how_it_works.org

Disclaimer: this doc is not finished but I wanted to write an overview of the script internal.

Design

Everything is done in the update script.

The script ensure the following things:

• all packages must have existing packages as dependencies
• internal libraries from emacs should be excluded from the dependencies
• a package is not updated if it already has the latest version (to avoid redownloading the emacs world)

Loop

The script works as follow:

1. We re-read already generated packages (to get package version)
   1. It reads the code source and parse it (there is no evaluation)
   2. It only reads packages defined with define-public
2. Then, it fetches https://melpa.org/archive.json and read it
3. Archives are then filtered
   1. We resolve dependencies to see if they exists
   2. This is done recursively
4. Once it’s done, we can download all new packages from melpa to compute the hashes
5. Then we write everything to ../emacs/packages/melpa.scm
   1. We are appending the define-module code + copyrights
   2. We are appending content of ../emacs/packages/melpa-overrides at the end of the file