Digesting MRI by Studying Alternative Ruby Implementations
summary
Reimplements hash maps in JRuby (prototype in Ruby then port to Java), tweaks Rubinius and Opal behavior, and lands a small fix in Artichoke; shows that array.minmax is 1.8× slower than separate calls, that Ruby 2.4’s new approach makes hash tables 40% faster, and that MRI’s C patch is ~1500 lines while the JRuby work is a 700‑line JRuby patch that needs a temporary revert for concurrency robustness; also covers learning to write a few C changes and the actual speed and correctness outcomes.
AI-written summary · about
chapters (22)
- 0:00Intro
- 1:30Experience
- 1:43Leadership
- 1:55Mentor
- 2:09Programming
- 2:30How?
- 2:44Digesting MRI by Studying Alternative Ruby Implementations
- 4:31Ruby is a computer program
- 6:27Embrace Failure
- 6:45Challenged
- 12:16Three Lines of Code
- 13:00It aims to be a complete, correct and fast implementation of Ruby
- 13:4740 % Faster
- 18:04Cache Locality
- 20:42Refactoring
- 21:32Release Early, Release Often
- 23:09Performance optimisation is a game in which you want to achieve a highscore
- 24:52Prototype
- 25:12Ask for Help
- 25:27What defines a "Senior Developer"?
- 25:46Never stop learning
- 26:06Community
