Key Practices in Software Release, Testing, and Continuous Delivery

This article aggregates ten of the most popular 2020 posts, each discussing practical software engineering practices such as release pipelines, testing strategies, feedback mechanisms, and continuous delivery.

1. Google release process – The workflow starts with TAP automatically fetching recent green‑status changelists, building a new version, deploying to QA, running automated and manual tests, fixing P0/P1 bugs, promoting to Canary for 3‑6 hours while monitoring eye3, and finally rolling out to production if no errors are reported.

2. Microsoft “test left shift” – Emphasizes pragmatic unit testing, allowing SQL‑resource‑based tests for legacy code, tracks each engineer’s unresolved defect count (teams with >5 defects must fix before new features), monitors long‑standing security bugs, and measures CI/CD pipeline speed to accelerate idea‑to‑production cycles.

3. Microsoft “test right shift” (TIP) – TIP (Production‑environment testing) comprises two practices: protecting the production environment and continuously validating its health and quality, acknowledging the difficulty of reproducing production load and the ever‑changing infrastructure.

4. Feedback loops for developer efficiency – Presents key metrics that differentiate high‑efficiency from low‑efficiency organizations, suggests using these metrics as concrete goals, and encourages teams to identify which cycle times and indicators are most impactful for their context.

5. Atomic commits as a core continuous‑delivery skill – Highlights the principle of “All or Nothing” change lists, urging that each commit contain a single, indivisible task, feature, fix, or refactor, and discusses why small CLs matter, how to shrink them, and when larger CLs may be acceptable.