You can use rrepl (the Rust REPL, built from crates/reussir-repl) to run the Reussir REPL. Build it with cmake --build build --target rrepl; the binary is at build/target/<profile>/rrepl (e.g. build/target/release/rrepl).

It has several CLI options (rrepl --help for the full list):

Usage: rrepl [OPTIONS]

  -O, --opt-level <OPT_LEVEL>   none, default, aggressive, size, tpde
  -l, --log-level <LOG_LEVEL>   error, warning, info, debug, trace  [default: warning]
  -i, --input <INPUT>           Input file for line-by-line execution (script mode)
      --no-tui                  Use the plain line-based prompt (implied by script mode / piped I/O)

The REPL can read from a file (-i) or interactively. It provides :dump commands to inspect interpreter state.

Assume the input file tests/integration/repl-rs/polymorphic_math.repl:

fn cos_squared(x: f64) -> f64 { let y = core::intrinsic::math::cos(x, 0); y * y }
fn sin_squared(x: f64) -> f64 { let y = core::intrinsic::math::sin(x, 0); y * y }
:dump context
fn poly_add<T : Num>(x : T) -> T { (cos_squared(x as f64) + sin_squared(x as f64)) as T }
poly_add(127)
poly_add(127.0)
:dump compiled

We can run it with:

build/target/release/rrepl -i tests/integration/repl-rs/polymorphic_math.repl -lerror

And get output like:

Definition added.
Definition added.
fn #cos_squared(v0 (x): f64) -> f64 {
    let v1 (y) = intrinsic#math#cos#0(v0 : f64) : f64;
    (v1 : f64 * v1 : f64) : f64
}

fn #sin_squared(v0 (x): f64) -> f64 {
    let v1 (y) = intrinsic#math#sin#0(v0 : f64) : f64;
    (v1 : f64 * v1 : f64) : f64
}

Definition added.
1 : i64
1.0 : f64
=== Compiled Functions ===
 - _RC11cos_squared
 - _RC11sin_squared
 - _RIC8poly_adddE
 - _RIC8poly_addxE

The REPL's own lit suite lives under tests/integration/repl-rs/ and runs as part of cmake --build build --target check once rrepl has been built.

npx skills add smithery/schrodingerzhu --skill run-repl