Hacker News Reader

Comment by sigjuice | Hacker News Reader

I have never understood the rampant practice of using ~/.emacs.d/ for one's personal Elisp files. Most of what is in this directory appears to be machine generated files from various packages. To me, it seems much cleaner to have a separate directory for my own files, instead of maintaining ~/.emacs.d in Git and having to constantly update my .gitignore as I install and use new packages.

EDIT: Also, C-h v user-emacs-directory RET says programs in Emacs store information in this directory.

View on HN

1. You shouldn't be updating your `.gitignore` continuously. You need to ignore everything and explicitly define exceptions e.g.

  *
  !.gitignore
  !init.el

2. The reason I prefer ~/.emacs.d/ is that I have it in a git repo and on every system, I'll just clone the repo and soft link it in the home directory. This way, I have only one directory inside home about emacs within which my files are explicitly separated and tracked.

P.S. I've recently tried to move from Spacemacs to Emacs and by no means a pro, so the process I've described may not be optimal.

I solve this issue by using my own config directory; this way I soft-link two things from my dotfiles repo:

- .emacs - which defines my config dir and loads everything in there

- my own config dir (called .emacs.conf), in which I have lots of .el files with specific configurations

Emacs packages still write by default to .emacs.d, and this way I don't have to care about any of that when moving between machines, because my configuration and Emacs's ongoing state are separate.

To be honest, though, I'll probably be migrating into a single-file literate solution soon, which will render my own config directory unnecessary.

I do this as well! I have this repo [0] at ~/cfg, then I symlink what I need [1] as an init script. My end goal is to be able to run a single script and get everything [2] set up on a new machine (I change machines somewhat frequently and I got the manual process to under 45 minutes, so there's room for improvement). More on setting up Macs [3].

There is also GNU Stow [4] that does something similar, and probably better.

0: https://github.com/andreis/cfg

1: https://github.com/andreis/cfg/blob/master/init.sh#L64-L69

2: brew and brew cask, dotfiles ofc, and a _lot_ of system settings on mac os can be changed via the CLI, see

3: https://github.com/iCHAIT/awesome-macOS#setup

4: http://brandon.invergo.net/news/2012-05-26-using-gnu-stow-to...

I used to do it that way. I've switched to `vcsh` to let me keep various files directly in my ~/ directory in various different git repos without needing a ~/.git/ marking the whole thing as a working directory.

So each time you launch emacs you have to point it to your init file?

No. My .emacs has a code snippet that points Emacs to another directory for my own configs. I softlink both .emacs and the config directory.

1. I still need to grow my .gitignore file as I add more .el files of my own.

2. A similar scheme would be to maintain a directory called ~/elisp in git and make ~/.emacs a symlink to ~/elisp/emacs. And pretend that ~/.emacs.d/ does not exist.

1. Perhaps make a new directory `.emacs.d/elisp`, append it to your path and add your files in it and have .gitignore not ignore the directory?

2. You could!

Using ~/.emacs.d this way came about well before git was invented. :)

Really? C-h v user-emacs-directory RET says programs in Emacs store information in this directory. To me, this says ~/.emacs.d/ is not intended for humans.

  user-emacs-directory is a variable defined in ‘subr.el’.
  Its value is "~/.emacs.d/"

    This variable may be risky if used as a file-local variable.

  Documentation:
  Directory beneath which additional per-user Emacs-specific files are placed.
  Various programs in Emacs store information in this directory.
  Note that this should end with a directory separator.
  See also ‘locate-user-emacs-file’.

It happened in 2002, and git was started in 2005. See commit a04d5983:

    commit a04d59830e46664faa4d534366a6b07e3b7c12f2
    Author: Richard M. Stallman <rms@gnu.org>
    Date:   Sat Oct 26 22:41:33 2002 +0000
        (command-line): Look for .emacs under ~/.emacs.d after looking in ~.

Interestingly it was changed to look for ~/.emacs.d/init.el in 2005, six months after git got started.

See also chapter 51.4 of the Emacs manual (which is often preferred over the documentation of variables simply because it can be more explanatory):

    51.4 The Emacs Initialization File
    ==================================
    
    When Emacs is started, it normally tries to load a Lisp program from an
    “initialization file”, or “init file” for short.  This file, if it
    exists, specifies how to initialize Emacs for you.  Emacs looks for your
    init file using the filenames ‘~/.emacs’, ‘~/.emacs.el’, or
    ‘~/.emacs.d/init.el’; you can choose to use any one of these three names
    (*note Find Init::).  Here, ‘~/’ stands for your home directory.

So yes, it's long been used for things like the list of recently opened files, and configuration changes you made with Custom, and the UIDVALIDITY of your IMAP folders and so on, but it's also been a place to put your init file for quite a long time as well. Once you get a longish init.el file, it's a pretty short step to making several .el files that you load from init.el, and from there you can get as fancy as you like.

Programs and humans can peacefully coexist. Emacs looks for your init.el in that directory so it is most definitely also meant for people to use. I put my own libraries in ~/.emacs.d/my-lisp (and I put this directory on the load path). I've had zero trouble with this setup.

I don't think there is a right and wrong way to customize emacs, whatever works is good. What I do is have a .emacs and my custom elisp files in a Dropbox folder. Then I have a simple ~/.emacs file that loads the Dropbox one. If I didn't use Dropbox for various other purposes I think I'd use git as well.

Reply on news.ycombinator.com