Ruby Persistence in MagLev
summary
MagLev stores Ruby objects as a shared repository and persists by following references, using persist by reachability with copy-on-read pages and commit-by-reachability; transaction logs plus warm-standby replay provide durability and optimistic concurrency with optional explicit locks. Ruby compiles to Smalltalk bytecodes, uses bridge methods for variadic calls, and native generation yields about 2–3x native speed versus the bytecode interpreter with a typical 20-instruction send path. An alpha to a small group is imminent.
AI-written summary · about
chapters (25)
- 0:00Intro
- 3:20Why MagLev
- 6:15Status
- 8:25Ruby Spec Update
- 10:42Ruby Object Persistence
- 14:34MagLev Execution Technology
- 15:20Object Persistence
- 17:26Execution Technology
- 22:26Compilation Technology
- 23:00MagLev VM
- 25:54Native Code
- 26:46Message Send
- 27:39Bridge Methods
- 28:48Ruby Message Send
- 30:15Method Context
- 31:48Object Memory
- 33:47Variable Size Objects
- 35:03Special Objects
- 36:55VariableSize Objects
- 37:43Dynamic Instances
- 39:22Small Doubles
- 40:22Reuse Small Talk
- 41:02Message Environments
- 41:55Hash
- 44:51Shared Memory Cache
related talks
33mData Migrations with MagLev
50mAllow me to reintroduce myself, my name is MagLev
34mMagLev - From Download to Deploy
43mTANSTAAFL - Implementing Maglev Ruby on top of GemStone
33mRuby Changed My Life
30mWould you like to forget about the database?
35mPersisting Relations Across Time and Space
42mRuby 124c41+