Updated practice for review articles and position papers in ArXiv CS category

Hard Rust requirements from May onward

SQLite concurrency and why you should care about it

GHC now runs in the browser

I spent a few hours last weekend testing whether AI can replace code by executing directly. Built a contact manager where every HTTP request goes to an LLM with three tools: database (SQLite), webResponse (HTML&#x2F;JSON&#x2F;JS), and updateMemory (feedback). No routes, no controllers, no business logic. The AI designs schemas on first request, generates UIs from paths alone, and evolves based on natural language feedback. It works—forms submit, data persists, APIs return JSON—but it&#x27;s catastrophically slow (30-60s per request), absurdly expensive ($0.05&#x2F;request), and has zero UI consistency between requests. The capability exists; performance is the problem. When inference gets 10x faster, maybe the question shifts from &quot;how do we generate better code?&quot; to &quot;why generate code at all?&quot;

Show HN: Why write code if the LLM can just do the thing? (web app experiment)

Claude Code can debug low-level cryptography

Visopsys: OS maintained by a single developer since 1997

SailfishOS: A Linux-based European alternative to dominant mobile OSes

The Smol Training Playbook: The Secrets to Building World-Class LLMs

How I use every Claude Code feature

<a href="https:&#x2F;&#x2F;labs.google.com&#x2F;pomelli&#x2F;about&#x2F;" rel="nofollow">https:&#x2F;&#x2F;labs.google.com&#x2F;pomelli&#x2F;about&#x2F;</a>

Pomelli

Beginner-friendly, unofficial documentation for Helix text editor

The hardest program I've ever written (2015)

From 400 Mbps to 1.7 Gbps: A WiFi 7 Debugging Journey

Dating: A mysterious constellation of facts

Word2vec-style vector arithmetic on docs embeddings

FFmpeg dealing with a security researcher

Anonymous credentials: rate-limit bots and agents without compromising privacy

3M Diskette Reference Manual (1983) [pdf]

RegEx Crossword