Graphical user interface testing

In software engineering, graphical user interface testing is the process of testing a product's graphical user interface to ensure it meets its specifications. This is normally done through the use of a variety of test cases.

To generate a set of test cases, test designers attempt to cover all the functionality of the system and fully exercise the GUI itself. The difficulty in accomplishing this task is twofold: to deal with domain size and with sequences. In addition, the tester faces more difficulty when they have to do regression testing.

Unlike a CLI (command line interface) system, a GUI has many operations that need to be tested. A relatively small program such as Microsoft WordPad has 325 possible GUI operations. In a large program, the number of operations can easily be an order of magnitude larger.

The second problem is the sequencing problem. Some functionality of the system may only be accomplished with a sequence of GUI events. For example, to open a file a user may have to first click on the File Menu, then select the Open operation, use a dialog box to specify the file name, and focus the application on the newly opened window. Increasing the number of possible operations increases the sequencing problem exponentially. This can become a serious issue when the tester is creating test cases manually.

Regression testing becomes a problem with GUIs as well. A GUI may change significantly, even though the underlying application does not. A test designed to follow a certain path through the GUI may then fail since a button, menu item, or dialog may have changed location or appearance.

These issues have driven the GUI testing problem domain towards automation. Many different techniques have been proposed to automatically generate test suites that are complete and that simulate user behavior.

Most of the testing techniques attempt to build on those previously used to test CLI (Command Line Interface) programs, but these can have scaling problems when applied to GUI’s. For example, Finite State Machine-based modeling — where a system is modeled as a finite state machine and a program is used to generate test cases that exercise all states — can work well on a system that has a limited number of states but may become overly complex and unwieldy for a GUI (see also model-based testing).

...
Wikipedia