Various regression test selection strategies have been developed to establish the correctness of software systems upon deployment of COTS upgrades. These strategies recommend a smaller regression test suite to be reexecuted. We present a process that identifies the patch impact on the system and localizes the source of potential errors without executing the system test suite. The impact is analyzed based on verifying the behavior of affected COTS API execution-sequences. An API executionsequence is called affected when one or more of the altered functions of the COTS upgrade are part of COTS API execution-sequence that interfaces with application components. A wrapper-based passive test oracle has been developed to check the correctness of the behavior of these execution-sequences. The changed behavior of these execution-sequences reflects the patch impact. A case study has been conducted and the results are encouraging.

Traditional approaches of executing complete system test suite to validate software systems on deploying the upgrades of Commercial Off-The-Shelf (COTS) software components is typically costly and time-consuming, resulting in delay in deploying the upgrades. In this paper, we present a process that examines the test case dependencies at finer granularity level of alteredfunctions in the upgrades of COTS software and recommends a smaller regression test suite. The process captures and analyzes the functional-level runtime interactions among the applications and COTS components to find the impact of the upgrades on the system. The process includes a procedure to identify the altered-functions in a component. We have developed a prototype based on the concepts proposed in the paper and performed a case study on a web-based application to illustrate efficacy of the prototype. The results of the case study show that up to 80% reduction in regression test efforts could be achieved.