3.13.1.1. Arithmetic Operation Mutant¶

The arithmetic operation mutant checks if an arithmetic operation is being tested by your test suite. This mutator replaces an arithmetic operation with one of its members. The mutator is composed of 2 sub-mutators, AOD_1 and AOD_2, that mutate the operation to its first and second member respectively.

If the test suite detects the change i.e. mutanting the code results in a test failing, then the test suite passes. The terminology is that the test suite “killed the mutation”. This results in a higher mutation coverage score. When the test suite fails to detect the mutant, this results in a lower mutation coverage score. In such case, we need to write additional test case assertions to test for the arithmetic operation in question.

For example

int a = b + c;

will be mutated to

1int a = b;        // replaced with first member

and to

1int a = c;        // replaced with second member

Here, b``is the first member and ``c is the second member.

This way, mutation testing ensures that the arithmetic operator is tested for its intended behavior.

3.13.1.2. Example Code 1: Arithmetic Operation Mutant¶

As an example, we want to write a function that takes two numbers and returns the sum.

1     public static int Addition(int num1, int num2) {
2             int sum = 0;
3
4             sum = num1 + num2;         // --> math operation
5
6             return sum;
7     }

Now if we execute mutation testing it will mutate the code as follows:

Replacing the arithmetic operation with first member:

1     public static int Addition(int num1, int num2) {
2             int sum = 0;
3
4             sum = num1;         // --> math operation
5
6             return sum;
7     }

Replacing the arithmetic operation with second member:

1     public static int Addition(int num1, int num2) {
2             int sum = 0;
3
4             sum = num2;         // --> math operation
5
6             return sum;
7     }

If we have no test cases execute this code, the test process will generate the following mutations in the LINES_NOT_TESTED group under the Mutations List tab. (The icons for unresolve mutants are shown as “red bugs”.)

Here, we can see that the HINT suggests that we write test case assertions to test the arithmetic operation for intended behavior.

The following test case will kill both of these mutations:

1     // testAddition tests for adding two numbers
2     @Test
3     public void testAddition() {
4             // testing if 5+10 == 15
5             assertEquals(15, SimpleExample.Addition(5, 10));
6     }

If we run the mutation testing again then we will not get any of the previous mutations in the LINES_NOT_TESTED group under the Mutations List tab.

3.13.1.3. Logical Expression Mutant (Remove Conditionals)¶

The logical expression mutator (a.k.a. remove conditionals mutator) checks if a logical expression is properly tested by your test suite. This mutator replaces the logical expression with either TRUE or FALSE and then runs your test suite with the mutant.

For example replacing logical expression with TRUE condition:

1if (a == b) {
2// do something
3}

will be mutated to

1if (true) {
2// do something
3}

For example replacing logical expression with FALSE condition:

1if (a == b) {
2// do something
3}

will be mutated to

1if (false) {
2// do something
3}

If there is more than one logical expression then each expression will be mutated in separate runs of the test suite. The logical expression mutator also mutates the bytecode instructions for order checks (e.g. <=, >).

If there are more than one logical expression in the same statement, then the generated mutants will be in order of the logical expressions in the statement. Keep in mind, for multiple logical expressions, you must test each and every one of the expressions.

3.13.1.4. Example Code 2: Logical Expression Mutant (Remove Conditionals)¶

As an example, we want to write a function that takes a number and returns TRUE if the number is positive and FALSE if the number is zero or negative.

1     public static boolean PositiveCheck(int number) {
2             if (number > 0) {                               // --> true or false (2 cases)
3                     return true;                            // positive number
4             }
5             else {
6                     return false;                           // zero or negative number
7             }
8     }

Executing mutation testing will mutate the code as follows.

Replacing the logical expression with TRUE:

1     public static boolean PositiveCheck(int number) {
2             if (true) {                             // --> true or false (2 cases)
3                     return true;                            // positive number
4             }
5             else {
6                     return false;                           // zero or negative number
7             }
8     }

Replacing the logical expression with FALSE:

1     public static boolean PositiveCheck(int number) {
2             if (false) {                            // --> true or false (2 cases)
3                     return true;                            // positive number
4             }
5             else {
6                     return false;                           // zero or negative number
7             }
8     }

Without tests to execute this code, it will generate the following mutations in the LINES_NOT_TESTED group under the Mutations List tab.

The icons for unresolve mutants are shown as “red bugs”.

In order to fix the mutations, we can write the following test case assertions:

 1     // testEvenOddCheckWithEvenNumber tests for positive number
 2     @Test
 3     public void testPositiveCheckWithPositiveNumber() {
 4             assertTrue(SimpleExample.PositiveCheck(10));
 5     }
 6
 7     // testEvenOddCheckWithOddNumber tests for zero
 8     @Test
 9     public void testPositiveCheckWithZero() {
10             assertFalse(SimpleExample.PositiveCheck(0));
11     }
12
13     // testEvenOddCheckWithOddNumber tests for negative number
14     @Test
15     public void testPositiveCheckWithNegativeNumber() {
16             assertFalse(SimpleExample.PositiveCheck(-5));
17     }

If we run the mutation testing again then we will not get any of the previous mutations in the LINES_NOT_TESTED group under the Mutations List tab.

3.13.1.5. Example Code 3: Multiple Mutants in One (EvenOddCheck)¶

We can have programming statements where we have both arithmetic operation(s) and logical expression(s). In such cases, mutation testing will return mutants for each type and list them under the Mutations List tab.

For example, we want to write a function that takes a number and returns TRUE if the number is even and FALSE if the number is odd.

1     public static boolean EvenOddCheck(int number) {
2             if (number % 2 == 0) {        // --> arithmetic operation (2 cases); logical expression (2 cases)
3                     return true;               // even number
4             }
5             else {
6                     return false;              // odd number
7             }
8     }

As a result, it will generate the following mutations under the Mutations List tab:

In order to fix the mutations, we can write the following test case assertions:

 1     // testEvenOddCheckWithEvenNumber tests for even number
 2     @Test
 3     public void testEvenOddCheckWithEvenNumber() {
 4             assertTrue(SimpleExample.EvenOddCheck(10));
 5     }
 6
 7     // testEvenOddCheckWithOddNumber tests for odd number
 8     @Test
 9     public void testEvenOddCheckWithOddNumber() {
10             assertFalse(SimpleExample.EvenOddCheck(5));
11     }

If we run the mutation testing again then we will not get any of the previous mutations in the LINES_NOT_TESTED group under the Mutations List tab.

3.13.1.6. Example Code 4: Loop Conditions (optional)¶

A loop contains a logical expression that needs testing. However, this is often optional and does not negatively impact mutation coverage.

For example:

1for (int i = 0; i < 10; i++)

Keep in mind, if you don’t test the terminating condition of a for loop then your code might get stuck in an infinite loop.