Grant: Engage Grants Program

To make sure that their software does what it is supposed to do, software creators test their products extensively. Cost-effective testing is crucial to a company's ability to deliver high-quality software to the marketplace before competitors can. Tests are carefully chosen and scripted in order to ensure that all features are tested and time is not wasted doing unnecessary tests. Software creators prefer to be able to re-use tests. This allows them to modify the software (to fix bugs or add new features) and check that all the old features still work. Tests can be re-used automatically by the use of ""record and playback"" software, which records the sequences of actions done by a human user, and stores them for automated playback later. However, these automated tests tend to be very inflexible, in the sense that if anything changes in the software's user interface, the test will indicate that a bug exists even if the change is not important. This kind of automated test is therefore referred to as ""brittle"". We propose to automatically derive less brittle tests from brittle, recorded tests. To do this, we will create tests that access the internal data and computations of the software, which are less likely to be affected by changes in the external user interface. The process of deriving such smart, flexible tests from brittle tests is something that software creators do now by hand. Our project, if successful, will therefore help improve the cost-effectiveness of the testing process, allowing our industrial partner to create more smart tests more quickly, and to run more tests automatically. This in turn will improve the quality and time-to-market of their software.