7 Automating work through continuous integration

Through the course of the previous chapters you’ve built up a repertoire of tasks you execute each time you change your package so that you can maintain functionality and code quality. This is a huge stride in building confidence in changes, but doing all this locally on your own computer is still a big limitation. You may have trouble remembering all the steps that go into verifying a change, and people just starting to work on the project may struggle even more. Even when they perform their due diligence, commands they run locally can’t be directly verified by you unless you’re supervising them. This is difficult enough with a team of a few people, but becomes impossible in the open-source world, where you may not even know the person contributing code changes.

In this chapter you’ll create a pipeline for your package to bring automation to nearly every aspect of the packaging process—aside from writing the code, of course. Before getting to deep into the details of setting up this pipeline, you first need to understand the high-level flow.

7.1 The continuous integration workflow

7.2 Continuous integration with GitHub Actions

7.2.1 A high-level GitHub Actions workflow

7.2.2 Understanding GitHub Actions terminology

7.2.3 Starting a GitHub Actions workflow configuration

7.3 Converting manual tasks to GitHub Actions

7.3.1 Running a job multiple times with a build matrix

7.3.2 Building Python package distributions for a variety of platforms

7.4 Publishing a package

7.5 Summary