The latest blog posts to take off on Hacker News. 49,196 posts.
When I first started writing for a real publication, I taught myself “reporting” with a simple self-made curriculum unfolding over six or seven articles. The first two pieces I wrote from my head, with reference to things I already knew or to books I’d read. For the third, I actually got out of the house, but didn’t…
In which we answer every question you've ever had about digital color, and some you haven't.
OAuth was first introduced in 2007.
The purpose of Standard Thermal is to make energy from solar PV available 24/7/365 at a price that is competitive with US natural gas.
At the time of writing, the most recent Adobe Reader 25.x.y.z 64-bit installer for Windows 11 weights 687,230,424 bytes . After installation, the program includes 'AI' (of course), an auto-updater, sprinkled ads for Acrobat online services everywhere, and 2 GUIs: 'new' and 'old'.
“Bro, ban me at the IP level if you don't like me!”
How to check your Mac is booting in Full Security, and how to read its log to verify all the key steps involved in that process.
A simple typo of ghcr.io to ghrc.io would normally be a small goof. You’d typically get a 404 or similar error, finally work out the issue, fix it, and move along. But in this case, that typo appears to be doing something very malicious, stealing GitHub credentials. What’s ghcr.io? First, a quick bit of background…
Railway Mania, dot-com, AI. Technology bubbles follow a pattern older than any of them. What's different this time is that everyone knows the script.
The DMCA online safe harbor is a notice-and-takedown scheme. Web hosts aren’t liable for copyright-infringing third-party uploads unless and until the copyright owner submits a proper takedown notice to the host, at which point the web host can remain legally...
Most of what modern software engineers do involves APIs: public interfaces for communicating with a program, like this one from Twilio. I’ve spent a lot of time…
SQLite doesn’t do fsync unless specified.
After 15 years as a software engineer, I realized I had never actually built and published a game.
written by Eric J. Ma on 2025-08-23 | tags: python runtime llm security namespace compilation execution functions toolbot monkeypatching
One of Germany's biggest ISPs changed how their DNS works, right after I exposed an organization that they’re part of.
Do we have to disclaim how we use AI ?
When I first started using Claude Code, I had a naive approach to working with it. I would describe the task directly in the prompt, press Enter, and cross my fingers. If the agent made mistakes, I would tell it how to fix them. For small tasks, this can be good enough, but as the task grows in complexity, this…
A visual introduction to big O notation.
How I use DeepWiki to speed up real coding work
I’ve been working on a CLI tool recently, and one of the things it does is manage files on disk. I have written a lot of file management tests for Bundler, and the two biggest reasons that the Bundler test suite is slow are exec and fstat. Knowing that, I thought I would try to get out ahead of the slow file stat…
Why buy a bespoke camera when you can do something much harder instead?
In the past few years, CPUs have gotten really fast. Shockingly fast! Yet most people are stuck on previous generation mobile chips (whether by choice, or by their companies choice), at a huge detriment to their productivity. Meanwhile, AI coding subscriptions like Cursor are all the rage these days. I'll skip the…
In a service with capacity $\mu$ requests per second and background load $\lambda_0$, the usable headroom is $H = \mu - \lambda_0 > 0$. When $M$ clients align, e.g., after a cache expiry, at a cron boundary, or as a service returns from an outage, the bucketed arrival rate can
My latest love letter to Linux has been published. It's called Omarchy, and it's an opinionated setup of the Arch Linux distribution and the Hyprland tiling window manager. With everything configured out-of-the-box to give you exactly the same setup that I now run every day. My Platonic ideal of what a developer…
It's not that hard to build a coding agent. 300 lines of code running in a loop with LLM tokens. You just keep throwing tokens at the loop, and then you've got yourself an agent.
My life is not a math Olympiad
LLM models are plateauing, but true AGI isn't about scaling the next breakthrough model—it's about engineering the right context, memory, and workflow systems. AGI is fundamentally a systems engineering problem, not a model training problem.
You’ve likely read lots of blog posts stating that it takes 23 minutes and 15 seconds to get back to work after an interruption, context switch, or meeting. Thus, “do you have five minutes” ends up not only costing those few minutes, but instead about half an hour. But where does that number come from?
