Skip to content

Latest commit

 

History

History
executable file
·
58 lines (39 loc) · 3.89 KB

README.md

File metadata and controls

executable file
·
58 lines (39 loc) · 3.89 KB
Raw

Summary

This "package" allows you to specify a complete dev env setup, including packages to install (homebrew, yay, pacman, npm, etc etc). You write a config.json, then run a python CLI tool. I use this to keep multiple machines (work mac, home Arch, Docker based arch dev boxes) all up to date with my dev environment.

Isn't there stuff already like this?

There are probably many. However, the well known ones; Puppet, Chef, Ansible, etc, are very heavyweight for my needs.

There are a lot of "dotfiles" repos on github; this depends on that --- it requires you to have a ~/dotfiles/ repo that can be symlinked against, but it goes a lot further (than dotfiles).

This was only a few hundred lines of python, and I've been using this to setup new computers (eg new Macs, new PC builds) and Dockerized development environments for years.

I am planning on making this more extensible so the user can provide functions that get run.

Prerequisites

  1. python3 (I am using 3.11) and poetry and pyenv (curl https://pyenv.run | bash)
  2. if on mac, homebrew, and if on Arch, yay
  3. If on mac, XCode command line tools (can't even build gcc from homebrew without this): sudo xcode-select --install
  4. make a directory called ~/dotfiles/ with all of your dotfiles. I personally have a private github dotfiles repo. I simply clone that to use this.
  5. Write config.json; see the next section

Optional:

  • if you want vim installed, your .vimrc should be using vim-plug.

Config.json

To use this package, you write a config.json to describe what you would like linked, installed, ran, etc. That is the only input, other than a few cmd line parameters.

A sample (actually a copy of my real one!) config is included; see sample_config.json

My personal config.json is in my private dotfiles repo, and I clone it in step 4 above.

Please see the (evolving) jsonschema for config.json in bootstrap/schema.py for the full schema. These are the major sections:

  1. comments: this is a list of strings for your own sanity, since JSON has no native commenting mechanism.
  2. initial_mkdirs: directories to be recursively made. For example, ~/.ssh/... You can specify whether to try deleting the directory first, which is helpful for "rebootstrapping"
  3. links: this is a list of softlinked dotfiles, from you ~/dotfiles directory, that can vary by "location" if you want. IE, if you have a .gitconfig that you use on "work" machines, and a seperate one you use on "private" machines, you can specify these seperately, and when you run the script, you specify the location. There is also a generic all key for specifying location-agnostic keys.
  4. commands: a list of arbitrary commands to run, which can be specified as os-agnostic, or by OS type. Warning, whatever you put here will be executed!. While I havent tested this on mac yet, on linux, if the command needs sudo, it will pause and ask for your sudo password using a STDIN pipe, so privleged commands are fine.
  5. packages: a list of packages to install, which can be specified as os-agnostic, or by OS type. Examples of agnostic installs include npm and pip. Examples of mac include brew. You can also include "agnostic" installs in the os-specific sections, for example, "I only want this NPM package installed on my mac".

Install and Running

poetry install
poetry run runboot [args]

You can also set CMD args on the command line:

poetry run runboot --prereqs t --name tommycarpenter --redovim n --systype mac --loctype work
poetry run runboot --prereqs t --name tommy --redovim n --systype arch --loctype private
/home/ubuntu/.local/bin/poetry run runboot --prereqs t --name ubuntu --redovim y --systype ubuntu --loctype work

It is normally safe to run this repeatedly in an additive manner; as in, you can add packages to config.json and re-run.

Major TODOs:

  1. User supplied functions, and vim() should move to this