This past week at RITLug, I give a quick overview of my favorite editor and customizations. Of course, I can’t imagine a text editor that doesn’t include a Tetris clone, a psychotherapist, or a Tower of Hanoi player.
I’m talking about Emacs of course: undoubtedly, the most feature-rich text editor in existence. But, also, one of the least user-friendly editors.
Emacs is quite the rabbit hole. So many packages, extensions, and customizations have been built for it over the years. While VS Code and Atom advertise extension, Emacs is essentially a Lisp machine with a screen: all of its internals are exposed, meaning customization is all but limitless.
How much freedom is too much?
That status endless customization comes at a cost, however. There aren’t really rules when writing Emacs code: in its Lisp dialect (Elisp), all variables and functions are global, so anything can modify anything at any time. Often, when I attempted to customize my Emacs, there would be an odd variable or list element that I could never pin down (or, even worse, was dynamically created/modified through macros), and eventually just give up.
Freedom to customize is important, but when that customization isn’t formal, it can lead to impracticality.
Maintaining my own Emacs config
I tried keeping my own dotfiles for Emacs for a couple years.
I came from Vim, so I wanted Vim bindings, and I heard Emacs did that well. That was my first concern, and my first pain point.
Plenty of Emacs documentation still discusses Viper, a very old package that’s been blown out of the water by Evil. Both of these provide vi-style bindings, but Evil does it much better, and is actively maintained.
So, once you’ve discovered Evil is better than Viper, then comes the question of configuring it. Evil provides its own functions for adding keybinds to different evil states. States are what Evil calls its vi modes (normal, visual, insert, etc.) because Emacs already has a concept of major and minor modes running in a buffer.
The main function for this is
evil-define-key. It asks for a map. What’s a map? A series of keybinds a mode uses. Or you can use
'global for all maps, according to the documentation. What does that tick mean? Ah, of course, that’s the name of a symbol (aka, a variable), rather than the value of the symbol itself.
Yikes. If this sounds like a lot, that’s because it is.
Coming from Vim, where batteries were included with every package, minimal configuration was needed, and documentation on a plugin was (generally) a page, this was a big paradigm shift. And I’m not alone in getting overwhelmed.
People go crazy with their Emacs configs. Hell, people build literate dotfiles in Org mode, then use its insane feature set to generate the actual config Emacs reads from that.
Many people create such complicated configs, they eventually throw them all out and start over. This practice is commont enough to have a name.
Why recommend Emacs then?
Despite the learning curve and insane amount of time it takes to configure, Emacs is still an incredibly powerful ecosystem, and integratable with any software system on the planet.
But, there are people way better at configuring it than I am. So good, in fact, they made configurable configurations.
There reaches a point where I want to do work, rather than messing with my editor. It’s fun, I learn a lot, and I can share what I create, but sometimes I just need autocomplete to work, or schoolwork just needs to get done.
Instead of working directly with packages and their configurations, Spacemacs abstracts these through layers. Layers are collections of packages with a standard configuration (such as Evil bindings, same defaults, hooks in the right place, etc.). Adding layers is simple, just add an entry to your
.spacemacs file. Add
rust for the Rust layer,
php for the PHP layer, and
auto-complete for autocomplete for both of them.
I swapped over to this several months ago: while I’ve had to adapt, it’s nice knowing I’m not the only maintainer of my config, and that I don’t have to be the only bugfixer either!
So, in short, what did I change to show off at RITLug, and what could I do?
All I did was add layers the Spacemacs community had already built, and show them off. So, not only was this a powerful way to config, it could also be reproduced in a matter of minutes, without learning any Elisp.
- Magit (via the git layer), for visual staging and unstaging of work
- GDB visualization, Spacemacs sets
gdb-many-windows, making GDB debugging beautiful. It also manages window sizes when adding and removing windows, which helps a lot when GDB pulls up its 6 windows.
- LaTeX rendering of math, in editor (via the latex layer)
- Remote editing (and building, and so worth) with TRAMP
So, why Spacemacs?
Spacemacs isn’t perfect: you don’t get fine-grained tuning, like you do with vanilla Emacs. But, quite frankly, I don’t need it. I would rather accept some unfamiliar configurations for ease of use, and still hold all the power of Emacs and its plugins.