Shelley Lambert

With the proliferation of OpenJDK binaries for a developer or a business to choose from, one factor in determining the selection is quality. How do you know your choice is up to snuff? The AdoptOpenJDK Quality Kit is an open and transparent verification story for OpenJDK binaries. A robust and adaptable test kit that can be utilized by any OpenJDK implementor, and represents the quality bar required by large-scale customers in enterprise environments. We test multiple freely available JDK implementations at AdoptOpenJDK and continue to refine this suite of tests to give the community access to high-quality binaries.

Beyond a frank discussion starter on the criteria by which we measure quality, this talk summarizes the real quality requirements of enterprise customers and presents a compelling story for verifying and/or selecting your OpenJDK implementation.

In this talk, we identify some well-known software verification problems, using real-world examples from open-source projects and see how we might apply some deep learning principles to address them. In the various points where we may test the Java runtime, we find candidates for deep learning. What is required? We need a problem to solve, a model that describes it, and a large amount of data to feed the neural network.