Run on Embedded Devices

Embedded devices often have tight limits on both program size and runtime memory (RAM).

The following techniques can reduce Rhai’s footprint on such a device, especially when scripts are prepared ahead of time.

Run Rhai Grain Bytecodes

The standard interpreter evaluates a script via its AST, which is much larger than the source text must all be kept in memory.

Instead, use Rhai Grain to transpile scripts into bytecodes and execute them with the VM.

The bytecode instruction stream is a compact representation of the script’s AST, avoiding the need to keep it on the device.

Scripts can be transpiled separately on a development or build machine, and the bytecodes shipped to the device for execution.

Enable the grain feature for Rhai Grain.

Build with no_ast

When scripts are transpiled to Rhai Grain bytecodes, enable no_ast in the device build.

This removes the AST, tokenizer, parser and optimizer, significantly reducing the runtime’s program size.

The device can only run bytecodes; it cannot compile or evaluate Rhai source text itself.

Lazy-Load the Standard Library

Use Engine::new_raw with StandardPackage::new_lazy to avoid eagerly loading the entire standard library:

Functions are looked up only when called, which can significantly reduce runtime heap usage on memory-constrained devices (lower by 15-20x). The trade-off is slower resolution the first time a function is used.

let mut engine = Engine::new_raw();

StandardPackage::new_lazy().register_into_engine(&mut engine);
//               ^^^^^^^^^^ lazy-loaded Standard library