Foundations of Predictive Test Selection

Developers typically change a vast minority of their codebases at any one time, and yet small changes often require many hours of testing time, even though very few tests are actually affected.

The time and energy costs of this phenomenon are monumental, which is why “Test Impact Analysis” is a trending concept in and around the realm of developer productivity.

In this talk we will discuss why Gradle decided to invest in Predictive Test Selection as a promising Test Impact Analysis solution, the tenets and tradeoffs we made intentionally to provide a solid developer and build engineering experience, and how we think this technology is best used by software engineering teams.