The CI server

This project makes a lot of milage from using GitHub Actions as the CI server. It uses the server for a lot of automated tasks including:

  • Checking whether all flake outputs are valid.

  • Checking if the flake lockfile is up-to-date.

  • Building the packages from this project and pushing it to a binary cache which speeds up fetching my packages.

  • Building the personalized console ISO mainly used for bootstrapping systems and publicly releasing it. [1]

  • Building the site project.

Not only this is great for them automation but also it prevents the magical it-works-on-my-machine-related problems.

Ezran
Ezran

Isn’t that the point of using Nix, though? To easily make reproducible builds preventing the it-works-on-my-machine problems.

foodogsquared
foodogsquared

Yeah but it can still happen especially that not all of the changes I push into the public repo is out.

To keep the CI workflows up-to-date, this project makes use of Dependabot which checks for GitHub actions versions weekly.

As a highlight, there are some nifty GitHub actions I found to fully take advantage of GitHub’s large ecosystem.


1. I can’t build the graphical installer since its size is more than 2GB which is the size limit for GitHub Releases.