511
We reverse-engineered Flash Attention 4
Asynchrony, fast approximate exponents, and 10x more efficient softmax.
New
The latest blog posts to take off on Hacker News. 1,748 posts.
512
A Postmark backdoor that’s downloading emails
Secure the modern software stack with Koi. Gain visibility, context, and real-time protection over extensions, MCPs, and autonomous AI agent endpoints.
513
How does lossless compression in Fuji RAF files work? (2020)
I spent 3 weeks earlier this year digging in to and reimplementing the lossless compression algorithm in Fuji’s RAF Camera RAW files. This post discusses the fundamentals of how it works.
514
Fast UDP I/O for Firefox in Rust
Motivation Around 20% of Firefox’s HTTP traffic today uses HTTP/3, which runs over QUIC, which in turn runs over UDP. This translates to substantial UDP I/O activity. Firefox uses NSPR for most of its network I/O. When it comes to UDP I/O, NSPR only offers a limited set of dated APIs, most relevant here PR_SendTo and…
515
Context is the bottleneck for coding agents now
Ship production-quality code at superhuman speed.
516
Translating a Fortran F-16 Simulator to Unity3D
I recently purchased the textbook “Aircraft Control and Simulation” by Brian L. Stevens, Frank L. Lewis, and Eric N. Johnson 1 . This book covers the control and simulation of aircraft. It’s really dense and frankly hard to understand. As far as aerodynamics texts go, it’s pretty typical.
517
`std::flip`
std::flip is a little-known utility from the C++ standard library header <functional>: it is a higher-order function that accepts a Callable and returns an equivalent Callable with the order of its parameters reversed (or “flipped”).
518
My Deus Ex lipsyncing fix mod
Back in 2021 I made a mod for Deus Ex 1 that fixes the lipsyncing and blinking, which, I betcha didn’t know, was broken since ship. Everything I wrote about it is on Twitter, and it oughta be somewhere else, so here’s a post about it. The mod itself can be downloaded here .
519
520
Redis is fast – I'll cache in Postgres
There are books & many articles online, like this one arguing for using Postgres for everything. I thought I’d take a look at one use case - using Postgres instead of Redis for caching. I work with APIs quite a bit, so I’d build a super simple HTTP server that responds with data from that cache. I’d start from Redis…
521
Improved Gemini 2.5 Flash and Flash-Lite
Google is releasing updated Gemini 2.5 Flash and Flash-Lite preview models with improved quality, speed, and efficiency.
522
The Theatre of Pull Requests and Code Review
I recently attended the Goatmire Elixir Conf and one of the standout talks for me was Saša Jurić’s “Tell Me a Story” . It was an incredible presentation that combined theatrical storytelling with practical technical advice. Saša performed parts of his talk in character, turning technical topics into a compelling…
523
Bundler Belongs to the Ruby Community
I’ve spent 15 years of my life working on Bundler. When I introduce myself, people say “oh, the Bundler guy?”, and I am forced to agree. I didn’t come up with the original idea for Bundler (that was Yehuda). I also didn’t work on the first six months worth of prototypes. That was all Carl and Yehuda together, back…
524
How fast is Go? Simulating particles on a smart TV
The challenge, simulate millions of particles in golang, multi-player enabled, cpu only, smart tv compatible.
525
Why is Windows still tinkering with critical sections? – The Old New Thing
The critical section may be an old dog, but it's still learning new tricks.
526
Zed's Pricing Has Changed: LLM Usage Is Now Token-Based
From the Zed Blog: We're moving Zed AI from prompt-based limits to token-based pricing.
527
SedonaDB: A new geospatial DataFrame library written in Rust
Apache Sedona is a cluster computing system for processing large-scale spatial data. Sedona extends existing cluster computing systems, such as Apache Spark, Apache Flink, and Snowflake, with a set of out-of-the-box distributed Spatial Datasets and Spatial SQL that efficiently load, process, and analyze large-scale…
528
Python on the Edge: Fast, sandboxed, and powered by WebAssembly
With AI workloads on the rise, the demand for Python support on WebAssembly on the Edge has grown rapidly. However, bringing Python to WebAssembly isn'...
529
Learning Persian with Anki, ChatGPT and YouTube
I’ve been learning Persian (Farsi) for a while now, and I’m using a bunch of tools for it. The central one is certainly Anki, a spaced repetition app to train memory. I’m creating my own never-ending deck of cards, with different types of content, for different purposes. The most frequent type of cards is grammar…
530
Python developers are embracing type hints
What is Typed Python? Why is it important for Python developers today? How to can you get started?
531
Preparing for the .NET 10 GC
In .NET 9 we enabled DATAS by default. But .NET 9 is not an LTS release so for many people they will be getting DATAS for the first time…
532
From Rust to reality: The hidden journey of fetch_max
Tracing Rust AtomicU64::fetch_max from macro expansion and rustc intrinsics through LLVM atomicrmw umax down to the final x86-64 CAS loop.
533
Context Engineering for AI Agents: Lessons
This post shares the local optima Manus arrived at through our own "SGD". If you're building your own AI agent, we hope these principles help you converge faster.
534
Ultra efficient vector extension for SQLite
Making vector search fast, memory-efficient, and natural in SQLite.
535
Libghostty is coming
Over two years ago, in one of my first public talks about Ghostty , I shared my vision for libghostty : an embeddable library for any application to embed their own fully functional, modern, and fast terminal emulator. Libghostty is finally starting to take shape, and I'm excited to share more details about my plans…
536
Mesh: I tried Htmx, then ditched it
Alex Moon is an Australian-British software engineer and artist based in the North East of England.
537
Wild performance tricks
Last week I had the pleasure of attending RustForge in Wellington, New Zealand. I gave a talk titled “Wild performance tricks”. You can watch a recording of my talk . If you’d prefer to read rather than watch, the rest of this post will cover more or less the same material. The talk shows some linker benchmarks, which…
538
Pairing with Claude Code to rebuild my startup's website
Adventures using AI agents, especially Claude Code, and MCP Servers
539
Testing is better than data structures and algorithms
People should spend less time learning DSA, more time learning testing.
540
Dear GitHub: no YAML anchors, please
This was a good thing . YAML anchors in GitHub Actions are (1) redundant with existing functionality, (2) introduce a complication to the data model that makes CI/CD human and machine comprehension harder, and (3) are not even uniquely useful because GitHub has chosen not to support the one feature (merge keys) that…