Programming Digest
Programming Digest
AI tools can generate messy code at a scale no rockstar developer ever could. Meanwhile, engineers who spot disasters in the making often stay quiet because speaking up feels risky. Both problems share a root: humans stepping back when they should lean in.
Jira's automation can run a Fibonacci generator — making it accidentally Turing-complete. Also: why dividing pixels by 255 vs 256 actually matters, and how memory layout can beat algorithmic improvements for raw speed.
AWS Lambda's networking team cut tunnel setup time from 150ms to 200 microseconds using eBPF. Stripe runs 100k test files but trims each CI run to just 5% by tracking which files tests actually open. Plus a solid intro to AI engineering covering RAG, agents, and real production challenges.
Google's IDE journey from scattered tools to one browser-based editor used by 80% of engineers by 2023 is a neat read. Also worth your time: how Pinterest spent months tracking mysterious Kubernetes crashes, only to find zombie memory cgroups from a rogue AWS agent starving network threads.
GPS uses satellite signals and Einstein's relativity fixes to pinpoint your location. Email was built for trust but runs on bolted-on fixes like SPF and DKIM. And if AI handles complex code, maybe Python's simplicity matters less than raw performance.
Bad software hides in plain sight — rigidity, fragility, immobility, viscosity. One dev built his own text editor from scratch and daily-drove it to fix real gaps. Plus, why AI agents are quietly breaking assumptions traditional databases were built on.
Shazam skips melody entirely — it maps sound to frequencies, picks the loudest peaks, and matches fingerprints across millions of songs in milliseconds. Also: an AI agent wiped a small business's entire database and backups in one API call, with two vendors sharing the blame.
Floating point rebuilt from scratch — from math to real silicon. Plus, why LLMs lack the laziness that makes good software, and a CIA-leaked git one-liner that actually saves time.
Good APIs expose little and promise less. Boring boundaries age better than clever ones. Plus: what really happens after you click "Login with Google" — tokens, redirects, and cryptographic signatures explained end to end.
Transistors power everything in GPUs. Jason built a game where you construct one from scratch. Also, five git commands that reveal codebase health before reading a single line — spotting risky files in minutes.
Too many approval layers slow teams down through waiting, not effort. AI speeds up coding but won't fix slow review pipelines. Also: shell shortcuts like CTRL+R and brace expansion that save real time, one trick at a time.
How JPEG compression exploits human vision to shrink images. Why queues don't fix traffic spikes—they just delay the pain. And how RollerCoaster Tycoon ran smoothly on 1999 hardware by turning technical limits into design features.
CPU branch prediction limits vary wildly by chip—AMD handles 30,000 learned branches, Apple 10,000, Intel just 5,000. Meanwhile, the IBM 9020 from the 1960s tracked flights across the US with sub-30-second failover, and its software outlasted the hardware by decades.
Docker turns 10 — it solved "works on my machine" by bundling apps with their dependencies using Linux process isolation. Containers aren't magic though: they share the host kernel, so skipping basics like non-root users and dropped capabilities leaves everything exposed.
Errors in software are either expected (bad input, network issues) or unexpected (bugs). Treat more as expected to build reliable software. Also: how Notion tracks offline pages using a two-table system that prevents accidental removal when one reason disappears.
This issue explores OAuth's role in secure online delegation, AI-assisted coding workflows, YAML's quirks, innovative rate limiting strategies, and ASCII rendering techniques for sharper images. Dive in for insightful reads.
In this issue, explore how programmers spend their time tackling bugs and build systems, the enduring growth of software in the AI era, Postgres locking challenges, and effective architecture decision-making strategies.
In this issue, explore AI adoption with Mitchell Hashimoto, discover semantic search innovations with Doug Turnbull, and master effective code reviews with Daniil Bastrich. Plus, learn about function inlining and the art of road design in games.
This issue dives into innovative software insights, from Tony's simplified Git implementation and Chad's "deletion test" to Antirez's thoughts on AI in programming. Discover memory leak fixes in Ghostty and core principles of performance engineering.
This issue explores the challenges of clock synchronization, the mechanics of Lobsters' ranking algorithm, the importance of soft skills for engineers, and a deep dive into DNS record order. Don't miss the latest on dithering visuals.