Regression Testing for Web Applications Using Reusable Constraint Values

Abstract

Companies that provide web applications need to perform frequent regression testing because companies often encounter various security attacks and frequent feature update demands from users. Typically, such applications require regression testing processes that require minimal test effort because they have already been deployed and used in the field. In our previous work, we presented an efficient regression testing approach that allows us to focus on the areas of code that have been changed and to regression test them to address this problem. While our experiment results showed that this approach can be efficient in saving the cost of regression testing by reducing the number of test paths necessary for the modified program, we also learned that resolving input constraints requires a lot of effort. In this paper, to accommodate further savings with regression testing, we present a technique that identifies reusable constraint values for regression test cases. The technique finds variables where input values from the previous version can be reused to execute the regression test path for the new version. Our empirical results show that a large number of variable constraints can be reused from the previous version's test cases, thus we can reduce a great deal of effort for resolving constraint values for those variables when we test a new version of the application.