Side Effects as Behavior
summary
Shows a concrete solution: replace scattered log calls and passing of parameters across 100 screens with an actor-owned flow logger that holds context (selected option, experiment bucket) and composes EventGroup protocol values into parameters sent via sync plugins (Firebase, OSLog, test stub). Demonstrates logging meaningful thresholds (25/50/100% scroll), explains how actors provide ownership, UI-independence and testability, and claims this pattern generalizes to networking, caching, and AB experiment state.
AI-written summary · about
chapters (12)
- 0:00Introduction and motivation
- 1:29Logging as a side effect
- 6:46Problems with unstructured logging
- 13:59Designing structured logging
- 15:26Actors and concurrency basics
- 17:56Actors for ownership and state
- 19:43Requirements for logging architecture
- 20:48Core logging architecture overview
- 23:42Event groups and protocols
- 25:30Concrete event group examples
- 27:21Extending the pattern beyond logging
- 28:05Takeaways and conclusion
