A history of IDEs at Google

A history of IDEs at Google
7 minutes by Laurent Le Brun

Google spent years with engineers using different IDEs, which created duplicate work to support internal tools across each one. Around 2016, a browser-based editor called Cider was built and slowly gained popularity, especially after adding smart code features powered by a backend that indexed the entire codebase. In 2020, the team switched to a VSCode frontend, and by 2023, 80% of Google development happened in this new editor. Having most engineers on one tool made investments more impactful and unlocked better AI features.

Fix the Shared Infrastructure Failure: Stop "Works on My Machine" From Stalling Your Team
sponsored by Flox

"Works on my machine" is a failure of shared infrastructure. For Fellow.ai, it stalled code reviews, slowed onboarding, and caused CI anomalies. With Flox, their platform team now defines reproducible project environments that travel with the repos. Running in native subshells, not heavy containers, Flox ensures the same pinned packages and environment variables run identically everywhere.

Programming still sucks
9 minutes by Steven Langbroek

Tech work was never the clean, orderly thing people imagined. Now it is a burning ship with broken tools, missing knowledge, and leaders chasing AI demos. Entire generations of junior engineers were cut before they could become the seniors who keep systems alive. The only people truly safe are the ones nobody noticed, quietly maintaining the systems that everything depends on.

How container filesystem works
24 minutes by Ivan Velichko

Learn how Linux containers are built from the ground up. Starting with the mount namespace and a root filesystem, see why PID, cgroup, UTS, and network namespaces naturally follow - and how this foundation makes concepts like bind mounts, volumes, and persistence in Docker or Kubernetes much easier to grasp.

Pushing and pulling: Three reactivity algorithms
18 minutes by Jonathan Frere

Jonathan explains three ways reactive systems work: push-based, pull-based, and push-pull reactivity. Using spreadsheet examples, he compares how each method handles updates, efficiency, dynamic dependencies, and glitches. Push systems are simple but can waste work, pull systems are flexible but less efficient, while push-pull combines the strengths of both. He concludes that push-pull reactivity offers the best balance for modern applications and user interfaces.

A real-world story of CPU bottlenecks
16 minutes by Vaibhav Shankar et al.

Pinterest engineers investigated repeated crashes in Ray-based ML training jobs caused by network failures in Kubernetes clusters. After months of debugging, they discovered that hidden “zombie” memory cgroups, created by a repeatedly crashing AWS ECS agent, were overloading CPU cores and starving network driver threads. By disabling the unnecessary ECS agent and improving profiling methods, the team restored system stability and highlighted the importance of monitoring, profiling, and consistent infrastructure configuration.

View the latest HubSpot Developer Platform updates in Spring Spotlight
sponsored by HubSpot

Spring Spotlight 2026 is live and we’ve rounded up the top updates for you. See what's new for the HubSpot Developer Platform! Ship faster with AI coding tools like Cursor, Claude Code, and Codex. Build MCP-powered AI connectors, run serverless functions with support for UI extensions, and use date-based versioning to streamline roadmap planning.

And the most popular article from the last issue was:

• Learning software architecture