This directory contains machine specific system wide configuration files. The directory hierarchy follows a convention that permits easily symlinking the right config file to the right location. Before this scheme, I had just been maintaining system configuration on an ad hoc basis (and sometimes via ~/bin/arch-install), but this was error prone and inefficient. Namely, I would frequently need to copy configs from other machines, or worse, would forget about a config before wiping a machine.

The general scheme is:

$HOME/system/etc/path/to/config-file-as-directory/{default,krusty,...}

Where /etc/path/to/config-file-as-directory is a directory that refers to a concrete file path in the /etc directory. Inside this directory is an optional file named default, along with other optional files named by hostname, e.g., krusty. If a hostname file is present, then that file is symlinked to the corresponding config file location. Otherwise the default file is. If no files are present, then no symlinks are created. Thus, if you want to keep a configuration checked into git but disable it, then rename it from {name} to {name}.disable.

symlinks are created by setup-system-links.