cliffy-test 0.3.1

Algebraic testing framework for Cliffy - tests as geometric invariants
Documentation

cliffy-test

Algebraic testing framework for geometric algebra applications.

Overview

cliffy-test provides property-based testing with geometric algebra semantics. Tests are expressed as geometric invariants with probabilistic guarantees.

Invariant Categories

Impossible (P = 0)

Properties that must never fail:

use cliffy_test::invariant_impossible;

invariant_impossible!(
    "Rotor normalization preserves unit magnitude",
    || {
        let r = random_rotor();
        let normalized = r.normalize();
        (normalized.magnitude() - 1.0).abs() < 1e-10
    }
);

Rare (0 < P << 1)

Properties with bounded failure probability:

use cliffy_test::invariant_rare;

invariant_rare!(
    "Hash collision rate",
    probability = 0.001,  // Must fail less than 0.1% of the time
    || {
        let a = random_state();
        let b = random_state();
        a.hash() != b.hash()  // Different states should have different hashes
    }
);

Emergent (P > 0)

Valid but unpredicted behaviors worth tracking:

use cliffy_test::emergent;

emergent!(
    "Convergence happens in fewer than 10 iterations",
    || {
        let iterations = run_consensus();
        iterations < 10
    }
);

Manifold Testing

Test that states lie on expected geometric manifolds:

use cliffy_test::{Manifold, unit_sphere, rotor_manifold};

// Unit sphere: pure vectors with magnitude 1
let sphere = unit_sphere();
assert!(sphere.contains(&unit_vector));

// Rotor manifold: unit magnitude elements
let rotor_m = rotor_manifold();
let result = rotor_m.verify(&rotor.coefficients());
assert!(result.is_pass());

// Custom manifolds
let custom = Manifold::new("MyManifold")
    .with_magnitude(2.0)
    .with_pure_vector()
    .with_tolerance(1e-8);

Random Generators

Generate random geometric objects for property testing:

use cliffy_test::{random_ga3, random_rotor, random_unit_vector};

let mv = random_ga3();           // Random multivector
let r = random_rotor();          // Random unit rotor
let v = random_unit_vector();    // Random unit vector

Geometric Assertions

use cliffy_test::{assert_ga3_equal, assert_magnitude, assert_zero};

// Compare multivectors within tolerance
let result = assert_ga3_equal(&a, &b, 1e-10);
assert!(result.is_pass());

// Check magnitude
let result = assert_magnitude(&rotor, 1.0, 1e-10);
assert!(result.is_pass());

// Check if approximately zero
let result = assert_zero(&difference, 1e-10);
assert!(result.is_pass());

Test Results

Results include geometric error information:

use cliffy_test::TestResult;

let result = manifold.verify(&point);
if result.is_fail() {
    let error = result.error().unwrap();
    println!("Distance from manifold: {}", error.distance);
    println!("Correction gradient: {:?}", error.gradient);
}

Integration with QuickCheck

Works with standard property testing:

use quickcheck_macros::quickcheck;
use cliffy_test::ArbitraryRotor;

#[quickcheck]
fn rotor_composition_is_associative(a: ArbitraryRotor, b: ArbitraryRotor, c: ArbitraryRotor) -> bool {
    let left = (a.0 * b.0) * c.0;
    let right = a.0 * (b.0 * c.0);
    (left - right).magnitude() < 1e-10
}

Testing Leptos Components

Test geometric state invariants in Leptos apps:

use cliffy_test::{invariant_impossible, test_impossible};
use cliffy_core::{Behavior, GeometricState};

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_rotation_state_invariants() {
        // Test that rotation state always stays normalized
        let report = test_impossible(
            "Rotation state magnitude is always 1",
            || {
                let state = GeometricState::from_vector(
                    rand::random::<f64>(),
                    rand::random::<f64>(),
                    rand::random::<f64>(),
                );
                let normalized = state.normalize();
                (normalized.magnitude() - 1.0).abs() < 1e-10
            },
            1000,
        );

        assert!(report.verified, "Invariant violated: {:?}", report);
    }
}

Testing Yew Components

Validate CRDT convergence in Yew collaborative apps:

use cliffy_test::{test_impossible, test_rare};
use cliffy_protocols::GeometricCRDT;

#[cfg(test)]
mod crdt_tests {
    use super::*;

    #[test]
    fn test_crdt_convergence() {
        // CRDTs must always converge regardless of merge order
        let report = test_impossible(
            "CRDT merge is commutative",
            || {
                let mut a = GeometricCRDT::new("a", rand::random());
                let mut b = GeometricCRDT::new("b", rand::random());

                a.add(rand::random());
                b.add(rand::random());

                let ab = a.clone().merge(&b);
                let ba = b.clone().merge(&a);

                (ab.value() - ba.value()).abs() < 1e-10
            },
            1000,
        );

        assert!(report.verified);
    }

    #[test]
    fn test_sync_latency() {
        // Sync should complete quickly in most cases
        let report = test_rare(
            "Sync completes in under 100ms",
            || {
                let start = std::time::Instant::now();
                // Simulate sync operation
                std::thread::sleep(std::time::Duration::from_millis(rand::random::<u64>() % 150));
                start.elapsed().as_millis() < 100
            },
            0.3,  // Allow up to 30% slow syncs
            100,
        );

        assert!(report.verified);
    }
}

License

MIT