Based On The Weights To Improve The Gui Testing Framework

The extensive use of graphic user interface (GUI) in applications brings great challenges to today's software development and testing. Normally, GUI has a portion of 60% in total codes of a single application. Therefore, GUI will play a significant role in software development and testing. Correctness, reliability and efficiency are emphasized in software quality assurance.This paper has analyzed a GUI testing framework called"Daily Automated Regression Tester"(DART) presented by A. M. Memon and Qing Xie of Maryland University, which provides efficient and usable solution for GUI testing. Also the performance of the framework with different factors is measured. This paper gives a quatitative analysis of influences on case generation and test oracle to GUI testing by studied the given empirical results.However the original framework is not perfect yet. In order to improve the performance and cost on error detection, an improved framework for GUI testing based on weights is introduced in this paper. The framework exploits the event driven GUI testing and DART test framework, while combining them with random testing. According to the characteristic of GUIs, the framework analyzes the complexity of GUI code. The weights of GUIs are given. Then events are distributes into cases in different sizes according to the GUI weights. Meanwhile, test oracles of different levels are also defined.The example of open-source software will prove that the improved framework will increase the number of error detected and decrease the cost of the testing. So the total performance is raise. The evaluating module embedded in the framework will judge the reliability of GUI in a single GUI testing by Bayes method.The improved framework for GUI testing based on weights overcome the defects and improves the performance and flexibility of the original framework. It is a new exploration in GUI testing.