Automating Product Deprecation
summary
Meta automates deprecation by running a dead-data cleanup system that scans production stores, notifies owners, blocks reads/writes after one month and then deletes and verifies records; and a dead-code removal with Glean that combines static analysis, runtime logs, a code-text index and job metadata to file automated change requests and has applied hundreds of thousands of removals with a low error rate. An internal deprecation tool maps endpoints, JS, APIs, business logic and data, shows blocking signals, and lets engineers accelerate or override automation.
AI-written summary · about
chapters (7)
- 0:00Start
- 3:54Process
- 4:49Dead data cleanup
- 5:22Measure static usage (code) and runtime usage (traffic)
- 7:16Dead code cleanup
- 10:35Multi-language
- 17:13Takeaways
related talks
31mLifecycle of Reusable Automations: Track, Maintain, Deprecate
44mWe Hate Code: The !Joy of Maintaining Dead Code
38mDeleting Code Is Hard And You Should Do It
31mHow We Deprecated 500+ Articles: Cleaning Up Documentation at Scale
29mDevTools for the decade to come
09mThe Robots are coming for your job, and that's okay
10mData Access Automation at Scale
33mKeep your dependencies in check