Search the archive

TIL: PYTHON'S SPLITLINES DOES A LOT MORE THAN JUST NEWLINES

Python's `str.splitlines()` method splits strings based on a wide range of Unicode line boundary characters, including more than just the common newline characters.

WHY SELF-HOST?

Self-hosting personal services like calendar, contacts, email, smart home, RSS feeds, and location tracking helps one regain privacy, sovereignty, and control over personal data, escaping reliance on Big Tech.

I'M BUILDING A BROWSER FOR REVERSE ENGINEERS

This article goes over the process of building a custom browser for web reverse-engineering to streamline the process of analyzing browser fingerprinting and anti-bot techniques. The project started with a simple browser extension for hooking JavaScript functions but evolved to use the Chrome Devtools Protocol for deeper access. Eventually, Chromium was forked to create custom events for stealthier instrumentation. The final browser includes features like deobfuscation, property overrides, and automated decryption of fingerprinting payloads.

HOW WE FOUND A BUG IN GO'S ARM64 COMPILER

Cloudflare discovered a race condition bug in Go's arm64 compiler due to high traffic volume. The bug caused sporadic panics on arm64 machines related to stack unwinding, which were initially thought to be rare stack memory corruption. Through deep investigation and debugging, it was found that async preemption during stack pointer adjustments in the `(*NetlinkSocket).Receive` function led to an invalid stack pointer during garbage collection, causing the unwinder to crash.

SVELTE REALLY IS THAT FAST

A study compared the rendering performance of Angular, React, Vue, Svelte, and Blazor JavaScript frameworks. It found that Svelte consistently outperformed the other frameworks in various benchmarks, especially at creating and updating elements. Svelte's performance is due to its reactivity system, optimizing compiler, and binding-based rendering.

SEEING LIKE A SOFTWARE COMPANY

Large software companies prioritize legibility (measurability and predictability) over efficiency. While small companies can thrive with illegible, engineer-driven work, large companies require legibility for planning, enterprise deals, and coordination, even if it slows down individual engineers. To balance this, companies create temporary "zones of illegibility" for urgent problems and rely on informal backchannels.

HOW TO *ACTUALLY* TEST YOUR README

Devs should test their software's installation instructions on a fresh virtual machine. When doing so, they should document every step taken during the installation process, including installing dependencies, and then transferring that documented history to the README file to guarantee accuracy and completeness.

THE PROGRAMMER IDENTITY CRISIS

The rise of AI and LLMs threatens the core identity of programmers as craftspeople. Vibe-coding and specification engineering have turned programmers into operators rather than creative problem-solvers. This shift removes a lot of the joy and fulfillment of programming.

ELEVENLABS UI

ElevenLabs UI is an open-source collection of customizable agent and audio components. It provides interactive visualizations like agent orbs that reflect agent states (idle, listening, and talking) and real-time audio visualizations with smooth scrolling waveforms. The UI also includes components for voice chat, music playback with controls, and a conversation interface with text and voice input options.

FLEETCODE

FleetCode is a desktop terminal application for running multiple CLI coding agents like Claude or Codex in isolated Git worktrees, allowing for parallel and persistent sessions. It has terminal theming, setup commands, and MCP server management, all configurable through a settings panel.

UT

ut is a fast and lightweight CLI utility toolkit for devs. It provides a variety of tools for encoding, hashing, data generation, text processing, development, web and network tasks, color manipulation, and Unicode reference.

MY FIRST CONTRIBUTION TO LINUX

This dev's first contribution to the Linux kernel was adding support for the hotkeys on their 2005 Fujitsu Lifebook S2110 laptop. They go over the process of identifying the relevant driver, studying the code to understand how it handles key events, and modifying it to recognize the media keys. They submitted their patch through the traditional email workflow and saw that it merged into the upstream kernel.

WHY WE NEED SIMD (THE REAL REASON)

SIMD (Single Instruction, Multiple Data) instructions give performance gains at a relatively low cost by reusing existing CPU infrastructure. This lets a single instruction perform multiple operations simultaneously. The adoption of SIMD has been hampered by the need for software updates to use new instructions. However, recent advancements are helping increase adoption.

THE EMAIL THEY SHOULDN'T HAVE READ

This dev helped Agency B and C transition to an open-source mail stack, angering their current provider, who used hidden clauses in their contract to retaliate. The provider potentially spied on their emails and threatened legal action, preventing the agencies from switching and ultimately raising their costs. Despite the ethical issues, the agencies were unable to break free, showing the challenges of escaping vendor lock-in.

NOW OPEN FOR BUILDING: INTRODUCING GEMINI CLI EXTENSIONS

The Gemini CLI now offers extensions, letting devs personalize their command-line experience by connecting Gemini to their ideal tools and workflows through integrations.

CSS HAS 42 UNITS

The increasing number of CSS units is a positive evolution driven by specific needs.

RECALL

Recall is a Redis-powered memory extension for Claude that provides persistent context across sessions.

YOU'RE DOING RAILS WRONG

Instead of adding and using numerous JavaScript tools and frameworks, you can just use Rails.

STATE OF THE SOFTWARE ENGINEERING JOBS MARKET, 2025: WHAT HIRING MANAGERS SEE

The 2025 software engineering job market is experiencing a paradox: companies are receiving a flood of applications (often 1,000+ per role) but are struggling to find quality candidates, with most actual hires coming from sourcing and referrals rather than inbound applications. The market shows increased demand for product engineers and founding engineers, with remote positions facing more competition but offering slightly lower compensation. Also, issues like AI-generated fake applications and interview cheating are becoming growing problems.

USING DENO AS MY GAME ENGINE

Microlandia is a city builder game inspired by SimCity with a focus on realistic simulation using real-world data. It was initially built in Go, but the project faced challenges in creating a 3D view and a full-featured UI. To overcome these hurdles, the game is being ported to Deno, taking advantage of its cross-platform capabilities and integration with web technologies like ThreeJS and React.

NODE.JS PERFORMANCE: PROCESSING 14GB FILES 78% FASTER WITH BUFFER OPTIMIZATION

A developer optimized Node.js performance for processing a 14.8GB weather data file containing 1 billion temperature measurements to achieve a 78% speedup. The main breakthrough involved bypassing JavaScript's string encoding overhead by parsing raw bytes directly, storing temperatures as integers (e.g., 25.3°C as 253 tenths), and implementing a hash-first approach that only decoded station names when absolutely necessary.

GPT-5-CODEX IS A BETTER AI RESEARCHER THAN ME

This dev uses OpenAI's GPT-5-Codex to automate AI research, and he found that Codex outperformed his solo efforts, especially in generating coherent stories. The process involved a feedback loop where Codex suggested experiments based on results, and Goedecke selected or contributed ideas. The most successful approach was distilling a transformer model from an n-gram model, which improved story coherence by initially focusing on grammatical correctness.

YOU ARE HOLDING BUILD FILES WRONG

BUILD files in Bazel are more than just dependency lists and should be used to encode a project's high-level architecture. By manually managing BUILD files and defining larger conceptual targets, developers can enforce module boundaries, detect architectural violations early, and guide AI tools better.

DEPLOY SAFETY: REDUCING CUSTOMER IMPACT FROM CHANGE

Slack implemented a "Deploy Safety Program" in 2023 to address the fact that 73% of customer-facing incidents were caused by their own code deployments, as the platform became increasingly mission-critical for users. The program focused on automated detection and remediation within 10 minutes. By January 2025, the initiative successfully reduced customer impact hours by 90% from peak levels through a combination of automated rollback systems, improved monitoring, and cultural changes across their engineering organization.

INTRODUCING THE GEMINI 2.5 COMPUTER USE MODEL

Google DeepMind has introduced the Gemini 2.5 Computer Use model, which allows agents to interact directly with user interfaces. This model, available via the Gemini API, can navigate web pages and applications by clicking, typing, and scrolling, outperforming existing solutions with lower latency. The model analyzes user requests and screenshots to generate UI actions, operating within a loop for iterative task completion while prioritizing safety through built-in features and developer controls.

ARC

Arc is a high-performance time-series data warehouse built on DuckDB, Parquet, and MinIO. It supports high-speed data ingestion and fast analytical queries with SQL.

/DEV/PUSH

/dev/push is an open-source platform that offers a self-hostable alternative to services like Vercel. It supports Git-based deployments, environment management, team collaboration, custom domains, and real-time monitoring.

WHAT MAKES 5% OF AI AGENTS ACTUALLY WORK IN PRODUCTION?

Successful AI agents in production depend on good context engineering, governance, memory architecture, and multi-model orchestration. Most AI deployments fail due to bad scaffolding around the models. Also, trust in AI is built through human-in-the-loop design and consistent, explainable behavior, positioning AI as an assistant rather than an autonomous decision maker.

VIBE ENGINEERING

Vibe Engineering is a responsible and stronger approach to software development using AI, whereas "vibe coding" is more irresponsible. Vibe engineering has experienced engineers using AI tools like coding agents to speed up their work while still having high standards. This approach requires strong software engineering practices such as automated testing, planning, documentation, and code review, while also relying on QA, research, and careful management of AI assistants.

WHO NEEDS GIT WHEN YOU HAVE 1M CONTEXT WINDOWS?

An AI developer accidentally deleted working code but was able to recover it using an LLM with a 1 million token context window, showing an unexpected benefit of LLMs remembering past interactions.

PDOC

Pdoc is a tool that automatically generates API documentation for Python projects with features like type annotation support, cross-linking, a built-in web server, and compatibility with various docstring styles.

BECOME UNBANNABLE FROM YOUR EMAIL/GMAIL

Email is a crucial digital asset, and you can gain control over it and become "unbannable" by using your own domain with a trusted email provider and backing up all emails to a local server.

PART I: HOW DOES GRADIENT DESCENT WORK?

This is a mathematical guide to gradient descent during deep learning, which is an optimization algorithm used to train machine learning models by finding the minimum of a cost function.

Try for free.

TYPESCRIPT VS ZOD: CLEARING UP VALIDATION CONFUSION

TypeScript and Zod serve complementary rather than competing roles in development. TypeScript provides compile-time type safety during development and building, while Zod validates untrusted data at runtime. This is very useful for external inputs like API requests, user forms, and third-party data.

THE NUANCE OF REACT RENDERING BEHAVIOUR AS IT RELATES TO CHILDREN

React child components behave differently when rendered directly versus passed through the `children` prop. The main difference is in object reference equality: when components are rendered directly, React creates new object references on each render, causing child components to re-render, but when passed via `children`, the same object reference is reused, preventing unnecessary re-renders.

WHY REACTIVE PROGRAMMING HASN'T TAKEN OFF IN PYTHON (AND HOW SIGNALS CAN CHANGE THAT)

Reactive programming hasn't gained traction in Python due to the complexity of existing tools like RxPY, which are better suited for event streams than state management. New libraries like reaktiv simplify state management similar to spreadsheet formulas. reaktiv uses signals, computed values, and effects to automatically manage dependencies and update values, removing manual coordination and reducing bugs.

STRUCTURED PROCRASTINATION

Procrastination can be a beneficial strategy. Procrastinators can become productive by prioritizing tasks and working on less important ones to avoid the most urgent ones. The key is to have a list topped with seemingly crucial but flexible deadlines, thus motivating action on other valuable activities.

YOU'RE ONLY USING 20% OF CLAUDE CODE - HERE'S HOW TO UNLOCK THE REST

Most developers are only using a small fraction of Claude Code's capabilities. To fully use it, devs should use Model Context Protocols (MCPs) correctly, set up a good CLAUDE.md configuration file, use Planning Mode for better project structure, and integrate the IDE plugin for real-time error detection and fixes.

BUILD YOUR ENGINEERING TEAM LIKE A DUNGEON PARTY

CHATGPT APPS SDK

OpenAI's Apps SDK is a framework that allows developers to build, test, and deploy custom apps within ChatGPT.

OPENAI CHATKIT

ChatKit is a framework from OpenAI to help create AI-powered chat experiences by providing a ready-to-use chat interface with advanced features. It allows devs to easily integrate conversational intelligence into their applications with a quick setup, offering deep UI customization, built-in response streaming, and interactive widgets.

WHY DO LLMS FREAK OUT OVER THE SEAHORSE EMOJI?

LLMs confidently believe a seahorse emoji exists, even though it doesn't. The author demonstrates this phenomenon across various models and notes that many people also mistakenly remember a seahorse emoji. By using a logit lens, the article shows that LLMs construct a "seahorse + emoji" residual, similar to how they handle real emojis, but because there's no actual seahorse emoji, the model selects a related but incorrect emoji and then can get stuck in a loop.

HELLOFRESH'S BROWNFIELD LEAP: ONE APP TO FEED THEM ALL!

HelloFresh Group launched Project PUMA (Project Unified Mobile App) to consolidate its fragmented mobile app infrastructure, which previously used separate native Swift/Kotlin codebases for meal kit brands and React Native for ready-to-eat brands. It chose a "brownfield" React Native approach that allows progressive migration while keeping existing native apps functional, enabling them to build features once and deploy across all brands.

WEBGPU AND THE PRICE OF COMPILING WGSL

This WebGPU diagnostics tool allows users to test shader compilation times and explore device/adapter limits by adjusting shader complexity via sliders.

THE LEAST AMOUNT OF CSS FOR A DECENT LOOKING SITE

The minimal amount of CSS needed to create a responsive and decent-looking website takes care of image overflow, good typography, dark mode support, and restrained content width for better readability.

GEM.COOP

gem.coop is a new, community-owned gem hosting server for the Ruby ecosystem, created by former RubyGems.org maintainers.