REPLs All The Way Up: A Rubric For Virtuous Feedback Loops
summary
Argues that the REPL is a practical archetype for better feedback loops and gives a concrete rubric—Read–Eval–Print–Loop—to evaluate them. Demonstrates tools (Lisp/Smalltalk REPLs, SQL shells, strace, debuggers, Honeycomb distributed tracing, feature flags, hot reload/Flutter, dev containers, CI and continuous deployment, generators/property tests) and practices (automate setup, make errors informative). Cites Dartmouth’s DTSS adoption (~80% of students) and a production trace-driven fix that speeds a query ~3x. Shows built artifacts—dev containers and feature-flagged deploys—that actually work.
AI-written summary · about
chapters (23)
- 0:00Feedback loops in software
- 1:04Types of feedback loops
- 3:08History of REPLs
- 10:14Test driven design
- 14:07Why REPLs feel good
- 16:13REPL mechanics
- 25:59Explorational learning
- 31:14Unit tests as exploration
- 32:39Avoiding Sherlock Holmes
- 34:22Tracing and debugging
- 36:32Reload driven development
- 37:16Developing services
- 38:42Development containers
- 39:46Design dialogues
- 41:49Automating test setup
- 43:12Continuous deployment
- 44:15Pull request reviews
- 45:40Production feedback loops
- 47:47Feature flags
- 49:34Tightening loops
- 51:00Enabling others
- 52:45Empowering business teams
- 54:10Chase new questions
