Bootstrap configuration - a temporary configuration for controlled pre-initialization

[Lockszmith-GH]

Ever found yourself needing to setup, your system BEFORE setting up chezmoi?
Like installing Homebrew or pulling, via .chezmoiexternals your secret/password vault binaries your envrionment needs to properly intialize.

Or, maybe you work on some systems that don't share enough to be managed in the same chezmoi source, but you still want them on the same repo, and have chezmoi just use one directory for system A and another for system B?

That's what I was looking to solve. I call my solution 'bootstrap chezmoi' - think of the Linux boot process: the bootloader loads initrd, a temporary "throwaway" OS, with the necessary drivers to initialize the hardware, so the OS can properly load.

Repo structure

The root dir of the repo is for documentation/orientation, as little content as possible.

It contains:

• A 'throwaway' chezmoi bootstrap environment:

  .chezmoiscripts/
  .chezmoi.toml.tmpl
  .chezmoiignore
  .gitignore
  

• This README.md and a docs directory for additional documentation.

• chezmoi.roots containing environment specific chezmoi roots.

Bootstrap

The key to the bootstrap process are the scripts in .chezmoiscripts/ and .chezmoiignore.

.chezmoiignore ^{(link to source)}

# The content at the root of the repo is for initial detection only.
# Thus, only .chezmoitscripts is required, because it will setup the
# correct .chezmoiroot, where the rest of the content resides.
#
*
! .chezmoiscripts/
! .chezmoiscripts/**

It ensures that chezmoi process ONLY .chezmoiscripts/ and it's content, allowing for bootstrap scripts to run and prepare the grounds for any requirements the actual chezmoi source-state will require to intiialize properly.

.chezmoiroot, which points to the actual root source-state, will be generated by these initialization scripts.
IMPORTANT: This file is NEVER part of the repo.

After the actual root has been setup, the bootstrap configuration needs to be discarded, and chezmoi init --apply will need to run again, this time intializing the user's envrionment.

To reiterreiterate, the initialization scripts will bootstrap the envrionment by taking the following steps:

• Detect the machine type
• Generates the .chezmoiroot
• Destroy the bootstrap's user configuration
• Initializes chezmoi (now that .chezmoiroot exists.)

.gitignore ensures .chezmoiroot is never added to git, as it is different on every machine.

The key to the process, is the initialization script, below are the bare-bone mandatory code in bash:

cd "$CHEZMOI_SOURCE_DIR"

#
# Make decisions
#

# Logic to setup .chezmoiroot
echo '<chezmoi root flavor>' > .chezmoiroot

# Brute force remove chezmoistate lock and initialize chezmoi
# `exec` replaces the existing process with the new command,
# breaking hezmoi's lock on the configuration - allowing to
# delete it.
exec sh -c ' 
    rm ~/.config/chezmoi/chezmoistate.boltdb \
    && chezmoi init \
    && chezmoi apply
`

Check out my implementation for a fully working solution (including initialization on Windows).

Additional tip- if you DO have stuff you want to share, you can still symlink within the chezmoi source.

[Lockszmith-GH]

Change log:

• 2025-04-26 Elaborated on .chezmoiignore's content.