Skip to content

Latest commit

 

History

History
1624 lines (1325 loc) · 63.3 KB

getting_started.org

File metadata and controls

1624 lines (1325 loc) · 63.3 KB
Raw

Getting Started Guide

GNU Emacs is one grand ol’ adventure, let alone Doom Emacs. Before you start you’ll need to set up Emacs, Doom, and its packages, then learn how to take care of your new puppy/operating system. This guide will walk you through installing, using, configuring and troubleshooting all of these things, to smooth you into your Emacs journey.

This guide will gloss over many technicalities so you can get up and running as soon as possible. A more technical user manual is in the works for aspiring contributors who want a deeper understanding of how Doom Emacs works.

If you feel like we’ve missed something, join us on our Discord server and let us know!

Table of Contents

Install

This is what you’ll have installed by the end of this section:

  • Git 2.23+
  • Emacs 26.1+ (27.x is recommended)
  • ripgrep 11.0+
  • GNU Find
  • (Optional) fd 7.3.0+ (known as fd-find on Debian, Ubuntu & derivatives) – improves performance for many file indexing commands

These packages ought to be available through the package managers of your operating system; i.e. homebrew & macports on macOS, scoop/chocolatey on Windows, or pacman/aptitude/etc on the various Linux distributions.

While Doom does claim to support 26.x, 27.x is recommended because it is faster, especially for LSP users. The installation guides below will touch on installing 27 if there is a simple way to do so on that particular operating system.

Emacs & dependencies

On Linux

In the unusual case that Emacs 26.x is unavailable through your package manager, you’ll have to build it from source. Otherwise:

Ubuntu

Since only Emacs 25.3 is available on Ubuntu 18.04 (and 24.3 on Ubuntu 14 or 16), extra steps are necessary to acquire 26.3:

add-apt-repository ppa:kelleyk/emacs
apt-get update
apt-get install emacs26

Then install the dependencies:

# required dependencies
apt-get install git ripgrep
# optional dependencies
apt-get install fd-find

To install Emacs 27 on Ubuntu, you’ll need to build it from source.

Fedora

# required dependencies
dnf install emacs git ripgrep
# optional dependencies
dnf install fd-find    # is 'fd' in Fedora <28

Arch Linux

# required dependencies
pacman -S git emacs ripgrep
# optional dependencies
pacman -S fd

The above installs Emacs 27 (at the time of writing).

NixOS

On NixOS Emacs 26.3 can be installed via nix-env -Ai nixos.emacs, or permanently with the following added to etc/nixos/configuration.nix:

environment.systemPackages = with pkgs; [
  # required dependencies
  git
  emacs    # Emacs 26.3
  ripgrep
  # optional dependencies
  coreutils # basic GNU utilities
  fd
  clang
];

Installing Emacs 27 will require nix-community/emacs-overlay:

nixpkgs.overlays = [
  (import (builtins.fetchTarball https://github.com/nix-community/emacs-overlay/archive/master.tar.gz))
];

environment.systemPackages = with pkgs; [
  emacsUnstable # Installs Emacs 27
];

openSUSE

Emacs 26.3

Emacs can be installed from the package list, or manually via zypper.

For example, to install on openSUSE Leap 15.1 (requires root):

zypper addrepo https://download.opensuse.org/repositories/editors/openSUSE_Leap_15.1/editors.repo
zypper refresh
zypper install emacs

If you already have an older version of Emacs installed, you will be prompted to install the update candidate (Emacs 26.3).

ripgrep

Download ripgrep 11.0.2 from the package list or installed manually (requires root).

zypper addrepo https://download.opensuse.org/repositories/openSUSE:Factory/standard/openSUSE:Factory.repo
zypper refresh
zypper install ripgrep

Only ripgrep 0.8.1 is officially available on Leap 15.1 and 15.2, so you will need to install Rust to build ripgrep from source. Rust can be downloaded from the package list or installed manually via zypper (requires root), e.g.

zypper addrepo https://download.opensuse.org/repositories/openSUSE:Leap:15.1:Update/standard/openSUSE:Leap:15.1:Update.repo
zypper refresh
zypper install rust

See the ripgrep documentation for instructions on building from source.

On macOS

MacOS users have many options for installing Emacs, but not all of them are well suited to Doom. Before we get to that you’ll need either the Homebrew or MacPorts package manager installed (you only need one):

With Homebrew

First, Doom’s dependencies:

# required dependencies
brew install git ripgrep
# optional dependencies
brew install coreutils fd
# Installs clang
xcode-select --install

For Emacs itself, these three formulas are the best options, ordered from most to least recommended for Doom (based on compatibility).

  • emacs-plus: 
    brew tap d12frosted/emacs-plus
brew install emacs-plus
ln -s /usr/local/opt/emacs-plus/Emacs.app /Applications/Emacs.app
  • emacs-mac is another acceptable option. It offers slightly better integration with macOS, native emojis and better childframe support. However, at the time of writing, it lacks multi-tty support (which impacts daemon usage): 
    brew tap railwaycat/emacsmacport
brew install emacs-mac --with-modules
ln -s /usr/local/opt/emacs-mac/Emacs.app /Applications/Emacs.app
  • emacs is another acceptable option, **but does not provide a Emacs.app**: 
    brew install emacs
Where not to install Emacs from

These builds/forks have known compatibility issues with Doom and are very likely to cause issues later on. They are not recommended:

  • emacsformacosx.com
  • brew cask install emacs (installs from emacsformacosx.com)
  • AquaMacs
  • XEmacs

With MacPorts

There are four ports (at time of writing) available through MacPorts, and they are all acceptable options:

Some of these ports do not add an emacs binary to your PATH, which is necessary for Doom’s installation process. You’ll have to do so yourself by adding this to your shell config:

# Add this to ~/.zshrc or ~/.bash_profile
export PATH="/Applications/MacPorts/Emacs.app/Contents/MacOS:$PATH"

Or by replacing /usr/local/bin/emacs with a shim script containing:

#!/bin/sh
/Applications/MacPorts/Emacs.app/Contents/MacOS/Emacs "$@"

On Windows

WARNING: Emacs on Windows is much slower than its Linux or macOS counterparts. There are some suggestions on how to speed it up later in this section.

There are three methods for installing Emacs 26.x on Windows, each with their pros and cons:

  • With chocolatey/scoop
  • With a precompiled binary + Git Bash
  • With WSL + Ubuntu

If you don’t know which to choose, I recommend WSL; it produces the fastest and most stable environment of the three, but has the most complex installation process.

Before moving on to installing Emacs et co, a few steps to prepare Windows for Emacs are necessary:

  1. Create a HOME system environment variable.

    Set it to C:\Users\USERNAME\, otherwise Emacs will treat C:\Users\USERNAME\AppData\Roaming as your HOME, which will cause issues later.

  2. Add C:\Users\USERNAME\.emacs.d\bin to your PATH.

    This way, you don’t have to type all of C:\Users\USERNAME\.emacs.d\bin\doom every time you need to run this script (and you’ll need to, often).

    A pre-existing PATH variable should already exist among your system variables. It contains a string of file paths separated by colons; pathA:pathB:pathC. Prepend the path to bin/doom to that string, like so: C:\Users\username\.emacs.d\bin:pathA:pathB:pathC

  3. Restart your system so your new values for HOME and PATH take effect.

Now we’re ready to move on!

With chocolatey / scoop

Chocolatey is a package manager for Windows, and is the simplest way to install Emacs and Doom’s dependencies:

choco install git emacs ripgrep fd llvm

Scoop will work too, but because Emacs is a GUI application you’ll need to enable the ‘extras’ Scoop bucket:

scoop bucket add extras
scoop install git emacs ripgrep fd llvm

With a precompiled binary + Git Bash

(Credit goes to @earvingad and his fantastic tutorial for informing this guide)

  1. Download and install Git from https://git-scm.com/download/win
  2. Download and extract Emacs, ripgrep and fd where you want them, but in different folders:
  3. Add the three folders from step 2 to your PATH
    • Go to Control panel -> User Accounts -> Change my environment variables.
    • Click “New”, type HOME and set your C:\Users\USERNAME and OK.
    • Select “Path”, click “edit”, prepend C:\path\to\the\emacs\bin: to it and click OK.
    • Select “Path”, click “edit”, prepend C:\path\to\the\ripgrep: to it and click OK.
    • Select “Path”, click “edit”, prepend C:\path\to\the\fd: to it and click OK.
    • Click Ok.

And done! Keep git-bash.exe open, you’ll need it for the rest of this guide.

IMPORTANT: you’ll need to open git-bash.exe whenever you want to run a bin/doom command.

With WSL + Ubuntu 18.04 LTS

(Credit goes to @lunias and <a href=”https://ethanaa.com/blog/switching-to-doom-emacs/#installing-on-windows-10 “>his fantastic tutorial for informing this guide)

  1. Install Powershell as admin (Windows key + x) with: 
    Enable-WindowsOptionalFeature -Online -FeatureName Microsoft-Windows-Subsystem-Linux
  2. Restart your Computer
  3. Download and install Ubuntu 18.04 L>TS from the Microsoft Store
  4. Launch Ubuntu 18.04 LTS
  5. Update and upgrade Ubuntu 
    sudo apt update && sudo apt upgrade
  6. Then install Emacs: 
    sudo add-apt-repository ppa:kelleyk/emacs
sudo apt update
sudo apt install emacs26
  7. Then Doom’s dependencies: 
    # required dependencies
sudo apt-get install git ripgrep
# optional dependencies
sudo apt-get install fd-find

And done! Keep Ubuntu open, you’ll need it for the rest of this guide.

Doom Emacs

With Emacs and Doom’s dependencies installed, next is to install Doom Emacs itself:

git clone https://github.com/hlissner/doom-emacs ~/.emacs.d
~/.emacs.d/bin/doom install

doom install will set up your DOOMDIR at ~/.doom.d (if it doesn’t already exist) and will work you through the first-time setup of Doom Emacs. Carefully follow any instructions it puts out.

If this is your first time, you should run doom doctor. This will diagnose common issues with your system or config.

If you’d like a more technical break down of doom install, it’s been translated into shell commands below, in the “Install Doom Manually” section.

The bin/doom utility

This utility is your new best friend. It won’t spot you a beer, but it’ll shoulder much of the work associated with managing and maintaining your Doom Emacs configuration, and then some. Not least of which is installation of and updating Doom and your installed packages.

It exposes a variety of commands. bin/doom help will list them all, but here is a summary of the most important ones:

  • doom sync: This synchronizes your config with Doom Emacs. It ensures that needed packages are installed, orphaned packages are removed and necessary metadata correctly generated. Run this whenever you modify your doom! block or packages.el file. You’ll need doom sync -u if you override the recipe of package installed by another module.
  • doom upgrade: Updates Doom Emacs (if available) and all its packages.
  • doom env: (Re)generates an “envvar file”, which is a snapshot of your shell environment that Doom loads at startup. If your app launcher or OS launches Emacs in the wrong environment you will need this. **This is required for GUI Emacs users on MacOS.**
  • doom doctor: If Doom misbehaves, the doc will diagnose common issues with your installation, system and environment.
  • doom purge: Over time, the repositories for Doom’s plugins will accumulate. Run this command from time to time to delete old, orphaned packages, and with the -g switch to compact existing package repos.

Use doom help to see an overview of the available commands that doom provides, and doom help COMMAND to display documentation for a particular COMMAND.

I recommend you add ~/.emacs.d/bin to your PATH so you can call doom directly and from anywhere. Accomplish this by adding this to your .bashrc or .zshrc file: export PATH="$HOME/.emacs.d/bin:$PATH"

Install Doom Manually

If you’d rather install Doom yourself, instead of rely on the magic of doom install, here is its equivalent in bash shell commands (assuming hlissner/doom-emacs has been cloned to ~/.emacs.d):

# So we don't have to write ~/.emacs.d/bin/doom every time
PATH="$HOME/.emacs.d/bin:$PATH"

# Create a directory for our private config
mkdir ~/.doom.d  # or ~/.config/doom

# The init.example.el file contains an example doom! call, which tells Doom what
# modules to load and in what order.
cp ~/.emacs.d/init.example.el ~/.doom.d/init.el
cp ~/.emacs.d/core/templates/config.example.el ~/.doom.d/config.el
cp ~/.emacs.d/core/templates/packages.example.el ~/.doom.d/packages.el

# You might want to edit ~/.doom.d/init.el here and make sure you only have the
# modules you want enabled.

# Then synchronize Doom with your config:
doom sync

# If you know Emacs won't be launched from your shell environment (e.g. you're
# on macOS or use an app launcher that doesn't launch programs with the correct
# shell) then create an envvar file to ensure Doom correctly inherits your shell
# environment.
#
# If you don't know whether you need this or not, there's no harm in doing it
# anyway. `doom install` will have prompted you to generate one. If you
# responded no, you can generate it later with the following command:
doom env

# Lastly, install the icon fonts Doom uses:
emacs --batch -f all-the-icons-install-fonts
# On Windows, `all-the-icons-install-fonts` will only download the fonts, you'll
# have to install them by hand afterwards!

To understand the purpose of the ~/.doom.d directory and ~/.doom.d/init.el file, see the Configure section further below.

Install Doom alongside other configs (with Chemacs)

Chemacs is a bootloader for Emacs. It allows you to switch between multiple Emacs configurations. Here is a quick guide for setting it up with Doom Emacs as the default config:

  1. First, install Doom somewhere: 
    git clone https://github.com/hlissner/doom-emacs ~/doom-emacs
~/doom-emacs/bin/doom install
  2. Download the Chemacs’ startup script to ~/.emacs: 
    wget -O ~/.emacs https://raw.githubusercontent.com/plexus/chemacs/master/.emacs

    WARNING: the ~/.emacs.d directory must not exist for this to work.

  3. Create ~/.emacs-profiles.el with a list of your Emacs profiles. This file is structured like a .dir-locals.el file. Here is an example with Doom (as the default), Spacemacs, and Prelude: 
    (("default"   . ((user-emacs-directory . "~/doom-emacs")))
 ("spacemacs" . ((user-emacs-directory . "~/spacemacs")))
 ("prelude"   . ((user-emacs-directory . "~/prelude"))))

To start Emacs with a specific config, use the --with-profile option:

emacs --with-profile spacemacs

If no profile is specified, the default profile is used.

External/system dependencies

Doom is comprised of approximately 160 modules which provide its features, language support and integration with external tools. Many of these have external dependencies that you must install yourself. You’ll find what a module needs and how to install them in that module’s README.org file or by running bin/doom doctor.

The Module Index lists all Doom’s available modules, with links to their documentation. Documentation is a work-in-progrees; some modules may not have README.org files yet!

Use M-x doom/help-modules (bound to SPC h d m or C-h d m) to jump to a module’s documentation from within Doom, otherwise, place your cursor on a module in your doom! block (in ~/.doom.d/init.el) and press K to jump to its documentation (or gd to jump to its source code). C-c g k and C-c g d for non-evil users, respectively.

Update & Rollback

Doom is an active project and many of its 300+ packages are in active development as well. It is wise to occasionally update:

doom upgrade   # or 'doom up'

If you want to update Doom manually, doom upgrade is equivalent to:

cd ~/.emacs.d
git pull        # updates Doom
doom clean      # Ensure your config isn't byte-compiled
doom sync       # synchronizes your config with Doom Emacs
doom update     # updates installed packages

To upgrade only your packages (and not Doom itself):

doom upgrade --packages

To minimize issues while upgrading, avoid modifying Doom’s source files in ~/.emacs.d. All your customization should be kept in your DOOMDIR (e.g. ~/.doom.d). Read the Configure section for more on configuring Doom.

Rollback

The bin/doom script doesn’t currently offer rollback support for Doom or its packages (yet).

Up/Downgrading Emacs

You may encounter errors after up/downgrading Emacs. Run doom sync on the command line after changing the installed version of Emacs. If you’ve changed the major version (e.g. 26 -> 27 or vice versa) run doom build too.

  • doom sync will re-index any built-in/site loaddef files. This is especially necessary if paths to built-in libraries have changed.
  • doom build will recompile all your installed packages, which is necessary because Emacs bytecode not generally forward compatible across major releases (e.g. 26 -> 27). Alternatively, reinstall all your packages by deleting ~/.emacs.d/.local, then run doom sync.

Migrate

If you’re here from another Emacs distribution (or your own), here are a few things to be aware of while you convert your old config to Doom:

  • Doom does not use package.el to manage its packages, but use-package does! You will see errors if you have :ensure ... properties in your use-package blocks. Remove these and, instead, add package! declarations to ~/.doom.d/packages.el to install your packages.

    See “Package Management”, further in this guide.

(This section is incomplete)

From vanilla Emacs

Have you migrated from your own config? Help me flesh out this section by letting me know what kind of hurdles you faced in doing so. You’ll find me on our Discord server.

From Spacemacs

Have you migrated from Spacemacs? Help me flesh out this section by letting me know what kind of hurdles you faced in doing so. You’ll find me on our Discord server.

Configure

You can configure Doom by tweaking the files found in your DOOMDIR. Doom expects this directory to be found in one of:

  1. ~/.config/doom (respects $XDG_CONFIG_HOME)
  2. or ~/.doom.d

This directory is referred to as your DOOMDIR. Only one of these directories should exist (Doom will only recognize one).

Change the DOOMDIR environment variable to change where Doom looks for this directory. Symlinks will work as well.

When you ran doom install, it deployed a simple Doom configuration to your DOOMDIR, comprised of these three files:

init.el
Where you’ll find your doom! block, which controls what Doom modules are enabled and in what order they will be loaded.

This file is evaluated early when Emacs is starting up; before any other module has loaded. You generally shouldn’t add code to this file unless you’re targeting Doom’s CLI or something that needs to be configured very early in the startup process.

config.el
Here is where 99.99% of your private configuration should go. Anything in here is evaluated after all other modules have loaded, when starting up Emacs.
packages.el
Package management is done from this file; where you’ll declare what packages to install and where from.

Note: do not use M-x customize or the customize API in general. Doom is designed to be configured programmatically from your config.el, which can conflict with Customize’s way of modifying variables.

If you’re concerned about defcustom setters, Doom has a setq! macro that will trigger them.

Modules

Doom consists of around 160 modules and growing. A Doom module is a bundle of packages, configuration and commands, organized into a unit that can be toggled easily by tweaking your doom! block (found in $DOOMDIR/init.el).

If $DOOMDIR/init.el doesn’t exist, you haven’t run doom install yet. See the “Install” section above.

Your doom! block should look something like this:

;; To comment something out, you insert at least one semicolon before it and the
;; Emacs Lisp interpreter will ignore everything until the end of the line.
(doom! :lang
       python        ; this module is not commented, therefore enabled
       ;;javascript  ; this module is commented out, therefore disabled
       ;;lua         ; this module is disabled
       ruby          ; this module is enabled
       php)          ; this module is enabled

It controls what modules are enabled and in what order they are loaded. Some modules have optional features that can be enabled by passing them flags, denoted by a plus prefix:

(doom! :completion
       (company +auto)
       :lang
       (csharp +unity)
       (org +brain +dragndrop +gnuplot +hugo +jupyter)
       (sh +fish))

Different modules support different flags. You’ll find a comprehensive list of available modules and their supported flags in the Module Index. Flags that a module does not recognize will be silently ignored.

IMPORTANT: any changes to your doom! block won’t take effect until you run doom sync on the command line.

doom doctor will detect issues with your doom! block, such as duplicate or misspelled modules and flags.

Package management

**Doom Emacs does not use package.el** (the package manager built into Emacs). Instead, it uses its own declarative package manager built on top of straight.el.

Packages are declared in packages.el files. You’ll find one in your DOOMDIR and in many of Doom’s modules. Read on to learn how to use this system to install your own packages.

WARNING: Do not install packages directly (with M-x package-install or M-x straight-use-package). Without an accompanying package! declaration somewhere these packages will be forgotten when you restart Emacs and uninstalled the next time you run doom sync or doom purge.

WARNING: If you’re here from another Emacs distro (or vanilla Emacs), be wary of the :ensure property in use-package blocks, because it will attempt (and fail) to install packages through package.el. Tutorials will recommend you install packages this way too!

Installing packages

To install a package, add a package! declaration for it to DOOMDIR/packages.el:

;; Install a package named "example" from ELPA, MELPA, or Emacsmirror
(package! example)

If a package could not be found in any known repo you will get an error like:

Could not find package X in recipe repositories: (org-elpa melpa gnu-elpa-mirror emacsmirror-mirror)

The most likely cause for this is either:

  • You’ve misspelled the package’s name.
  • Or the package really doesn’t exist on ELPA, MELPA, or EmacsMirror and you’ll need to specify a recipe for it.

package! will return non-nil if the package is cleared for install and hasn’t been disabled elsewhere. Use this fact to chain package dependencies together. e.g.

(when (package! example)
  (package! plugin-that-example-depends-on))

IMPORTANT: New packages won’t be installed until you run doom sync.

Installing packages from external sources

To install a package straight from an external source (like github, gitlab, etc), you’ll need to specify a MELPA-style straight recipe:

Here are a few examples:

;; Install it directly from a github repository. For this to work, the package
;; must have an appropriate PACKAGENAME.el file which must contain at least a
;; Package-Version or Version line in its header.
(package! example
  :recipe (:host github :repo "username/my-example-fork"))

;; If the source files for a package are in a subdirectory in said repo, use
;; `:files' to target them.
(package! example :recipe
  (:host github
   :repo "username/my-example-fork"
   :files ("*.el" "src/lisp/*.el")))

;; To grab a particular branch or tag:
(package! example :recipe
  (:host gitlab
   :repo "username/my-example-fork"
   :branch "develop"))

;; If a package has a default recipe on MELPA or emacsmirror, you may omit
;; keywords and the recipe will inherit the rest of the recipe from their
;; original.
(package! example :recipe (:branch "develop"))

;; If the repo pulls in many unneeded submodules, you can disable recursive cloning
(package! example :recipe (:nonrecursive t))

;; A package can be installed straight from a git repo by setting :host to nil:
(package! example
  :recipe (:host nil :repo "https://some/git/repo"))

The specification for the package! macro’s :recipe is laid out in Straight.el’s README.

IMPORTANT: Run bin/doom sync whenever you modify packages.el files to ensure your changes take effect.

Pinning packages to specific commits

All of Doom’s packages are pinned by default. A pinned package is a package locked to a specific commit, like so:

(package! evil :pin "e00626d9fd")

To unpin a package, use the unpin! macro:

(unpin! evil)

;; It can be used to unpin multiple packages at once
(unpin! evil helm org-mode)

;; Or to unpin all packages in modules
(unpin! (:lang python ruby rust) (:tools docker))

;; Or to unpin an entire category of modules
(unpin! :completion :lang :tools)

;; This will work too, if you prefer the syntax, but it provides no concise
;; syntax for unpinning multiple packages:
(package! helm :pin nil)

Though it is highly discouraged, you may unpin all packages and make Doom Emacs rolling release:

(unpin! t)

Unpinning all packages is discouraged because Doom’s modules are designed against the pinned versions of its packages. More volatile packages (like lsp-mode, ein and org) change rapidly, and are likely to cause breakages if unpinned.

Instead, it’s a better to selectively unpin packages, or repin them to the exact commit you want.

Disabling packages

The package! macro possesses a :disable property:

(package! irony :disable t)
(package! rtags :disable t)

Once a package is disabled, use-package! and after! blocks for it will be ignored, and the package is removed the next time you run bin/doom sync. Use this to disable Doom’s packages that you don’t want or need.

There is also the disable-packages! macro for conveniently disabling multiple packages:

(disable-packages! irony rtags)

IMPORTANT: Run bin/doom sync whenever you modify packages.el files to ensure your changes take effect.

Changing a recipe for a included package

If a Doom module installs package X from one place, but you’d like to install it from another (say, a superior fork), add a package! declaration for it in your DOOMDIR/packages.el. Your private declarations always have precedence over modules (even your own).

;; in modules/editor/evil/packages.el
(package! evil) ; installs from MELPA

;; in DOOMDIR/packages.el
(package! evil :recipe (:host github :repo "username/my-evil-fork"))

To install a package only if a built-in package doesn’t exist, use :built-in 'prefer:

(package! so-long :built-in 'prefer)

IMPORTANT: Remember to run doom sync -u after changing recipes for existing packages. At the time of writing, doom sync alone will not pick up on recipe changes.

Using/loading local packages

Say you are developing an Emacs package locally and want to “install” it for live testing. To do this specify a :local-repo in that package’s recipe:

(package! my-package
  :recipe (:local-repo "/path/to/my/package"))

;; Relative paths are expanded to ~/.emacs.d/.local/straight/repos/{local-repo}
;; or ~/.doom.d/{local-repo} -- the first that is found.
(package! my-package
  :recipe (:local-repo "my/package"))  ; looks for ~/.doom.d/my/package/my-package.el

(package! my-package
  :recipe (:local-repo "/path/to/my/package"

           ;; By default, the package manager grabs all *.el files at the root
           ;; of the project and nothing else. To include other files, or
           ;; accommodate unconventional project structures, specify what :files
           ;; you want:
           :files ("*.el" "src/lisp/*.el")

           ;; With ':no-byte-compile t' you can avoid having to run 'doom sync'
           ;; every time you change the package.
           :no-byte-compile t))

Alternatively, add the package’s location to Emacs’ load-path. Do this if you don’t need/care for autoload cookies or byte-compilation:

;; Doom has modified `use-package's `:load-path' to expand relative paths from
;; your DOOMDIR. e.g. ~/.doom.d/lisp/package
(use-package my-package
  :load-path "lisp/package")

;; or

(add-load-path! "lisp/package")

IMPORTANT: Remember to run doom sync to rebuild your package after you’ve changed it, and to re-index any autoloads in it.

Configuring Doom

Configuring packages

If your configuration needs are simple, the use-package!, after!, add-hook! and setq-hook! macros are your bread and butter.

;;; ~/.doom.d/config.el (example)
(setq doom-font (font-spec :family "Fira Mono" :size 12))

;; Takes a feature symbol or a library name (string)
(after! evil
  (setq evil-magic nil))

;; Takes a major-mode, a quoted hook function or a list of either
(add-hook! python-mode
  (setq python-shell-interpreter "bpython"))

;; These are equivalent
(setq-hook! 'python-mode-hook python-indent-offset 2)
(setq-hook! python-mode python-indent-offset 2)

(use-package! hl-todo
  ;; if you omit :defer, :hook, :commands, or :after, then the package is loaded
  ;; immediately. By using :hook here, the `hl-todo` package won't be loaded
  ;; until prog-mode-hook is triggered (by activating a major mode derived from
  ;; it, e.g. python-mode)
  :hook (prog-mode . hl-todo-mode)
  :init
  ;; code here will run immediately
  :config
  ;; code here will run after the package is loaded
  (setq hl-todo-highlight-punctuation ":"))

For more flexibility, the use-package-hook! is another option, but should be considered a last resort (because there is usually a better way). It allows you to disable, append/prepend to and/or overwrite Doom’s use-package! blocks. These are powered by use-package’s inject-hooks under the hood.

use-package-hook! must be used before that package’s ~use-package!~ block. Therefore it must be used from your private init.el file.

;;; ~/.doom.d/init.el (example)
;; If a :pre-init / :pre-config hook returns nil, it overwrites that package's
;; original :init / :config block. Exploit this to overwrite Doom's config.
(use-package-hook! doom-themes
  :pre-config
  (setq doom-neotree-file-icons t)
  nil)

;; ...otherwise, make sure they always return non-nil!
(use-package-hook! evil
  :pre-init
  (setq evil-magic nil)
  t)

;; `use-package-hook' also has :post-init and :post-config hooks

Reloading your config

You may find it helpful to have your changes take effect immediately. For things that don’t require a complete restart of Doom Emacs (like changing your enabled modules or installed packages), you can evaluate Emacs Lisp code on-the-fly.

  • Evil users can use the gr operator to evaluate a segment of code. The return value is displayed in the minibuffer or in a popup (if the result is large enough to warrant one).

    gr works for most languages, but using it on Elisp is a special case; it’s executed within your current session of Emacs. You can use this to modify Emacs’ state on the fly.

  • Non-evil users can use C-x C-e to run eval-last-sexp, as well as M-x +eval/buffer-or-region (on SPC c e).
  • Another option is to open a scratch buffer with SPC x, change its major mode (M-x emacs-lisp-mode), and use the above keys to evaluate your code.
  • An ielm REPL is available by pressing SPC o r (+eval/open-repl-other-window).
  • There’s also M-: or SPC ;, which invokes eval-expression, which you can use to run elisp code inline.

While all this is helpful for reconfiguring your running Emacs session, it can also be helpful for debugging.

Binding keys

  • define-key
  • global-set-key
  • map!
  • undefine-key!
  • define-key!

Writing your own modules

To create your own module you need only create a directory for it in ~/.doom.d/modules/abc/xyz, then add :abc xyz to your doom! block in ~/.doom.d/init.el to enable it.

In this example, :abc is called the category and xyz is the name of the module. Doom refers to modules in one of two formats: :abc xyz and abc/xyz.

If a private module possesses the same name as a built-in Doom module (say, :lang org), it replaces the built-in module. Use this fact to rewrite modules you don’t agree with.

Of course, an empty module isn’t terribly useful, but it goes to show that nothing in a module is required. The typical module will have:

  • A packages.el to declare all the packages it will install,
  • A config.el to configure and load those packages,
  • And, sometimes, an autoload.el to store that module’s functions, to be loaded when they are used.

These are a few exceptional examples of a well-rounded module:

The remainder of this guide will go over the technical details of a Doom module.

File structure

Doom recognizes a handful of special file names, none of which are required for a module to function. They are:

category/
  module/
    test/*.el
    autoload/*.el
    autoload.el
    init.el
    cli.el
    config.el
    packages.el
    doctor.el

init.el

This file is loaded early, before anything else, but after Doom core is loaded. It is loaded in both interactive and non-interactive sessions (it’s the only file, besides cli.el that is loaded when the bin/doom starts up).

Do:

  • Configure Emacs or perform setup/teardown operations that must be set early; before other modules are (or this module is) loaded.
  • Reconfigure packages defined in Doom modules with use-package-hook! (as a last resort, when after! and hooks aren’t enough).
  • Configure behavior of bin/doom in a way that must also apply in interactive sessions.

Don’t:

  • Configure packages with use-package! or after! from here
  • Preform expensive or error-prone operations; these files are evaluated whenever bin/doom is used; a fatal error in this file can make Doom unbootable (but not irreversibly).
  • Define new bin/doom commands here. That’s what cli.el is for.

config.el

The heart of every module. Code in this file should expect dependencies (in packages.el) to be installed and available. Use it to load and configure its packages.

Do:

  • Use after! or use-package! to configure packages. 
    ;; from modules/completion/company/config.el
(use-package! company  ; `use-package!' is a thin wrapper around `use-package'
                       ; it is required that you use this in Doom's modules,
                       ; but not required to be used in your private config.
  :commands (company-mode global-company-mode company-complete
             company-complete-common company-manual-begin company-grab-line)
  :config
  (setq company-idle-delay nil
        company-tooltip-limit 10
        company-dabbrev-downcase nil
        company-dabbrev-ignore-case nil)
   [...])
  • Lazy load packages with use-package’s :defer property.
  • Use the featurep! macro to make some configuration conditional based on the state of another module or the presence of a flag.

Don’t:

  • Use package!
  • Install packages with package.el or use-package’s :ensure property. Doom has its own package manager. That’s what packages.el is for.

packages.el

This file is where package declarations belong. It’s also a good place to look if you want to see what packages a module manages (and where they are installed from).

Do:

  • Declare packages with the package! macro
  • Disable single packages with package!’s :disable property or multiple packages with the disable-packages! macro.
  • Use the featurep! macro to make packages conditional based on the state of another module or the presence of a flag.

Don’t:

  • Configure packages here (definitely no use-package! or after! in here!). This file is read in an isolated environment and will have no lasting effect. The only exception is configuration targeting straight.el.
  • Perform expensive calculations. These files are read often and sometimes multiple times.
  • Produce any side-effects, for the same reason.

The ”Package Management” section goes over the package! macro and how to deal with packages.

autoload/*.el OR autoload.el

These files are where you’ll store functions that shouldn’t be loaded until they’re needed and logic that should be autoloaded (evaluated very, very early at startup).

This is all made possible thanks to these autoload cookie: ;;;###autoload. Placing this on top of a lisp form will do one of two things:

  1. Add a autoload call to Doom’s autoload file (found in ~/.emacs.d/.local/autoloads.el, which is read very early in the startup process).
  2. Or copy that lisp form to Doom’s autoload file verbatim (usually the case for anything other then def* forms, like defun or defmacro).

Doom’s autoload file is generated by scanning these files when you execute doom sync.

For example:

;; from modules/lang/org/autoload/org.el
;;;###autoload
(defun +org/toggle-checkbox ()
  (interactive)
  [...])

;; from modules/lang/org/autoload/evil.el
;;;###autoload (autoload '+org:attach "lang/org/autoload/evil" nil t)
(evil-define-command +org:attach (&optional uri)
  (interactive "<a>")
  [...])

doctor.el

When you execute doom doctor, this file defines a series of tests for the module. These should perform sanity checks on the environment, such as:

  • Check if the module’s dependencies are satisfied,
  • Warn if any of the enabled flags are incompatible,
  • Check if the system has any issues that may interfere with the operation of this module.

Use the warn!, error! and explain! macros to communicate issues to the user and, ideally, explain how to fix them.

For example, the :lang cc module’s doctor checks to see if the irony server is installed:

;; from lang/cc/doctor.el
(require 'irony)
(unless (file-directory-p irony-server-install-prefix)
  (warn! "Irony server isn't installed. Run M-x irony-install-server"))

cli.el

This file is read when bin/doom starts up. Use it to define your own CLI commands or reconfigure existing ones.

test/**/test-*.el

Doom’s unit tests go here. More information on them to come…

Additional files

Any files beyond the ones I have already named are not given special treatment. They must be loaded manually to be loaded at all. In this way modules can be organized in any way you wish. Still, there is one convention that has emerged in Doom’s community that you may choose to adopt: extra files in the root of the module are prefixed with a plus, e.g. +extra.el. There is no syntactical or functional significance to this convention.

These can be loaded with the load! macro, which will load an elisp file relative to the file it’s used from. e.g.

;; Omitting the file extension allows Emacs to load the byte-compiled version,
;; if it is available:
(load! "+git")   ; loads ./+git.el

This can be useful for splitting up your configuration into multiple files, saving you the hassle of creating multiple modules.

Load order

A module’s files have a precise load-order, which differs slightly depending on what kind of session it is. Doom has three types of sessions:

Interactive session
the typical session you open when you intend to use Emacs (e.g. for text editing). This loads the most, because you will likely be using a lot of it.
Batch session
this is a non-interactive session, loaded when you execute Emacs commands on the command line with no UI, e.g. emacs --batch --eval '(message "Hello world")'.

The expectation for these sessions is that it should quickly spin up, run the command then quit, therefore very little is loaded in this session.

CLI session
this is the same as a batch session except it is what starts up when you run any bin/doom command.

With that out of the way, here is the load order of Doom’s most important files:

FileInteractiveBatchCLI
~/.emacs.d/early-init.el (Emacs 27+ only)yesnono
~/.emacs.d/init.elyesnono
$DOOMDIR/init.elyesyesyes
{~/.emacs.d,$DOOMDIR}/modules/*/*/init.elyesyesyes
$DOOMDIR/cli.elnonoyes
{~/.emacs.d,$DOOMDIR}/modules/*/*/cli.elnonoyes
{~/.emacs.d,$DOOMDIR}/modules/*/*/config.elyesnono
$DOOMDIR/config.elyesnono

Flags

A module’s flag is an arbitrary symbol. By convention, these symbols are prefixed with a + or a - to denote the addition or removal of a feature, respectively. There is no functional significance to this notation.

A module may choose to interpret flags however it wishes, and can be tested for using the featurep! macro:

;; Has the current module been enabled with the +my-feature flag?
(when (featurep! +my-feature) ...)

;; It can be used to check the presence of flags in other modules:
(when (featurep! :lang python +lsp) ...)

Use this fact to make aspects of a module conditional. e.g. Prevent company plugins from loading if the :completion company module isn’t enabled.

Doom cookies

Autoload cookies were mentioned earlier. A couple more exist that are specific to Doom Emacs. This section will go over what they do and how to use them.

;;;###if

Any file in a module can have a ;;;###if FORM cookie at or near the top of the file (must be within the first 256 bytes of the file). FORM is evaluated to determine whether or not to include this file for autoloads scanning (on doom sync) or byte-compilation (on doom compile).

i.e. if FORM returns nil, Doom will neither index its ;;;###autoload cookies nor byte-compile the file.

Use this to prevent errors that would occur if certain conditions aren’t met. For example, say file.el is using a certain function that won’t be available if the containing module wasn’t enabled with a particular flag. We could safe guard against this with:

;;;###if (featurep! +particular-flag)

This will prevent errors at compile time or if/when that file is loaded.

Another example, this time contingent on so-long not being present:

;;;###if (not (locate-library "so-long"))

Keep in mind that FORM runs in a limited, non-interactive sub-session. I don’t recommend doing anything expensive or especially complicated in them.

;;;###package

This cookie exists solely to assist the doom/help-packages command. This command shows you documentation about packages in the Emacs ecosystem, including the ones that are installed. It also lists a) all the modules that install said package and b) all the places it is configured.

It accomplishes A by scanning for at package! declarations for that package, but it accomplishes B by scanning for:

  • after! calls
  • use-package! or use-package calls
  • and ;;;###package X cookies, where X is the name of the package

Use it to let doom/help-packages know where to find config for packages where no after! or use-package! call is involved.

;;;###autodef

An autodef is a special kind of autoloaded function (or macro) which Doom guarantees will always be defined, whether or not its containing module is enabled (but will no-op if it is disabled).

If the containing module is disabled the definition is replaced with a macro that does not process its arguments, so it is a zero-cost abstraction.

You can browse the available autodefs in your current session with M-x doom/help-autodefs (SPC h d u or C-h d u).

An autodef cookie is used in exactly the same way as the autoload cookie:

;;;###autodef
(defun set-something! (value)
  ...)

An example would be the set-company-backend! function that the :completion company module exposes. It lets you register company completion backends with certain major modes. For instance:

(set-company-backend! 'python-mode '(company-anaconda))

And if :completion company is disabled, this call and its arguments are left unprocessed and ignored.

Common mistakes when configuring Doom Emacs

Having helped many users configure Doom, I’ve spotted a few recurring oversights that I will list here, in the hopes that it will help you avoid the same mistakes:

Packages are eagerly loaded

Using use-package! without a deferring keyword (one of: :defer :after :commands :defer-incrementally :after-call) will load the package immediately. This causes other packages to be pulled in and loaded, which will compromise many of Doom’s startup optimizations.

This is usually by accident. Choosing which keyword to use depends on the needs of the package, so there is no simple answer to this.

Manual package management

A lot of Emacs documentation and help will contain advice to install packages with package.el’s API (e.g. package-install) or with use-package’s :ensure keyword). You are free to do this, if it is your preference, but otherwise, Doom has its own package management system.

Migrating use-package code to Doom is usually a case of removing the :ensure keyword and adding a (package! PACKAGENAME) to ~/.doom.d/packages.el (and running doom sync to sync your config).

Using org-babel-do-load-languages to load your babel packages

You don’t need org-babel-do-load-languages. Doom lazy loads babel packages based on the language name in #+BEGIN_SRC blocks needed. As long as the babel plugin is installed and the plugin is named after its language (e.g. #+BEGIN_SRC rust will load ob-rust), you don’t need to do anything else.

There may be some special cases, however. Doom tries to handle a couple of them (e.g. with ob-jupyter, ob-ipython and ob-async). If you are experiencing errors while trying to use a certain language in org src blocks, check out the :lang org module documentation for details on how to add support for it.

Using delete-trailing-whitespaces or whitespace-cleanup to manage leftover whitespace

(add-hook 'after-save-hook #'delete-trailing-whitespace)
;; or
(add-hook 'after-save-hook #'whitespace-cleanup)

These two lines are a common sight in Emacs configs, but they are unnecessary for Doom Emacs. We already use the more sophisticated wsbutler to manage extraneous whitespace. However, you might have the impression that it isn’t working. That’s because wsbutler works in two unusual ways, meant to be less imposing than its alternatives:

  1. It only cleans up trailing whitespace on lines that you’ve touched (but always strips newlines at EOF).

    Why do this? Because I believe file-wide reformatting should be a deliberate act (and not blindly automated). If it is necessary, chances are you’re working on somebody else’s project – or with other people, but here, large scale whitespace changes could cause problems or simply be rude. We don’t endorse PRs that are 1% contribution and 99% whitespace!

    However, if it’s truly deliberate, M-x delete-trailing-whitespaces and M-x whitespace-cleanup are available to be called deliberately, instead.

  2. wsbutler replaces trailing whitespace and newlines with virtual whitespace. This is whitespace that only exists in the Emacs buffer, but isn’t actually written to the file.

    Why do this? Because you might have wanted to use that space for something in your current editing session, and it would be inconvenient for the editor to delete it before you got to it.

    If you use it, it’s there. If you don’t, it isn’t written to the file.

Troubleshoot

When problems arise, you should be prepared to collect information in order to solve them, or for the bug report you’re about to write. Both Emacs and Doom provide tools to make this easier. Here are a few things you can try, first:

  • Investigate the *Messages* log for warnings or error messages. This log can be opened with SPC h e, C-h e or M-x view-echo-area-messages.
  • Look up errors/warnings on the FAQ and Doom’s issue tracker. It is possible that a solution for your issue already exists. The FAQ can be searched from inside Doom with SPC h d f (or C-h d f for non-evil users).
  • Run bin/doom doctor on the command line to diagnose common issues with your environment and config. It will suggest solutions for them as well.
  • bin/doom clean will ensure the problem isn’t stale bytecode in your private config or Doom core. If you haven’t used bin/doom compile, there’s no need to do this.
  • bin/doom sync will ensure the problem isn’t missing packages or outdated autoloads files
  • bin/doom build will ensure the problem isn’t stale package bytecode or broken symlinks.
  • bin/doom update will ensure that your packages are up-to-date, eliminating issues that originate from upstream.
  • If you happen to know what module(s) are relevant to your issue, check their documentation (press <leader> h m to jump to a module’s documentation). Your issue may be documented.
  • If possible, see if the issue can be reproduced in vanilla Emacs (Emacs without Doom) and/or vanilla Doom (Doom without your private config). Doom’s sandbox can help you check.
  • Ask for help on our Discord server. It is the quickest way to get help, sometimes straight from Doom’s maintainer, who is very active there.

If none of these things have helped you, then it’s time to open a bug report. See ”Reporting Issues” in the contributing guidelines on how to file an effective bug report.

Looking up documentation and state from within Emacs

…

Variables, functions, faces, etc.

Emacs is a Lisp interpreter whose state you can access on-the-fly with tools provided to you by Emacs itself. They’re available on the SPC h prefix by default. Use them to debug your sessions.

Here are some of the more important ones:

  • describe-variable (SPC h v)
  • describe-function (SPC h f)
  • describe-face (SPC h F)
  • describe-bindings (SPC h b)
  • describe-key (SPC h k)
  • describe-char (SPC h ')
  • find-library (SPC h P)

You can also evaluate code with eval-expression (M-; or SPC ;).

For Doom Modules, packages, autodefs, etc.

doom/open-news (SPC h n)
…
doom/help (SPC h d h)
Open the index of Doom’s manual.
doom/help-modules (SPC h d m)
Jumps to a module’s documentation.
doom/help-autodefs (SPC h u)
Jumps to the documentation for an autodef function/macro. These are special functions that are always defined, whether or not their containing modules are enabled.
doom/help-packages (SPC h p)
Look up packages that are installed, by whom (what modules) and where jump to all the places it is being configured.
doom/info
…

How to extract a backtrace from an error

If you encounter an error while using Doom Emacs, you’re probably about to head off and file a bug report (or request help on our Discord server). Before you do, please generate a backtrace to include with it.

To do so you must enable debug-on-error then recreate the error.

Enabling debug-on-error

There are three ways to enable debug-on-error:

  1. Start Emacs with emacs --debug-init. Use this for errors that occur at startup.
  2. Evil users can press SPC h d d and non-evil users can press C-h d d.
  3. If the above don’t work, there’s always: M-x toggle-debug-on-error

Now that debug-on-error is on, recreate the error. A window should pop up with a backtrace.

A backtrace from bin/doom

If the error you’ve encountered is emitted from bin/doom, you can re-run the same command with the -d or --debug switches to force it to emit a backtrace when an error occurs. The DEBUG environment variable will work to.

doom -d sync
doom --debug install
DEBUG=1 doom update

Note: switch order is important. -d / --debug must come right after doom and before the subcommand. This will be fixed eventually.

Evaluating Elisp on-the-fly

Often, you may find it helpful for debugging to evaluate some Emacs Lisp. Here are couple things you can do:

  • Use M-; (bound to eval-expression),
  • SPC x will open a scratch buffer. M-x emacs-lisp-mode will change it to the appropriate major mode, then use +eval:region (gr) and +eval:buffer (gR) to evaluate code,

How to determine the origin of a bug

Testing in Doom’s sandbox

“The sandbox” is one of Doom Emacs’ features; it is a test bed for running elisp in a fresh instance of Emacs with varying amounts of Doom loaded (none at all, all of it, or somewhere in between). This can be helpful for isolating bugs to determine who you should report a bug to.

If you can recreate a bug in vanilla Emacs than it should be reported to the developers of the relevant packages or, perhaps, the Emacs devs themselves.

Otherwise, it is best to bring it up on the Doom Emacs issue list, rather than confusing and inundating the Emacs community with Doom-specific issues.

Opening the sandbox

There are three common ways to access the sandbox:

  • SPC h E (for evil users)
  • C-h E (for non-evil users)
  • M-x doom/sandbox

Doing any of the above will pop up a *doom:sandbox* window. What you enter into this buffer will be executed in the new instance of Emacs when you decide to launch it.

Launching the sandbox

You have four options when it comes to launching the sandbox:

C-c C-c
This launches “vanilla Emacs”. Vanilla means nothing is loaded; purely Emacs and nothing else. If you can reproduce an error here, then the issue likely lies in the plugin(s) you are testing or in Emacs itself.
C-c C-d
This launches “vanilla Doom”, which is vanilla Emacs plus Doom’s core. This does not load your private config, nor any of Doom’s (or your) modules.
C-c C-p
This launches “vanilla Doom+”. That is, Doom core plus the modules that you have specified in the doom! block of your private config (in ~/.doom.d/init.el). This does not load your private config, however.
C-c C-f
This launches “full Doom”. It loads Doom’s core, your enabled modules, and your private config. This instance should be identical to the instance you launched it from.

All new instances will inherit your load-path so you can access any packages you have installed.

Testing packages in the sandbox

Instances of Emacs launched from the sandbox have inherited your load-path. This means you can load packages – even in Vanilla Emacs – without worrying about installing or setting them up. Just (require PACKAGE) and launch the sandbox. e.g.

(require 'magit)
(find-file "~/some/file/in/a/repo")
(call-interactively #'magit-status)

Bisecting your private config

Bisecting Doom Emacs