chrash0 @ chrash0 @lemmy.world

Posts

2

Comments

179

Joined

2 yr. ago

as someone who used to work on “expert models” i’m excited that not everyone has abandoned them for “what if we just had a model that knows everything (that doesn’t exist) and costs a billion dollars to run”

honestly, the brag document is a great takeaway. so many times people will subtly or implicitly question documentation or refactors, and when shit hits the fan and the readability refactor or documentation or logging/tracing PR becomes the clincher in quickly resolving the issue i love calling it out. documenting those cases themselves as a way to sell documentation and code quality to others feels like an amazing idea (if not exactly a “brag document”)

weird. i’d probably good around for drive checks personally, any test should be pretty definitive. how did you install Discord? the AUR? maybe a reinstall is in order.

i don’t think this is Node specific. if i were writing logs, i’d want to be pretty sure they get written, either by creating the directories necessary, putting them in a place i was pretty dang sure existed, and/or crashing as early as possible if the location didn’t exist, which makes me think it’s a weird Discord config error.

error is right here:

    
EROFS: read-only file system

  

this line is concerning:

At random times

this sounds like a drive failure. Discord’s logger can’t write to the filesystem; it’s reporting as read-only.

what’s the drive like where /opt lives?

three, maybe four things:

1. as mentioned: Obsidian. i pay for Sync cuz i like the product and want them to succeed and want reliable offsite backups and conflict resolution. use a ton of links and tags. i’ve been into using DataView to make tables of IoT devices, services, todo items, etc based on tags and other YAML frontmatter.
2. chezmoi. manages my dotfiles so my machines are consistent. i have scripts that are heavily commented that show how to access MQTT, how to read and parse logs from journald, how to inspect my network, etc. i do think of them as code as documentation, even if they’re also just convenient.
3. NixOS. this has been my code as config as documentation silver bullet. i use it as a replacement for Docker, k8s, Ansible, etc as it contains definitions for my machines and all the services and configuration they run, including any package dependencies and user configurations. no more statting an assortment of files to figure out the state of the system. it’s in flake.nix
4. honorable mention to git and whatever git hosting provider is not on your network. track your work over time, and you’ll thank yourself when things go wrong.

some things are resistant to documentation and have a lot of stateful components (HomeAssitant is my biggest problem child from an infra perspective), but mainly being in that graph mindset of “how would i find a path here if i forgot where this was” helps a lot

in addition to what others have said, i’d say a lot of civil infrastructure—hospitals, clinics, government facilities, etc—are locked in either because of bad politics or weird vendor lock in. my dad ran his own dental clinic, and he had to run a Windows server because it was required by his software vendor that did everything from appointment reminders, to the web portal, to billing, to showing which of your teeth were missing, to integrating with scanners or other equipment. it was shit software that looked like Windows 3.1 well into the 2020s, but it did the job and 24hr support was reliable. just an anecdote, but as a software engineer i was fascinated by it.

it’s not stupid. i have pretty successfully done some NixOS work flying basically blind with an LLM guiding the way.

1. ask follow up questions. “can you show me in the docs where this is defined”, “why did you add this line here”, etc
2. you’re going to have to understand this config eventually. the LLM will start to get confused if you’re trying to squash a weird bug and you’re just chastising it. it will always tell you you’re right even when you aren’t.
3. document everything with comments and in git
4. Caddy is better :P

definitely. Qualcomm provides the SoC and drivers for what comes on that package, but you’ll want to add a battery controller, power controls, and other embedded systems onto the motherboard to make it act like a real system. it’s also a way different boot process in my experience than a normal x86 platform. the difference between ARM and x86 isn’t just the instruction set. plus at this level nothing is ever plug and play.

as for how Valve was able to ship an ARM device, they stuck to the normal kinds of IO a mobile device with a SD8gen3 would have and already have a great OS for fast iteration that they have tight controls over.

i’m excited for this XElite line, but i can see how it’s not in Qualcomm’s best interest to spend their engineering labor on porting to desktop Linux, not with Microsoft and Dell etc already having bids on that time. as long as Qualcomm is upstreaming and not actively blocking open source development, i don’t understand the kind of resentment i see for them. because they work with Google? i see them becoming more open as they become more prolific outside of embedded systems and Android. i see it as an exposure problem.

he’s been salty about this for years now and frustrated at companies throwing training and compute scaling at LLMs hoping for another emergent breakthrough like GPT-3. i believe he’s the one that really tried to push the Llama models toward multimodality

i don’t understand. don’t they operate in one of the largest Linux platforms around, Android? if you mean they don’t support your desktop wifi chipset or publish modules for their SoCs, then i guess that’s fair to say. but i think a deeper integration with Linux can only be a good thing. i guess my perspective on Qualcomm is colored by the fact that i worked with them briefly on an embedded project, have seen their docs, and have booted their dev kits into a full Ubuntu environment.

sucks

(but also maybe yay for Linux on ARM?)

this is amazing. i might have to try it!

i know JetBrains owning the whole Kotlin developer experience is kind of the point, but as it stands i’d rather not use Kotlin than go back to IntelliJ these days. it’s not worth the context switch just to use an otherwise pretty great language.

honestly, where NixOS shines for me is in my homelab. i don’t always have time to fully document what i’m doing, but my NixOS config is code-as-documentation for when work burns all of my memories away and has a git log and conflict management so i can manage multiple systems that share common config.

and once you find out you can have services run on systemd with syntax like services.jellyfin.enabled = true you’ll never want to go back to containers, although it has ways to manage those as well.

it’s overall a great OS for tinkering and deploying small services across small networks. not sure how it scales, but for my use case it’s damn near perfect

i guess the point that’s being missed is that when i say “hard” i mean practically impossible

my point is that it’s hard to program someone’s subjective, if written in whatever form of legalese, point of view into a detection system, especially when those same detection systems can be used to great effect to train systems to bypass them. any such detection system would likely be an “AI” in the same way the ones they ban are and would be similarly prone to mistakes and to reflecting the values of the company (read: Jack Dorsey) rather than enforcing any objective ethical boundary.

but what are the criteria? just because you think you have a handle on it doesn’t mean everyone else does or even shares your conclusion. and there’s no metric here i can measure, to for example block it from my platform.

what about the neural networks that power the DSP modules in all modern cell phones cameras? does a neural network filter that generates a 3D mesh or rather imposes a 3D projection, eg putting dog ears on yourself or Memojis, count? what if i record a real video and have Gemini/Veo/whatever edit the white balance? i don’t think it’s as cut and dry as most people think

it’s already the case that the distinction between what’s “AI” and what isn’t is a subjective, aesthetic difference and not a technical one

modifier for window manager nav and general OS controls like wofi/rofi

pretty sure it’s SteamOS, an Arch Linux derivative, on a fairly popular Snapdragon platform. probably not too difficult to hack on it.