Is OOXML Artifically Complex?
OOXML is less an act of sabotage than the continuation of a flawed, inelegant lineage.
The latest blog posts to take off on Hacker News. 21,819 posts.
OOXML is less an act of sabotage than the continuation of a flawed, inelegant lineage.
In a real world benchmark, the default musl allocator caused a 7x slowdown compared to other allocators. I recommend all Rust projects immediately swap to a different allocator in a musl environment.
This book is a reference. Here's my summary of its salient points so you understand why it's so reco
The data on the page is presented in a hierarchichal structure. The movie page contains a director, a list of genres, a list of actors, and each actor in the list contains a list of characters they played in the movie. You can't sensibly fit all of this into a single flat structure like a relation.
C++ seems to finally converge with their contracts proposal, I decided to give it a try and come up with ideas how such a thing would look for C. This is in early stages, not a full proposal yet, a…
Fil-C uses a parallel concurrent on-the-fly grey-stack Dijkstra accurate non-moving garbage collector called FUGC (Fil's Unbelievable Garbage Collector). You can find the source code for the collector itself in fugc.c , though be warned, that code cannot possibly work without lots of support logic in the rest of the…
Sourcing data directly from disk IS faster than caching in memory. I brought receipts. Because hardware got wider but not faster, the old methods don't get you there. You need new tools to use what is scaling and avoid what isn't.
This post goes through how to maximize file IO performance on linux using zig with io_uring.
On trying to mash up SQLite with ideas stolen from Accountants, Clojure, Datomic, XTDB, Rama, and Local-first-ers, to satisfy Henderson's Tenth Law. Viz., to make a sufficiently complicated data system containing an ad-hoc, informally-specified, bug-ridden, slow implementation of half of a bitemporal database…
Our lab investigated whether frontier models can write optimized GPU kernels for Apple devices to speed up inference. We found that they can: our AI-generated Metal kernels were 1.24x faster across KernelBench v0.1 problems, and 1.87x faster across KernelBench v0 problems.
Visualizing the internal state of a Transformer model
From the Zed Blog: You asked, and here it is. Use Claude Code in public beta directly in Zed, built on the new Agent Client Protocol.
There is a pattern I’ve learned while using Zig which I’ve never seen used in any other language. It’s an extremely simple trick which - when applied to a data structure - reduces memory usage, reduces memory allocations, speeds up accesses, makes freeing instantaneous, and generally makes everything much, much…
This started as an internal Sanity workshop where I demoed how I actually use AI. Spoiler: it's running multiple agents like a small team with daily amnesia.
Discover Ada through a fun, project-based tutorial! Learn the language’s clarity, safety, and modern features while building an SVG rosetta generator. A…
A deep-dive into the challenges and misconceptions surrounding async programming in Python
Hey, it's finally happened. I've decided to write a blog post. And if you're reading this, I've also finished one. I have wanted to do this for a long time,…
Frameworks for LLM Agents have been surprisingly disappointing. I want to offer some principles for building agents based on our own trial & error, and…
The blog for CocoaPods.org the Cocoa Dependency Manager
Note: Everything that follows is purely my personal opinion as an individual. It should not be seen as any sort of policy of the Meson buil...
CVE-2025-7783 is a very recent vulnerability affecting a lot of applications in the Node.js ecosystem including those which use axios or the deprecated request library. In all honesty, this vulnera…
I like to answer this frequent question by way of an example. An example of two companies. It goes something like this:
Ultrassembler is a superfast and complete RISC-V assembler library that I'm writing as a component of the bigger Chata signal processing project.
We all know the current US President is one hell of an orator and often assures us that he has “the best words”:
Humans have a tendency toward binary thinking (pardon the pun). If it's not A, it's B. Perhaps because Lisps have REPLs, they are often thought of from the outside as being dynamic, interpreted languages. Our years of Python have taught us that such languages don't really have strong typing - it's all a wild guess…
You see a series of checkboxes checked. Schedules met. Requirements satisfied. Demos delivered. It's a good day. Good job, you, good job! A promotion is in sight.
Writing a Hypervisor in 1,000 Lines
When designing software systems, do the simplest thing that could possibly work. It’s surprising how far you can take this piece of advice. I genuinely think…