[pdimitar]

Yeah, sadly true. While I am not a platform engineer I've witnessed their plight many times and I truly sympathize.

Now more than ever because I started making an effort to self-host much more than before... the amount of scripts I have to write just to achieve idempotency, nevermind immutability, is staggering, and I am already questioning my approach. Will likely start making use of ZFS or BTRFS snapshots, or I don't know, I'll just start snapshotting manually the entire filesystem on my Linux machines (like store all dir/file paths with their sizes and modification dates; it's a start and you can diff against such "snapshots").

I am just not comfortable with running commands and not having an idea what and where changed. It's insane that everyone is just accepting this! I am not okay with it, I want to see an exact breakdown on what changed and where and how.

IMO working on this and bringing it to the mainstream is loooong overdue.

[throwawaaarrgh]

I think it's that few people can see its potential. When I first started using immutable infra like 10 years ago, and saw how many problems it solved, my mind was blown. Until I saw the difference myself, it just looked like some trivial CS concept.

It's not apparent that problems X, Y and Z will be solved by immutability. Once it's applied everywhere, whole classes of problems just disappear. But until people see the problems disappear, they won't implement it. Catch-22.

[pdimitar]

True, plus not many devs are directly exposed to the problems and thus the will to fix the problem never has a chance to materialize.

One of the best-oiled teams I was in had devs and sysadmins work together closely. If Jim made a huge Python mess out of its small throwaway project (that the CEO needed because he wanted a nice chart for an investor meeting) that required several virtual environments and a particular (older) version of something then the sysadmin had the power to call him out and question his methods. While not many programmers appreciate that, those that do make for a more positive workplace IMO.

RE: idempotency / immutability in general, I heard about Nix many times but I have been put off every time I tried it: cutesy (and rather dumb) terminology like pills and flakes and such, a Haskell dialect the world really did not need, tight binding between things (forgot which at this point, sorry), and the list kept growing until I just gave up. With all their quirkiness and edge cases my scripts still beat the pants off of Nix for my own goals. I mean, pacman/yay have a flag that says "only install this package if not already installed" so... ¯\_(ツ)_/¯

But I really do want something like Nix (and no, not Guix either). Not only for packages -- for the entire system. I want to be able to plug an USB drive and issue a command that says "show me new devices plugged in the last 5 minutes, or last time I checked".

We don't have stuff like that. Or if we do, I am blissfully unaware of it. Can't we just start writing them and push their adoption? Every sysadmin team invents magic from scratch. Surely we can and should collectively do better...