How We Automated Code Maintenance and You Can Too!

Let’s face it: as developers, we dedicate a third of our time to code maintenance, which includes tasks such as upgrading dependencies, addressing security vulnerabilities, and removing obsolete code. This is tedious and repetitive. Neglecting regular maintenance can lead to costly outcomes, including unexpected crashes, and it makes the codebase more difficult to understand and evolve.

However, automation of these tasks is not always straightforward. Existing tools such as security scanners and feature flag systems warn you about the issues or obsolete code, but fall short of automatically rectifying these problems. Tools that upgrade dependencies merely increase the version number, leaving engineers to handle any API compatibility issues. Automating code changes is hard, and the polyglot nature of modern development makes it harder.

In this talk, we delve into code rewriting techniques such as pattern matching, program analysis, and AI. We illustrate how we leveraged the complementing power of these tools to generate over 1,800 automated pull requests, eliminating or refactoring more than 500,000 lines of code. In this talk, you also learn how to harness the power of these tools to drive down tech debt, ensuring your codebase is not only functional but also future-proof.

Presented by Ameya Ketkar and Gautam Korlam (Gitar) at DPE Summit 2024, an event developed and hosted by Gradle.