Introducing TLA-ts A 2× faster drop-in replacement for TLC, written 100% in TypeScript Why? TLA+ for formal verification and model checking is great!... but is written in Java (ew), unusable in browser, and slow

The question that is stopping me from buying an M5 Ultra: why does a Mac with 1.2 TB/s of bandwidth lose to DGX Spark at long context and with multiple users? With one user, decode is memory-bound. Every token means reading all the active weights once, and the GPU mostly waits for data. More bandwidth means more tok/s, and that's where the Mac shines. 👍 Batching changes everything. With 8 users you read the weights once and use them for 8 tokens. Same bytes, 8× the math. Throughput scales almost for free… until compute becomes the bottleneck. 👀 That's where the Mac hits a wall: good bandwidth, lower low-precision compute. The Spark's Blackwell tensor cores with native FP4 give far more FLOPs per GB/s of bandwidth, so it keeps scaling while the Mac stops. Real numbers on DeepSeek V4.1 Flash with TensorFold, total tok/s from 1 to 8 users (chart below): 2× Spark: 77 → 105 → 150 → 194 (@rwojo) 1× M5 Ultra: 63 → 78 → 98 → 99 (@TechMDAI, first runs, capped at 4 slots) Long context is even worse. Prefill is pure compute, attention grows with context length, and the KV cache is per-user: no way to share it. On DeepSeek V4 Flash the Mac's decode drops from ~60 to ~27 tok/s, while 2 Sparks hold ~51 (tests by @satsuki2694). Bottom line: the Mac is unbeatable for fitting huge models (MoE) in a single box, but the Spark wins when you need compute: long prompts, many users, agents. Different tools, different jobs. I run both, and I was planning a deep dive on the M5 Ultra to really understand its limits, but at this price I don't know if it makes sense spending all these money just to make tests and share data. I'll try to figure out something, otherwise I'll wait for others to test these concurrency limits in depth. 💪

Node.js 26 “Lithium” is now Long Term Support 🎉 v26.13.0 is out, with updates through April 2029. Heads-up for Linux users coming from an earlier LTS: libatomic may need to be installed separately.

Today at Universe we talked about the changes we are making to git to support collaboration at machine speed. The new Git architecture will support 7,200 writes / sec!!! Coming early next year.

Why are many companies red / security teams do not want to report vulnerabilities via HackerOne? We built most Node.js automation for security release on topn of it, and current climate makes it somewhat suboptimal.

I'm starting to sense a divide in the tech industry: - People thriving on autonomy who are reinventing how they work and what tools they use with AI: LOTS of work, but they're enjoying it and getting very efficient - People used to being mentored / trained / or simply told what to do and how. Frustrated with the lack of guidance from management on what to do with AI; also frustrated by seeing some of their peers thriving; increasingly bitter about AI and the state of the industry, as a whole Neither group seems to be aware that there's another group, or have empathy for the other group

@geistgrail @matteocollina @GraphQL Open the network tab (inspect in chrome), refresh this page, search for "graphql", notice that you're currently using GraphQL. Then go to AirBnB, Reddit, Booking, Instagram, WhatsApp, and so many more websites and you realize that many of the top websites use it.

With highly profitable companies using AI to clean room reimplement existing projects, it’s more important than ever to support independent nonprofit open source projects like ESLint. We’ve been doing the work for 13 years.

Reflections on the recent @deno_land acquisition by @Cloudflare and what is says about the state of JS runtimes and the overall fragility of the web. As always thx to my @openjsf friends, James Snell, @matteocollina, @rginn206

Could you or your company sponsor an x86-64 server and hosting for FFmpeg’s CI? RISC-V, PowerPC or MIPS systems would also be welcome! We currently test these using QEMU on our Ampere ARM server. Please get in touch!

What is making GitHub explode => massive surge of agent based traffic, and I expect the majority if this to be on the OSS/free plans. What is the reaction: celebrate paid-only git hosting. The only thing paid-only competition would do is to make GitHub business model harder to justify, worsening what everybody gets for free?