deterministic-rng

Seedable, platform-identical PRNG for deterministic game simulations.

Produces bit-identical output on x86, ARM, and WASM for the same seed. Designed for lockstep multiplayer, deterministic replays, and save/load reproducibility.

Status

⚠️ Early development — API is unstable but core generator works.

Features

• GameRng — xoshiro256** with version-pinned, forever-stable output
• Range sampling with zero modulo bias (Lemire's method)
• Damage spread for RTS combat: spread(base, variance)
• Deterministic Fisher-Yates shuffle
• #![no_std] compatible

Planned

• Weighted selection from slices
• Weighted index selection for probability tables
• Serde serialization for save/load

Usage

use deterministic_rng::GameRng;

let mut rng = GameRng::from_seed(42);

// Same seed always produces same values — on every platform.
let damage = rng.spread(100, 15); // 85..=115
let crit = rng.next_bounded(100) < 5; // 5% chance

Design Documents

Design rationale is in the Iron Curtain Design Documentation:

• D009 — Deterministic simulation
• D076 — Standalone crate extraction

License

Licensed under either of:

• Apache License, Version 2.0 (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
• MIT License (LICENSE-MIT or http://opensource.org/licenses/MIT)

at your option.

Contributing

Contributions require a Developer Certificate of Origin (DCO) — add Signed-off-by to your commit messages (git commit -s).

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in this crate by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.