Splitbrain: Automatic PR Splitting

Splitbrain is a research system aiming to compute disjoint changelists from a single code diff, and apply them to a user's workspace.

The motivation of this work follows from the intuition that smaller CLs have many benefits:

• Faster turnaround time in code reviews
• Smaller, atomic units of code committed in each CL
• Easier to rollback
• Less blocking on any particular code review
• and more...

This repository contains a series of tools and experiments designed to further understand this relationship, however it is not a usable tool and is exclusively an algorithmic research system.

Overview

At a high level, Splitbrain interprets a subset of the Bazel and Kythe graphs relevant to a single CL, and builds a linear chain of CLs based on structure of the original CL.

For example, a medium sized code change with 250 LoC could split into 5 changes of 50 LoC each.

These changes can be reviewed and potentially submitted in parallel.

Documentation

The original design doc is bundled with the source code. See docs/DESIGN.md for details.

Build & Test

Splitbrain uses bazel as it's build system.

Quick start:

bazel build //... && bazel test //...

Contributing

See CONTRIBUTING.md for details.

License

Apache 2.0; see LICENSE for details.

Disclaimer

This project is not an official Google project. It is not supported by Google and Google specifically disclaims all warranties as to its quality, merchantability, or fitness for a particular purpose.