Property-based testing generates random inputs and checks whether general properties hold, reducing unit-test maintenance while increasing assurance. section 6.10 section 6.3
Use it with these practical constraints:
- Use proptest to generate test data, verify results, and report the minimum input that reproduces an error. section 6.3
- Keep tests for well-known values when checking specification compliance; generated cases are not a complete replacement. section 6.3
- Tune the number of generated cases because broader random testing can consume more CPU without covering every possible outcome. section 6.3
- Watch arithmetic semantics: debug-mode Rust tests use checked arithmetic, so generated integer inputs can expose overflow. section 6.3
Finding an overflowing adder
A proptest test compares a generic adder with ordinary addition over generated i64 values. The test finds overflow and reports a minimal failing pair; changing the implementation and oracle to wrapping addition makes the test pass. section 6.3
Listing 6.3 Code listing of adder with proptest
pub fn add<T: std::ops::Add<Output = T>>(a: T, b: T) -> T {
a + b
}
#[cfg(test)]
mod tests {
use super::*;
use proptest::prelude::*; #1
proptest! {
#[test]
fn test_add(a: i64, b: i64) { #2
assert_eq!(add(a, b), a + b); #3
}
}
}
Inspect the focused Rust implementation using the proptest macro, generated i64 arguments, and an equality property. section 6.3
