gkfasdfasdf

But how does it do on the balls benchmark???

https://benjdd.com/languages/

I love it. Reverse minesweeper describes it perfectly.

The latest version of Claude Code is the compiled with the Rust version of Bun, I've been using it and haven't noticed any regressions.

In addition to an assert, a 'nonnull' attribute on compilers that support it seems like a good idea.

The lib is a joke I know, but these will absolutely get your prototype greenlit.

Having used both Anthropic and OpenAI models at $work via copilot extensively, I have to say GPT 5.5 currently is best at getting work done with minimal mistakes. However, Claude Code is way ahead of OpenAI Codex in terms of harness features and tooling. MCPs, skills, sub agents, these all were pioneered in Claude Code first. Perhaps that contributed to Anthropic's success.

Odd they used GPT-5.2 and not GPT-5.2-codex. i.e. the one optimized for coding agent tasks.

Someone should fork yt-dlp and rewrite it in Rust...and have it use bun exclusively

Yes, it’s actually similar to discriminating based on race or religion, in the sense that it’s an arbitrary, meaningless criterion to discriminate on. If the Rust Bun port is better in every measurable way — passes all tests, has the same performance or better, and fixes existing bugs — then who cares what language it’s written in or how it was implemented? The point is that it’s higher quality. If you don’t trust the Bun team when they release a Rust version and give it their stamp of approval, why did you trust them when they released the Zig version two weeks ago? It makes no logical sense, and it makes the yt-dlp devs look foolish.

Idiotic. Why do they care what language it's written in? If it works it works.

Putting the markdown in the URL, that is very clever!

Looks wonderful, is there a skill or prompt that can teach agents how to use this format?

Ghostty split panes - coding TUI in one pane, neovim in the other, maybe a third pane for shell.

Someone needs to go to prison for this.

The grok button on twitter is pretty awesome. Instantly summarize / explain any tweet, even memes, including replies. Ask follow up questions. Not sure many people know it's there.

Also grok in the Tesla is fun, get answers to questions without looking at a phone. I once had it search up a blog post and read it out to me while driving. The NSFW mode is pretty...disgusting so I leave that off.

I hope they find a way with Optimus or something. FSD is incredible. More competition is a good thing.

I would love to know how they built this. Did they use json-render [0], openui [1], or rolled their own?

[0]: https://github.com/vercel-labs/json-render

[1]: https://github.com/thesysdev/openui

> But here's where it gets tricky

Pretty sure this comment is AI

Tesla FSD will change lanes when you use the blinker. It will also accelerate and remain engaged if you press the pedal, e.g. if you want to coax it forward at an intersection.

Congrats on the upgrade - what exactly did they upgrade and what version of FSD are you running? Hardware and software definitely matters. My FSD experience applies to 2026 Model Y, latest FSD (v14.x).

> And to add insult to injury, even GM Super Cruise is widely renowned as better and safer than Tesla’s current “FSD”.

Do you have any sources for that claim? I can attest that current iteration of FSD is very, very good, and very likely is a safer driver than I am. At least one major insurance company agrees [0]. I don't have any experience with Super Cruise though.

[0] - https://www.lemonade.com/fsd

Tesla FSD is awesome. I use it almost all the time now, it feels safer than me driving. It's like having a private chauffeur. My disengagements are mostly nav related.

See also https://skills.sh

Can you elaborate? What is it about Haskell that makes it better?

TIL Nix flakes work on macos - is this a legit alternative to homebrew?

Interesting trick about the constant value, and thank you for the detailed write up!

Is it possible to use Fil-C as a replacement for valgrind/address sanitizer/leak sanitizer? I.e. say I have a C program that does manual memory management already. Can I then compile it with Fil-C and have it panic/assert on heap use after free, uninitialized memory read (including stack), array out of bounds read, etc?

Does Fil-C catch uninitialized memory reads?

> image-manip squash: This is the key to reclaiming disk space and the core of our strategy to squash the image layers. The tool creates a temporary container, applies all 272 layers in sequence to an empty root filesystem, and then exports the final, merged filesystem as a single new layer. This flattens the image's bloated history into a lean, optimized final state.

Wouldn't a multistage Dockerfile have accomplished the same thing? smth like

FROM bigimage

RUN rm bigfile

FROM scratch

COPY --from=0 / /

What were you using for RAG? Did you build your own or some off the shelf solution (e.g. openwebui)

You can specify python version requirements in the comment, as the standard describes