Crate quickcfg

source ·
Expand description

github crates.io docs.rs

Apply a base configuration to a system, quickly!

It reads a configuration and template structure from a dotfiles directory and tries to normalize the machine that it is run base on this configuration.

WARNING: This project is currently in development. I’ve tried my best to make all operations non-destructive, but beware of bugs!

Example


Getting started

To get started, you can have quickcfg clone the configuration for you in the default location:

qc --init <git-url>

This will put the checked out configuration in the default config location for you platform. For example:

  • Windows - %APPDATA%\quickcfg
  • Linux - $HOME/.config/quickcfg

To find out where the various quickcfg directories are, use:

qc --paths

Features

Zero dependencies, All you need is the quickcfg binary and your configuration repo.

Blazingly fast, multi-threaded and uses a simple dependency graph to determine when things can run in parallel.

Flexible but opinionated manifests, There are a couple of powerful primitives available (e.g. copy-dir), which does a lot of work with very little configuration.

Uses fast checksumming, to reduce the amount of unnecessary work. Only applies changes when it has to.


Automatically applying updates

If you want quickcfg to periodically check your git repositories for updates, you can add the following to your .zshrc or .bashrc:

if command -v qc > /dev/null 2>&1; then
    qc --updates-only
    alias upd="qc"
fi

Every time you open a shell quickcfg will not check if your dotfiles are up-to-date.

You control how frequently by setting the git_refresh option in quickcfg.yml:

git_refresh: 3d

Configuration

Create a repository with a quickcfg.yml in its root:

git_refresh: 1d

hierarchy:
  - secrets.yml
  - db/common.yml
  - db/{distro}.yml

systems:
  # System to ensure that a set of packages are installed.
  - type: install

You also want to add a .gitignore file that looks like this:

/secrets.yml
/.state.yml
/.state

Then populate secrets.yml with your secret information - this you DO NOT check into git. Any variables you put in here can be used in future templates since they are part of the hierarchy.

The hierarchy specifies a set of files that should be looked for. These can use variables like {distro}, which will be expanded based on the facts known of the system you are running on.

You can use my dotfiles repository as inspiration.

The following section will detail all the systems which are available.


Hierarchy

The hierarchy is a collection of files which contain data.

Some systems query the hierarchy for information, like the key setting in install. This then determines which packages should be installed.

Hierarchy variables can also be made available in templates by adding a quickcfg: tag at the top of the template.


Systems


copy-dir

Copies a directory recursively.

type: copy-dir
from: ./some/dir
to: home://some/dir
templates: false

Will copy a directory recursively.


Links a directory recursively.

type: link-dir
# Directory to link from.
from: ./some/dir
# Directory to link towards.
to: home://some/dir

Will create the corresponding directory structure, but all files will be symbolic links.


git-sync

System that syncs a single git repository to some path.

type: git-sync
# Where to clone.
path: home://.oh-my-zsh
# Remote to clone.
remote: https://github.com/robbyrussell/oh-my-zsh.git
# Refresh once per day.
refresh: 1d

install

Compares the set of installed packages, with a set of packages from the hierarchy to install and installs any that are missing.

Will use sudo if needed to install packages.

type: install
# The provider of the package manager to use.
provider: pip3
# Hierarchy key to lookup for packages to install.
key: pip3::packages

The simplest example of this system is the one that uses the primary provider:

systems:
  - type: install

This will look up packages under the packages key and install it using the primary provider for the system that you are currently running.

These are the supported providers:

  • debian: For Debian-based systems. This is a primary provider.
  • pip: The Python 2 package manager.
  • pip3: The Python 3 package manager.
  • gem: The Ruby package manager.
  • cargo: Install packages using cargo.
  • rust components: Rust components using rustup.
    • Key: rust::components
  • rust toolchains: Rust toolchains using rustup.
    • Key: rust::toolchains

By default, any primary provider will be the default provider of the system if it can be detected.

Explicitly configured providers look up packages based on the hierarchy key <provider>::packages. Default providers use the key packages.


download

Downloads a single file into the given target path.

type: download
id: plug-vim
# Url to download the command from.
url: https://raw.githubusercontent.com/junegunn/vim-plug/master/plug.vim
# Path to download to.
path: home://.config/nvim/autoload/plug.vim

The id is to uniquely identify that this system has only been run once.


download-and-run

Downloads a script of the internet and runs it once.

type: download-and-run
id: install-oh-my-zsh
# Url to download the command from.
url: https://raw.githubusercontent.com/robbyrussell/oh-my-zsh/master/tools/install.sh
# Set to `true` if the downloaded command requires interaction. (default: false)
interactive: true
# Set to `true` if the command must be run through a shell (`/bin/sh`). (default: false).
shell: true

The id is to uniquely identify that this system has only been run once.


Creates a symlink.

type: link
path: home://.vimrc
link: .vim/vimrc

This creates a symbolic link at path which contains whatever is specified in link.


only-for

Limit a set of systems based on a condition.

type: only-for
os: windows
systems:
  # Download and install Rust
  - type: download-and-run
    name: rustup-init.exe
    id: install-rust
    url: https://win.rustup.rs/x86_64
    args: ["-y"]

Templating

Some systems treats files as templates, like copy-dir when the templating option is enabled. Any file being copied is then treated as a handlebars template.

Any template file can make use of hierarchy data, by specifying their dependencies using a quickcfg: tag at the top of the file, like this:

# quickcfg: name, hobbies:array

Hi, my name is {{name}}

My hobbies are:
{{#each hobbies}}
- {{this}}
{{/#each}}

This will load the name and hobbies variables out of the hierarchy. hobbies will be loaded as an array, causing all values in the hierarchy for that value to be loaded.

Modules

  • Helpers for accessing environment variables.
  • Loading facts about the system that we are currently running on.
  • Git abstraction.
  • Dealing with the hierarchy of data.
  • Set up options.
  • Package abstraction.
  • Utilities to process a set of units into a set of inter-dependent stages.
  • Things to do.
  • A unit of work. Does a single thing and DOES IT WELL.

Structs

  • A command wrapper that simplifies interaction with external commands.
  • Configuration model.
  • The way the state is serialized.
  • Helper and tracker of any filesystem modifications.
  • State model. This keeps track of any changes with the dirty flag, which is an indication whether it should be serialized or not.
  • A loaded template string.
  • A timestamp.

Traits