第9章ソフトウェア工学

A crucial role in software development, the process of unit testing involves testing individual units of code, such as methods and functions, and it is usually done by developers during the development of the software or when updates are made. Software developers write scripts or code that test the functionality of a specific piece of code. Unit tests are typically added to a regression test suite, so they can be run again after each change to the source code to verify that the change did not break any existing functionality. Imagine a function that calculates the area of a rectangle. A unit test would verify that the function returns the correct area for different width and height values.

Unit testing offers several benefits:

• Early bug detection: Unit tests help identify bugs early in the development process, when they are easier and less expensive to fix.

• Improved code quality: The process of writing unit tests encourages developers to write clean, modular, and well-documented code.

• Maintainability: Unit tests serve as living documentation, clarifying the intended behavior of code components and making future modifications easier.

In the automobile industry, most new cars now include a lane detection system (LDS). This system includes code that verifies video images that are received by the automobile. The software checks to see if the vehicle crosses one of the lines painted on the road. If so, it activates a warning system to let the driver know. Units test for such a system include the following:

• An appropriate line is recognized.

• A warning occurs if a line appears to be crossed.

• The warning ends after the designated period of time.

• When the LDS system is turned off, the warning is not triggered.

Software engineers as well as testers perform these tests when the code for the LDS system is created as well as any time any updates are made to the system.

Proper unit testing improves the quality of a software system and helps discover bugs early. In simple terms, a software bug is an issue or error with software programming. To find bugs, it is important to review and run the code.

The range of a unit test is typically measured in terms of code coverage, which is the measurement of the percentage of code that is activated or reviewed by the test. Common types of code coverage are statement coverage, line coverage, and path coverage. Measurement of the percentage of statements that are activated at least once when you run all tests is statement coverage. Measurement of the percentage of activated lines of source code that are tested is line coverage. Measurement of the percentage of paths through source code that you go through is path coverage. As shown in Figure 9.18, with path coverage, if you have two conditional statements, one after the other, there are typically four paths through this code:

1. Both conditions are true.

2. The first condition is true and the second is false.

3. The first condition is false and the second is true.

4. Both conditions are false.

Path coverage is more demanding and typically requires more tests than statement or line coverage, but tests are better. You need a minimum of four tests to achieve 100% path coverage, but a minimum of two tests to achieve the same level of statement coverage.

Unit testing is so common that major programming languages provide some support for automated unit testing. An example is JUnit, a unit testing framework for the Java programming language, which is shown in Figure 9.19. JUnit facilitates the writing and managing of tests, and it can be run in some development environments quickly, typically via a single click. Many of the other major programming languages have similar frameworks or tools. There are also third-party software products that can be used to help with unit testing.