Stay up to date with practical insights on backend engineering, system design, and building reliable software.
A unit test suite that takes 10 minutes to run will stop being run. Slow tests accumulate through specific, fixable causes — hidden I/O, over-reliance on containers, and tests that are integration tests wearing unit test clothing.
Read moreMost unit tests verify that code does what it already does — they pass when the code is written and continue passing through every refactor, catching nothing. Here is how to write tests that fail when something breaks and survive when nothing does.
Read moreOver-engineered pipelines built before teams understand their actual needs are a major source of CI/CD dysfunction. The path to a mature pipeline runs through a simple, working one — not around it.
Read moreA working Dockerfile is not the same as a good one. Most images carry megabytes of unnecessary weight that slows builds, bloats registries, and widens the attack surface — and the fixes are straightforward once you know where to look.
Read moreMost unit test suites are heavily weighted toward testing implementation details rather than behavior. When the implementation changes — even correctly — the tests break, and the suite becomes a maintenance burden instead of a safety net.
Read more