This repository contains my configuration for GNU Emacs.

It is intended to be used only by me, not by other people. Anyway, here are the main features of my config.

• No Org-mode for emacs config because good old .el files are much more convenient for me.
• Unusable key bindings: I use hundreds and hundreds of non-standard key bindings that are suitable for me. Besides, I use dvorak layout, so my bindings may look really weird for other people. I use my code (from keys.el) based on bind-key to manage those tons of keys.
• I prefer to split my config into multiple files instead of keeping everything in a single one. And it is really easy to search them using imenus. I just press s-s (which is bound to al/imenus-search-elisp-dir from al-imenus.el) and jump to with-eval-after-load clause or perform isearch or occur if I need to search for something.
• I use an insane mix of quelpa/melpa/GNU Guix to install/upgrade Emacs packages. Do not try this at home ☺.
• My Emacs config is intended to be robust. I mean it does not require any third-party packages to be started successfully. So I can easily deploy my config on a fresh system: I just clone the repo and make ~/.emacs.d/init.el file a symlink to my init.el. Actually, this process is even more simple, as I don’t do this manually — I use my config script to fetch and deploy all my configs including this Emacs config.
• And I don’t use use-package anymore (see commit 2e49f03). I used to do it until I realized the importance of the previous point. When you use use-package in your config, it becomes a required dependency for starting Emacs, so on a “fresh” system, you need to install it at first, before you can use your config. Also someday it may be changed in a way that will break your system, and you’ll have to fix your config without using it.

  Update: use-package is shipped with Emacs nowadays but I don’t see any benefits to use it again.

• My Emacs start time is always less than a second according to M-x emacs-init-time (actually, I have this value in my initial-scratch-message so I know the init time very well), because:
  • I use some hacks to reduce startup time in my early-init.el.
  • I don’t require packages without a reason as some people do. Instead, I use with-eval-after-load everywhere.
  • I don’t load …-autoloads.el files for every installed package (Emacs does it by default). I disable this autoloading by setting package-enable-at-startup to nil in early-init.el. Instead, I generate and load only a couple of autoloads:
    • one file for all Emacs packages (installed from MELPA, etc.),
    • one file for my utils,
    • one file for my packages.

    Search for al/generate-autoloads at al-autoload.el and init.el to see how autoloads are generated. I also use al/generate-autoloads-from-presets command to quickly update these autoloads when I need.

As you can see, this repository consists of the following directories:

• init: files that are loaded on Emacs start. As described above, init.el is the “main” init file. It defines some major things and loads the rest files from this directory.
• utils: various utilities and additional code (for Emacs itself and for external packages) that is not needed for start-up. All of the code here has al/ prefix and is placed in files that provide al-… features. My init files are full of things like this:

  (with-eval-after-load 'eshell
    ;; ...
    (when (require 'al-eshell nil t)
      ;; ...
      ))
      

  So these utils are loaded only if and when they are needed. Also there are many interactive commands there which are autoloaded.

  In the past these utils were placed in a separate repository and had “utl-” prefix.

• packages: small packages that do not deserve their own repositories. I also clone all my “big” packages to this directory (they are ignored by .gitignore).
• unused: files with old and unused code.
• data: just some Emacs related data (like eshell aliases, auto save and backup files, gnus archive, elpa packages, etc.).