News for hackers. Stuff that matters.

yevbar

Re:We sped up bun by 100x on Thursday April 02, 2026 @04:32PM (#47619791)
I'm actually assured to hear the git CLI is better covered than libgit2 since the CLI test suite is what I used as my "validation" for progress on meeting git's functionality

As for what happened with Bun and libgit2, my best guess honestly is smth to do with zig-c interops but don't doubt there are optimizations everywhere to be done

Re:We sped up bun by 100x on Thursday April 02, 2026 @04:31PM (#47619783)
Bun's attempted to integrate with libgit2 instead of spawning calls to the git CLI and found it to be consistently 3x slower iirc

The micro-benchmarks are for the internal git operations that bun rn delegates to CLI calls. Overall, network time (ie round trip to GitHub and back) is what balances the performance when evaluating `bun install` but there are still places where ziggit has better visible wins like on arm-based Macs https://github.com/hdresearch/ziggit/blob/master/BENCHMARKS....

Re:We sped up bun by 100x on Thursday April 02, 2026 @04:29PM (#47619759)
If this were 2+ years ago perhaps, with industry adopting more agents in their SDLCs (ie Stripe minions or Ramp background agents), I think we're more a matter of time before we treat agent/human built products the same unless we're branding smth as artisanal human-crafted software
Re:We sped up bun by 100x on Thursday April 02, 2026 @04:27PM (#47619736)
> Sure, if you have a complete test suite for a library or CLI tool, it is possible to prompt Claude Opus 4.6 such that it creates a 100% passing, "more performant", drop-in replacement.

This was the "validation" used for determining how much progress was made at a given point in time. Re training data concerns, this was done and shipped to be open source (under GPLv2) so there's no abuse of open source work here imo

Re the tradeoffs you highlight - these are absolutely true and fair. I don't expect or want anyone to just use ziggit because it's new. The places where there performance gains (ie internally with `bun install` or as a better WASM binary alternative) are places that I do have interest or use in myself

_However_, if I could interest you in one thing. ziggit when compiled into a release build on my arm-based Mac, showed 4-10x faster performance than git's CLI for the core workflows I use in my git development

Re:We sped up bun by 100x on Thursday April 02, 2026 @04:24PM (#47619707)
This was orchestrated and developed by agents with verifications like the codebase compiling or git's CLI test suite passing.

That was so the commit authors don't all appear like blank accounts on GitHub

Re:We sped up bun by 100x on Thursday April 02, 2026 @04:23PM (#47619692)
The original target was bun since it itself is written in zig, not because of anything specific to the language

When it was clear that there were benefits in filling in more of git's capabilities (ie targeting WASM), I then went and filled in more git features.

It's not by any means a universal win over everything but it does have notable wins like having git operations be between 4-10x faster on arm-based MacBooks than git itself

Re:We sped up bun by 100x on Thursday April 02, 2026 @04:21PM (#47619666)
The reason being bun actually tested both using the git CLI as well as libgit2. Across the board the C library was 3x slower than just spawning calls to the git CLI.

Under the hood, bun's calling these operations when doing a `bun install` and these are the places where integrating 100% gives the most boost. When more and more git deps are included in a project, these gains pile up.

However, the results appear more at 1x parity when accounting for network times (ie round trip to GitHub)

Re:We sped up bun by 100x on Thursday April 02, 2026 @04:19PM (#47619642)
I cross-posted to my personal blog too https://yev.bar/ziggit
Re:We sped up bun by 100x on Thursday April 02, 2026 @04:18PM (#47619636)
Edge cases certainly apply with scripts depending on specific git CLI args or stdout strings may not suffice with ziggit.

_However_, for the use cases that most developers or agents are looking for, ziggit should have enough features covered. Happy to fix issues or bugs if that's not the case

Re:We sped up bun by 100x on Thursday April 02, 2026 @04:17PM (#47619623)
I ran the test suite specifically for git's CLI as that was the target I wanted to build towards (Anthropic's C compiler failed to make an operating system since that was never in their original prompts/goals)

The way it gets organized is there are "scripts" which encompass different commands (status, diff, commit, etc) however each of these scripts themselves contain several hundred distinct assertions covering flags and arguments.

The test suite was my way of validating I not only had a feature implemented but also "valid" by git's standards

Re:We sped up bun by 100x on Thursday April 02, 2026 @04:15PM (#47619593)
> maintain a separate git implementation

If git were a rapidly evolving project then I'd think this'd be a stronger issue.

With git being more of an established protocol that projects can piggy-back off of from GitHub to jj, filling a library in a new language seems like something that contributes

Show HN: I spent a billion tokens bridging Elixir and WebAssembly

https://yev.bar/firebird

If you'd like to learn about how I corralled the agents to do this, check out this blog post https://vers.sh/blog/elixir-webassembly-billion-tokensAnd, of course, here's the GitHub https://github.com/hdresearch/firebirdI also posted a thread to Twitter where I ship nifty stuff if fun uses of technology is interesting to you https://x.com/itisyev/status/2028543436388016510

( Read More... | 1 comment | 2 points | on HN )

Show HN: I spent a billion tokens and all I got was this repo

https://vers.sh/blog/elixir-webassembly-billion-tokens

If you'd like to learn about what Elixir and WebAssembly are or why even integrate the two together, check out this blog post https://yev.bar/firebirdAnd, of course, here's the GitHub https://github.com/hdresearch/firebirdI also posted a thread to Twitter where I ship nifty stuff if fun uses of technology is interesting to you https://x.com/itisyev/status/2028543436388016510

( Read More... | no comments | on HN )

Show HN: I vibed a better OCaml parser than Jane Street in 69 steps

https://github.com/hdresearch/parsexp

*for some casesUsing cloud sandboxes to run them in I tested: - A single coding agent just told to make a better parser - An agent told to write a better parser within the constraints of tests/benchmarks - An agent swarm that self-improved the premise with extra tests/benchmarks in order to more "truly" write a better parserThe results were a success! I was able to end up with both performance...

( Read More... | 1 comment | 2 points | on HN )

Show HN: 15% of Forbes 30 under 30 winners did fraud

https://30u30.rip

15% of Forbes 30 under 30 winners can be attributed to fraud. Here is an interactive visualization of 8215 winners and whether or not they're in some controversy.I originally vibe scraped the winners and hit my weekly Claude limit, manually Googled over 2500 times (around 2400 I actually got an internal error), then realized Gemini has a free API with search supported.Few are funny, some may...

( Read More... | 3 comments | 2 points | on HN )

Show HN: Run Your Own ChatGPT Agent on Cloudflare Containers

https://github.com/lsd-so/agentflare

Hi HN!I was disappointed when the ChatGPT Agent announcement came with the note that there'd be limited usages available for something that's architecturally simple:> Pro users have 400 messages per month, while other paid users get 40 messages monthly, with additional usage available via flexible credit-based options.So assembled this with Cloudflare's recent Containers API. Here's a link...

( Read More... | no comments | on HN )
We recently updated our LSD MCP server to be able to control a local browser as well as accomplish function interpolation via MCP tools.

For a more serious showcase of how this looks: we turned scraping Substack metrics data into a simple zod/TypeScript project:

https://x.com/getlsd/status/1919808557115179414

Feel free to reach out with issues or requests to yev at lsd dot so

Hey HN,

We made a dynamically typed programming language specifically for interacting with web browsers and made a parody of the song "No Type" by Rae Sremmurd to accompany our TypeScript SDK

Re: on Wednesday April 16, 2025 @07:02PM (#43711256)
Hey HN!

We scraped the Shopify GraphQL docs with code examples so it'd be easier to vibe code using Shopify's API. Enjoy!

Re:Show HN: Gave Claude LSD SQL on Monday February 03, 2025 @03:14AM (#42916055)

Show HN: Gave Claude LSD SQL

https://github.com/lsd-so/lsd-mcp

LSD SQL is a DSL for the web that can self-correct as an LLM traverses the internet. Here's what it looks like now that Claude is connected to the internet similar to OpenAI's Deep Researcher.Want to be a Claudestine Chemist? Follow the quickstart instructions in the README to get started! https://github.com/lsd-so/lsd-mcpCheck out @getlsd on Twitter to see some of our other work or see our...

( Read More... | 11 comments | on HN )
Re:Show HN: Automate iMessage with Python on Monday September 02, 2024 @11:34AM (#41426314)
Some options:

- Don't use the python package at all - Get a second macbook solely for automated iMessaging - Review the code before disabling/installing

If you don't care that much about programmatically getting blue bubbles then this post isn't for you

Show HN: Turn pages into CSVs with the bicycle browser

https://lsd.so

Hey HN!TLDR: if you're non-technical you can think of this as a highlighter for the internet; if you're technical it's a way to treat websites as structured SQL tables.I've been mostly tweeting updates to what I'm working on but know a few folks here liked my previous ships. The landing page has a Loom recording (https://www.loom.com/share/a7a0fd1bbca741558450548f54dba364?...) of using the...

( Read More... | no comments | on HN )

Show HN: Made a bookmarklet to pluck things from the internet into PSQL

https://lsd.so

Howdy HN,I previously did a "Show HN" (https://news.ycombinator.com/item?id=40966268) when I got my custom database to a point where you could query the internet like a structured database by treating CSS selectors/URLs as valid identifiers ie the following:``` SELECT span.titleline > a AS post_title, span.titleline > a@href AS post_link FROM https://news.ycombinator.com; ```And now...

( Read More... | no comments | 2 points | on HN )

Older »