already earned its keep. The display wasn't coming back from suspend: the screens woke up black. The cause turned out to be SELinux denying the driver's write to /var/tmp while the system was going to sleep. On a box without SELinux I'd never have seen that denial, and I'd have shipped code that assumes a write like that just works. Production would have taught me the hard way.

And if your servers enforce SELinux, your dev box should too. Otherwise you're testing against a policy you don't actually run.

What I kept

The move was about the base system, so my desktop and habits came along.

  • Hyprland is still the session. I tried the others, niri included, and my hands kept reaching for the same binds. I packaged Hyprland myself instead of using a COPR (that's its own post).
  • My keybinds came straight over from the Arch config.

What I left on the Arch side

The second machine stays on Arch. A rolling release is the right call when you want the newest of everything and don't mind keeping up with it.

The airgapped system stays on Arch too. That's all I'll say about that one.

What you can take from this

Pick each machine's OS by the job that machine does, and ignore the meme. Before your next distro hop, ask three questions:

  1. Do you need this machine to change only when you say so? If yes, pick something with releases.
  2. Does production enforce something your dev box doesn't? SELinux, AppArmor, FIPS, a hardened kernel. Turn it on locally.
  3. What do you deploy to? If you can match its family on the box where you write the code, that's a free bonus.

The switch had its share of surprises, and a couple of them could have hurt. I wrote those up in what breaks when you switch distros, and how I split the work across machines in one Linux per job.

Arch btw. Just not on this box.

// tags:
  • Linux
  • Fedora
  • Arch Linux
  • Workflow

Back to top