cafkafk

Reply

biblically accurate ace chaser

Found this on the trash site, and I want you all to suffer with me :)

The #nixos SC unanimously voted in favor of adopting home-manager as an official project.

It's worth stopping to appreciate what this means. We are much more than a distribution for servers. Even if we're also the best at servers!

statement that works out of context for anything

I love this post andrewkelley.me/post/my-thou...

Andrew: "I don't have any personal criticisms of Jarred"

Also Andrew: except he's a stinky manager, with poor communication, unrealistic expectations, and no empathy; who was writing slop before he even had LLMs to blame it on... (cont)

francis fukuyama, 2026

And the maintainers took that at face value and had a nuanced and respectful discussion about the topic

Reply

I think you can love computers deeply and use language models. I love computers deeply and I enjoy talking with my computer about my computer.

Most programmers didn't love computers before LLM. LLMs only made them more visible by multiplying everyone. That put them above the baseline to be seen.

if you're 20-30 & your agent council isn't discussing:

- Tokenmaxxing
- AI psychosis
- Wet Claude
- Goatse of gnosis
- Desideratic AI

Then it's time to find a new council.
Your TOKEN SPEND is your GNOSTIC ASCENT!

Nix(OS) is such a mind-breakingly good idea and no one understands that it literally makes you a demi-god it's so hard to understand the scale of the power it gives you to people that have only tasted vercel

I love how the UK having a government crisis is universally believable at this point LOL

Tim Duffy @timfduffy.com

I also tested a version where I made it clear that I was testing credulity, and that 90% of presented prompts would be real, and asked for explicit credences. Here, Gemini disbelieved almost every story despite being told the high base rate, and Opus was in the middle of the pack

I kinda want to add all the vibeslopper names like claude and chatgpt and copilot etc as git coauthors on my organic code just to be trifling

dude i really don't want to keep glazing AI stuff but it's just like when the haters are so delusional it's just so tempting. Feels like I have the moderate opinion it's really useful but falls short sometimes but apparently that's like an ai psychosis extremist take at this point

Sorry but the point is just policy like this is just bullshit.

It effectively just allowing closeted LLM users as long as they keep to "don't ask don't tell" about their usage.

Like yes, you win internet points. Probably not a good choice for that community long term.

NixOS 26.05 “Yarara” is here #nixos.

20442 new packages, 85 new modules, and 1547 options!

Systemd is now the default Stage 1. x86_64-darwin support ends completely in 26.11. GNOME is updated to 50. GCC is upgraded to 15; and more.

Read the rest here:

It's honestly so good that we rejected flakes in NixOS back in 2019. I think if we'd approved them they'd still have been experimental to this day.

Lowkey there is something cynical about a take like this from an account only following one other account.

Buried the lead, likely so they could posture more, but actual policy includes:

> No autonomous AI agent use or vibe coding
> No use of AI to generate substantial pieces of code

So you're allowed to use AI, but only "non-substantially" and no "vibecoding".

Unfalsifiable review criteria. Ouch.

Godot Engine @godotengine.org

For years we have been overwhelmed by the volume of code contributions we have to review, especially from new contributors. AI has made the problem much worse. That's why we are taking steps to reduce the burden on maintainers while still welcoming new contributors: godotengine.org/article/cont...

Changes to our Contribution Policies – Godot Engine

Godot is growing fast, so here is how we are dealing with the huge increase in contributions.

Fil-C is memory safe if you think software that crashes is "safe"

Shoutout to Arch Linux for helping recruit so many new NixOS users

fuck math conjectures I'm using Claude Mythos to make Garlic 2.0

frontier models own benchmarks be like

turgut keskintürk @tkeskinturk.bsky.social

data visualization is my passion.