Tim Caswell

@rikarends Yeah, most people won't have that much. Not too crazy though

@creationix m3 max 128 + 5090

I made codex make a Rust port of MLX based on its own Rust port of GGML. Apparently its 33% faster than mlx stock for gemma 4. Another is porting 'stable-diffusion cpp' to this new MLX/GGML stack. 40% faster than stock for Flux 1 on apple.

@rikarends Also the agent frameworks are all optimized for mega models with massive contexts. I want to see more optimized to perform best on smaller local models on consumer hardware.

@rikarends What kind of computer do you have locally. I tried some things on M4 pro and the latest AMD and Intel consumer ai chips all 32+GB ram

When Opus 4.5 came out, it was a one-way door to a new way of engineering. Agents now do most of our coding. Knowing the inherent flaws and over-confidence of LLMs, we sent a clear message to our teams. Vibing and mission-critical infrastructure don’t go together. We’re sharing some of our early internal guidance in how we’re “agenting responsibly”, prioritizing security, durability, and availability at all times. vercel.com/blog/agent-res…

@fat @pierrecomputer Can we get a we socket based protocol to fetch git objects in batches? Something like one side sends a list of hashes in a message and the other side sends the corresponding objects as messages. Think lazy loaded VFS for ultra low latency clones.

Code[dot]Storage A new Git provider for machines by @pierrecomputer. In Oct, Github shared they were averaging ~230 new repos per minute. Last week we hit a sustained peak of > 15,000 repos per minute for 3 hours. And in the last 30 days customers have created > 9m repos🧵

@r0ktech user-id

As a developer,what will you choose ?

@mitchellh Welcome to the party 🎉

Excited to share that I've joined Vercel's Board of Directors. Vercel is made up of builders and tastemakers that continually ship things that deeply impact how developers work: Next.js, AI SDK, v0, etc. I can't think of a more exciting place to be. Let's fucking ship. ▲ My relationship with Vercel goes back to the earliest days. HashiCorp was an early adopter of NextJS and Vercel (~10 years ago!) and it remains my default tech stack and deployment platform to this day. Ghostty's website is all on Vercel, too! Beyond that, I've been continually impressed with the teams relentless focus on shipping meaningful software. And importantly, software that has incredible taste. Now we are in the age of agentic software development. Vercel is building agentic infrastructure that I think every app and agent will need (I certainly need it!) and I can't think of a more exciting place to be. Huge thanks to @rauchg , Jeanne, Marten, @cramforce, @tomocchino and the entire Vercel team for the warm welcome. Time to work.

@wesbos How does it compare to sveltekit for svelte apps that are frontend only.

Vite+ is SO GOOD I’ve moved 3 sites over and scaffolded out a few more. So sold

@cramforce @royalicing But great minds think alike. My first version did exactly what you suggest and walk forwards and back to find the newline delimiters. It worked, but I didn't like the complexity of the code. Pre-scanning makes the hot-path simpler and cheaper.

@cramforce @royalicing We do a quick scan for newlines when we first fetch the file into the worker and cache the full bytes along with an array of line offsets in the memory cache. Then when doing the binary search we use the cached line offsets to greatly simplify the logic.

🚨 Blog post about bloom filters. What is the first rule about blog posts about bloom filters? You need to read all blog posts about bloom filters

@dCodes03 @vercel_dev @vercel @rauchg Bloom filters are simpler, the encoding is plenty small as-is, and we need to optimize for CPU cost and simplicity when doing the lookups.

@vercel_dev @vercel @rauchg why are you not trying ribbon filters? They are practically much smaller than bloom filters. And can achieve competitive latency. We can connect to discuss it in depth.

Bulk redirects on Vercel now scale to 1 million per project with near-constant lookup time. We shipped the simplest thing first, profiled it in production, and iterated from there ↓ vercel.com/blog/scaling-r…

This was a fun project to work on! Again we balanced technical performance with operational complexity and found a really nice solution.

@thdxr I'm finding that it lets us experiment with new directions quickly, but we need to be OK throwing away lots of code as well once we decide it wasn't a good idea.

sent this to the team today everything great comes from being able to delay gratification for as long as possible and it feels like we're collectively losing our ability to do that

@robpalmer2 I still hope this lands in the browser eventually

The Hermes JS engine gains runtime support for type-stripping 🎉 The feature is similar to the support in Node.js & TypeScript's "erasableSyntaxOnly" flag.

@theo Now the real question, Claude code vs open code vs copilot?

Which model do you actually use for code every day?

@justjs14 The anime PFP is fairly accurate actually. My old profile was quite old from 2009-ish

@creationix this is a spin off project. very slow progress - mostly for fun and experimentation. i'll try to write some things up about it in next while. PS - you look different! 🤣

i was intrigued by the very poor throughput of node.js on macos here, so i did a bit of digging. results are on the gist below. should be easy to repro. i'll see if it's possible to work this into a PR for node.js or libuv.

@dillon_mulroy @tecchirp But what from! My goodness

@tecchirp you read it correctly

your claude could never

@justjs14 At first I thought you meant io.js and was very confused. How long has your project had the new name? Very cool stuff BTW

highlights: - with changes to libuv, node.js gets 25x faster on macos - next version of Bun will be ~1.5x faster than this, so still room for removing some overhead in node.js - lo runtime, which exposes event loop to JS, is 1.75x faster than patched Bun gist.github.com/billywhizz/67f…

@robpalmer2 @TwiMeta Exactly, that's like saying that node.js is just a wrapper on V8.

@TwiMeta Things that host JS engines are normally called runtimes.

Electrobun is a new cross-platform runtime for desktop apps ✨ Electron mostly uses C++ to embed Node.js & Chromium. Electrobun uses Zig to embed Bun and - like Tauri - relies on the OS default web renderer. 🔷 Write apps in TypeScript 🧱 Process isolation ⚖️ 12MB base install

@thdxr And of course couple this with a giant shared git object cache and fetch objects on-demand for random access and instant "clones"

@thdxr I've always wanted a git native filesystem. Submodules let you link and protect dependencies instead of random lock files. Then to make it more useful, add more metadata to .gitmodules such as branch or semver range so that tools know how to automatically update them.

what if in opencode.json you could specify project references references: ["git@github.com:Effect-TS/effect.git"] these would get cloned to a global cache and kept updated and opencode would have a subagent that could answer questions about them would you use this?