The Joys and Perils of Interactive Development
summary
NASA launches Deep Space 1 on Oct 24, 1998 — a $150M, four-year, minimum-viable mission that runs an autonomous Remote Agent written in Lisp; on May 17, 1999 the agent controls the spacecraft but later stalls and a race condition 45–60 million miles away is diagnosed and patched via a live REPL. The practical claims: tools.namespace parses ns forms to reload dependents, Component lifecycle wraps stateful resources for stop/start, and the reloaded pattern provides a single reset that stops the system, refreshes code, and restarts it.
AI-written summary · about
chapters (12)
- 0:00Intro
- 8:09Deep Space One
- 10:45Clean State
- 15:16Unique to Closure
- 19:11Reloading Code
- 26:52Component
- 29:25Function Reset
- 31:15Defining Functions
- 32:48Reuse
- 34:43Records Maps
- 37:17Source Code vs Program
- 38:20Conclusion
related talks
33mAgent Experience
14mAgent-Driven Software Development
13mRun, Build and Grow Small Systems Without Leaving Your Text Editor
30mExperiments in Engineering without reading code
15mLive Programming a Live Programming Environment: An Experience Report
51mAI-Powered Development: Hands-On Techniques for Immediate Impact
01h 29mTooling for the JavaScript Era
18mCollaborative AI Engineering: One Dev, Two Dozen Agents, Zero Alignment