Let's talk concurrency
summary
Explains why moving from many processes to single-process multi-core concurrency matters as machines go from 4 cores to 16–24, shows a CSV-sum example split in halves, and demonstrates a three-increment race that produces 2 when using unsynchronized threads. Compares common approaches: mutexes (deadlocks, developer-placed locks), software transactional memory (optimistic retries and commit overhead used by Clojure), and message-passing actors (Erlang/Go channels/Elixir) with Ruby libraries like Celluloid and notes which designs are easy to distribute.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 3:26Declarative model
- 9:43Dataflow variables
- 13:42Side effects
- 17:20How it works
- 20:25Transactional Memory
- 25:48Server
- 30:41Dataflow
- 32:40Final considerations
