Table of contents
Open Table of contents
Why do I need it
I’ve recently found myself switching between computers regularly. I have a work laptop, a personal laptop and a personal desktop, although I often use all three for both personal and professional work. This has driven me to try and standardise my configuration across all of my devices and one of the tools that I need to be configured correctly on all devices is Git.
My workflow until now has been something like this:
- Clone or start a project
- Commit and push to remote
- See in PR that I’m commit to a work repo with a personal email or vice versa
- Run
git config --local1.user.email ...with the correct email - Force push the changes
- Go to sleep knowing I will do the same thing the next day
Running git config ... regularly is becoming tiresome, so I decided to create a .gitconfig file that I can take anywhere with me. Even spinning up and working in a container or a VM, I can quickly configure Git just by adding the .gitconfig file.
Also, Git just has some weird and verbose commands. I mean, think about listing tags, branches and remotes:
git tag -lgit branch -agit remote -v
Surely if we are just trying to list things it would be more intuitive for all three to have the same list command! And yes, I know git branch can accept -l too.
Configuring Git
There are a few places where we can configure Git and in most of them you will probably already find some basic configuration already.
System
/etc/gitconfig holds your system-wide git configuration. You probably don’t ever have to touch this file unless you have a specific system level use case. I haven’t really found the need to. You can use git config --system to update this file.
User
You’ll find user level config files in a couple of places:
$XDG_CONFIG_HOME/config$HOME/.gitconfig(if$XDG_CONFIG_HOMEdoesn’t exist)
If both of these files exist, then they are read in that order. This is the equivalent of git config --global ....
Local
For repository level configurations you can find that at $GIT_DIR/config (normally .git/config) and that’s the equivalent of using git config --local.
Aliases
One of the most powerful things that you can configure in your .gitconfig file are aliases.
You can set them like this:
[alias]
cm = commit
This alias allows you to use “cm” instead of writing the full “commit” command. Personally, I don’t use these aliases much since I tend to use tab to autocomplete and I don’t feel it saves much time at all. Victor Zhou did the maths and found that with an average typing speed of 135WPM, his git aliases only saved him about 1 minute of typing per day.
I prefer to use aliases to quickly run longer commands, here are some of my favourites:
[alias]
# Lists
aliases = config --get-regexp alias
tags = tag -l
branches = branch -a
remotes = remote -v
Basically running git branches will actually run git branch -a.
We can also pass arguments to our aliases:
[alias]
# Amending
amend = commit --amend --no-edit
rename = "!f() {commit --amend -m $1}"
Notice that the amend alias doesn’t take any arguments, but the rename command does. So if I run git rename "foo" then it will amend my last commit message to “foo”.
So here’s a pretty sensible (in my opinion) .gitconfig file to get you started. Obviously, update your editor, name and email before using this.
[core]
# Line endings
autocrlf = false
# Use custom `.gitignore` and `.gitattributes`
excludesfile = ~/.gitignore
attributesfile = ~/.gitattributes
# Default editor for .git files
editor = vim
[alias]
# Lists
aliases = config --get-regexp alias
tags = tag -l
branches = branch -a
remotes = remote -v
# Amending
amend = commit --amend --no-edit
rename = "!f() {commit --amend -m $1}"
graph = log --oneline --graph
[help]
# Correct and execute mistyped commands
autocorrect = 1
[init]
defaultBranch = main
[user]
name = Benjamin Rae
email = ben@benjaminrae.dev
[pull]
ff = only
[fetch]
prune = true
[color]
ui = auto
Conditional Configuration
Now to the best part. I want to be able to have different configurations for personal and work repositories. How you do that?
One option would be to have two different users on your machine, one for work and one for personal use. Then each can have their own dedicated .gitconfig. But I don’t like that option: too much logging in and out.
I recently came across these two approaches:
https://www.brantonboehm.com/code/conditional-git-config/ (using remotes) https://blog.scottlowe.org/2023/12/15/conditional-git-configuration/ (using directories)
The first option uses remotes to update your .gitconfig. I think this is a good option if you are working at several companies and need to fine-tune configuration for each remote.
The next uses directories, so if you would would want to keep all your personal projects in one directory, and the work ones in another. I assume that most of you already do this.
Both options seem adaptable for most scenarios, but I think I’m a little more attracted to remotes, especially if you’re likely to clone a work project to a personal directory on your machine. This also helps with the portability of our .gitconfig file since remotes should stay the same across machines but our folder structure may not. However. you will need to either make sure you don’t mix HTTP and SSH remotes, or include configuration for both of them.
So as well as our .gitconfig file, let’s create one for work. I’m going to call it .gitconfig.work but you call call it whatever you want:
[user]
email = work@benjaminrae.dev
And in the original .gitconfig we’ll add this to the end:
[includeIf "hasconfig:remote.*.url:https://gitlab.benjaminrae.dev/*/**"]
path = .gitconfig.work
Now all you need to do is put these files into a remote repository and then clone them to any machines that you plan on using.
Finally, RTFM
I don’t know why, but for a very long time I found Git’s documentation to be the most intimidating and I never really delved into it. If I had, then I’d have probably set up this configuration a long time ago. Remember, everything I’ve covered is also in the official Git documentation: