rbrb

A library for building RoBust RollBack-based networked games.

rbrb is heavily inspired by GGPO and GGRS, but aims to be more reliable and capable.

Assumptions

This library assumes your game is a deterministic Fn(&State, Set<Input>) -> State. We (will) have an additional testing mode that will spend extra cycles on checking that the state is consitent between players and deterministic on the same logical update.

Roadmap

Core Functionality

• Multi-party sync
• Consistent disconnection
• Reconnect disconnected player

Robustness

• Determinism checks
• Checksum propagation
• Debugging failed checks
• Fake a bad network
• Confirmation state

Features

• In-game replays
• Out of game replays
  • Headless
• Spectators
  • Drop in/out

Performance

• Input delta encoding
• Hub and spoke network

License: MIT

lib.rs:

A library for building RoBust RollBack-based networked games.

rbrb is heavily inspired by GGPO and GGRS, but aims to be more reliable and capable.

Assumptions

This library assumes your game is a deterministic Fn(&State, Set<Input>) -> State. We (will) have an additional testing mode that will spend extra cycles on checking that the state is consitent between players and deterministic on the same logical update.

Roadmap

Core Functionality

• Multi-party sync
• Consistent disconnection
• Reconnect disconnected player

Robustness

• Determinism checks
• Checksum propagation
• Debugging failed checks
• Fake a bad network
• Confirmation state

Features

• In-game replays
• Out of game replays
  • Headless
• Spectators
  • Drop in/out

Performance

• Input delta encoding
• Hub and spoke network