International Journal of Advances in Computer Science and Its Applications
Author(s) : AASTHA KAUR , NAVDEEP SINGH
To reduce the effort, testing cycle time & % of human errors that can easily creep in while comparing the results of Regression Test Suite, a thought process was put into designing & implementing an Automation Framework for the purpose. A lot of work and research has already being done for the Execution phase of Regression Testing wherein two parallel sides – Test & Prod are setup & Test Cases executed by firing the same one after the another & results stored. A large number of Regression Automation Tools are available in market like, QTP, Selenium, WATIR etc, to cover this up. Contrary to this very less work is available & very less has been thought about the Comparison phase wherein Test Results thus generated have to be compared to produce a summary report for QA Testers to analyze which they can further categorize into Expected & Unexpected Breaks & then reach out to Development for investigation & thus complete the end-to-end life cycle of Regression Testing. With advent of IT and shift of focus toward Financial Banks & Institutions, a need is felt to have some faster & feasible way to compare records with high volume. That is the starting point for this paper under which an Automation Framework for Comparison Phase of Regression Testing is built in Perl, that could easily cover records of any volume. Use of Industry Compliant Methodology, named Best Match, made the framework even more flexible for scenarios having duplicate records on either of the two parallel sides. Best practice Data Structures like Hash are being used in the implementation that have fasten up the parsing & key pattern filtering, hence lowering down the overall comparison & summary generation time. Use of programming language Perl has made the framework platform or operating system independent as the implementation code can easily be run on any OS, like Unix, Sun, Windows.Comparing results of Regression Test Suite is far more complex than it seems. The below paper aims toward designing and implementing a framework that could simplify this complexity.