PhD ceremony - Gunnar Kudrjavets, Computer Science

Title: The need for speed: increasing the code review velocity

Abstract:

One of the software industry’s critical goals is to increase code velocity. Code velocity is the speed at which code changes published for the code review are merged until they are committed to the intended destination branch. This thesis focuses on the part of the development process where engineers review and critique each other’s code (Modern Core Review) and how to shorten that period.

We find a negligible correlation between the pull request size, composition, and time-to-merge. We identify two primary categories of time delays during the code reviews: (a) the wait time from the proposal of code changes until the first response and (b) the wait time between acceptance and merging. Our findings indicate that kernel code is the primary optimization target for increasing code velocity in the BSD family of operating systems.

We discover no public evidence documenting the benefits of removing dead code from the active code bases. Similarly, no evidence demonstrates that fixing compiler warnings benefits the software project.

We conduct a qualitative survey to study the code velocity-related beliefs and practices in place. We find that (a) the industry and open-source community hold a similar set of beliefs, (b) quick reaction time is of utmost importance, (c) time-to-merge is the essential code review metric to improve, (d) engineers do not see evidence of the benefits of increased code velocity for their career growth, and (e) the controlled application of the commit-then-review model can increase code velocity.