WYGIWYG

  • 0 Posts
  • 494 Comments
Joined 1 year ago
cake
Cake day: September 24th, 2024

help-circle



  • LOL, probably.

    I was Linux on the desktop at work for over a decade, changed jobs, they forced me to go Windows or Mac. Mac infuriated me with it’s half Linux half locked down BS, so I went windows, WSL came out and I was relatively pacified, kept Windows for a couple more years after I changed jobs again. Finally said WTF and tried Debian, bookworm just worked, steam, laptop, dual video cards, Got the hardest windows only software running in wine, worked around the mandatory no stuff. I ran it for 6 months, got bored. Looked at immutable. Watched someone on peertube doing a live sesh working on a mastodon or lemmy cluster in NixOS, fell down a rabbit hole.

    Now, by default I have nothing installed I don’t need immediately, I have nix shells for yt-dlp, qemu, video editing, I need something it gets installed just for that shell and then it’s gone. Almost any package I could want is available. I clone my home directory and a couple config files, I can move my entire setup box to box. configs are stored in git.

    But then, updates are scary. A lot of my packages are in the hands of people who didn’t create them. Probably 90% of the stuff I run is just a .deb file unpacked by someone who wants to run the app in Nix and decided to share the work. One of my core modules got stale after an update last week and it too me most of a night to figure out why it took me 10 minutes to boot.

    It’s complicated enough I never feel like I understand it all. It breaks just often enough that I’m forced to learn more about it. It has a few super-powers that make me not want to swap it out with something else.









  • The NixOS, it callllssss usssssss

    come in they ssssaid, itssss delcarativvvveee they ssssssaid.

    Wait i just put an environment variable in conifguration.nix and moved home manager back out of my home folder to a central spot why does sddm take 5 minutes to give me Wayland now?

    edit: OMG 6 hours later and I have it working. I have a configuration.nix that i re-grew with my 2025 backup and a configuration.nix.slow that is still broken if i switch it out. SDDM timeouts all over the place

    the diff between them give 0 indication why sddm would fail.

    I kinda want to go back through line by line and find out what did it, but I kinda also want to sleep, eat and go to work in a few hours :)

    edit: edit: no rest for the wicked. I ran it through Meld, and there was very little there. Best I can tell, my home manager was synlinked to the wrong config in the store. I’m running it modular, so the nixos-rebuild “should” have moved its configs. The defunct home manager somehow broke QT6 and I lost my file/edit menus in qt apps, the fix for that was a template override env var in configuration.nix. When i fixed the borked home symlink, that failure stopped being a failure and the QT override somehow gave SDDM heartburn. I hadn’t seen it because I rarely change home manager, and whatever was wrong sat that way since 25.11.

    Removing the line for QT to ignore the template stopped SDDM/Portal from loading and crashing for 5 minutes straight.


  • It depends.

    A 2-5 year-old laptop, you want to web browse, maybe watch some videos, use google docs or open office, you probably never need a terminal

    If it’s a really new laptop or you want to get the most out of video drivers and push it harder, you’ll probably need to be ready for some light terminal crap. Gets a little janky if you have a dual-video-card setup. Nothing hard to handle, but if you’re not looking to have to handle anything…

    I think the numnber of available packages is better on the Debian side. Mint or Kubuntu run newer hotter stuff, debian runs older more stable stuff.



  • Not accounting for any interesting custom choices you made under the hood, the default file browser for both os’s use libfuse3 for MTP. My point is, it shouldn’t have crashed, there are open issues in libfuse3 for possible crashes, so you might just have hit one at the wrong time, but at that, it REALLY shouldn’t have f’d over your journaling filesystem enough to keep you from logging in. A breaking read/write to fuse should not have been able to f your journal over beyond a simple automatic recovery. Most of the design choices in Linux over the last decade have been made specifically to prevent that kind of thing from happening on a healthy system. One can argue that one distro is more stable than another because they take, or refuse to take newer packages, but for your specific issue, they use the same piece of software under the hood.

    The wipe and new OS might have just moved the problems to a less visible area.

    My primary anger with Wayland is the security issues that broke AHK that they’re just now considering. There’s been lots of finger pointing over the years, but now that most OS’s are ditching X11 support all together, we’re going to see a lot more compatibility coming in the next year or two.



  • Time->TLS errors aren’t handled well anywhere.

    As critical as they are to 2fa and TLS, you’d think every OS out there would poke around a few time servers and scream bloody murder if the time was off.

    Honestly, I think we, as a society, have leaned a little too hard into time as a precise critical failure point. It’s fine for things like GPS that actually require it. but our clocks don’t need to be precisely the same to tell how recent a request and response are and we can certainly make better hashing algos


  • Oddly enough, giving the general public exact error messages ends up costing you in support and reputation.

    They obscure the messages because the inexperienced masses start digging up red herrings. Knowledge to someone with zero experience causes a lot of confusion.

    The experienced and capable users look up the codes and think about it for a minute, check their vpn, maybe a health dashboard, maybe reboot.

    Just about every complex machine out there give error codes instad of real messages, even when they have large displays capable of telling you exactly what the condition is.