Skip to main content

Git

Here are my notes on Git.

Resources​

High quality resources to learn Git

Introduction​

Git is the go-to tool for professional developers.

The ability to use Git is an essential skill for almost every developer. Git sounds simple enough, but it does have a bit of a learning curve. And if you get stuck, it can be hard to get unstuck if you lack some understanding.

Over the years I've derived at a relatively simple and foolproof workflow, with accompanied config. I'll first describe the basic steps. Then I'll describe how to deal with specific situations, like merge conflicts.

note

I'm using Git for Windows and opt-in ot the linux tools for bash during the installation, which allows using linux commands in Windows (PowerShell 7+)

One-time setup​

Setup key​

(once per machine)

I use SSH keys to authenticate with GitHub. The private key requires a passphrase, so I'm the only one who can use it.

To set up a new key I always use this excellent guide from GitHub.

Setup git config​

(once per machine)

Over the years I've collected a few useful settings for my git config. Take what you need.

# ~/.gitconfig

# First things first, always set up your user info.
[user]
name = my_username
email = [email protected]

# Display colours in the console
[color]
ui = auto
branch = auto
status = auto

# Always rebase when using `git pull`, except for `main, master, develop` branches.
[branch]
autosetuprebase = always

[branch "main"]
rebase = false

[branch "master"]
rebase = false

[branch "develop"]
rebase = false

# Automatically set up remote tracking branches for all new local branches.
[push]
autoSetupRemote = true

# Important setting in windows that allows working on both Windows and cross-platform projects
[core]
eol = native
autocrlf = input

# Common typos and aliases for long commands
[alias]
chekcout = checkout
ocmmit = commit
statsu = status
sttaus = status
fa = fetch --all
pushf = push --force-with-lease
stsah = stash

# Common LFS Settings
[filter "lfs"]
smudge = git-lfs smudge -- %f
process = git-lfs filter-process
required = true
clean = git-lfs clean -- %f

# How private registries are handled
[credential "https://my_company_vcs.example.com"]
provider = generic

Clone​

(once per repository)

I usually clone using SSH, because my private key has a passphrase for extra security.

git clone [email protected]:webbertakken/takken.io.git
note

LFS repositories really don't work well with SSH, so in those cases I clone using HTTPS instead.

git clone https://github.com/webbertakken/takken.io.git

The Commit tool window​

(discovering this feature: once in your life)

To review the changes I use WebStorms built-in commit tool window.

You can enable this tool window by clicking View > Tool Windows > Commit.

If the tool window is still not visible or looks very different, you might have to enable the non-modal commit interface in the settings.

Basic workflow​

info

This way of working with Git strikes a balance between maximum flexibility and keeping a very easy to understand rationale.

Most commonly I work with repositories that I have direct access to, so I can push and pull.

When using your own fork you have to create a pull request to the upstream repository instead of to the main branch. The rest works the same.

1. Create a local branch​

Mostly I try to make every change or group of changes in a separate (local) branch so that I can create a pull request when the changes are finished. create Pull Requests.

git checkout -b my-feature
note

Sometimes I make changes first, but I always switch from main to a feature branch before committing the first changes

2. Make changes​

Make changes to the code, add files, remove files, etc.

tip

When editing code, or any file for that matter, use a proper editor like VS Code, Intellij (e.g. WebStorm or Rider), Google IDX etc.

I mostly use Intellij editors because of their superior refactoring capabilities and because of their really neat commit tool window. All Intellij editors are free if you contribute to open source.

3. Review the changes locally​

Then review those changes (the diff) inside the IDE.

Once reviewed and satisfied, I commit the changes, then repeat steps 2 and 3 as needed.

4. commit the changes​

I use two ways to commit changes.

  1. Simply committing all changes at once from the command line, using git commit -am "feat: my feature description"
  2. Only committing partial changes. I do this using the commit tool window, simply by checking all files, and then unchecking some of them.
tip

Using the commit tool window you can even select specific lines of code to commit when you click show diff.

5. Push the changes​

When the changes are ready to be shared, I push the changes to the remote repository.

git push
note

Since my git config is set up to automatically set up remote tracking branches for all new local branches, that is all.

If that is not the case the command would be

git push -u origin my-feature

6. Pull request​

When the changes are pushed to the remote repository, I create a pull request to the main branch.

When working on a repository by myself, I usually work on getting the pull request merged immediately.

7. Getting ready to work on the next feature​

Update the main branch before switching back to it.

git fetch -u origin main:main

Then switch to it

git checkout main
info

This can be especially quick if you have just merged the feature into main, since no files need to be changed locally - because main is already up-to-date with remote (and thus the feature that you just merged).

Complete flow in practice​

Here's what the complete happy flow looks like in practice.

Workflow: Remote changes, rebase using stash​

Sometimes you have changes on the remote repository that you want to pull in, but you also have local changes that you want to keep.

info

This is common when working on the same branch with multiple people, or when you are working from multiple machines, or when you haven't switched into a feature branch yet.

If you haven't committed your local changes yet, you can simply stash them, pull the remote changes, and then re-apply the stashed changes.

1. Stash local changes​

git stash

2. Pull remote changes​

git pull

3. Re-apply stashed changes​

git stash pop

Complete flow in practice​

Here's what the complete flow looks like in practice.

What are your thoughts?