Skip to main content
Version: 2.x

Getting Started With Property Checking

The fundamental idea behind property checking is to test the properties of the target function using random inputs.

So to test a system using property checking, two things are required:

  1. Properties
  2. Generators

A property of a system is a predicate that is always true regardless of the system's input. For example, the addition of two numbers is commutative. So it doesn't matter what numbers we pass to the addition function, for any pair of a and b, the result of add(a, b) is always the same as add(b, a):

def add(a: Int, b: Int): Int = ???

def is_add_commutative(a: Int, b: Int): Boolean =
add(a, b) == add(b, a)

The is_add_commutative predicate takes two inputs and checks if the add function is commutative or not. To check this property, we need some random integer pairs. This is where generators come in.

The Gen[A] data type is used to generate random values of type A. ZIO Test provides numerous Gen instances for common types:

import zio.test._

val intGen: Gen[Any, Int] = Gen.int
val stringGen: Gen[Sized, String] = Gen.string

It is also composable, so we can combine them to generate random values of more complex types:

val stringIntGen: Gen[Sized, (String, Int)] = stringGen <*> intGen

case class Person(name: String, age: Int)
val personGen: Gen[Sized, Person] = stringIntGen.map(Person.tupled)

ZIO Test provides the check function for this purpose. It takes a list of generators and provides them to another taken function, which is a property checker:

def property[T1, T2](input1: T1, input2: T2, ...): Boolean = ???

val input1Gen: Gen[_, T1] = ???
val input2Gen: Gen[_, T2] = ???

check(input1Gen, input2Gen, ...) { (input1, input2, ...) =>
assertTrue(property(input1, input2, ...))
}

In our example, the is_add_commutative predicate takes two inputs. So we need to pass two generators of type Int to the check function:

def add(a: Int, b: Int): Int = ???

test("add is commutative") {
check(Gen.int, Gen.int) { (a, b) =>
assertTrue(add(a, b) == add(b, a))
}
}

Number of Samples

In the previous example, we used check to test if the add function is commutative. In other words, we try to generate samples of random pairs of integers and try to falsify the is_add_commutative predicate. If we find a pair of integers that falsifies the predicate, then we know that the property is violated.

By default, the check function, try to generate 200 samples. We can change this by using the sample test aspect:

import zio.test._

object AdditionSpec extends ZIOSpecDefault {
def spec =
test("add is commutative") {
check(Gen.int, Gen.int) { (a, b) =>
assertTrue(add(a, b) == add(b, a))
}
} @@ TestAspect.samples(10)
}

To debug the test, we added a println statement inside the check function to see the generated samples.