Why and When GUI Test Automation Fails in Practice and Our Solutions to The Problem

• Main Authors: Tzu-Chao Wang, 汪子超
• Other Authors: Yung-Pin Cheng
• Format: Others
• Language: en_US
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/93ypue

碩士 === 國立中央大學 === 資訊工程學系 === 105 === GUI (graphic user interface) testing and integration testing play a pivotal position in software industry. The GUI test automation, should ideally be a powerful solution to GUI / integration test. It has been actually adopted by some commercial companies for formal use. However, the adoption often appears to be problematic and difficult in practice. In a GUI automation test case, the interactions between user and the system under test can be recorded with “Capture / Replay” software, or be manually written into test scripts. Afterwards, testers are required to add assertion to determine the correctness of the test runs. By regressing these GUI tests, the quality and correctness of software product can be ensured by performing test runs before each iterative software production release. However, when it comes to frequent changes of UI elements, evolvement of user interface, and the replaying environments are different from the ones when scripts were recorded or written, the usability, robustness, and flexibility of most of the GUI test automation tools remain problematic and are yet not sufficient enough for practical usage. This paper describes and analyzes the problems of GUI test automation tools, and the real successful cases of applying a capture / replay software – Korat to the industry 4.0 production line of industrial personal computer (IPC) company – ADLINK and the Taiwan Semiconductor Manufacturing Company – TSMC. The experiences and works presented in this paper are provided as a reference for the development and the commercial application of GUI test automation tools.