F# in Social Gaming
summary
Backend for social slots and mobile games runs on F# and an actor-based agent/mailbox processor model supporting ~1,000,000 DAU; the slots engine and bonus-game logic are expressed as discriminated unions and units of measure to eliminate class-of-bugs, a custom DynamoDB SQL-like DSL is built, and a state-for-server actor design yields ~5x efficiency and ~60% average latency reduction while removing a Couch cluster.
AI-written summary · about
chapters (19)
- 0:00Social gaming at GAM
- 1:33What is F#
- 2:58F# in slot engine
- 5:23Game rules with unions
- 7:10Game state with records
- 8:15Bonus mechanics in F#
- 14:28Hippie Monsters overview
- 18:15State challenges in Hippie Monsters
- 19:20Actor model for servers
- 21:04Actor model in F#
- 22:27Worker and gatekeeper actor types
- 24:12Actors with mailbox processors
- 26:03Async workflows for I/O
- 28:56Actor concurrency benefits
- 30:20Quest and achievement design
- 33:32Fact processors for data
- 36:01Complex systems with unions
- 36:43DSL for DynamoDB
- 39:12Q&A on F#
related talks
01h 01mYou got FP in my Enterprise!
01h 00mF# for C# programmers
01h 01mStay with C# or switch to F#? Let's fight! Let's vote!
07mOver 6 Months of ReasonML at Codeheroes
54mEmbracing Change
51mSide Projects: The Unsung Heroes of Software Development
42mBuilding a Company on Scala
35mBuilding a Player-Focused Data Team