As much as we have embraced test automation, there are still too many manual processes required in daily test activities, such as analyzing test failures to match with known issues and maintaining third-party application test code. At the Eclipse AQAvit and Eclipse OpenJ9 projects, we target some of these manual processes as opportunities to apply various technologies and collaborate with Outreachy interns to see if we could improve our daily workflow.
This presentation covers our approaches to using machine learning to match test failures with possible related issues (our Deep AQAtik project) and the simplification and gamification of AQA external tests using the Twitter API (our Tweetest project). For Deep AQAtik, we introduce our initial prototype development, data collection and pre-processing pipeline, machine learning model selection, model deployment, and user feedback collection for improvement. For Tweetest, we present the idea of creating a generic external_custom test target and how we used Twitter to interact our testing framework to invoke third-party test material.
With the Deep AQAtik and Tweetest projects, we stretched our testing framework in new ways to be more intelligent, playful and simple. Perhaps an even greater outcome of these projects was what we gained from our collaboration and incorporation of new and diverse viewpoints and skillful contributions of our interns.