bme's recent activity

  1. Comment on Has anyone ever used NixOS as daily-drive distro? in ~tech

    bme
    (edited )
    Link Parent
    What does that workflow look like? Are you using something like vscode dev containers? The thing that I really really like about nix is how it combines configuration and code in a single closure...

    the rest I run in Docker

    What does that workflow look like? Are you using something like vscode dev containers? The thing that I really really like about nix is how it combines configuration and code in a single closure which once you get the hang of it is difficult to imagine living without. The snippet below is a heavily redacted snippet of what I have to install my shell (fish). It is effectivly a function that takes many arguments one of which has to be a package set, which returns a config map which causes

    1. fish to be installed
    2. fish greeting to be disabled
    3. creates a fish function (similar to bash alias) which calls terraform. The ${pkgs.terraform} string interpolation syntax expands to the input-hash addressed path that terraform is stored under and installs terraform at the same time. The beauty of this being that in order to refer to the path of a package it has to be installed. Configuration + closure. This snippet will never be missing it's dependencies, even if terraform somehow isn't available on the path. For funsies I also include a bit cross platform stuff. I run linux at home and travel with a macbook. I use the same set of nix expressions to manage both. isDarwin checks which platform I am on and selects the right way for me to get my md5 hash.
    { pkgs, ... }:
let
  md5 = if (!pkgs.stdenv.isDarwin) then "${pkgs.outils}/bin/md5 -qs" else "md5 -s";
in
{
  programs.fish = {
    enable = true;

    interactiveShellInit = ''
      set fish_greeting
    '';

    functions = {
      "ta" = {
        description = "Apply terraform";
        body = ''
          ${pkgs.terraform}/bin/terraform apply /tmp/(${md5} "$PWD").plan $argv
        '';
      };
    };
  };
}

    I'll do another one: have you ever just wanted to refer to a script and not wanted to care about naming it? I do all of the time, and nix frees me from having to care about it. Another redacted snippet, this time setting up the MFA process to pull OTP codes out of a nitrokey for AWS. This generate my .aws/config file and installs both aws-vault and pynitrokey and a small script that tidies up the nitrokey script output. I don't care where the script lives, and with nix I don't have to name a path, it's just input-hashed. It's also guaranteed to exist as long as it's referenced by the config. Win win win.

    { config, pkgs, ... }:
let
  credentialProcess =
      "${pkgs.aws-vault}/bin/aws-vault export --pass-dir=${config.xdg.dataHome}/v --backend=pass --format=json default"
  getOtp = pkgs.writeText "get-otp" ''
    set -l totpout (${pkgs.pynitrokey}/bin/nitropy nk3 secrets get-otp aws 2>/dev/tty)
    # get-otp outputs two lines, the otp is on the second line
    echo $totpout[2]
  '';
in
{
  home.file.".aws/config".text = ''
    [default]
    region=eu-west-1
    credential_process = ${credentialProcess}
    mfa_serial = arn:aws:iam::XXX:mfa/XXX
    mfa_process = ${config.programs.fish.package}/bin/fish ${getOtp}
  '';
}

    I don't know about you, but my computer is littered with this stuff, and now it's trivial to evolve, track and rollback from (switching config takes about 10 seconds typically). I can't see myself going back to anything else that doesn't at least bear a passing resemblence to nix.

  2. Comment on What the hell is a Typescript or: Creation ideas above my skill level in ~tech

    bme
    Link Parent
    I don't know what the right way to push the project forwards for you is, but this sounds like an emminently feasible project. Especially if you have put thought into the design / ui elements /...

    I don't know what the right way to push the project forwards for you is, but this sounds like an emminently feasible project. Especially if you have put thought into the design / ui elements / user stories, and have some semi-formal description of how scoring should work, or at least what the objective of scoring should be and what data it should be derived from. As others have already suggested you just need to find one talented-enough dev that already likes art fight and is stoked on pixel clash.

    If you ever want any architecture/code review if you move the coding side of it along shoot me a dm, and if I can help I will (creds: ex-google / ex-investment bank distributed systems guy (currently doing startup things)). Sounds like a fun project.

    2 votes

  3. Comment on What the hell is a Typescript or: Creation ideas above my skill level in ~tech

    bme
    (edited )
    Link
    To answer the question on cope: hakuna matata. Life is too short to master everything, so there is no point in worrying about it. Either accept it's a multiyear project and what will be important...

    To answer the question on cope: hakuna matata. Life is too short to master everything, so there is no point in worrying about it. Either accept it's a multiyear project and what will be important is doing a little bit consistently, or drop it. Everyone over estimates what they can do in the short term and under estimates just as badly what they can achieve with steady application on a longer timescale.

    On the "just get a dev to do it": this is only gross if you have no skin in the game. You've already invested a significant amount in doing. You'd continue to be doing, even if you grabbed a dev to join you on the journey. I read the wiki page on Art Fight. I am mildly curious (but not enough to do more research myself). This looks really trivial to scale out: 2 teams, keep a running total. There's a CRDT for that. The only complex part seems to be scoring attacks. How is that done today? I'm way too busy to be able commit to implementation, but I'd be willing to do some light mentoring on how to structure the thing.

    Lastly, on Art Fight being too big for a hobbiest open source competitor. The wiki page says 400k people participated in 2024. Assuming that scoring isn't obscenely expensive then this could totally run on a pretty limited budget. Assuming every participant never slept and interacted with the site every 10 seconds, 40k requests per second can be had on a single boring mid-tier VPS. Assuming scoring is chunky it's still going to be multiple orders of magnitude less. 10s per second, (assumes someone submits art once per 12 hours)? Ez.

    EDIT: it's self-scored???? Ok. 400k users could be handled pretty easily, and if you wanted to make it work for millions, sure. I'd say getting people to use it would be harder than making it usable. I make no judgement on UI complexity, that's not my bag.

    8 votes

  4. Comment on Has anyone ever used NixOS as daily-drive distro? in ~tech

    bme
    Link Parent
    This depends on entirely how you use your computer. Do you mostly use things out-of-the-box? If so, I'd agree, this probably doesn't sound particularly strong. If you take the "system-crafter"...

    This depends on entirely how you use your computer. Do you mostly use things out-of-the-box? If so, I'd agree, this probably doesn't sound particularly strong. If you take the "system-crafter" approach of investing in customisation with the long view in mind you might accumulate much more state. I run various bits of nix on all of the computers I use, and it is far far far far cheaper to replicate a few kb of text than gigabytes of full system backup. I also use nix for work, so if I get issued a new corp laptop I spend zero time setting it up.

    If you are a dev then it gets even more compelling in terms of reproducing evironments sharing scripts / packages across operating systems etc. If you don't have any problems it solves then it's probably just niche for niche's sake, but I'd say anyone that uses a computer for long enough will one day wish they had a succinct way of capturing the essence of the state of their computer without the attendent cruft. Nix is that.

    1 vote

  5. Comment on What advantages does Linux have over other operating systems? in ~tech

    bme
    Link Parent
    I guess I don't really see these as kernel modules, because if any program living in userspace that interacts with the OS via a systemcall interface is a "kernel module" then reasonably it seems...

    macOS 10.15 or later enables developers to extend the capabilities of macOS by installing and managing system extensions that run in user space rather than at the kernel level.

    I guess I don't really see these as kernel modules, because if any program living in userspace that interacts with the OS via a systemcall interface is a "kernel module" then reasonably it seems like every program in MacOS could be labelled as a kernel module. I get that given the hybrid nature of the mach kernel this is perhaps unfair, but I only mentioned this as a counter to the idea that you are getting to peek behind the curtains. I'll concede it's not a particularly strong point :)

    1 vote

  6. Comment on What advantages does Linux have over other operating systems? in ~tech

    bme
    Link Parent
    You can't develop kernel modules for (modern) MacOS: anything you want to do has to run is userspace. We were definitely all forcefully reminded recently (crowdstrike) though of how possible it is...

    You can't develop kernel modules for (modern) MacOS: anything you want to do has to run is userspace. We were definitely all forcefully reminded recently (crowdstrike) though of how possible it is to write windows kernel modules.

    Windows has logs

    It is my general experience that the system event viewer on windows is barely usable compared to the breadth of information available in journald, in part because I don't think (but happy to be corrected) you have anything like the linux kernel logs directly interleaved and filterable with respect to application level logs. Admittedly it's been a long time since I developed anything for windows, so I am happy to be corrected. It's also true that typically everything I have loaded in my userspace has source available so drilling directly to the source of some interesting log message is always possible. However I'll concede the for the average consumer of PC hardware it's pretty unlikely that this leads to actionable insight. On the other hand it does lead I think to a far more readily available group of people who have direct knowledge of how the sausage is made, and I think that's where a lot of this "you can't fix windows problems" mentality comes from. I don't have direct access to anyone in Redmond. Vendors of most prop software put up insurmountable barriers to getting to talk to someone who actually knows what's up. Just not true in the world of FOSS. You either can interact with the devs, or if you have the skills you can just fix it yourself.

    I mean... https://www.youtube.com/watch?v=iyA9DRTJtyE&t=397s watch 10 seconds of this video (at the timestamp provided). This is where this perception comes from, imo. AMD has no such advice for linux, because all of the scheduling is already just working, and if it wasn't AMD could (and does) contribute fixes to the kernel to make it work.

    This whole exaggeration of "nothing is possible on Windows and MacOS" is not helpful in these topics.

    Agree.

    2 votes

  7. Comment on What advantages does Linux have over other operating systems? in ~tech

    bme
    Link Parent
    But advanced user of what? I know exactly what you are saying (and you're not wrong), but I really wish these frustrated "advanced users" would take a step back from time to time and realise how...

    advanced user

    But advanced user of what? I know exactly what you are saying (and you're not wrong), but I really wish these frustrated "advanced users" would take a step back from time to time and realise how complete silly their expectations are. No violin player, no matter how advanced they were at violin would throw down a trumpet with intense disgust if they weren't able to play it immediately at the same level as their violin, or at least if they did everyone would call them crazy. Sure, their sense of pitch, and understanding of musical tone is transferable, and sure: the musical notation is identical, so they have a head start over a total novice. Everything else needs to be learnt again. This confuses no one. And yet 9/10 someone tries linux who has ever used some other operating system the first thing that comes out of their mouth is going to be a complaint that can be replaced roughly with "I had to learn". Well no shit, Sherlock. You have to decide if what you are getting is worth the effort, but there will be effort, and you will need to go back down the hill your are on to get to the top of the mountain yonder.

    5 votes

  8. Comment on What advantages does Linux have over other operating systems? in ~tech

    bme
    Link
    Headlines No global registry No weird "oh my computer seems to be getting slower and slower, I guess it's time for a fresh windows install" No completely arbitrary user-hostile offer you can't...

    Headlines

    1. No global registry
    2. No weird "oh my computer seems to be getting slower and slower, I guess it's time for a fresh windows install"
    3. No completely arbitrary user-hostile offer you can't refuse bullshit like "now you have to use a ms accounts and sign in online" stuff.

    Detail

    1. Ok, maybe if you are using gnome really heavily then you could argue gnome settings is kind of like the windows registry, but that's a stretch. For the most part applications configure themselves in an isolated way. You want to to know where the settings are for $app, well look in ~/.config/$app if it's well behaved and if not read the docs / man pages and at absolute worst open it using strace and see what it's reading on boot. Once you've located the config it's very likely just a text file. Nice! No need to go dumpster diving in some sprawling barely documented global database where one wrong move can nuke your system.
    2. Due to 1 and "package managers" it is trivial to query your computer for everything that's ever been installed on it, and spring clean whenever you fancy. If you are blessed enough to use a rolling distribution you can keep the same install for as long as you'd like. I think the last time I installed fresh linux on the computer I am writing this on was 10 years ago. It's undergone many kernel upgrades, hardware transplants, and even distro changes (home on a separate partition (ok btrfs subvol) ftw). I just keep pulling the latest packages, no stress. Aside: opportunity to shill my current love: nix. You want to keep your computer fresh? Try a distro which forces you to name every bit of software you have installed and will delete everything that isn't covered by that config file. Your computer will will exist in a perennial state of cleanliness. People have taken it really far, but I have not gone quite that extreme yet. It's quite freeing: I install stuff all the time to try it out without sticking it in config, knowing that at some point it's just going to get nuked. If I find myself coming back to it, I add it to my config and back that up. Now if the worst happens it will take me ~ 15 minutes of investment and probably an hour of waiting to have the computer. Are there drawbacks: yes, you are niche of a niche, so it's not for everyone, but I can't see myself using anything else that doesn't offer this feature (completely reproducible machines from config).
    3. This is fairly obvious, and has been pointed out by others: when the users are the developers, and the profit motive isn't there, you get much better alignment. Linux remains simple enough (despite some people's concerted effort) that there are a whole slew of different ways to interact with the kernel from login, to window manager, to application suites. It makes it very hard to hate your users and have anyone use your stuff.
    4 votes

  9. Comment on Distrohoppers, what's your flavor this week? in ~comp

    bme
    Link Parent
    The best way to get to the inner workings of nixos I think depends on what type of person you are. If you are a programmer type, I really recommend doing something like installing nix on whatever...

    NixOS

    The best way to get to the inner workings of nixos I think depends on what type of person you are. If you are a programmer type, I really recommend doing something like installing nix on whatever distro you currently have and messing about with the repl. Before going any further, I want to say that I deliberately ran all of these examples on a box running arch linux + nix. You can just dip your toes in! Let me show off something dumb: I want to package a shell script, but not only that, I want a function that given someone's name produces a package that installs a shell script that greets them, and I want to figure this out on the fly

    ~ ben@lamorna
❯ nix repl
Welcome to Nix 2.15.0. Type :? for help.

nix-repl> :l <nixpkgs> # load up nixpkg repository (bringing `pkgs` into scope)                           
Added 17839 variables.

nix-repl> hai = name: pkgs.writeShellScript "hello tildes" ''
          echo Hai ${name}
          ''

nix-repl> haipkg = hai "geniusraunchyassman" # apply function to yield a package

nix-repl> :b haipkg # build my package

This derivation produced the following outputs:
  out -> /nix/store/9r73ihnlybzxbnyaiy6ij4sxxnk352xz-hello-tildes

nix-repl> 

~ ben@lamorna
❯ /nix/store/9r73ihnlybzxbnyaiy6ij4sxxnk352xz-hello-tildes
Hai geniusraunchyassman

    Maybe that isn't compelling. Maybe you like experimenting with different versions of a package that doesn't match what is on the system, but you'd like the install to be native and not conflict with what the system is providing. Ok, here is me wanting to get a version of fish that doesn't match the version in nixpkgs:

    ~ ben@lamorna
❯ nix-prefetch-url --type sha256 https://github.com/fish-shell/fish-shell/releases/download/3.5.1/fish-3.5
.1.tar.xz
path is '/nix/store/6zisgncm6j3m4cnjnjck4i84mcc57qpy-fish-3.5.1.tar.xz'
0a39vf0wqq6asw5xcrwgdsc67h5bxkgxzy77f8bx6pd4qlympm56

~ ben@lamorna
❯ nix repl
Welcome to Nix 2.15.0. Type :? for help.

nix-repl> :l <nixpkgs>
Added 17839 variables.

nix-repl> fishy = pkgs.fish.overrideAttrs (old: rec {                                                     
            version = "3.5.1";                                                                            
            src = pkgs.fetchurl {                                                                         
              url = "https://github.com/fish-shell/fish-shell/releases/download/${version}/${old.pname}-${version}.tar.xz";
              sha256 = "0a39vf0wqq6asw5xcrwgdsc67h5bxkgxzy77f8bx6pd4qlympm56";                            
            };    
          })

nix-repl> :b fishy

This derivation produced the following outputs:
  doc -> /nix/store/3skdvch8sk970d0l6lr9dn9gjpfnafmy-fish-3.5.1-doc
  out -> /nix/store/qnn7gp4q6la4ikgi3bw9g59ix10va7xp-fish-3.5.1

nix-repl> 

~ ben@lamorna
❯ ls /nix/store/qnn7gp4q6la4ikgi3bw9g59ix10va7xp-fish-3.5.1
bin/  etc/  nix-support/  share/

    Now in both of the above cases, I'm just building stuff in the store, which to be usable just gets symlinked to discoverable places by slightly higher level tools. There are many linux distributions, there are even multiple distributions that give you atomic roll forwards and backwards package sets, but nixos and guix stand alone as being supremely hackable by dint of creating a composable package abstraction which

    1. Makes it nearly impossible to miss a dependency (complete)
    2. Makes it nearly impossible for packages to clash (isolation)
    3. Exposes existing packages in a way that makes them open to extension either by composing them or by overriding parts of them in a principled way

    Nix / NixOS is not without its downsides, but everything else seems basically crippled in comparison once you climb up the cliff into productivity.

    2 votes

  10. Comment on How IoT betrays us: Today, Sonos speakers. Tomorrow, Alexa and electric cars? in ~tech

    bme
    Link Parent
    I think something that has ruined a lot of things that used to "last" is the insertion of software everywhere. Remember when cars mostly had compatible slots for aftermarket radios? Then we had...

    I think something that has ruined a lot of things that used to "last" is the insertion of software everywhere. Remember when cars mostly had compatible slots for aftermarket radios? Then we had external GPS navigation, then it started to be integrated. No more swapping out the tech in the centre console. Now you need a new car if you want an upgrade, and now many fancy new things have arrived which were never available externally (adaptive cruise control etc). Driving a car even a few years old now means substantial features are missing and can't be acquired after the fact.

    8 votes

  11. Comment on How IoT betrays us: Today, Sonos speakers. Tomorrow, Alexa and electric cars? in ~tech

    bme
    Link
    Specifically on the Sonos front I am never buying another product of theirs again, and in general I have stopped buying anything that needs an internet connection to a component that I can't host....

    Specifically on the Sonos front I am never buying another product of theirs again, and in general I have stopped buying anything that needs an internet connection to a component that I can't host. I haven't got an answer for everything yet, and on the multiroom audio front it's disappointing to see traditional speaker companies follow Sonos' lead with closed ecosystems. There is no multiroom equivalent of line-in or toslink. There is playfi but it's basically dead as far as I can see.

    My next startup after current obligations expire is going to be pluggable multiroom sound and video on published standards. It's not a hard problem. There are multiple DIY solutions already, i.e. SnapCast, it just needs someone to invest in a business around it with some pucks a la chromecast audio (discontinued). Shift the brains onto a server which you could sell too and the pucks at least would be good for the life of the silicon (presumably a DAC + microcontroller + wifi reciever). Publish the protocols and hopefully grow a decently compatible ecosystem.

    7 votes

  12. Comment on <deleted topic> in ~tech

    bme
    Link
    I think this is probably a metaphor for how almost all software is trending: towards a lowest common denominator, most useful for most people. For those hold-outs that that can do more with tools...

    I think this is probably a metaphor for how almost all software is trending: towards a lowest common denominator, most useful for most people. For those hold-outs that that can do more with tools that require investment there will always be options and alternatives. I run plex and keep all my media locally and with an offsite backup. Do I think this is best for most people? Nope. I do it because I like dicking around with computers. For people that want files, there will always be NAS systems and that jazz.

    This isn't even unique to software, I bet every activity under the sun has a control / convenience trade-off.

    17 votes

  13. Comment on Please tell me what you think about this idea for a text editor/Linux Distribution combo in ~comp

    bme
    Link Parent
    Completely agree with void / alpine recommendations. I haven't run alpine but void basically leaves you with nothing, it's trivial to manage the ttys.

    Completely agree with void / alpine recommendations. I haven't run alpine but void basically leaves you with nothing, it's trivial to manage the ttys.

    2 votes

  14. Comment on Please tell me what you think about this idea for a text editor/Linux Distribution combo in ~comp

    bme
    (edited )
    Link Parent
    Eh, I dig what this guy is trying to do. Often a barrier is enough. Do you still lock your door? You can make it pretty annoying to escape the jail if you fancy it: Install the stuff, get the...

    Eh, I dig what this guy is trying to do. Often a barrier is enough. Do you still lock your door? You can make it pretty annoying to escape the jail if you fancy it: Install the stuff, get the chain all working for some locked down user then run a shell that changes the root password to some unknown random 20 char string and reboots the computer. Congrats, you've made it reasonably challenging for yourself, which is often enough to get yourself into the flow of things.

    @mrbig: If I were you i'd look at customizing something like bspwm to basically launch the editor of your choice and simply don't launch anything else, have root own the config and make it read only. change the login shell to something suitable neutered. I guess you'd also need to kill all the ttys. Change root password to something else and that's a low-effort lock down that should be good enough for what you are trying to do. For extra fun: consider pairing a rock64 or pi with waveshare epaper display or something like that.

    5 votes

  15. Comment on What are some startup scripts you have on your daily driver? in ~comp

    bme
    (edited )
    Link
    I don't think most peoples innovations are driven by imagination, they are driven problems that they look to solve. What currently bugs you about your workflow? Is there a collection of windowing...

    I don't think most peoples innovations are driven by imagination, they are driven problems that they look to solve. What currently bugs you about your workflow? Is there a collection of windowing tasks that you regularly perform? Would it make sense to automate them? Do you manually place your applications into windows / workspaces on startup?

    I'll give you one example of a typical problem:
    You have some environment variables relevant to a given project, which only need to exist within the project folder. This is easily solved with direnv. Ok fine. What if you want those values to encrypted because they are secrets and only decrypted on demand? I use pass so naturally I came up with the following: I will encode collections of env vars into yaml files that I will keep in my pass directory. I'll write a script to decode them and stick it in my ~/.direnvrc. So in my ~/.direnvrc I have the following function:

    pass-export() {

  [[ -n "$1" ]] || { >&2 echo "Must pass pass arg"; return 63; }
  
  while read -r key; do
    key=${key#'"'}
    key=${key%'"'}

    read -r value
    value=${value#'"'}
    value=${value%'"'}

    export $key="$value"
  done < <(pass "$1" | yq 'to_entries | .[] | .key, .value')

}

    and in a .envrc file I might have

    #!/bin/bash

export AWS_DEFAULT_REGION=eu-west-1
pass-export work/outw/web/aws-keys

    which exports my secret keys for a given aws project so they can be used with terraform or aws cli or whatever. This is typical of my collection of random bits and bobs. Search is another regular problem. Learn how to wield things like fzf well and you'll get around your computer with ease, but it can be inserted in all kinds of places. For instance if I have more than one git remote I want to be prompted to pick where to push an upstream branch for the first time otherwise just send it! For this I use a little git alias:

    [alias]
	p  = !just-push-it

    and a dumb script:

    #!/bin/bash

if [[ -n "$(git for-each-ref --format '%(upstream:short)' $(git symbolic-ref -q HEAD))" ]]; then
   exec git push 
else
   exec git push -u "$(git remote | fzf -1)" HEAD 
fi

    The script basically does the right thing if there is just one option and if there are multiple it gives me the choices piped into a fuzzy finder and then does the right thing once the choice is made. This pattern can be repeated all over the place. The fzf wiki is full of them.

    The point of all this is that efficiency doesn't require imagination, it only requires the will to be lazy enough to learn how to not do things. I don't have time to not be intentionally improving my workflow. I doubt many people do, and yet somehow they make time for it by refusing to learn the skills they need to stop wasting time.

    4 votes

  16. Comment on Plain Text Accounting | Double-Entry Accounting with Plain Text Files in the Command Line in ~comp

    bme
    Link Parent
    I used to use it a bunch, but while I didn't mind much of the classification work, the thing that I really needed to do any couldn't find the time for was automating the feed from my bank and...

    I used to use it a bunch, but while I didn't mind much of the classification work, the thing that I really needed to do any couldn't find the time for was automating the feed from my bank and implementing some simple statistical learning for matching up recurring payments.

  17. Comment on <deleted topic> in ~comp

    bme
    Link
    I think this is an interesting one. I think one of the most obvious self serving shifts on the web came from Google. When it lacked the ability to meaningfully parse out natural language to feed...

    I think this is an interesting one. I think one of the most obvious self serving shifts on the web came from Google. When it lacked the ability to meaningfully parse out natural language to feed it's knowledge graph there was a big push to get everyone onto microformats. Now that it has the tech to skip that the push is gone, leaving anyone else without that muscle out in the cold. Just one more aspect of the web being weaponized for material gain I guess, and a relatively small transgression in the larger scheme of things.

    5 votes

  18. Comment on Reinventing Home Directories in ~tech

    bme
    (edited )
    Link Parent
    That's super cool of you, but I'll pass. I intend to "retire" early by paying off my house in the next couple of years through a mix of contracting gigs and investments, and then go full time OSS....

    If you're serious about this, I'm happy to try to help guide you through finding a job in the linux desktop space. Red Hat is the obvious one but working at KDE is also a very good bet if you want to work to improve these things. Canonical often has open jobs as well and they do some good work. There's many ways to be paid to contribute.

    That's super cool of you, but I'll pass. I intend to "retire" early by paying off my house in the next couple of years through a mix of contracting gigs and investments, and then go full time OSS. I don't yet know which projects I'd like to contribute to.

    I can't help with your systemd issues…

    That's the thing, I'm not asking for help. I solve my own problems. Eventually I solved them completely by ditching systemd entirely where possible, and limiting usage of it's features where I can't (I run a lot of rhel / centos boxes).

    The feature would work as intended, what you put into it doesn't.

    This is I guess where I completely disagree with the kinds of UX choices systemd makes. Lets be concrete: this particular issue was a fun confluence of things: I had a working socket activated user service of gnupg. Arch at some point decided they would also provide that service. Conflict! Fine. Whatever. What I take issue with is how this manifested: I'd login, then 30 seconds later I'd get logged out. What. The. Fuck. After scrabbling around for 10 minutes I put on my thinking cap, logged in via a tty, worked my way through journalctl and figured it out. What would have been much easier is if systemd had just let the service fail and left me to notice my gnupg setup was broken when I went to use it, and I'd have had the luxury of my whole DE to investigate it. This and other papercuts too mundane to remember eventually got me off arch on my personal boxes.

    I know you can't randomly know the skills and abilities of random people on the internet but I've written hundreds of systemd unit files and know many of the man pages off heart. Systemd has weird failure modes and fails in more cryptic ways than just about any other supervision suite. I will freely admit I push more edges many people would (I am a distributed systems engineer with a strong SRE / ops bent). I'd take runit or s6 any day of the week.

    3 votes

  19. Comment on Reinventing Home Directories in ~tech

    bme
    (edited )
    Link Parent
    The big problem that I have with many of Poettering's projects is that he is able to do what he does because the people that came before him made simple text / file based tooling. Some of it not...

    I wish the Linux community would shut the fuck up about this guy because this approach is the only one that has even a hope of continuing to compete with modern operating systems.

    The big problem that I have with many of Poettering's projects is that he is able to do what he does because the people that came before him made simple text / file based tooling. Some of it not great, but in any case, it is easy to extend most of linux because it's often just a few files and maybe some shell scripts or whatever. His projects on the other hand basically pull up the ladder behind them by creating sprawling interfaces that stick their fingers in everywhere. I have a problem with this because it will make it very hard for us to ever escape this local minima if something better rolls along. His works complete disdain for modularity is maddening.

    One of the things I find massively frustrating is that my time and talents are wrapped up solving other problems. I'm super jealous of his 40 hours a week that he gets to spend creating software that I often have to actively avoid. I shit you not, I am actively trying to achieve financial independence s.t. I throw my hat in the ring on this stuff because I have completely lost faith in the Poettering-led vision for the end state of linux userspace.

    I guess I should also add that I don't care much for

    compet[ing] with modern operating systems

    whatever that is supposed to mean. I want to be able to enjoy my computer without it puking all over me. I lost many hours to systemd (oh your user service failed to start in 30 seconds, I guess I'll log you out of your session, and other such nonsense). And it's not the man, it's the code. PulseAudio has definitively made things better. Systemd is a mess. If he can find someway to design something s.t. that any other project has a cat in hells chance of using it without dragging in the entire jungle I'll be pleased to see it, but I won't hold my breath.

    Lastly:

    Or join one of the couple that are led by similarly-cranky people.

    It's a real dick move to suggest that everyone that is dissatisfied with systemd is a crank. There are plenty of reasons to not want to use it, and you undermine everything else you say by mixing in such commentary.

    14 votes

  20. Comment on <deleted topic> in ~games

    bme
    Link Parent
    I'm pretty annoyed. I'm not installing EGS though.

    I'm pretty annoyed. I'm not installing EGS though.