Research Of Software Testing And Automatic Model

Software automation testing, as a newly emerging technology in software testing research field, has got preliminary and rapid development, but it still remains many problems to solve, Especially with the rapid development of Internet-Tech, B/S structure and Net-based platform software systems become increasingly popular. Before they are put into actual use, they should be effectively tested firstly, or even performed automatic testing as early as possible. How to check the function of B/S constructed software application system by software automation testing? How to verify that whether the users'requirements on the performance index of software system have been satisfied?Through analyzing the performance test results of application system,testers can easily find out the performance bottleneck and defects of the Web applications,and decide how to optimize the system performance and boost its working efficiency. All of these problems will be researched in this paper.Different web application systems have different orientation and test requirements, for instance, some B/S structure systems require less users'interaction but have to implement more function, so the test working emphasis should be focused on the functional testing, while to those highly-interactive websites, the whole system performance should be put onto the first place and tested in order to find its performance bottleneck in time. Based on all above mentioned requirements, this paper mainly researches the functional testing and performance testing through two different practical projects respectively.For the functional automation testing, this paper firstly specifies the related theory knowledge in the course of implementing automation testing on the B/S constructed software system, and presents the automatic testing methods and content, as well as its process pattern of functional automatic testing, then performs functional automatic testing on a software project that I ever participated--college through the specialized automation test tools, including excellent testing management platform HP mercury Center and the specialized B/S-based automatic test tool HP QuickTest Professional 10.0.First of all, Analyze test requirements of the system through the client specification, create an empty project under Quality Center, design testcases by Equivalent-division method, set up the test steps, then use QTP to record the test script, set checkpoints to inspect the status of the objects and judge whether the operation has been executed or not.After configuring the test parameters of the data-driven application system, we continue to modify and better the script through VBS technology, including importing test data source to perform automatic functional testing.The replay (Execution) process usually has two stages:the test development stage and recession test stage.During the testing development stage,firstly,the Windows application in the same version should be recorded and then replay the test script to verify whether it works as expected,All the work developed in this stage is deemed to be the baseline of checking the whole expected action of Application system.Recession testing is used to expose any difference caused by imported program since that is constructed for the final time,and it can evaluate these difference and judge between existed Bugs and deliberate changes.After analyzing the test results through replaying the test script,We find that the automatic testing efficiency is greatly elevated and reach the aims of funcational automatic testing.For the performance test project case, a university BBS forum Website is selected as the test object, it owns a high interaction degree and attracts a very huge click-rate. In this thesis, an excellent and newly released performance tool, IBM Rational Performance Tester 8.1(RPT 8.1) is chosen for the performance test tool after comparing with other test tools. Create a performance test plan first, and configure the test scenario, then produce 200 virtual users on the basis of 1000 real users on-line through RPT, when the test run finishes, a series of test result data is acquired, such as system thruput, the average response time of the whole pages, the response time of some single pages, standard variance of response time for both system and single pages, as well as the virtual concurrence users'number that the website can afford. According to these test data, this paper analyzes their corresponding performance index, and finally catches its performance data of the whole system for the further positioning in real-time.