JRuby in the Wild
summary
Describes concrete moves: swaps the Ruby C search stack for Lucene bindings via JRuby, implements a Java trie search for spell correction, and packages Java jars as gems. Reports object-creation benchmark numbers: MRI ~13s, JRuby ~10.9s with objectspace on, Java ~4.1s, and Ruby memory blowup when nodes reach millions. Deployment runs on Debian with Sun JVM as standalone daemons. Recommends bindings over ports and using Java for heavy memory and performance hotspots.
AI-written summary · about
chapters (32)
- 0:00Introduction
- 1:20About Casttv
- 3:33What is JRuby
- 4:46Enterprise Motivation
- 5:38Rails vs Java
- 7:24Picking the right tool
- 8:36Ruby
- 9:14Performance
- 10:04Library Support
- 11:10Improving Ruby
- 11:58Expanding the tool set
- 13:03Rails
- 14:24Search
- 16:10Solar
- 17:59Porting
- 18:51API
- 19:26Bindings
- 19:43Spell checker
- 20:39Algorithm performance
- 21:05ternary search
- 22:06pure ruby
- 23:30JRuby in line
- 25:44Dictionary performance
- 28:04Installation
- 28:34subversion
- 28:49rubygems
- 30:24sane build strategy
- 31:48jar to gem
- 36:46ruby to drb
- 37:28Resources
- 40:03Integration
- 42:08Deployment
