Revision Control (Using Git & GitHub)
Git-Mastery lessons are divided into several 'tours', each aiming to cover the knowledge required to perform a general revision control use case.
Target Usage: To use Git to systematically record the history of a folder in your own computer. More specifically, to use Git to save a snapshot of the folder at specific points of time.
Motivation: Recording the history of files in a folder (e.g, code files of a software project, case notes, files related to an article/book that you are authoring) can be useful in case you need to refer to past versions.
Lesson plan:
T1L1. Introduction to Revision Control covers that part.
T1L2. Preparing to Use Git covers that part.
T1L3. Putting a Folder Under Git's Control covers that part.
T1L4. Specifying What to Include in a Snapshot covers that part.
T1L5. Saving a Snapshot covers that part.
T1L6. Examining the Revision History covers that part.
Before you start learning Git, you need to install some tools on your computer.
Installing Git
Git is a free and open source software used for revision control. To use Git, you need to install Git on your computer.
Download the Git installer from the official Git website.
Run the installer and make sure to select the option to install Git Bash when prompted.
Screenshots given below provide some guidance on the dialogs you might encounter when installing Git. In other cases, go with the default option.
When running Git commands, we recommend Windows users to use the Git Bash terminal that comes with Git. To open Git Bash terminal, hit the key and type git-bash.
It may be possible that the installation didn't add a shortcut to the Start Menu.
You can navigate to the directory where git-bash.exe is (most likely C:\Program Files\Git\git-bash.exe), double click git-bash.exe to open Git Bash.
You can also right-click it and choose Pin to Start or Pin to taskbar.
SIDEBAR: Git Bash Terminal
Git Bash is a terminal application that lets you use Git from the command line on Windows. Since Git was originally developed for Unix-like systems (like Linux and MacOS), Windows does not come with a native shell that supports all the commands and utilities commonly used with Git.
Git Bash provides a Unix-like command-line environment on Windows. It includes:
- A Bash shell (Bash stands for Bourne Again SHell), which is a widely used command-line interpreter on Linux and MacOS.
- Common Unix tools and commands (like
ls,cat,ssh, etc.) that are useful when working with Git and scripting.
When copy-pasting text onto a Git Bash terminal, you will not be able to use the familiar Ctrl+V key combo to paste. Instead, use Shift+Insert, or, right-click on the terminal and use the Paste menu option.
On Windows, you might need to close and open the terminal again for it to recognise changes done elsewhere in the computer (e.g., newly-installed software, changes to system variables, etc.).
Install homebrew if you don't already have it, and then, run brew install git
Use your Linux distribution's package manager to install Git. Examples:
- Debian/Ubuntu, run
sudo apt-get updateand thensudo apt-get install git. - Fedora: run
sudo dnf updateand thensudo dnf install git.
Verify Git is installed, by running the following command in a terminal.
git --version
git version 2._._
The output should display the version number.
Configuring user.name and user.email
Git needs to know who you are to record changes properly. When you save a snapshot of your work in Git, it records your name and email as the author of that change. This ensures everyone working on the project can see who made which changes. Accordingly, you should set the config settings user.name and user.email before you start Git for revision control.
Configuring init.defaultBranch
Git has a property named init.defaultBranch that specifies a branch name that it uses for a couple of things (you'll learn more about Git branches in later lessons). Git uses the default value master for this but these days it is more common to use main instead. Git-Mastery uses main too. To make your Git behaviour match our lessons better, you should set this property to main, as described in the panel below:
Interacting with Git: CLI vs GUI
Git is fundamentally a command-line tool. You primarily interact with it through its by typing commands. This gives you full control over its features and helps you understand what’s really happening under the hood.
clients for Git also exist, such as Sourcetree, GitKraken, and the built-in Git support in editors like Intellij IDEA and VS Code. These tools provide a more visual way to perform some Git operations.
If you're new to Git, it's best to learn the CLI first. The CLI is universal, always available (even on servers), and helps you build a solid understanding of Git’s concepts. You can use GUI clients as a supplement — for example, to visualise complex history structures.
Mastering the CLI gives you confidence and flexibility, while GUI tools can serve as helpful companions.
Optionally, you can install a Git GUI client. e.g., Sourcetree (installation instructions).
Our Git lessons show how to perform Git operations in Git CLI, and in Sourcetree -- the latter just to illustrate how Git GUIs work. It is perfectly fine for you to learn the CLI only.
[image credit: https://www.sourcetreeapp.com]
Installing the Git-Mastery App
In these lessons, we will be using a companion app called Git-Mastery that we have developed to help Git learners. In particular, it provides exercises that you can do to self-test your Git knowledge, and also verifies if your solution is correct.
If you are new to Git, we strongly recommend that you install and use the Git-Mastery app.
To be able to save snapshots of a folder using Git, you must first put the folder under Git's control by initialising a Git repository in that folder.
Normally, we use Git to manage a revision history of a specific folder, which gives us the ability to revision-control any file in that folder and its subfolders.
To put a folder under the control of Git, we initialise a repository (short name: repo) in that folder. This way, we can initialise repos in different folders, to revision-control different clusters of files independently of each other e.g., files belonging to different projects.
You can follow the hands-on practical below to learn how to initialise a repo in a folder.
What is this? HANDS-ON panels contain hands-on activities you can do as you learn Git. If you are new to Git, we strongly recommend that you do them yourself (even if they appear straightforward), as hands-on usage will help you internalise the concepts and operations better.
Preparation Choose a folder to put under Git's control. The folder may or may not contain any files. For this practical, let us create a folder named things for this purpose.
You can use the Git-Mastery app to for doing this practical, or create the sandbox manually. Instructions for both options are given below.
Avoid putting Git repos inside cloud-synced (e.g., OneDrive, Dropbox) folders. Reason: Multiple tools trying to detect/sync changes in the same folder can cause conflicts and unexpected behaviors.
If you want to access project files from multiple computers, use Git to do that (rather than cloud syncing tools).
1 Then, cd into it. For example,
cd hp-init-repo/things
2 Run the git status command to check the status of the folder.
git status
fatal: not a git repository (or any of the parent directories): .git
Don't panic. The error message is expected. It confirms that the folder currently does not have a Git repo.
3 Now, initialise a repository in that folder.
Use the command git init which should initialise the repo.
git init
Initialized empty Git repository in <path-to-repo>/things/.git/
Note how the output mentions the repo being created in things/.git/ (not things/). More on that later.
Click File → Clone/New… → Click on + Create button on the top menu bar.
Enter the location of the directory and click Create.
New... → Create Local Repository (or Create New Repository) → Click ... button to select the folder location for the repository → click the Create button.
done!
Initialising a repo results in two things:
- First, Git now recognises this folder as a Git repository, which means it can now help you track the version history of files inside this folder.
To confirm, you can run the git status command. It should respond with something like the following:
git status
On branch main
No commits yet
nothing to commit (create/copy files and use "git add" to track)
Don't worry if you don't understand the output (we will learn about them later) or if your output is slightly different (e.g., master instead of main); what matters is that it no longer gives an error message as it did before.
done!
- Second, Git created a hidden subfolder named
.gitinside thethingsfolder. This folder will be used by Git to store metadata about this repository.
A Git-controlled folder is divided into two main parts:
- The repository – stored in the hidden
.gitsubfolder, which contains all the metadata and history. - The working directory – everything else in that folder, where you create and edit files.
What is this? EXERCISE panels contain a Git-Mastery exercise that you can download using the Git-Mastery app, and you can use the same app to verify that your solution is correct.
What is this? DETOUR panels contain related directions you can optionally explore. We recommend that you only skim them the first time you are going through a tour (i.e., just to know what each detour covers); you can revisit them later, to deepen your knowledge further, or when you encounter a use case related to the concepts covered by the detour.
To save a snapshot, you start by specifying what to include in it, also called staging.
Git considers new files that you add to the working directory as 'untracked' i.e., Git is aware of them, but they are not yet under Git's control. The same applies to files that existed in the working folder at the time you initialised the repo.
A Git repo has an internal space called the staging area which it uses to build the next snapshot. Another name for the staging area is the index.
We can stage an untracked file to tell Git that we want its current version to be included in the next snapshot (in Git terminology, such a snapshot is called a commit). Once you stage an untracked file, it becomes 'tracked' (i.e., under Git's control). A staged file can be unstaged to indicate that we no longer want it to be included in the next snapshot.
In the example below, you can see how staging files change the status of the repo as you go from (a) to (c).
other Git metadata ...
staging area
[empty]
├─ fruits.txt (untracked!)
└─ colours.txt (untracked!)
(a) State of the repo, just after initialisation, and creating two files. Both are untracked.
other Git metadata ...
staging area
└─ fruits.txt
├─ fruits.txt (tracked)
└─ colours.txt (untracked!)
(b) State after staging
fruits.txt.other Git metadata ...
staging area
├─ fruits.txt
└─ colours.txt
├─ fruits.txt (tracked)
└─ colours.txt (tracked)
(c) State after staging
colours.txt.Preparation
Option 2: Continue with the sandbox from the previous hands-on practical
1 Add a file (e.g., fruits.txt) to the things repo folder.
Here is an easy way to do that with a single terminal command.
Windows users: Use the git-bash terminal to run the commands given in these lessons. Some of them might not work in other terminals such as the PowerShell.
echo -e "apples\nbananas\ncherries" > fruits.txt
Explanation of the echo -e "apples\nbananas\ncherries" > fruits.txt command
apples
bananas
cherries
To see the content of the file, you can use the cat command (or open it in your favorite text editor):
cat fruits.txt
2 Stage the new file.
2.1 Check the status of the folder using the git status command.
git status
On branch main
No commits yet
Untracked files:
(use "git add <file>..." to include in what will be committed)
fruits.txt
nothing added to commit but untracked files present (use "git add" to track)
Git commands targeting specific repo should be run inside the repo folder. For example, to check the status of the things repo, you need to navigate to the things folder in your terminal before you run the git status command.
Remember this for future Git commands too.
2.2 Use the git add <file> command to stage the file.
git add fruits.txt
You can replace the add with stage (e.g., git stage fruits.txt) and the result is the same (they are synonyms).
Windows users: When using the echo command to write to text files from Git Bash, you might see a warning LF will be replaced by CRLF the next time Git touches it when Git interacts with such a file. This warning is caused by the way line endings are handled differently by Git and Windows. You can simply ignore it, or suppress it in future by running the following command:
git config --global core.safecrlf false
2.3 Check the status again. You can see the file is no longer 'untracked'.
git status
On branch main
No commits yet
Changes to be committed:
(use "git rm --cached <file>..." to unstage)
new file: fruits.txt
As before, don't worry if you don't understand the content of the output (we'll unpack it in a later lesson). The point to note is that the file is no longer listed as 'untracked'.
2.1 Note how the file is shown as ‘unstaged’. The question mark icon indicates the file is untracked.
If the newly-added file does not show up in Sourcetree UI, refresh the UI (: F5
| ⌥+R)
Sourcetree screenshots/instructions: vs
Note that Sourcetree UI can vary slightly between Windows and Mac versions. Some of the screenshots given in our lessons are from the Windows version while some are from the Mac version.
In some cases, we have specified how they differ.
In other cases, you may need to adapt if the given screenshots/instructions are slightly different from what you are seeing in your Sourcetree.
2.2 Stage the file:
Select the fruits.txt and click on the Stage Selected button.
Staging can be done using tick boxes or the ... menu in front of the file.
2.3 Note how the file is staged now i.e., fruits.txt appears in the Staged files panel now.
If Sourcetree shows a \ No newline at the end of the file message below the staged lines (i.e., below the cherries line in the above screenshot), that is because you did not hit enter after entering the last line of the file (hence, Git is not sure if that line is complete). To rectify, move the cursor to the end of the last line in that file and hit enter (like you are adding a blank line below it). This new change will now appear as an 'unstaged' change. Stage it as well.
done!
If you modify a staged file, it goes into the 'modified' state i.e., the file contains modifications that are not present in the copy that is waiting (in the staging area) to be included in the next snapshot. If you wish to include these new changes in the next snapshot, you need to stage the file again, which will overwrite the copy of the file that was previously in the staging area.
The example below shows how the status of a file changes when it is modified after it was staged.
other Git metadata ...
staging area
Alice
Alice
(a) The file names.txt is staged. The copy in the staging area is an exact match to the one in the working directory.
other Git metadata ...
staging area
Alice
Alice
Bob
(b) State after adding a line to the file. Git indicates it as 'modified' because it now differs from the version in the staged area.
other Git metadata ...
staging area
Alice
Bob
Alice
Bob
(c) After staging the file again, the staging area is updated with the latest copy of the file, and it is no longer marked as 'modified'.
Preparation
1 Now, verify that Git sees that file as 'modified'.
Use the git status command to check the status of the working directory.
git status
On branch main
No commits yet
Changes to be committed:
(use "git rm --cached <file>..." to unstage)
new file: fruits.txt
Changes not staged for commit:
(use "git add <file>..." to update what will be committed)
(use "git restore <file>..." to discard changes in working directory)
modified: fruits.txt
Note how fruits.txt now appears twice, once as new file: ... (representing the version of the file we staged earlier, which had only three lines) and once as modified: ... (representing the latest version of the file which now has a fourth line).
Note how fruits.txt appears in the Staged files panel as well as 'Unstaged files'.
2 Stage the file again, the same way you added/staged it earlier.
3 Verify that Git no longer sees it as 'modified', similar to step 1.
done!
Staging applies regardless of whether a file is currently tracked.
- Staging an untracked file will both begin tracking the file and include it in the next snapshot.
- Staging an already tracked file will simply mark its current changes for inclusion in the next commit.
Git also supports fine-grained selective staging i.e., staging only specific changes within a file while leaving other changes to the same file unstaged. This will be covered in a later lesson.
Git does not track empty folders. It tracks only folders that contain tracked files.
You can test this by adding an empty subfolder inside the things folder (e.g., things/more-things) and checking if it shows up as 'untracked' (it will not). If you add a file to that folder (e.g., things/more-things/food.txt) and then staged that file (e.g., git add more-things/food.txt), the folder will now be included in the next snapshot.
PRO-TIP: Applying a Git command to multiple files in one go
When a Git command expects a list of files or paths as a parameter (as the git add command does), these parameters are known as pathspecs — patterns that tell Git which files or directories to operate on. Pathspecs can be simple file names, directory names, or more complex patterns.
Here are some common ways to write them, shown with examples using the git add <pathspec> command:
Specify multiple files, separated by spaces:
git add f1.txt f2.txt data/lists/f3.txt # stages the specified three filesUse a glob pattern:
git add '*.txt' # stages all .txt files in the current directoryWhen using glob patterns in Git commands, putting them inside quotes (
'*.txt'instead of*.txt) is recommended, to avoid your shell expanding the pattern before Git sees it.Use
.to indicate 'all in the current directory and subdirectories':git add . # stages all files in current directory and its subdirectoriesSpecific directory, to indicate 'this directory and its subdirectories':
git add path/to/dir # stages all files in path/to/dir and its subdirectoriesNegated pathspecs, to indicate 'except these':
git add . ':!*.log' # stage everything except .log files
Git supports combining these features — for example, you could add all .txt files except those in a certain folder using:
git add '*.txt' ':!docs/*.txt'
DETOUR: Unstaging Changes
You can unstage a staged file, which simply removes it from the staging area but keeps the changes in your working directory. This is useful if you later realise that you don’t actually want to include a staged file in the next commit — perhaps you staged it by mistake, or you want to include that change in a later commit.
To unstage a file you added or modified, run
git restore --staged <pathspec>. This command removes the file from the staging area, leaving your working directory untouched.git restore --staged plan.txt budget.txt data/list.txtIf your repo does not have any commits yet,
git restore --stagedwill fail with the error fatal: could not resolve HEAD.
The remedy is to usegit reset <pathspec>instead.git reset plan.txtIn fact,
git resetis an alternative way of unstaging files, and it works regardless of whether you have any commits.Wait. Then why does
git restore --stagedexists at all, given it is more verbose and doesn't even work in some special cases?
Answer: It is still considered the "modern" way of unstaging files (it was introduced more recently), because it is more intuitive and purpose-specific -- whereas thegit resetserves multiple purposes and, if used incorrectly, can cause unintended consequences.The
restorecommand can accept multiple files/paths as input, which means you can use the notation for specifying multiple files. For example, to unstage all changes you've staged, you can use thegit restore --staged .To unstage a file deletion (staged using
git rm), use the same command as above. It will unstage the deletion and restore the file in the staging area.
If you also deleted the file from your working directory, you may need to recover it separately withgit restore <file-name(s)>git restore data/list.txt data/plan.txtTo 'nuke' all changes (i.e., get rid of all staged and unstaged changes to tracked files), you can add the
--worktreeflag to thegit restore --staged <pathspec>command.git restore --staged --worktree . # nuke all changes in current folder and subfolders
To unstage a file, locate the file among the staged files section, click the ... in front the file, and choose Unstage file:
After staging, you can now proceed to save the snapshot, aka creating a commit.
Saving a snapshot is called committing and a saved snapshot is called a commit.
A Git commit is a full snapshot of your working directory based on the files you have staged, more precisely, a record of the exact state of all files in the staging area (aka index) at that moment -- even the files that have not changed since the last commit. This is in contrast to the intuitive expectation that only the is stored in a commit. Consequently, a Git commit has all the information it needs to recreate the snapshot of the working directory at the time the commit was created.
A commit also includes metadata such as the author, date, and an optional commit message describing the change.
C1 ← a commit
e.g., author, date, commit message
staging area
[empty, if all staged changes were committed]
├─ fruits.txt (tracked)
└─ colours.txt (tracked)
A Git commit is a snapshot of all tracked files, not simply a delta of what changed since the last commit.
Target To create a commit based on staged changes.
Preparation
Option 2: Continue with the sandbox from the previous hands-on practical
1 First, let us do a sanity check using the git status command, to confirm there are staged files.
git status
On branch main
No commits yet
Changes to be committed:
(use "git rm --cached <file>..." to unstage)
new file: fruits.txt
2 Now, create a commit using the commit command. The -m switch is used to specify the commit message.
git commit -m "Add fruits.txt"
[main (root-commit) d5f91de] Add fruits.txt
1 file changed, 4 insertions(+)
create mode 100644 fruits.txt
3 Verify the staging area is empty using the git status command again.
git status
On branch main
nothing to commit, working tree clean
Note how the output says nothing to commit which means the staging area is now empty.
1 Ensure the new file fruits.txt has been staged.
2 Click the Commit button at the top of the window.
3 Enter a commit message (e.g. add fruits.txt) into the text box.
4 Click Commit.
done!
It is useful to be able to visualise the commits timeline, aka the revision graph.
Git commits form a timeline, as each corresponds to a point in time when you asked Git to take a snapshot of your working directory. Each commit links to at least one previous commit, forming a structure that we can traverse.
A timeline of commits is called a branch. By default, Git names the initial branch master -- though many now use main instead. You'll learn more about branches in future lessons. For now, just be aware that the commits you create in a new repo will be on a branch called main (or master) by default.
gitGraph
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main (or master)'}} }%%
commit id: "Add fruits.txt"
commit id: "Update fruits.txt"
commit id: "Add colours.txt"
commit id: "..."
Git can show you the list of commits in the Git history.
Preparation
Option 2: Continue with the sandbox from the previous hands-on practical
1 View the list of commits, which should show just the one commit you created just now.
Navigate into the repo folder and run the git log command to see the commit history.
cd hp-list-commands/things
git log
commit ... (HEAD -> main)
Author: ... <...@...>
Date: ...
Add fruits.txt
Use the Q key to exit the output screen of the git log command.
Note how the output has some details about the commit you just created. You can ignore most of it for now, but notice it also shows the commit message you provided.
Expand the BRANCHES menu and click on the main to view the history graph, which contains only one node at the moment, representing the commit you just added. For now, ignore the label main attached to the commit.
2 Create a few more commits (i.e., a few rounds of add/edit files → stage → commit), and observe how the list of commits grows.
Here is an example list of bash commands to add two commits while observing the list of commits after each commit.
echo "figs" >> fruits.txt # add another line to fruits.txt
git add fruits.txt # stage the updated file
git commit -m "Insert figs into fruits.txt" # commit the changes
git log # check commits list
echo "a file for colours" >> colours.txt # add a colours.txt file
echo "a file for shapes" >> shapes.txt # add a shapes.txt file
git add colours.txt shapes.txt # stage both files in one go
git commit -m "Add colours.txt, shapes.txt" # commit the changes
git log # check commits list
You can copy-paste a list of commands (such as commands given above), including any comments, to the terminal. After that, hit enter to run them in sequence.
zsh users: If the terminal errors out due to comments starting with #, you may need to enable interactive comments feature in your terminal (e.g., add the line setopt INTERACTIVE_COMMENTS to your ~/.zshrc).
The output of the final git log should be something like this:
commit ... (HEAD -> main)
Author: ... <...@...>
Date: ...
Add colours.txt, shapes.txt
commit ...
Author: ... <...@...>
Date: ...
Insert figs into fruits.txt
commit ...
Author: ... <...@...>
Date: ...
Add fruits.txt
SIDEBAR: Working with the 'less' pager
Some Git commands — such as git log— may show their output through a pager. A pager is a program that lets you view long text one screen at a time, so you don’t miss anything that scrolls off the top. For example, the output of git log command will temporarily hide the current content of the terminal, and enter the pager view that shows the output one screen at a time. When you exit the pager, the git log output will disappear from view, and the previous content of the terminal will reappear.
command 1
output 1
git log
→
commit f761ea63738a...
Author: ... <...@...>
Date: Sat ...
Add colours.txt
By default, Git uses a pager called less. Given below are some useful commands you can use inside the less pager.
| Command | Description |
|---|---|
q | Quit less and return to the terminal |
↓ or j | Move down one line |
↑ or k | Move up one line |
Space | Move down one screen |
b | Move up one screen |
G | Go to the end of the content |
g | Go to the beginning of the content |
/pattern | Search forward for pattern (e.g., /fix) |
n | Repeat the last search (forward) |
N | Repeat the last search (backward) |
h | Show help screen with all less commands |
If you’d rather see the output directly, without using a pager, you can add the --no-pager flag to the command e.g.,
git --no-pager log
It is possible to ask Git to not use less at all, use a different pager, or fine-tune how less is used. For example, you can reduce Git's use of the pager (recommended), using the following command:
git config --global core.pager "less -FRX"
Explanation: -FRX is a shorthand for combining the following three flags.
-F: Quits if the output fits on one screen (don’t show pager unnecessarily)-R: Shows raw control characters (for coloured Git output)-X: Keeps content visible after quitting the pager (so output stays on the terminal)
To see the list of commits, click on the History item (listed under the WORKSPACE section) on the menu on the right edge of Sourcetree.
After adding two more commits, the list of commits should look something like this:
done!
The Git data model consists of two types of entities: objects and refs (short for references). In this lesson, you will encounter examples of both.
A Git revision graph is a visualisation of a repo's revision history, consisting of one or more branches. First, let us learn to work with simpler revision graphs consisting of one branch, such as the one given below.
- Nodes in the revision graph represent commits. A commit is one of four main types of Git objects. For completeness, the other three are:
- blob (short for binary large object): stores the contents of a file
- tree: represents a directory and records the hierarchy of its contents by referencing blobs and other trees
- tag (specifically, annotated tag): a label-like object that can store additional metadata and point to a specific commit
- A commit is identified by its SHA value. A SHA (Secure Hash Algorithm) value is a unique identifier generated by Git to represent each commit. It is produced by using SHA-1 (i.e., one of the algorithms in the SHA family of cryptographic hash functions) on the entire content of the commit. It's a 40-character hexadecimal string (e.g.,
f761ea63738a67258628e9e54095b88ea67d95e2) that acts like a fingerprint, ensuring that every commit can be referenced unambiguously. That is, every commit has a unique SHA-1 hash value. - A commit is a full snapshot of the working directory, constructed based on the previous commit, and the changes staged. That means each commit (except the initial commit) is based on another 'parent' commit. Some commits can have multiple parent commits -- we’ll cover that later.
Given every commit has a unique hash, the commit hash values you see in our examples will be different from the hash values of your own commits, for example, when following our hands-on practicals.
Edges in the revision graph represent links between a commit and its parent commit(s). In some revision graph visualisations, you might see arrows (instead of lines) showing how each commit points to its parent commit.
Git uses refs to name and keep track of various points in a repository’s history. These refs are essentially 'named-pointers' that can serve as bookmarks to reach a certain point in the revision graph using the ref name.
In the revision graph above, there are two refs main and ←HEAD.
- main is a branch ref. A branch ref points to the latest commit on a branch. In this visualisation, the commit shown alongside the ref is the one it points to i.e.,
C3.
When you create a new commit, the branch ref of the branch moves to the new commit.
You'll be learning more about Git branches in a later lesson. - ←HEAD is a special ref that typically points to the current branch and moves along with that branch ref. In this example, it is pointing to the
mainbranch.
In certain cases, theHEADmay point directly to a specific commit instead of a branch. This situation is called a 'detachedHEAD', which will be covered in a later lesson.
Target Use Git features to examine the revision graph of a simple repo.
Preparation
Option 2: Continue with the sandbox from the previous hands-on practical
1 First, use a simple git log to view the list of commits.
git log
commit f761ea63738a... (HEAD -> main)
Author: ... <...@...>
Date: Sat ...
Add colours.txt, shapes.txt
commit 2bedace69990...
Author: ... <...@...>
Date: Sat ...
Insert figs into fruits.txt
commit d5f91de5f0b5...
Author: ... <...@...>
Date: Fri ...
Add fruits.txt
Given below is the visual representation of the same revision graph. As you can see, the log output shows the refs slightly differently, but it is not hard to see what they mean.
2 Use the --oneline flag to get a more concise view. Note how the commit SHA has been truncated to first seven characters (first seven characters of a commit SHA is enough for Git to identify a commit).
git log --oneline
f761ea6 (HEAD -> main) Add colours.txt, shapes.txt
2bedace Insert figs into fruits.txt
d5f91de Add fruits.txt
3 The --graph flag makes the result closer to a graphical revision graph. Note the * that indicates a node in a revision graph.
git log --oneline --graph
* f761ea6 (HEAD -> main) Add colours.txt, shapes.txt
* 2bedace Insert figs into fruits.txt
* d5f91de Add fruits.txt
The --graph option is more useful when examining a more complicated revision graph consisting of multiple parallel branches (branches will be covered in a later lesson).
Click the History to see the revision graph.
- In some versions of Sourcetree, the
HEADref may not be shown -- it is implied that theHEADref is pointing to the same commit the currently active branch ref is pointing.
done!
At this point: You should now be able to initialise a Git repository in a folder and commit snapshots of its files at times of your choice. So far, you did not learn how to actually make use of those snapshots (other than to show a list of them) -- we will do that in later tours.
What's next: Tour 2: Backing up a Repo on the Cloud
Target Usage: To back up a Git repository on a cloud-based Git service such as GitHub.
Motivation: One (of several) benefits of maintaining a copy of a repo on a cloud server: it acts as a safety net (e.g., against the folder becoming inaccessible due to a hardware fault).
Lesson plan:
T2L1. Remote Repositories covers that part.
T2L2. Preparing to use GitHub covers that part.
T2L3. Creating a Repo on GitHub covers that part.
T2L4. Linking a Local Repo With a Remote Repo covers that part.
T2L5. Updating the Remote Repo covers that part.
T2L6. Omitting Files from Revision Control covers that part.
To use GitHub, you need to sign up for an account, and configure related tools/settings first.
GitHub is a web-based service that hosts Git repositories and adds collaboration features on top of Git. Two other similar platforms are GitLab and Bitbucket. While you use Git to manage version-control locally, you can use such a platform in conjunction to utilise additional features such as shared access to repositories, issue tracking, code reviews, and permission controls. They are widely used in software development projects, for both open-source software (OSS) and closed-source software projects.
On GitHub, a Git repo can be put in one of two spaces:
- A GitHub user account represents an individual user. It is created when you sign up for GitHub and includes a username, profile page, and personal settings. With a user account, you can create your own repositories, contribute to others’ projects, and manage collaboration settings for any repositories you own.
- A GitHub organisation (org for short) is a shared account used by a group such as a team, company, or open-source project. Organisations can own repositories and manage access to them through teams, roles, and permissions. Organisations are especially useful when managing repositories with shared ownership or when working at scale.
Every GitHub user must have a user account, even if they primarily work within an organisation.
Before you can interact with GitHub from your local Git client, you need to set up authentication. In the past, you could simply enter your GitHub username and password, but GitHub no longer accepts passwords for Git operations. Instead, you’ll use a more secure method — such as a Personal Access Token (PAT) or SSH keys — to prove your identity.
A Personal Access Token (PAT) is essentially a long, random string that acts like a password, but it can be scoped to specific permissions (e.g., read-only or full access) and revoked at any time. This makes it more secure and flexible than a traditional password.
Git supports two main protocols for communicating with GitHub: HTTPS and SSH .
- With HTTPS, you connect over the web and authenticate using your GitHub username and a Personal Access Token.
- With SSH, you connect using a cryptographic key pair you generate on your machine. Once you add your public key to your GitHub account, GitHub recognises your machine and lets you authenticate without typing anything further.
GitHub associates a commit to a user based on the email address in the commit metadata. When you push a commit, GitHub checks if the email matches a verified email on a GitHub account. If it does, the commit is shown as authored by that user. If the email doesn’t match any account, the commit is still accepted but won’t be linked to any profile.
GitHub provides a no-reply email (e.g., 12345678+username@users.noreply.github.com) that you can use as your Git user.email to hide your real email while still associating commits with your GitHub account.
If you prefer not to include your real email address in commits, you can do the following:
Find your no-reply email provided by GitHub: Navigate to the email settings of your GitHub account and select the option to
Keep my email address private. The no-reply address will then be displayed, typically in the formatID+USERNAME@users.noreply.github.com.
Update your
user.emailwith that email address e.g.,git config --global user.email "12345678+username@users.noreply.github.com"
GitHub offers its own clients to make working with GitHub more convenient.
- The GitHub Desktop app provides a GUI for performing most GitHub operations from your desktop, without needing to visit the GitHub web UI.
Git-Mastery lessons do not use the GitHub Desktop -- so, there is no need for you to install it. - The GitHub CLI (
gh) brings GitHub-specific commands to your terminal, letting you perform most GitHub operations through your terminal.
If you are using Git-Mastery exercises (strongly recommended), you need to install and configure GitHub CLI because it is needed by Git-Mastery exercises involving GitHub.
The first step of backing up a local repo on GitHub: create an empty repository on GitHub.
You can create a remote repository based on an existing local repository, to serve as a remote copy of your local repo. For example, suppose you created a local repo and worked with it for a while, but now you want to upload it onto GitHub. The first step is to create an empty repository on GitHub.
Target Create an empty repo named gitmastery-things in your GitHub account.
1 Login to your GitHub account and choose to create a new repo.
2 In the next screen, provide gitmastery-things as the name for your repo. Refer to the screenshot below on some guidance on how to provide the required information.
Click Create repository button to create the new repository.
If you enable any of the three Add _____ options shown above, GitHub will not only create a repo, but will also initialise it with some initial content. That is not what we want here. To create an empty remote repo, keep those options disabled.
3 Note the URL of the repo. It will be of the form
https://github.com/{your_user_name}/{repo_name}.git.
e.g., https://github.com/johndoe/gitmastery-things.git (note the .git at the end)
done!
The second step of backing up a local repo on GitHub: link the local repo with the remote repo on GitHub.
A Git remote is a reference to a repository hosted elsewhere, usually on a server like GitHub, GitLab, or Bitbucket. It allows your local Git repo to communicate with another remote copy — for example, to upload locally-created commits that are missing in the remote copy.
By adding a remote, you are informing the local repo details of a remote repo it can communicate with, for example, where the repo exists and what name to use to refer to the remote.
The URL you use to connect to a remote repo depends on the protocol — HTTPS or SSH:
- HTTPS URLs use the standard web protocol and start with
https://github.com/(for GitHub users). e.g.,https://github.com/username/repo-name.git - SSH URLs use the secure shell protocol and start with
git@github.com:. e.g.,git@github.com:username/repo-name.git
A Git repo can have multiple remotes. You simply need to specify different names for each remote (e.g., upstream, central, production, other-backup ...).
Target Add the empty remote repo you created on GitHub as a remote of a local repo you have.
Preparation
Option 2: Continue with the things local repo from the previous hands-on practical
1 In a terminal, navigate to the folder containing the local repo things.
2 List the current list of remotes using the git remote -v command (-v stands for 'verbose'), for a sanity check. No output is expected if there are no remotes yet.
3 Add a new remote repo using the git remote add <remote-name> <remote-url> command.
Format of the <remote-url>:
https://github.com/<owner>/<remote-repo>.git # using HTTPS
git@github.com:<owner>/<remote-repo>.git # using SSH
The full command:
git remote add origin https://github.com/JohnDoe/gitmastery-things.git # using HTTPS
git remote add origin git@github.com:JohnDoe/gitmastery-things.git # using SSH
To find the URL of a repo on GitHub, you can click on the Code button:
4 List the remotes again to verify the new remote was added.
git remote -v
origin https://github.com/johndoe/gitmastery-things.git (fetch)
origin https://github.com/johndoe/gitmastery-things.git (push)
The same remote will be listed twice, to indicate that the remote supports two operations (fetch and push). You can ignore that for now. The important thing is the remote you added is being listed.
1 Open the local repo in Sourcetree.
2 Open the dialog for adding a remote, as follows:
Choose Repository → Repository Settings menu option.
Choose Repository → Repository Settings... → Choose Remotes tab.
3 Add a new remote to the repo with the following values.
Remote name: the name you want to assign to the remote repo i.e.,originURL/path: the URL of your remote repohttps://github.com/<owner>/<repo>.git # using HTTPS git@github.com:<owner>/<repo>.git # using SSHhttps://github.com/JohnDoe/things.git # using HTTPS git@github.com:JohnDoe/things.git # using SSH
To find the URL of a repo on GitHub, you can click on the Code button:
Username: your GitHub username
4 Verify the remote was added by going to Repository → Repository Settings again.
5 Add another remote, to verify that a repo can have multiple remotes. You can use any name (e.g., backup and any made-up <owner>/<remote-repo> for this).
done!
The third step of backing up a local repo on GitHub: push a copy of the local repo to the remote repo.
You can push content of one repository to another, usually from your local repo to a remote repo. Pushing transfers recorded Git history (such as past commits), but it does not transfer unstaged changes or untracked files.
- To push, you need to have to the remote repo.
- Pushing is performed one branch at a time; you must specify which branch you want to push.
You can configure Git to track a pairing between a local branch and a remote branch, so in future you can push from the same local branch to the corresponding remote branch without needing to specify them again. For example, you can set your local main branch to track the main branch on the remote repo origin i.e., local main branch will track the branch origin/main.
In the revision graph above, you see a new type of ref ( origin/main). This is a remote-tracking branch ref that represents the state of a corresponding branch in a remote repository (if you previously set up the branch to 'track' a remote branch). In this example, the main branch in the remote origin is also at the commit C3 (which means you have not created new commits after you pushed to the remote).
If you now create a new commit C4, the state of the revision graph will be as follows:
Explanation: When you create C4, the current branch main moves to C4, and HEAD moves along with it. However, the main branch in the remote origin remains at C3 (because you have not pushed C4 yet). That is, the remote-tracking branch origin/main is one commit behind the local branch main (or, the local branch is one commit ahead). The origin/main ref will move to C4 only after you push your local branch to the remote again.
Target Upload (i.e., push) the revision history from a local repo to a remote repo.
Preparation
Option 2: Continue with the things local repo and the gitmastery-things remote repo from the previous hands-on practical
1 Push the main branch to the remote. Also instruct Git to track this branch pair.
Navigate inside the things folder.
Use the git push -u <remote-repo-name> <local-branch-name> to push the commits to a remote repository.
git push -u origin main
Explanation:
push: the Git sub-command that pushes the current local repo content to a remote repoorigin: name of the remotemain: branch to push-u(or--set-upstream): the flag that tells Git to track that this localmainis trackingorigin/mainbranch
Click the Push button on the buttons ribbon at the top.
In the next dialog, ensure the settings are as follows, ensure the Track option is selected, and click the Push button on the dialog.
2 Observe the remote-tracking branch origin/main is now pointing at the same commit as the main branch.
Use the git log --oneline --graph to see the revision graph.
* f761ea6 (HEAD -> main, origin/main) Add colours.txt, shapes.txt
* 2bedace Insert figs into fruits.txt
* d5f91de Add fruits.txt
Click the History to see the revision graph.
- In some versions of Sourcetree, the
HEADref may not be shown -- it is implied that theHEADref is pointing to the same commit the currently active branch ref is pointing. - If the remote-tracking branch ref (e.g.,
origin/main) is not showing up, you may need to enable theShow Remote Branchesoption.
done!
The push command can be used repeatedly to send further updates to another repo e.g., to update the remote with commits you created since you pushed the first time.
Target Add a commit to the same local repo, and push it to the remote repo.
Preparation
Option 2: Continue with the things local repo and the gitmastery-things remote repo from the previous hands-on practical
1 Commit some changes in your local repo. Example:
echo "Elderberries" >> fruits.txt
git commit -am "Update fruits list"
-am is a shorthand for -a -m. The -a option stages any changes to tracked files, and -m is for specifying the commit message. See here for a longer explanation.
Optionally, you can run the git status command, which should confirm that your local branch is 'ahead' by one commit (i.e., the local branch has commits that are not present in the corresponding branch in the remote repo).
git status
On branch main
Your branch is ahead of 'origin/main' by 1 commit.
(use "git push" to publish your local commits)
nothing to commit, working tree clean
You can also use the git log --oneline --graph command to see where the branch refs are. Note how the remote-tracking branch origin/main is one commit behind the local main.
e60deae (HEAD -> main) Update fruits list
f761ea6 (origin/main) Add colours.txt, shapes.txt
2bedace Insert figs into fruits.txt
d5f91de Add fruits.txt
Create commits as you did before.
Before pushing the new commit, Sourcetree will indicate that your local branch is 'ahead' by one commit (i.e., the local branch has one new commit that is not in the corresponding branch in the remote repo).
2 Push the new commits to your fork on GitHub.
To push the newer commit(s) in the current branch main to the remote origin, you can use any of the following commands:
git push origin maingit push origin
→ Git will assume you are pushing the current branch (e.g.,main) even if you don't specify it.git push
→ Git will assume you are pushing the current branch (e.g.,main). Due to tracking you've set up earlier, Git will assume that you want to push it to the matching branch onorigin.
After pushing, the revision graph should look something like the following (note how both local and remote-tracking branch refs are pointing to the same commit again).
e60deae (HEAD -> main, origin/main) Update fruits list
f761ea6 Add colours.txt, shapes.txt
2bedace Insert figs into fruits.txt
d5f91de Add fruits.txt
To push, click the Push button on the top buttons ribbon, ensure the settings are as follows in the next dialog, and click the Push button on the dialog.
After pushing the new commit to the remote, the remote-tracking branch ref should move to the new commit:
done!
Note that you can push between two repos only if those repos have a shared history among them (i.e., one should have been created by copying the other).
DETOUR: Pushing to Multiple Repos
You can push to any number of repos, as long as the target repos and your repo have a shared history.
- Add the GitHub repo URL as a remote while giving a suitable name (e.g.,
upstream,central,production,backup...), if you haven't done so already. - Push to the target repo -- remember to select the correct target repo when you do.
e.g., git push backup main
Git allows you to specify which files should be omitted from revision control.
You can specify which files Git should ignore from revision control. While you can always omit files from revision control simply by not staging them, having an 'ignore-list' is more convenient, especially if there are files inside the working folder that are not suitable for revision control (e.g., temporary log files) or files you want to prevent from accidentally including in a commit (files containing confidential information).
A repo-specific ignore-list of files can be specified in a .gitignore file, stored in the root of the repo folder.
The .gitignore file itself can be either revision controlled or ignored.
- To version control it (the more common choice – which allows you to track how the
.gitignorefile changes over time), simply commit it as you would commit any other file. - To ignore it, simply add its name to the
.gitignorefile itself.
The .gitignore file supports file patterns e.g., adding temp/*.tmp to the .gitignore file prevents Git from tracking any .tmp files in the temp directory.
SIDEBAR: .gitignore File Syntax
Blank lines: Ignored and can be used for spacing.
Comments: Begin with
#(lines starting with # are ignored).# This is a commentWrite the name or pattern of files/directories to ignore.
log.txt # Ignores a file named log.txtWildcards:
*matches any number of characters, except/(i.e., for matching a string within a single directory level):abc/*.tmp # Ignores all .tmp files in abc directory**matches any number of characters (including/)**/foo.tmp # Ignores all foo.tmp files in any directory?matches a single characterconfig?.yml # Ignores config1.yml, configA.yml, etc.[abc]matches a single character (a, b, or c)file[123].txt # Ignores file1.txt, file2.txt, file3.txt
Directories:
- Add a trailing
/to match directories.logs/ # Ignores the logs directory - Patterns without
/match files/folders recursively.*.bak # Ignores all .bak files anywhere - Patterns with
/are relative to the.gitignorelocation./secret.txt # Only ignores secret.txt in the root directory
- Add a trailing
Negation: Use
!at the start of a line to not ignore something.*.log # Ignores all .log files !important.log # Except important.log
Example:
# Ignore all log files
*.log
# Ignore node_modules folder
node_modules/
# Don’t ignore main.log
!main.log
.gitignore is a 'hidden' file!
Files with a name starting with . (such as .gitignore) are considered as hidden files by MacOS and Linux. It is very likely the Git tools used in Windows also marked the .gitignore file as a hidden file. Therefore, if the .gitignore files is not visible to you, you'll need to look for it among 'hidden' files.
How to do that in: Windows | MacOS | Linux
Target Get Git to ignore some files in a repo.
Preparation
1 Configure Git to ignore those files:
Create a file named .gitignore in the working directory root and add the text temp.txt into it.
echo "temp.txt" >> .gitignore
temp.txt
Observe how temp.txt is no longer detected as 'untracked' by running the git status command (but now it will detect the .gitignore file as 'untracked'.
Update the .gitignore file as follows:
temp.txt
*.tmp
Observe how .tmp files are no longer detected as 'untracked' by running the git status command.
The file should be currently listed under Unstaged files. Right-click it and choose Ignore.... Choose Ignore exact filename(s) and click OK.
Also take note of other options available e.g., Ignore all files with this extension etc. They may be useful in future.
Note how the temp.text is no longer listed under Unstaged files. Observe that a file named .gitignore has been created in the working directory root and has the following line in it. This new file is now listed under Unstaged files.
temp.txt
Right-click on any of the .tmp files you added, and choose Ignore... as you did previously. This time, choose the option Ignore files with this extension.
Note how .temp files are no longer shown as unstaged files, and the .gitignore file has been updated as given below:
temp.txt
*.tmp
2 Optionally, stage and commit the .gitignore file.
done!
Files recommended to be omitted from version control
- Binary files generated when building your project e.g.,
*.class,*.jar,*.exe
Reasons:- no need to version control these files as they can be generated again from the source code
- Revision control systems are optimized for tracking text-based files, not binary files.
- Temporary files e.g., log files generated while testing the product
- Local files i.e., files specific to your own computer e.g., local settings of your IDE (
.idea/) - Sensitive content i.e., files containing sensitive/personal information e.g., credential files, personal identification data (especially if there is a possibility of those files getting leaked via the revision control system).
At this point: You should now be able to create a copy of your repo on GitHub, and keep it updated as you add more commits to your local repo. If something goes wrong with your local repo (e.g., disk crash), you can now recover the repo using the remote repo (this tour did not cover how exactly you can do that -- it will be covered in a future tour).
What's next: Tour 3: Working Off a Remote Repo
Target Usage: To work with an existing remote repository.
Motivation: Often, you will need to start with an existing remote repository. In such cases, you may have to create your own copies of that repository, and keep those copies updated when more changes appear in the remote repository.
Lesson plan:
T3L1. Duplicating a Remote Repo on the Cloud covers that part.
T3L2. Creating a Local Copy of a Repo covers that part.
T3L3. Downloading Data Into a Local Repo covers that part.
GitHub allows you to create a remote copy of another remote repo, called forking.
A fork is a copy of a remote repository created on the same hosting service such as GitHub, GitLab, or Bitbucket. On GitHub, you can fork a repository from another user or organisation into your own space (i.e., your user account or an organisation you have sufficient access to). Forking is particularly useful if you want to experiment with a repo but don't have write permissions to the original -- you can fork it (which gives you your own remote copy of the repo) and work with the forked repo without affecting the original repository.
Preparation Create a GitHub account if you don't have one yet.
1 Go to the GitHub repo you want to fork e.g., samplerepo-things
2 Click on the 
button in the top-right corner. In the next step,
- choose to fork to your own account or to another GitHub organization that you are an admin of.
- un-tick the
[ ] Copy the main branch onlyoption, so that you get copies of other branches (if any) in the repo.
done!
Forking is not a Git feature, but a feature provided by hosted Git services like GitHub, GitLab, or Bitbucket.
GitHub does not allow you to fork the same repo more than once to the same destination. If you want to re-fork, you need to delete the previous fork.
The next step is to create a local copy of the remote repo, by cloning the remote repo.
You can clone a repository to create a full copy of it on your computer. This copy includes the entire revision history, branches, and files of the original, so it behaves just like the original repository. For example, you can clone a repository from a hosting service like GitHub to your computer, giving you a complete local version to work with.
Cloning a repo automatically creates a remote named origin which points to the repo you cloned from.
The repo you cloned from is often referred to as the upstream repo.
1 Clone the remote repo to your computer. For example, you can clone the samplerepo-things repo, or the fork you created from it in a previous lesson.
Note that the URL of the GitHub project is different from the URL you need to clone a repo in that GitHub project. e.g.
https://github.com/git-mastery/samplerepo-things # GitHub project URL
https://github.com/git-mastery/samplerepo-things.git # the repo URL
You can use the git clone <repository-url> [directory-name] command to clone a repo.
<repository-url>: The URL of the remote repository you want to copy.[directory-name](optional): The name of the folder where you want the repository to be cloned. If you omit this, Git will create a folder with the same name as the repository.
git clone https://github.com/git-mastery/samplerepo-things.git # if using HTTPS
git clone git@github.com:se-edu/samplerepo-things.git # if using SSH
git clone https://github.com/foo/bar.git my-bar-copy # also specifies a dir to use
For exact steps for cloning a repo from GitHub, refer to this GitHub document.
File → Clone / New ... and provide the URL of the repo and the destination directory.
File → New ... → Choose as shown below → Provide the URL of the repo and the destination directory in the next dialog.
2 Verify the clone has a remote named origin pointing to the upstream repo.
Use the git remote -v command that you learned earlier.
Choose Repository → Repository Settings menu option.
done!
When there are new changes in the remote, you need to pull those changes down to your local repo.
There are two steps to bringing over changes from a remote repository into a local repository: fetch and merge.
- Fetch is the act of downloading the latest changes from the remote repository, but without applying them to your current branch yet. It updates metadata in your repo so that it knows what has changed in the remote repo, but your own local branch remains untouched.
- Merge is what you do after fetching, to actually incorporate the fetched changes into your local branch. It combines your local branch with the changes from the corresponding branch from the remote repo.
Scenario You have cloned a remote repo. After you have cloned, two new commits have been added to it. R and L1 in the diagram below represents this scenario.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "add loans.txt"
commit id: "add loan to Ben"
commit id: "add assets.txt"
commit id: "add goals.txt"
commit id: "[HEAD → main] add loan to Chang"
origin]
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "add loans.txt"
commit id: "add loan to Ben"
commit id: "[HEAD → main][origin/main] add assets.txt"
2 commits behind the remote]
→
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "add loans.txt"
commit id: "add loan to Ben"
commit id: "add assets.txt"
commit id: "add goals.txt"
commit id: "[HEAD → main][origin/main] add loan to Chang"
the missing commits]
Target Now, you wish to bring over those missing commits to your clone, taking it from the state L1 to state L2 (as given in the diagram above).
Preparation
1 Verify the local repo is unaware of the extra commits in the remote.
git status
On branch main
Your branch is up to date with 'origin/main'.
nothing to commit, working tree clean
The revision graph should look like the below:
If it looks like the below, it is possible that Sourcetree is auto-fetching data from the repo periodically.
2 Fetch from the new remote.
Use the git fetch <remote> command to fetch changes from a remote. If the <remote> is not specified, the default remote origin will be used.
git fetch origin
remote: Enumerating objects: 8, done.
... # more output ...
afbe966..b201f03 main -> origin/main
Click on the Fetch button on the top menu:
3 Verify the fetch worked i.e., the local repo is now aware of the two missing commits. Also observe how the local branch ref of the main branch, the staging area, and the working directory remain unchanged after the fetch.
Use the git status command to confirm the repo now knows that it is behind the remote repo.
git status
On branch main
Your branch is behind 'origin/main' by 2 commits, and can be fast-forwarded.
(use "git pull" to update your local branch)
nothing to commit, working tree clean
Now, the revision graph should look something like the below. Note how the origin/main ref is now two commits ahead of the main ref.
4 Merge the fetched changes.
Use the git merge <remote-tracking-branch> command to merge the fetched changes. Check the status and the revision graph to verify that the branch tip has now moved by two more commits.
git merge origin/main
Updating afbe966..b201f03
Fast-forward
goals.txt | 1 +
loans.txt | 1 +
2 files changed, 2 insertions(+)
create mode 100644 goals.txt
Verify the status of the repo is as expected:
git status
On branch main
Your branch is up to date with 'origin/main'.
git log --oneline --decorate
b201f03 (HEAD -> main, origin/main, origin/HEAD) Add loan to Chang
1b923a4 Add goals.txt
afbe966 Add assets.txt
0434002 Add loan to Ben
fd96227 Add loans.txt
To merge the fetched changes, right-click on the latest commit on origin/remote branch and choose Merge.
In the next dialog, choose as follows:
The final result should be something like the below (same as the repo state before we started this hands-on practical):
Note that merging the fetched changes can get complicated if there are multiple branches or the commits in the local repo conflict with commits in the remote repo. We will address them when we learn more about Git branches, in a later lesson.
done!
Pull is a shortcut that combines fetch and merge — it fetches the latest changes from the remote and immediately merges them into your current branch. In practice, Git users typically use the pull instead of the fetch-then-merge.
pull = fetch + merge
Scenario Same as previous hands-on practical.
Target Same as the previous, but this time we intend to fetch and merge in one step.
Preparation
1 Pull the newer commits from the remote, instead of a fetch-then-merge.
Use the git pull <remote> <branch> command to pull changes.
git pull origin main
remote: Enumerating objects: 8, done.
remote: Counting objects: 100% (8/8), done.
remote: Compressing objects: 100% (3/3), done.
remote: Total 6 (delta 1), reused 6 (delta 1), pack-reused 0 (from 0)
Unpacking objects: 100% (6/6), 557 bytes | 69.00 KiB/s, done.
From https://github.com/git-mastery/samplerepo-finances-2
* branch main -> FETCH_HEAD
afbe966..b201f03 main -> origin/main
Updating afbe966..b201f03
Fast-forward
goals.txt | 1 +
loans.txt | 1 +
2 files changed, 2 insertions(+)
create mode 100644 goals.txt
The following works too. If the <remote> and <branch> are not specified, Git will pull to the current branch from the remote branch it is tracking.
git pull
Click on the Pull button on the top menu:
2 Verify the outcome is same as the fetch + merge steps you did in the previous hands-on practical.
done!
You can pull from any number of remote repos, provided the repos involved have a shared history. This can be useful when the upstream repo you forked from has some new commits that you wish to bring over to your copies of the repo (i.e., your fork and your local repo).
Scenario You have forked and cloned a remote repo. Since then, new commits have been added to the original remote repo.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "add loans.txt"
commit id: "add loan to Ben"
commit id: "add assets.txt"
commit id: "add goals.txt"
commit id: "[HEAD → main] add loan to Chang"
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "add loans.txt"
commit id: "add loan to Ben"
commit id: "[HEAD → main] add assets.txt"
2 commits behind the remote]
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "add loans.txt"
commit id: "add loan to Ben"
commit id: "[HEAD → main][origin/main] add assets.txt"
2 commits behind]
Target Now, you wish to bring over new commits to your clone, and also update your fork with those commits.
Preparation
1 Confirm your local repo is behind by two commits. For example, you can examine the remote-tracking branches for this.
git log --oneline --decorate --graph --all
* b201f03 (upstream/main, upstream/HEAD) Add loan to Chang
* 1b923a4 Add goals.txt
* afbe966 (HEAD -> main, origin/main, origin/HEAD) Add assets.txt
* 0434002 Add loan to Ben
* fd96227 Add loans.txt
2 Pull from the upstream repo. If there are new commits, those will come over to your local repo. For example:
git pull upstream main
3 Push to your fork. Any new commits you pulled from the upstream repo will now appear in your fork as well. For example:
git push origin main
The method given above is the more 'standard' method of synchronising a fork with the upstream repo. In addition, platforms such as GitHub can provide other ways (example: GitHub's Sync fork feature).
done!
SIDEBAR: Distributed vs Centralised Revision Control
Revision control can be done in two ways: the centralized way and the distributed way.
Centralized RCS use a single central (server-hosted) repository that is shared by the team. Developers check out a working copy, make changes locally, and then commit directly to the central repository. Instead of having their own copy of the entire repository history, they only have a working copy of files. One advantage of this model is having a clear and single "source of truth." One big disadvantage is that the central server becomes a critical dependency: if it's down, most operations (commits and history queries beyond the local working copy) are blocked. Older RCS tools such as CVS, Subversion, and Perforce follow this model.
The centralized RCS approach
Distributed RCS (also known as decentralized RCS) allow multiple remote and local repositories to work together. The workflow can vary from team to team. For example, every team member can have their own remote repository in addition to a local repository. This architecture enables offline work, fast local operations, and more flexible workflows. It also supports multiple integration points (e.g., forks or alternative remotes) and uses cryptographic checksums to ensure history integrity. The trade-offs include more conceptual complexity (multiple repositories, remotes, and sync patterns) and the need for conventions to establish an authoritative integration flow. Git and Mercurial are prominent RCS tools that support the distributed approach.
The decentralized RCS approach
Given its use of multiple copies of a repository, Git is considered a distributed revision control software, as opposed to a centralised revision control software that keep only a single repository.
At this point: Now you can create your own remote and local copies of any repo on GitHub, and update your copy when there are new changes in the upstream repo.
What's next: Tour 4: Using the Revision History of a Repo
Target Usage: To make use of the revision history stored by Git.
Motivation: Having put in effort to record the revision history of the working folder, it only makes sense that we use the revision history to our benefit. For example, to be able to answer questions such as "What did I change in this file since last Monday?"
Lesson plan:
T4L1. Examining a Commit covers that part.
T4L2. Tagging Commits covers that part.
T4L3. Comparing Points of History covers that part.
T4L4. Traversing to a Specific Commit covers that part.
T4L5. Rewriting History to Start Over covers that part.
T4L6. Reverting a Specific Commit covers that part.
It is useful to be able to see what changes were included in a specific commit.
When you examine a commit, normally what you see is the 'changes made since the previous commit'. This does not mean that a Git commit contains only the changes made since the previous commit. As you recall, a Git commit contains a full snapshot of the working directory. However, tools used to examine commits typically show only the changes, as that is the more informative part.
Git shows changes included in a commit by dynamically calculating the difference between the snapshots stored in the target commit and the parent commit. This is because Git commits store snapshots of the working directory, not changes themselves.
Although each commit represents a copy of the entire working directory, Git uses space efficiently in two main ways:
- Reuse of unchanged data: If a file hasn’t changed since a previous commit, the commit simply points to the already stored version of that file instead of making another copy. This means only new or changed files take up extra space, while unchanged files are reused.
- Compression: Git also compresses all the files and data it stores using an algorithm (zlib). So, even the objects that are stored (whether reused or new) take up less disk space because they are saved in a compressed format.
To address a specific commit, you can use its SHA (e.g., e60deaeb2964bf2ebc907b7416efc890c9d4914b). In fact, just the first few characters of the SHA is enough to uniquely address a commit (e.g., e60deae), provided the partial SHA is long enough to uniquely identify the commit (i.e., only one commit has that partial SHA).
Naturally, a commit can be addressed using any ref pointing to it too (e.g., HEAD, main).
Another related technique is to use the <ref>~<n> notation (e.g., HEAD~1) to address the commit that is n commits prior to the commit pointed by <ref> i.e., "start with the commit pointed by <ref> and go back n commits".
A related alternative notation is HEAD~, HEAD~~, HEAD~~~, ... to mean HEAD~1, HEAD~2, HEAD~3 etc.
HEAD or mainHEAD~1 or main~1 or HEAD~ or main~HEAD~2 or main~2Git uses the diff format to show file changes in a commit. The diff format was originally developed for Unix. It was later extended with headers and metadata to show changes between file versions and commits. Here is an example diff showing the changes to a file.
diff --git a/fruits.txt b/fruits.txt
index 7d0a594..f84d1c9 100644
--- a/fruits.txt
+++ b/fruits.txt
@@ -1,6 +1,6 @@
-apples
+apples, apricots
bananas
cherries
dragon fruits
-elderberries
figs
@@ -20,2 +20,3 @@
oranges
+pears
raisins
diff --git a/colours.txt b/colours.txt
new file mode 100644
index 0000000..55c8449
--- /dev/null
+++ b/colours.txt
@@ -0,0 +1 @@
+a file for colours
A Git diff can consist of multiple file diffs, one for each changed file. Each file diff can contain one or more hunk i.e., a localised group of changes within the file — including lines added, removed, or left unchanged (included for context).
Given below is how the above diff is divided into its components:
File diff for fruits.txt:
diff --git a/fruits.txt b/fruits.txt
index 7d0a594..f84d1c9 100644
--- a/fruits.txt
+++ b/fruits.txt
Hunk 1:
@@ -1,6 +1,6 @@
-apples
+apples, apricots
bananas
cherries
dragon fruits
-elderberries
figs
Hunk 2:
@@ -20,2 +20,3 @@
oranges
+pears
raisins
File diff for colours.txt:
diff --git a/colours.txt b/colours.txt
new file mode 100644
index 0000000..55c8449
--- /dev/null
+++ b/colours.txt
Hunk 1:
@@ -0,0 +1 @@
+a file for colours
Here is an explanation of the diff:
| Part of Diff | Explanation |
|---|---|
diff --git a/fruits.txt b/fruits.txt | The diff header, indicating that it is comparing the file fruits.txt between two versions: the old (a/) and new (b/). |
index 7d0a594..f84d1c9 100644 | Shows the before and after the change, and the file mode (100 means a regular file, 644 are file permission indicators). |
--- a/fruits.txt+++ b/fruits.txt | Marks the old version of the file (a/fruits.txt) and the new version of the file (b/fruits.txt). |
| This hunk header shows that lines 1-6 (i.e., starting at line 1, showing 6 lines) in the old file were compared with lines 1–6 in the new file. |
-apples+apples, apricots | Removed line apples and added line apples, apricots. |
bananascherriesdragon fruits | Unchanged lines, shown for context. |
-elderberries | Removed line: elderberries. |
figs | Unchanged line, shown for context. |
| Hunk header showing that lines 20-21 in the old file were compared with lines 20–22 in the new file. |
oranges+pearsraisins | Unchanged line. Added line: pears.Unchanged line. |
diff --git a/colours.txt b/colours.txt | The usual diff header, indicates that Git is comparing two versions of the file colours.txt: one before and one after the change. |
new file mode 100644 | This is a new file being added. 100644 means it’s a normal, non-executable file with standard read/write permissions. |
index 0000000..55c8449 | The usual SHA hashes for the two versions of the file. 0000000 indicates the file did not exist before. |
--- /dev/null+++ b/colours.txt | Refers to the "old" version of the file (/dev/null means it didn’t exist before), and the new version. |
@@ -0,0 +1 @@ | Hunk header, saying: “0 lines in the old file were replaced with 1 line in the new file, starting at line 1.” |
+a file for colours | Added line |
Points to note:
+indicates a line being added.
-indicates a line being deleted.- Editing a line is seen as deleting the original line and adding the new line.
Target View contents of specific commits in a repo.
Preparation
1 Locate the commits to view, using the revision graph.
git log --oneline --decorate
e60deae (HEAD -> main) Update fruits list
f761ea6 Add colours.txt, shapes.txt
2bedace Add elderberries and figs into fruits.txt
d5f91de Add fruits.txt
2 Use the git show command to view specific commits.
git show # shows the latest commit
commit e60deaeb2964bf2ebc907b7416efc890c9d4914b (HEAD -> main)
Author: damithc <...@...>
Date: Sat Jun ...
Update fruits list
diff --git a/fruits.txt b/fruits.txt
index 7d0a594..6d502c3 100644
--- a/fruits.txt
+++ b/fruits.txt
@@ -1,6 +1,6 @@
-apples
+apples, apricots
bananas
+blueberries
cherries
dragon fruits
-elderberries
figs
To view the parent commit of the latest commit, you can use any of these commands:
git show HEAD~1
git show main~1
git show e60deae # first few characters of the SHA
git show e60deae..... # run git log to find the full SHA and specify the full SHA
To view the commit that is two commits before the latest commit, you can use git show HEAD~2 etc.
Click on the commit. The remaining panels (indicated in the image below) will be populated with the details of the commit.
done!
PRO-TIP: Use Git Aliases to Work Faster
The Git alias feature allows you to create custom shortcuts for frequently used Git commands. This saves time and reduces typing, especially for long or complex commands. Once an alias is defined, you can use the alias just like any other Git command e.g., use git lodg as an alias for git log --oneline --decorate --graph.
To define a global git alias, you can use the git config --global alias.<alias> "<command>" command. e.g.,
git config --global alias.lodg "log --oneline --graph --decorate"
You can also create shell-level aliases using your shell configuration (e.g., .bashrc, .zshrc) to make even shorter aliases. This lets you create shortcuts for any command, including Git commands, and even combine them with other tools. e.g., instead of the Git alias git lodg, you can define a shorter shell-level alias glodg.
1. Locate your .bash_profile file (likely to be in : C:\Users\<YourName>\.bash_profile -- if it doesn’t exist, create it.)
1. Locate your shell's config file e.g., .bashrc or .zshrc (likely to be in your ~ folder)
1. Locate your shell's config file e.g., .bashrc or .zshrc (likely to be in your ~ folder)
Oh-My-Zsh for Zsh terminal supports a Git plugin that adds a wide array of Git command aliases to your terminal.
2. Add aliases to that file:
alias gs='git status'
alias glod='git log --oneline --graph --decorate'
3. Apply changes by running the command source ~/.zshrc or source ~/.bash_profile or source ~/.bashrc, depending on which file you put the aliases in.
When working with many commits, it helps to tag specific commits with custom names so they’re easier to refer to later.
Git lets you tag commits with names, making them easy to reference later. This is useful when you want to mark specific commits -- such as releases or key milestones (e.g., v1.0 or v2.1). Using tags to refer to commits is much more convenient than using SHA hashes. In the diagram below, v1.0 and interim are tags.
A tag stays fixed to the commit. Unlike branch refs or HEAD, tags do not move automatically as new commits are made. As you see below, after adding a new commit, tags stay in the previous commits while main←HEAD has moved to the new commit.
Git supports two kinds of tags:
- A lightweight tag is just a ref that points directly to a commit, like a branch that doesn’t move.
- An annotated tag is a full Git object that stores a reference to a commit along with metadata such as the tagger’s name, date, and a message.
Annotated tags are generally preferred for versioning and public releases, while lightweight tags are often used for less formal purposes, such as marking a commit for your own reference.
Target Add a few tags to a repository.
Preparation
1 Add a lightweight tag to the current commit as v1.0:
git tag v1.0
2 Verify the tag was added. To view tags:
git tag
v1.0
To view tags in the context of the revision graph:
git log --oneline --decorate
507bb74 (HEAD -> main, tag: v1.0, origin/main, origin/HEAD) Add donuts
de97f08 Add cake
5e6733a Add bananas
3398df7 Add food.txt
3 Use the tag to refer to the commit e.g., git show v1.0 should show the changes in the tagged commit.
4 Add an annotated tag to an earlier commit. The example below adds a tag v0.9 to the commit HEAD~2 with the message First beta release. The -a switch tells Git this is an annotated tag.
git tag -a v0.9 HEAD~2 -m "First beta release"
5 Check the new annotated tag. While both types of tags appear similarly in the revision graph, the show command on an annotated tag will show the details of the tag and the details of the commit it points to.
git show v0.9
tag v0.9
Tagger: ... <...@...>
Date: Sun Jun ...
First beta release
commit ....999087124af... (tag: v0.9)
Author: ... <...@...>
Date: Sat Jun ...
Add banana
diff --git a/fruits.txt b/fruits.txt
index a8a0a01..7d0a594 100644
# rest of the diff goes here
Right-click on the commit (in the graphical revision graph) you want to tag and choose Tag….
Specify the tag name e.g., v1.0 and click Add Tag.
Configure tag properties in the next dialog and press Add. For example, you can choose whether to make it a lightweight tag or an annotated tag (default).
Tags will appear as labels in the revision graph, as seen below. To see the details of an annotated tag, you need to use the menu indicated in the screenshot.
done!
If you need to change what a tag points to, you must delete the old one and create a new tag with the same name. This is because tags are designed to be fixed references to a specific commit, and there is no built-in mechanism to 'move' a tag.
Preparation
Option 2: Continue with the sandbox from the previous hands-on practical
Move the v1.0 tag to the commit HEAD~1, by deleting it first and creating it again at the destination commit.
Delete the previous v1.0 tag by using the -d . Add it again to the other commit, as before.
git tag -d v1.0
git tag v1.0 HEAD~1
The same dialog used to add a tag can be used to delete and even move a tag. Note that the 'moving' here translates to deleting and re-adding behind the scene.
done!
Tags are different from commit messages, in purpose and in form. A commit message is a description of the commit that is part of the commit itself. A tag is a short name for a commit, which you can use to address a commit.
Pushing commits to a remote does not push tags automatically. You need to push tags specifically.
Target Push tags you created earlier to the remote.
Preparation
Option 2: Continue with the sandbox from the previous hands-on practical
You can go to your remote on GitHub link https://github.com/{USER}/{REPO}/tags (e.g., https://github.com/johndoe/samplerepo-preferences/tags) to verify the tag is present there.
Note how GitHub assumes these tags are meant as releases, and automatically provides zip and tar.gz archives of the repo (as at that tag).
1 Push a specific tag in the local repo to the remote (e.g., v1.0) using the git push <remote> <tag-name> command.
git push origin v1.0
In addition to verifying the tag's presence via GitHub, you can also use the following command to list the tags presently in the remote.
git ls-remote --tags origin
2 Delete a tag in the remote, using the git push --delete <remote> <tag-name> command.
git push --delete origin v1.0
3 Push all tags to the remote repo, using the git push <remote> --tags command.
git push origin --tags
To push a specific tag, use the following menu:
To push all tags, you can tick the Push all tags option when pushing commits:
done!
Git can tell you the net effect of changes between two points of history.
Git's diff feature can show you what changed between two points in the revision history. Given below are some use cases.
Usage 1: Comparing two commits at different points of the revision graph
Example use case: Suppose you’re trying to improve the performance of a piece of software by experimenting with different code tweaks. You commit after each change (as you should). After several commits, you now want to review the overall effect of all those changes on the code.
Target Compare two commits in a repo.
Preparation
You can use the git diff <commit1> <commit2> command for this.
- You may use any valid way to refer to commits (e.g., SHA, tag, HEAD~n etc.).
- You may also use the
..notation to specify the commit range too e.g.,0023cdd..fcd6199,HEAD~2..HEAD
git diff v0.9 HEAD
diff --git a/colours.txt b/colours.txt
index 55c8449..435e81d 100644
--- a/colours.txt
+++ b/colours.txt
@@ -1 +1,4 @@
a file for colours
+blue
# rest of the diff ...
Swap the commit order in the command and see what happens.
git diff HEAD v0.9
diff --git a/colours.txt b/colours.txt
index 435e81d..55c8449 100644
--- a/colours.txt
+++ b/colours.txt
@@ -1,4 +1 @@
a file for colours
-blue
# rest of the diff ...
As you can see, the diff is directional i.e., diff <commit1> <commit2> shows what changes you need to do to go from the <commit1> to <commit2>. If you swap <commit1> and <commit2>, the output will change accordingly e.g., lines previously shown as 'added' will now be shown as 'deleted'.
Select the two commits: Click on one commit, and Ctrl-Click (or Cmd-Click) on the second commit. The changes between the two selected commits will appear in the other panels, as shown below:
The same method can be used to compare the current state of the working directory (which might have uncommitted changes) to a point in the history.
done!
Usage 2: Examining changes in the working directory
Example use case: To verify the next commit will include exactly what you intend it to include.
Preparation
Option 2: Continue with the things repo from the previous hands-on practical
1 Do some changes to the working directory. Stage some (but not all) changes. For example, you can run the following commands.
echo -e "blue\nred\ngreen" >> colours.txt
git add . # a shortcut to stage all changes
echo "no shapes added yet" >> shapes.txt
2 Examine the staged and unstaged changes.
The git diff command shows unstaged changes in the working directory (tracked files only). The output of the diff command, is a diff view (introduced in this lesson).
git diff
diff --git a/shapes.txt b/shapes.txt
index 4bc044e..1971ab8 100644
--- a/shapes.txt
+++ b/shapes.txt
@@ -3,3 +3,4 @@ circle
oval
rectangle
square
+no shapes added yet
The git diff --staged command shows the staged changes (same as git diff --cached).
git diff --staged
Select the two commits: Click on one commit, and Ctrl-Click (or Cmd-Click) on the second commit. The changes between the two selected commits will appear in the other panels, as shown below:
done!
Usage 3: Examining changes to a specific file
Example use case: Similar to other use cases but when you are interested in a specific file only.
Target Examine the changes done to a file between two different points in the version history (including the working directory).
Preparation
1 Examine changes to a specific file, between specific points in history.
Add the -- path/to/file to a previous diff command to narrow the output to a specific file. Some examples:
git diff -- andy/history.txt # unstaged changes to andy/history.txt
git diff --staged -- list.txt # staged changes to list.txt
git diff HEAD~2..HEAD -- list.txt # changes to list.txt between commits
Sourcetree UI shows changes to one file at a time by default; just click on the file to view changes to that file. To view changes to multiple files, Ctrl-Click (or Cmd-Click) on multiple files to select them.
done!
Another useful feature of revision control is to be able to view the working directory as it was at a specific point in history, by checking out a commit created at that point.
Suppose you added a new feature to a software product, and while testing it, you noticed that another feature added two commits ago doesn’t handle a certain edge case correctly. Now you’re wondering: did the new feature break the old one, or was it already broken? Can you go back to the moment you committed the old feature and test it in isolation, and come back to the present after you found the answer? With Git, you can.
To view the working directory at a specific point in history, you can check out a commit created at that point.
When you check out a commit, Git:
- Updates your working directory to match the snapshot in that commit, overwriting current files as needed.
- Moves the
HEADref to that commit, marking it as the current state you’re viewing.
→
[check out commit C2...]
Checking out a specific commit puts you in a "detached HEAD" state: i.e., the HEAD no longer points to a branch, but directly to a commit (see the above diagram for an example). This isn't a problem by itself, but any commits you make in this state can be lost, unless certain follow-up actions are taken. It is perfectly fine to be in a detached state if you are only examining the state of the working directory at that commit.
To get out of a "detached HEAD" state, you can simply check out a branch, which "re-attaches" HEAD to the branch you checked out.
→
[check out main...]
Target Checkout a few commits in a local repo, while examining the working directory to verify that it matches the state when you created the corresponding commit
Preparation
1 Examine the revision tree, to get your bearing first.
git log --oneline --decorate
Reminder: You can use aliases to reduce typing Git commands.
6304a59 (HEAD -> main) shapes.txt: Add some shapes
9f68246 (tag: 1.0) colours.txt: Add some colours
2ef8852 Update fruits list
8ca5cc6 (tag: 0.9) Add colours.txt, shapes.txt
542668f Add elderberries and figs into fruits.txt
ec49b17 Add fruits.txt
2 Use the checkout <commit-identifier> command to check out a commit other than the one currently pointed by HEAD. You can use any of the following methods:
git checkout v1.0: checks out the commit taggedv1.0git checkout 0023cdd: checks out the commit with the hash0023cddgit checkout HEAD~2: checks out the commit 2 commits behind the most recent commit.
git checkout HEAD~2
Note: switching to 'HEAD~2'.
You are in 'detached HEAD' state.
# rest of the warning about the detached head ...
HEAD is now at 2ef8852 Update fruits list
3 Verify HEAD and the working directory have updated as expected.
HEADshould now be pointing at the target commit- The working directory should match the state it was in at that commit (i.e., changes done after that commit should not be in the folder).
git log --oneline --decorate
2ef8852 (HEAD) Update fruits list
8ca5cc6 (tag: 0.9) Add colours.txt, shapes.txt
542668f Add elderberries and figs into fruits.txt
ec49b17 Add fruits.txt
HEAD is indeed pointing at the target commit.
But note how the output does not show commits you added after the checked-out commit.
The --all switch tells git log to show commits from all refs, not just those reachable from the current HEAD. This includes commits from other branches, tags, and remotes.
git log --oneline --decorate --all
6304a59 (main) shapes.txt: Add some shapes
8ca5cc6 (tag: 0.9) Add colours.txt, shapes.txt
9f68246 (tag: 1.0) colours.txt: Add some colours
2ef8852 (HEAD) Update fruits list
542668f Add elderberries and figs into fruits.txt
ec49b17 Add fruits.txt
4 Go back to the latest commit by checking out the main branch again.
git checkout main
In the revision graph, double-click the commit you want to check out, or right-click on that commit and choose Checkout....
Click OK to the warning about ‘detached HEAD’ (similar to below).
The specified commit is now loaded onto the working folder, as indicated by the HEAD label.
To go back to the latest commit on the main branch, double-click the main branch.
If you check out a commit that comes before the commit in which you added a certain file (e.g., temp.txt) to the .gitignore file, and if the .gitignore file is version controlled as well, Git will now show it under ‘unstaged modifications’ because at Git hasn’t been told to ignore that file yet.
done!
If there are uncommitted changes in the working directory, Git proceeds with a checkout only if it can preserve those changes.
- Example 1: There is a new file in the working directory that is not committed yet.
→ Git will proceed with the checkout and will keep the uncommitted file as well. - Example 2: There is an uncommitted change to a file that conflicts with the version of that file in the commit you wish to check out.
→ Git will abort the checkout, and the repo will remain in the current commit.
The Git stash feature temporarily sets aside uncommitted changes you’ve made (in your working directory and staging area), without committing them. This is useful when you’re in the middle of some work, but need to switch to another state (e.g., checkout a previous commit), and your current changes are not yet ready to be committed or discarded. You can later reapply the stashed changes when you’re ready to resume that work.
Git can also reset the revision history to a specific point so that you can start over from that point.
Suppose you realise your last few commits have gone in the wrong direction, and you want to go back to an earlier commit and continue from there — as if the “bad” commits never happened. Git’s reset feature can help you do that.
Git reset moves the tip of the current branch to a specific commit, optionally adjusting your staged and unstaged changes to match. This effectively rewrites the branch's history by discarding any commits that came after that point.
Resetting is different from the checkout feature:
- Reset: Lets you start over from a past state. It rewrites history by moving the branch ref to a new location.
- Checkout: Lets you explore a past state without rewriting history. It just moves the
HEADref.
→
[reset to C2...]
main branch!There are three types of resets: soft, mixed, hard. All three move the branch pointer to a new commit, but they vary based on what happens to the staging area and the working directory.
- soft reset: Moves the cumulative changes from the discarded commits into the staging area, waiting to be committed again. Any staged and unstaged changes that existed before the reset will remain untouched.
- mixed reset: Cumulative changes from the discarded commits, and any existing staged changes, are moved into the working directory.
- hard reset: All staged and unstaged changes are discarded. Both the working directory and the staging area are aligned with the target commit (as if no changes were done after that commit).
Scenario Imagine the following scenario. After working with the things repo for a while, you realised that you made the following mistakes.
i) First, you added four 'bad' commits (i.e., commits that shouldn't have been created) -- shown as B1 to B4 in the revision graph given below.
iii) Last, you did a 'bad' change in `shapes.txt`, but didn't stage it yet.
Target To rewrite the history of the repo in way that gets rid of the 'bad' commits/changes listed above.
Preparation
Now we have some 'bad' commits and some 'bad' changes in both the staging area and the working directory. Let's use the reset feature to get rid of all of them, but do it in three steps so that you can learn all three types of resets.
1 Do a soft reset to B2 (i.e., discard last two commits). Verify,
- the
mainbranch is now pointing atB2, and, - the changes that were in the discarded commits (i.e.,
B3andB4) are now in the staging area.
Use the git reset --soft <commit> command to do a soft reset.
git reset --soft HEAD~2
You can run the following commands to verify the current status of the repo is as expected.
git status # check overall status
git log --oneline --decorate # check the branch tip
git diff # check unstaged changes
git diff --staged # check staged changes
Right-click on the commit that you want to reset to, and choose Reset <branch-name> to this commit option.
In the next dialog, choose Soft - keep all local changes.
2 Do a mixed reset to commit B1. Verify,
- the
mainbranch is now pointing atB1. - the staging area is empty.
- the accumulated changes from all three discarded commits (including those from the previous soft reset) are now appearing as unstaged changes in the working directory.
Note howincorrect.txtappears as an 'untracked' file -- this is because unstaging a change of type 'add file' results in an untracked file.
Use the git reset --mixed <commit> command to do a mixed reset. The --mixed flag is the default, and can be omitted.
git reset HEAD~1
Verify the repo status, as before.
Similar to the previous reset, but choose the Mixed - keep working copy but reset index option in the reset dialog.
3 Do a hard reset to commit C4. Verify,
- the
mainbranch is now pointing atC4i.e., all 'bad' commits are gone. - the staging area is empty.
- there are no unstaged changes (except for the untracked files
incorrect.txt-- Git leaves untracked files alone, as untracked files are not meant to be under Git's control).
Use the git reset --hard <commit> command.
git reset --hard HEAD~1
Verify the repo status, as before.
Similar to the previous reset, but choose the Hard - discard all working copy changes option.
done!
Rewriting history can cause your local repo to diverge from its remote counterpart. For example, if you discard earlier commits and create new ones in their place, and you’ve already pushed the original commits to a remote repository, your local branch history will no longer match the corresponding remote branch. Git refers to this as a diverged history.
To protect the integrity of the remote, Git will reject attempts to push a diverged branch using a normal push. If you want to overwrite the remote history with your local version, you must perform a force push.
Scenario You have a local repo that is linked to a remote repo on the GitHub. You have pushed all local commits to the remote repo (i.e., the two are in sync).
Target You wish to rewrite the last commit in the local repo and update the remote repo to match the local repo.
Preparation
1 Rewrite the last commit as follows: Reset the current branch back by one commit, and add a new commit.
For example, you can use the following commands.
git reset --hard HEAD~1
echo "water" >> drinks.txt
git add .
git commit -m "Add drinks.txt"
2 Observe how the local branch is diverged.
git log --oneline --graph --all
* 9ff4d04 (HEAD -> main) Add drinks.txt
| * d086f15 (origin/main) shapes.txt: Add some shapes
|/
* 01472c4 colours.txt: Add some colours
* 4d6714c Update fruits list
* 55be601 Add colours.txt, shapes.txt
* fb8d75d Add elderberries and figs into fruits.txt
* 16cad65 Add fruits.txt
3 Attempt to push to the remote. Observe Git rejects the push.
git push origin main
To https://github.com/.../things.git
! [rejected] main -> main (non-fast-forward)
error: failed to push some refs to 'https://github.com/.../gitmastery-samplerepo-things.git'
hint: Updates were rejected because the tip of your current branch is behind
hint: its remote counterpart. If you want to integrate the remote changes,
hint: ...
4 Do a force-push.
You can use the --force (or -f) flag to force push.
git push -f origin main
A safer alternative to --force is --force-with-lease which overwrites the remote branch only if it hasn’t changed since you last fetched it (i.e., only if remote doesn't have recent changes that you are unaware of):
git push --force-with-lease origin main
done!
DETOUR: Resetting Uncommitted Changes
At times, you might need to get rid of uncommitted changes so that you have a fresh start to the next commit.
To get rid of uncommitted changes, you can reset the repo to the last commit (i.e., HEAD):
The command git reset (without specifying a commit) defaults to git reset HEAD.
git reset: moves any staged changes to working directory (i.e., unstage).git reset --hard: get rid of any staged and unstaged changes.
DETOUR: Undoing/Deleting Recent Commits
How do you undo or delete the last few commits if you realise they were incorrect, unnecessary, or done too soon?
Undoing or deleting recent n commits is easily accomplished with Git's reset feature.
- To delete recent
ncommits and discard the those changes entirely, do ahardreset the commitHEAD~ne.g.,git reset --hard HEAD~3 - To undo recent
ncommits, but keep changes staged, do asoftreset the commitHEAD~ne.g.,git reset --soft HEAD~3 - To undo recent
ncommits, and move changes to the working directory, do amixedreset the commitHEAD~ne.g.,git reset --mixed HEAD~3
To do the above for the most recent commit only, use HEAD~1 (or just HEAD~).
To undo the last commit, right-click on the commit just before it, and choose Reset current branch to this commit.
In the next dialog, choose the mode Mixed - keep working copy but reset index option. This will make the offending commit disappear but will keep the changes that you included in that commit intact.
If you use the Soft - ... mode instead, the last commit will be undone as before, but the changes included in that commit will stay in the staging area.
To delete the last commit entirely (i.e., undo the commit and also discard the changes included in that commit), do as above but choose the Hard - ... mode instead.
To undo/delete last n commits, right-click on the commit just before the last n commits, and do as above.
At this point: You should now be able to use a repository’s revision history to understand how the working directory evolved over time -- and use that insight to inform your work.
How useful this history is depends greatly on how well it was constructed -- for example, how frequently and meaningfully commits were made -- we’ll explore that in a later tour.
What's next: Tour 5: Fine-Tuning the Revision History
Target Usage: To maintain a clean and meaningful revision history.
Motivation: The usefulness of the revision history increases if it consists of well-crafted and well-documented commits.
Lesson plan:
T5L1. Controlling What Goes Into a Commit covers that part.
T5L2. Writing Good Commit Messages covers that part.
T5L3. Reorganising Commits covers that part.
To create well-crafted commits, you need to know how to control which precise changes go into a commit.
Crafting a commit involves two aspects:
- What changes to include in it: deciding what changes belong together in a single commit — this is about commit granularity, ensuring each commit represents a meaningful, self-contained change.
- How to include those changes: carefully staging just those changes — this is about using Git’s tools to precisely control what ends up in the commit.
SIDEBAR: Guidelines on what to include in a commit
A good commit represents a single, logical unit of change — something that can be described clearly in one sentence. For example, fixing a specific bug, adding a specific feature, or refactoring a specific function. If each commit tells a clear story about why the change was made and what it achieves, your repository history becomes a valuable narrative of the project’s development. Here are some (non-exhaustive) guidelines:
- No more than one change per commit: Avoid lumping unrelated changes into one commit, as this makes the history harder to understand, review, or revert (if each commit contains one standalone change, to reverse that change can be done by deleting or reverting that specific commit entirely, without affecting any other changes).
- Make the commit standalone: Don’t split a single logical change across multiple commits unnecessarily, as this can clutter the history and make it harder to follow the evolution of an idea or feature.
- Small enough to review easily, but large enough to stand on its own: For example, fixing the same typo in five files can be one commit — splitting it into five separate commits is excessive. Conversely, implementing a big feature may be too much for one commit — instead, break it down into a series of commits, each containing a meaningful yet standalone step towards the final goal.
Git can let you choose not just which files, but which specific changes within those files, to include in a commit. Most Git tools — including the command line and many GUIs — let you interactively select which "hunks" or even individual lines of a file to stage. This allows you to separate unrelated changes and avoid committing unnecessary edits. If you make multiple changes in the same file, you can selectively stage only the parts that belong to the current logical change.
This level of control is particularly useful when:
- You noticed and fixed a small, unrelated issue while working on something else.
- You experimented with multiple approaches in the same file and now want to commit only the final, clean solution.
- You want your commit history to clearly separate concerns, even when the edits touch the same files.
Preparation You can use any repo for this.
1 Do several changes to some tracked files. Change multiple files. Also change multiple locations in the same file.
2 Stage some changes in some files while keeping other changes in the same files unstaged.
As you know, you can use git add <filename> to stage changes to an entire file.
To select which hunks to stage, you can use the git add -p command instead (-p stands for 'by patch'):
git add -p
This command will take you to an interactive mode in which you can go through each hunk and decide if you want to stage it. The video below contains a demonstration of how this feature works:
To stage a hunk, you can click the Stage button above the hunk in question:
Unstaging can be done similarly:Most git operations can be done faster through the CLI than equivalent Git GUI clients, once you are familiar enough with the CLI commands.
However, selective staging is one exception where a good GUI can do better than the CLI, if you need to do many fine-grained staging operations (e.g., frequently staging only parts of hunks).
done!
Detailed and well-written commit messages can increase the value of Git revision history.
Every commit you make in Git also includes a commit message that explains the change. While one-line messages are fine for small or obvious changes, as your revision history grows, good commit messages become an important source of information — for example, to understand the rationale behind a specific change made in the past.
A commit message is meant to explain the intent behind the changes, not just what was changed. The code (or diff) already shows what changed. Well-written commit messages make collaboration, code reviews, debugging, and future maintenance easier by helping you and others quickly understand the project’s history without digging into the code of every commit.
A complete commit message can include a short summary line (the subject) followed by a more detailed body if needed. The subject line should be a concise description of the change, while the body can elaborate on the context, rationale, side effects, or other details if the change is more complex.
A commit message has the following structure (note how the subject and the body are separated by a blank line):
Subject line
<blank line>
Body
# lines starting with '#' are ignored (they will not be included in the commit message)
Here is an example commit message:
Find command: make matching case-insensitive
Find command is case-sensitive.
A case-insensitive find is more user-friendly because users cannot be
expected to remember the exact case of the keywords.
Let's,
* update the search algorithm to use case-insensitive matching
* add a script to migrate stress tests to the new format
Do some changes to a repo you have.
Commit the changes while writing a full commit message (i.e., subject + body).
When you are ready to commit, use the git commit command (without specifying a commit message).
git commit
This will open your default text editor (like Vim, Nano, or VS Code). Write the commit message inside the editor.
Save and close the editor to create the commit.
You can write your full commit message in the textbox you have been using to write commit messages already.
done!
Following a style guide makes your commit messages more consistent and fit-for-purpose. Many teams adopt established guidelines. These style guides typically contain common conventions that Git users follow when writing commit messages. For example:
- Keep the subject line (the first line) under 50–72 characters.
- Write the subject in the imperative mood (e.g.,
Fix typo in READMErather thanFixed typoorFixes typo). - Leave a blank line between the subject and the body, if you include a body.
- Wrap the body at around 72 characters per line for readability.
PRO-TIP: Configure Git to use your preferred text editor
Git will use the default text editor when it needs you to write a commit message. However, Git can be configured to use a different text editor of your choice.
You can use the following command to set the Git's default text editor:
git config --global core.editor "<editor command>"
Some examples for <editor command>
| Editor | Command to use |
|---|---|
| Vim (default) | vim |
| Nano | nano |
| VS Code | code --wait e.g., git config --global core.editor "code --wait"For this to work, your computer should already be configured to launch VS Code using the code command. See here to find how (refer to the 'Launching from command line' section). |
| Sublime Text | subl -n -w |
| Atom | atom --wait |
| Notepad++ | notepad++.exe (Windows only) |
| Notepad | notepad (Windows built-in) |
Why use --wait or -w? Graphical editors (like VS Code or Sublime) start a separate process, which can take a few seconds. Without --wait, Git may think editing is done before you actually write the message. --wait makes Git pause until the editor window is closed.
When the revision history gets 'messy', Git has a way to 'tidy up' the recent commits.
Git has a powerful tool called interactive rebasing which lets you review and reorganise your recent commits. With it, you can reword commit messages, change their order, delete commits, combine several commits into one (squash), or split a commit into smaller pieces. This feature is useful for tidying up a commit history that has become messy — for example, when some commits are out of order, poorly described, or include changes that would be clearer if split up or combined.
Preparation Run the following commands to create a sample repo that we'll be using for this hands-on practical:
mkdir samplerepo-sitcom
cd samplerepo-sitcom
git init -b main
echo "Aspiring actress" >> Penny.txt
git add .
git commit -m "C1: Add Penny.txt"
echo "Scientist" >> Sheldon.txt
git add .
git commit -m "C3: Add Sheldon.txt"
echo "Comic book store owner" >> Stuart.txt
git add .
git commit -m "C2: Add Stuart.txt"
echo "Engineer" >> Stuart.txt
git commit -am "X: Incorrectly update Stuart.txt"
echo "Engineer" >> Howard.txt
git add .
git commit -m "C4: Adddd Howard.txt"
Target Here are the commits that should be in the created repo, and how each commit needs to be 'tidied up'.
C4: Adddd Howard.txt-- Fix typo in the commit messageAdddd→Add.X: Incorrectly update Stuart.txt-- Drop this commit.C2: Add Stuart.txt-- Swap this commit with the one below.C3: Add Sheldon.txt-- Swap this commit with the one above.C1: Add Penny.txt-- No change required.
1 Start the interactive rebasing.
To start the interactive rebase, use the git rebase -i <start-commit> command. -i stands for 'interactive'. In this case, we want to modify the last four commits (hence, HEAD~4).
git rebase -i HEAD~4
pick 97a8c4a C3: Add Sheldon.txt
pick 60bd28d C2: Add Stuart.txt
pick 8b9a36f X: Incorrectly update Stuart.txt
pick 8ab6941 C4: Adddd Howard.txt
# Rebase ee04afe..8ab6941 onto ee04afe (4 commands)
#
# Commands:
# p, pick <commit> = use commit
# r, reword <commit> = use commit, but edit the commit message
# e, edit <commit> = use commit, but stop for amending
# s, squash <commit> = use commit, but meld into previous commit
# f, fixup [-C | -c] <commit> = like "squash" but keep only the previous
# commit's log message, unless -C is used, in which case
# keep only this commit's message; -c is same as -C but
# opens the editor
# x, exec <command> = run command (the rest of the line) using shell
# b, break = stop here (continue rebase later with 'git rebase --continue')
# d, drop <commit> = remove commit
# l, label <label> = label current HEAD with a name
# t, reset <label> = reset HEAD to a label
# m, merge [-C <commit> | -c <commit>] <label> [# <oneline>]
# create a merge commit using the original merge commit's
# message (or the oneline, if no original merge commit was
# specified); use -c <commit> to reword the commit message
# u, update-ref <ref> = track a placeholder for the <ref> to be updated
# to this position in the new commits. The <ref> is
# updated at the end of the rebase
#
# These lines can be re-ordered; they are executed from top to bottom.
#
# If you remove a line here THAT COMMIT WILL BE LOST.
#
# However, if you remove everything, the rebase will be aborted.
#
The command will take you to the text editor, which will present you with a wall of text similar to the above. It has two parts:
- At the top, the list of commits and the action to take on each, oldest commit first, with the action
pickindicated by default (pickmeans 'use this commit in the result') for each. - At the bottom, instructions on how to edit those lines.
2 Edit the commit list to specify the rebase actions, as follows:
pick 60bd28d C2: Add Stuart.txt
pick 97a8c4a C3: Add Sheldon.txt
drop 8b9a36f X: Incorrectly update Stuart.txt
reword 8ab6941 C4: Addddd Howard.txt
4 Once you save edits and exit the text editor, Git will perform the rebase based on the actions you specified, from top to bottom.
At some steps, Git will pause the rebase and ask for your inputs. In this case, it will ask you to specify the new commit message when it is processing the following line.
reword 8ab6941 C4: Addddd Howard.txt
To go to the interactive rebase mode, right-click the parent commit of the earliest commit you want to reorganise (in this case, it is C1: Add Penny.txt) and choose Rebase children of <SHA> interactively...
2 To indicate what action you want to perform on each commit, select the commit in the list and click on the button for the action you want to do on it:
3 To execute the rebase, after indicating the action for all commits (the dialog will look like the below), click OK.
The final result should be something like the following, 'tidied up' exactly as we wanted:
* 727d877 C4: Add Howard.txt
* 764fc29 C3: Add Sheldon.txt
* 08a965a C2: Add Stuart.txt
* 6436598 C1: Add Penny.txt
done!
Rebasing rewrites history. It is not recommended to rebase commits you have already shared with others.
DETOUR: Updating the Last Commit
Git allows you to amend the most recent commit. This is useful when you realise there’s something you’d like to change — e.g., fix a typo in the commit message, or to exclude some unintended change from the commit.
Updating the commit message
To change the commit message subject only, use the git commit --amend -m "<new commit message>" command.
git commit --amend -m "Fix bug that froze the GUI"
To change the entire commit message (not just the subject), run the git commit --amend command, which will open the text editor for you to edit the commit message. The commit will be updated when you close the text editor.
Click on the Commit button on the top menu. In the region that you use to enter the commit message, use one of the two methods given below to go into the 'Amend last commit' mode.
Updating changes in the commit
While there are multiple ways to do this, one method that will work universally is to do a 'soft reset' of the last commit, update the staging area as you wish, and commit again.
'Updating' a commit does not really update that commit -- it simply creates a new commit with the new data. The original commit remains and is 'left behind' in the repo, and will be garbage-collected after a while if it is not referenced by anything else.
At this point: You should now be able to create more meaningful commits from the start, and also refine them further after they’ve been created.
What's next: Tour 6: Branching Locally
Target Usage: To make use of multiple timelines of work in a local repository.
Motivation: At times, you need to do multiple parallel changes to files (e.g., to try two alternative implementations of the same feature).
Lesson plan:
T6L1. Creating Branches covers that part.
T6L2. Merging Branches covers that part.
T6L3. Resolving Merge Conflicts covers that part.
T6L4. Renaming Branches covers that part.
T6L5. Deleting Branches covers that part.
To work in parallel timelines, you can use Git branches.
Git branches let you develop multiple versions of your work in parallel — effectively creating diverged timelines of your repository’s history. For example, one team member can create a new branch to experiment with a change, while the rest of the team continues working on another branch. Branches can have meaningful names, such as main, release, or draft.
A Git branch is simply a ref (a named label) that points to a commit and automatically moves forward as you add new commits to that branch. As you’ve seen before, the HEAD ref indicates which branch you’re currently working on, by pointing to the corresponding branch ref.
When you add a commit, it goes into the branch you are currently on, and the branch ref moves to the new commit, effectively making the HEAD ref point to the new commit as well (via the branch ref).
Git creates a branch named master by default (Git can be configured to use a different name e.g., main -- which is the more common choice these days, and is the default used by Git-Mastery).
Given below is an illustration of how branch refs move as branches evolve. Refer to the text below it for explanations of each stage.
- There is only one branch (i.e.,
main) and there is only one commit on it. TheHEADref is pointing to themainbranch (as we are currently on that branch). - A new commit has been added. The
mainhas moved to the new commit.HEADcontinues to be attached to themainref, thus pointing to the new commit as well. - A new branch
fix1has been added. The repo has switched to the new branch too (hence, theHEADref is now attached to thefix1branch). - A new commit (
c) has been added. The current branch reffix1has moved to the new commit.HEADis also pointing to the new commit, as it remains attached to thefix1branch ref.
- The repo has switched back to the
mainbranch. Hence, theHEADis now pointing to themainbranch ref, and via that, to commitbthat is at the of that branch.
As a result, the repo's working directory now reflects the code at commitb(notc). - A new commit (
d) has been added. Themainand theHEADrefs now point to that commit. - The repo has switched back to the
fix1branch and added a new commiteto it. The branch reffix1(together withHEAD) are now pointing to the new commitewhile the branch refmainstill points tod.
Appearance of the revision graph (colors, positioning, orientation etc.) varies based on the Git client you use, and might not match the exact diagrams given above.
Preparation
1 Observe that you are on the branch called main.
git status
On branch main
2 Start a branch named feature1 and switch to the new branch.
You can use the branch command to create a new branch and the checkout command to switch to a specific branch.
git branch feature1
git checkout feature1
One-step shortcut to create a branch and switch to it at the same time:
git checkout -b feature1
Switched to branch 'feature1'
The new switch command
Git recently introduced a switch command that you can use instead of the checkout command given above.
To create a new branch and switch to it:
git branch feature1
git switch feature1
One-step shortcut (by using -c or --create flag):
git switch -c feature1
Click on the Branch button on the main menu. In the next dialog, enter the branch name and click Create Branch.
Note how the feature1 is indicated as the current branch (reason: Sourcetree automatically switches to the new branch when you create a new branch, if the Checkout New Branch was selected in the previous dialog).
3 Create some commits in the new branch, as follows.
- Add a file named
boxing.txt, stage it, commit it.echo -e "Muhammad Ali" > boxing.txt git stage boxing.txt git commit -m "Add boxing.txt" - Observe how commits you add while on
feature1branch will becomes part of that branch.
Observe how themainref and theHEADref move to the new commit.
As before, you can use the git log --oneline --decorate command for this.
- At times, the
HEADref of the local repo is represented as in Sourcetree, as illustrated in the screenshot below
. - The
HEADref is not shown in the UI if it is already pointing at the active branch.
- Add some more texts to
boxing.txt, stage the changes, and commit it. This commit too will be added to thefeature1branch.echo -e "Mike Tyson" >> boxing.txt git commit -am "Add Tyson to boxing.txt"
4 Switch to the main branch. Note how the changes you made in the feature1 branch are no longer in the working directory.
git switch main
Double-click the main branch.
Revisiting main vs origin/main
In the screenshot above, you see a main ref and a origin/main ref for the same commit. The former identifies the of the local main branch while the latter identifies the tip of the main branch at the remote repo named origin. The fact that both refs point to the same commit means the local main branch and its remote counterpart are with each other.
Similarly, origin/HEAD ref appearing against the same commit indicates that of the remote repo is pointing to this commit as well.
5 Add a commit to the main branch. Let’s imagine it’s a bug fix.
To keep things simple for the time being, this commit should not involve the boxing.txt file that you changed in the feature1 branch. Of course, this is easily done, as the boxing.txt file you added in the feature1 branch is not even visible when you are in the main branch.
echo -e "Martina Navratilova" >> tennis.txt
git commit -am "Add Martina to tennis.txt"
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
commit id: "m2"
branch feature1
commit id: "f1"
commit id: "[feature1] f2"
checkout main
commit id: "[HEAD → main] m3"
checkout feature1
6 Switch between the two branches and see how the working directory changes accordingly. That is, now you have two parallel timelines that you can freely switch between.
done!
You can also start a branch from an earlier commit, instead of the latest commit in the current branch. For that, simply check out the commit you wish to start from.
Preparation
Option 2: Continue with the sports repo from the previous hands-on practical
Scenario Suppose we want to create a branch containing an alternative version of the content we added in the feature1 branch.
Target Create a new branch that starts from the same commit the feature1 branch started from, as shown below:
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
commit id: "m2"
branch feature1
branch feature1-alt
checkout feature1
commit id: "f1"
commit id: "[feature1] f2"
checkout main
commit id: "[HEAD → main] m3"
checkout feature1-alt
commit id: "[HEAD → feature1-alt] a1"
Avoid this rookie mistake!
Always remember to switch back to the main branch before creating a new branch. If not, your new branch will be created on top of the current branch.
1 Switch to the main branch.
2 Checkout the commit at which the feature1 branch diverged from the main branch (e.g. git checkout HEAD~1). This will create a detached HEAD.
3 Create a new branch called feature1-alt. The HEAD will now point to this new branch (i.e., no longer 'detached').
PRO-TIP: Creating a branch based on another branch in one shot
Suppose you are currently on branch b2 and you want to create a new branch b3 that starts from b1. Normally, you can do that in two steps:
git switch b1 # switch to the intended base branch first
git switch -c b3 # create the new branch and switch to it
This can be done in one shot using the git switch -c <new-branch> <base-branch> command:
git switch -c b3 b1
4 Add a commit on the new branch. Example:
echo -e "Venus Williams" >> tennis.txt
git commit -am "Add Venus to tennis.txt"
done!
Most work done in branches eventually gets merged together.
Merging combines the changes from one branch into another, bringing their diverged timelines back together.
When you merge, Git looks at the two branches and figures out how their histories have diverged since their merge base (i.e., the most recent common ancestor commit of two branches). It then applies the changes from the other branch onto your current branch, creating a new commit. The new commit created when merging is called a merge commit — it records the result of combining both sets of changes.
Given below is an illustration of how such a merge looks like in the revision graph:
- We are on the
fix1branch (as indicated byHEAD). - We have switched to the
mainbranch (thus,HEADis now pointing tomainref). - The
fix1branch has been merged into themainbranch, creating a merge commitf. The repo is still on themainbranch.
The branch you are merging into is called the destination branch (other terms: receiving branch, target branch)
The branch you are merging is referred to as the source branch (other terms: incoming branch, merge branch).
In the above example, main is the destination branch and fix1 is the source branch.
A merge commit has two parent commits e.g., in the above example, the merge commit f has both d and e as parent commits. The parent commit on the destination branch is considered the first parent and the parent commit on the source branch is considered the second parent e.g., in the example above, fix1 branch is the source branch that is being merged into the destination branch main -- accordingly, d is the first parent and e is the second parent.
Scenario You have a repo with two unmerged branches main and feature1.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
commit id: "m2"
branch feature1
commit id: "f1"
commit id: "[feature1] f2"
checkout main
commit id: "[HEAD → main] m3"
checkout feature1
Target To practice branch merging, let's merge each branch to the other.
Preparation
1 Switch back to the feature1 branch.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
commit id: "m2"
branch feature1
commit id: "f1"
commit id: "[HEAD → feature1] f2"
checkout main
commit id: "[main] m3"
checkout feature1
2 Merge the main branch to the feature1 branch, giving an end-result like the following. Also note how Git has created a merge commit (shown as mc1 in the diagram below).
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
commit id: "m2"
branch feature1
commit id: "f1"
commit id: "f2"
checkout main
commit id: "[main] m3"
checkout feature1
merge main id: "[HEAD → feature1] mc1"
git merge main
Right-click on the main branch and choose merge main into the current branch. Click OK in the next dialog.
If a confirmation dialog pops up, choose as follows:
The revision graph should look like this now (colours and line alignment might vary but the graph structure should be the same):
Observe how the changes you made in the main branch (i.e., the imaginary bug fix in m3) is now available even when you are in the feature1 branch.
Furthermore, observe (e.g., git show HEAD) how the merge commit contains the sum of changes done in commits m3, f1, and f2.
3 Add another commit to the feature1 branch, in which you do some further changes to the boxing.txt.
echo -e "Manny Pacquiao" >> boxing.txt
git commit -am "Add Manny to boxing.txt"
Switch to the main branch and add one more commit.
git switch main
echo -e "Lionel Messi" >> football.txt
git commit -am "Add Messi to football.txt"
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
commit id: "m2"
branch feature1
commit id: "f1"
commit id: "f2"
checkout main
commit id: "m3"
checkout feature1
merge main id: "mc1"
commit id: "[feature1] f3"
checkout main
commit id: "[HEAD → main] m4"
4 Merge feature1 to the main branch, giving an end-result like this:
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
commit id: "m2"
branch feature1
commit id: "f1"
commit id: "f2"
checkout main
commit id: "m3"
checkout feature1
merge main id: "mc1"
commit id: "[feature1] f3"
checkout main
commit id: "m4"
merge feature1 id: "[HEAD → main] mc2"
git merge feature1
Right-click on the feature1 branch and choose Merge.... The resulting revision graph should look like this:
Now, any changes you made in feature1 branch are available in the main branch.
done!
When the destination branch hasn't diverged — meaning it hasn't had any new commits since the merge base commit — Git simply moves the branch pointer forward to include all the new commits in the source branch, keeping the history clean and linear. This is called a fast-forward merge because Git simply "fast-forwards" the branch pointer to the tip of the other branch. The result looks as if all the changes had been made directly on one branch, without any branching at all.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
commit id: "[HEAD → main] m2"
branch bug-fix
commit id: "b1"
commit id: "[bug-fix] b2"
checkout main
→
[merge bug-fix]
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
commit id: "m2"
commit id: "b1"
commit id: "[HEAD → main][bug-fix] b2"
checkout main
In the example above, the main branch has not changed since the merge base (i.e., m2). Hence, merging the branch bug-fix onto main can be done by fast-forwarding the main branch ref to the tip of the bug-fix branch (i.e., b2).
Scenario You have a repo with an unmerged branch add-swimming. The main branch has not diverged from the add-swimming branch yet.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "more commits ..."
commit id: "[main] mc2"
branch add-swimming
commit id: "a1"
commit id: "[HEAD → add-swimming] a2"
Target Do a fast-forward merge of the add-swimming branch into the main branch.
Preparation
Target Do a fast-forward merge of the add-swimming branch.
1 Ensure you are on the main branch.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "more commits ..."
commit id: "[HEAD → main] mc2"
branch add-swimming
commit id: "a1"
commit id: "add-swimming] a2"
2 Merge the add-swimming branch onto the main branch. Observe that there is no merge commit. The main branch ref (and the HEAD ref along with it) moved to the tip of the add-swimming branch (i.e., a2) and both branches now point to a2.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main (and add-swimming)'}} }%%
commit id: "more commits ..."
commit id: "mc2"
commit id: "a1"
commit id: "[HEAD → main][add-swimming] a2"
done!
It is possible to force Git to create a merge commit even if fast forwarding is possible. This is useful if you prefer the revision graph to visually show when each branch was merged to the main timeline.
To prevent Git from fast-forwarding, use the --no-ff switch when merging. Example:
git merge --no-ff add-swimming
Here are two other relevant options for the git merge command:
--ff-only: Merge goes ahead only if a fast-forward merge is possible.--ff: Explicitly specify a fast-forward merge is preferred, but allows creating a merge commit if a fast-forward merge is not possible. In fact, this is Git's the default behaviour even if you don't use this option. So, the option is not useful unless the relevant default merge behaviour has been changed previously.
Tick the box shown below when you merge a branch:
Trigger the branch operation using the following menu button:
In the next dialog, tick the following option:
To permanently prevent fast-forwarding:
- Go to Sourcetree
Settings. - Navigate to the
Gitsection. - Tick the box
Do not fast-forward when merging, always create commit.
A squash merge takes all the changes from the source branch and combines them into a single commit on the destination branch. It is especially useful when the source branch has many small or experimental commits that would otherwise clutter history.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "more commits ..."
commit id: "[HEAD → main] m1"
branch feature
checkout feature
commit id: "f1"
commit id: "[feature] f2"
→
[squash merge...]
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "more commits ..."
commit id: "m1"
branch feature
checkout feature
commit id: "f1"
commit id: "[feature] f2"
checkout main
commit id: "[HEAD → main] s1 (same as f1+f2)"
In the example above, the branch feature has been squash merged onto the main branch, creating a single 'squashed' commit s1 that combines all the commits in feature branch.
After a squash merge, you typically delete the source branch, so its individual commits are no longer kept. The destination branch's history stays linear, as the work done in the source branch is replaced by one commit on the destination branch. As a result, a squash merge commit is just a normal commit, and does not have a 'parent' reference to the source branch.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "more commits ..."
commit id: "[HEAD → main] m1"
branch feature
checkout feature
commit id: "f1"
commit id: "[feature] f2"
checkout main
merge feature
with a merge commit]
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "more commits ..."
commit id: "m1"
commit id: "f1"
commit id: "[HEAD → main][feature] f2"
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "more commits ..."
commit id: "m1"
commit id: "[HEAD → main] s1 (same as f1+f2)"
deleting the source branch thereafter]
The mechanics of doing a squash merge is covered in a separate detour.
DETOUR: Undoing a Merge
- Ensure you are in the .
- Do a hard reset of that branch to the commit that would be the tip of that branch had you not done the offending merge i.e., rewind that branch to the state it was in before the merge.
In the example below, you merged main to feature1.
If you want to undo that merge,
- Ensure you are in the
feature1branch (because that's the destination branch). - Reset the
feature1branch to the commit that was the tip of thefeature1branch just before you merged themainbranch to it.
Branches can be renamed, for example, to fix a mistake in the branch name.
Local branches can be renamed easily. Renaming a branch simply changes the branch reference (i.e., the name used to identify the branch) — it is just a cosmetic change.
Target You wish to rename fantasy (not yet merged) and textbooks (merged), as shown below:
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
branch textbooks
checkout textbooks
commit id: "[textbooks] t1"
checkout main
branch fantasy
checkout fantasy
commit id: "[fantasy] f1"
checkout main
merge textbooks id: "[HEAD → main] mc1"
→
[rename branches]
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
branch study-books
checkout study-books
commit id: "[study-books] t1"
checkout main
branch fantasy-books
checkout fantasy-books
commit id: "[fantasy-books] f1"
checkout main
merge study-books id: "[HEAD → main] mc1"
Preparation
steps:
To rename a branch, use the git branch -m <current-name> <new-name> command (-m stands for 'move'):
git branch -m fantasy fantasy-books
git branch -m textbooks study-books
git log --oneline --decorate --graph --all # verify the changes
* 443132a (HEAD -> main) Merge branch textbooks
|\
| * 4969163 (study-books) Add textbooks.txt
|/
| * 0586ee1 (fantasy-books) Add fantasy.txt
|/
* 7f28f0e Add horror.txt
Note these additional switches to the log command:
--all: Shows all branches, not just the current branch.--graph: Shows a graph-like visualisation (notice how*is used to indicate a commit, and branches are indicated using vertical lines).
Right-click on the branch name and choose Rename.... Provide the new branch name in the next dialog.
done!
SIDEBAR: Branch naming conventions
Branch names can contain lowercase letters, numbers, /, dashes (-), underscores (_), and dots (.).
You can use uppercase letters too, but many teams avoid them for consistency.
A common branch naming convention is to prefix it with <category>/. Some examples:
feature/login-form— for new features (origin/feature/login-formcould be the matching remote-tracking branch)bugfix/profile-photo— for fixing bugshotfix/payment-crash— for urgent production fixesrelease/2.0— for prepping a releaseexperiment/ai-chatbot— for “just trying stuff”
Although forward-slash (/) in the prefix doesn't mean folders, some tools treat it kind of like a path so you can group related branches when you run git branch. Shown below is an example of how Sourcetree groups branches with the same prefix.
Branches can be deleted to get rid of them when they are no longer needed.
Deleting a branch deletes the corresponding branch ref from the revision history (it does not delete any commits). The impact of the loss of the branch ref depends on whether the branch has been merged.
When you delete a branch that has been merged, the commits of the branch will still exist in the history and will be safe. Only the branch ref is lost.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
branch bug-fix
checkout bug-fix
commit id: "[bug-fix] b1"
checkout main
merge bug-fix id: "[HEAD → main] mc1"
→
[delete branch bug-fix]
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
branch _
checkout _
commit id: "b1"
checkout main
merge _ id: "[HEAD → main] mc1"
In the above example, the only impact of the deletion is the loss of the branch ref bug-fix. All commits remain reachable (via the main branch), and there is no other impact on the revision history.
In fact, some prefer to delete the branch soon after merging it, to reduce branch references cluttering up the revision history.
When you delete a branch that has not been merged, the loss of the branch ref can render some commits unreachable (unless you know their commit IDs or they are reachable through other refs), putting them at risk of being lost eventually.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "[HEAD → main] m1"
branch bug-fix
checkout bug-fix
commit id: "[bug-fix] b1"
checkout main
→
[delete branch bug-fix]
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "[HEAD → main] m1"
branch _
checkout _
commit id: "b1"
checkout main
In the above example, the commit b1 is no longer reachable, unless we know its commit ID (i.e., the SHA).
SIDEBAR: What makes a commit 'unreachable'?
Recall that a commit only has a pointer to its parent commit (not its descendent commits).
A commit is considered reachable if you can get to it by starting at a branch, tag, or other ref and walking backward through its parent commits. 'Reachable' is the normal state for commits — they are part of the visible history of a branch or tag.
If no branch, tag, or ref in the repo can be used as the starting point to reach a certain commit, that commit is unreachable. This often happens when you delete a branch or rewrite history (e.g., with reset or rebase), leaving some commits "orphaned" (or "dangling") without a ref pointing to them.
In the example below, C4 is unreachable (i.e., cannot be reached by starting at any of the three refs: v1.0 or main or ←HEAD), but the other three are all reachable.
Unreachable commits are not deleted immediately — Git keeps them for a while before cleaning them up. By default, Git retains unreachable commits for at least 30 days, during which they can still be recovered if you know their SHA. After that, they will be garbage-collected, and will be lost for good.
Scenario You have the following repo, named samplerepo-books-2.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
branch textbooks
checkout textbooks
commit id: "[textbooks] t1"
checkout main
branch fantasy
checkout fantasy
commit id: "[fantasy] f1"
checkout main
merge textbooks id: "[HEAD → main] mc1"
The work in the textbook branch has been completed, and the branch has been merged -- there is no need to keep that branch anymore.
The work in the fantasy branch is no longer needed. Hence, there is no need for the branch either.
Target Delete the textbooks (merged) and fantasy branches (unmerged).
Preparation
1 Delete the (the merged) textbooks branch.
Use the git branch -d <branch> command to delete a local branch 'safely' -- this command will fail if the branch has unmerged changes. In this case, it will succeed as the branch has no unmerged commits.
git branch -d textbooks
git log --oneline --decorate --graph --all # check the current revision graph
* 443132a (HEAD -> main) Merge branch textbooks
|\
| * 4969163 Add textbooks.txt
|/
| * 0586ee1 (fantasy) Add fantasy.txt
|/
* 7f28f0e Add horror.txt
Right-click on the branch name and choose Delete <branch>:
In the next dialog, click OK:
Observe that all commits remain. The only missing thing is the textbook ref.
2 Make a copy of the SHA of the tip of the (unmerged) fantasy branch.
3 Delete the fantasy branch.
Attempt to delete the branch. It should fail, as shown below:
git branch -d fantasy
error: the branch 'fantasy' is not fully merged
hint: If you are sure you want to delete it, run 'git branch -D fantasy'
As also hinted by the error message, you can replace the -d with -D to 'force' the deletion.
git branch -D fantasy
Now, check the revision graph:
git log --oneline --decorate --graph --all
* 443132a (HEAD -> main) Merge branch textbooks
|\
| * 4969163 Add textbooks.txt
|/
* 7f28f0e Add horror.txt
Attempt to delete the branch as you did before. It will fail because the branch has unmerged commits.
Try again but this time, tick the Force delete option, which will force Git to delete the unmerged branch:
Observe how the branch ref fantasy is gone, together with any unmerged commits on it.
4 Attempt to view the 'unreachable' commit whose SHA you noted in step 2.
e.g., git show 32b34fb (use the SHA you copied earlier)
Observe how the commit still exists and still is reachable using the commit ID, although it is not reachable by other means, and not visible in the revision graph.
done!
As this is the last lesson for this tour, this is a good time to attempt exercises that combine knowledge from multiple lessons in this tour:
At this point: Now you can create, maintain, and merge multiple parallel branches in a local repo. This tour covered only the basic use of Git branches. More advanced usage will be covered in other tours.
What's next: Tour 7: Keeping Branches in Sync
Target Usage: To keep branches in a local repository synchronised with each other, as needed.
Motivation: While working on one branch, you might want to have access to changes introduced in another branch (e.g., to take advantage of a bug fix introduced in another branch).
Lesson plan:
T7L1. Merging to Sync Branches covers that part.
T7L2. Rebasing to Sync Branches covers that part.
T7L3. Copying Specific Commits covers that part.
At this point: You should now be able to bring changes from one branch to another in your local repository.
What's next: Tour 8: Working with Remote Branches
Target Usage: To synchronise branches in the local repo with a remote repo's branches.
Motivation: It is useful to be able to have another copy of branches in a remote repo.
Lesson plan:
T8L1. Pushing Branches to a Remote covers that part.
T8L2. Pulling Branches from a Remote covers that part.
T8L3. Deleting Branches from a Remote covers that part.
T8L4. Renaming Branches in a Remote covers that part.
Local branches can be replicated in a remote.
Pushing a copy of local branches to the corresponding remote repo makes those branches available remotely.
In a previous lesson, we saw how to push the default branch to a remote repository and have Git set up tracking between the local and remote branches using a remote-tracking reference. Pushing any other local branch to a remote works the same way as pushing the default branch — you simply specify the target branch instead of the default branch. Pushing any new commits in any local branch to a corresponding remote branch is done similarly as well.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
branch bug-fix
checkout main
commit id: "[origin/main][HEAD → main] m2"
checkout bug-fix
commit id: "[bug-fix] b1"
checkout main
[bug-fix branch does not exist in the remote origin]
→
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
branch bug-fix
checkout main
commit id: "[origin/main][HEAD → main] m2"
checkout bug-fix
commit id: "[origin/bug-fix][bug-fix] b1"
checkout main
[after pushing bug-fix branch to origin,
and setting up a remote-tracking branch]
Preparation Fork the samplerepo-company to your GitHub account. When doing so, un-tick the Copy the main branch only option.
After forking, go to the fork and ensure both branches (main, and track-sales) are in there.
Clone the fork to your computer. It should look something like this:
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
commit id: "m2"
branch track-sales
checkout track-sales
commit id: "[origin/track-sales] s1"
checkout main
commit id: "[origin/main][origin/HEAD][HEAD → main] m3"
The origin/HEAD remote-tracking ref indicates where the HEAD ref is in the remote origin.
1 Create a new branch called hiring, and add a commit to that branch. The commit can contain any changes you want.
Here are the commands you can run in the terminal to do this step in one shot:
git switch -c hiring
echo "Receptionist: Pam" >> employees.txt
git commit -am "Add Pam to employees.txt"
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
commit id: "m2"
branch track-sales
checkout track-sales
commit id: "[origin/track-sales] s1"
checkout main
commit id: "[origin/main][origin/HEAD][main] m3"
branch hiring
checkout hiring
commit id: "[HEAD → hiring] h1"
The resulting revision graph should look like the one above.
2 Push the hiring branch to the remote.
You can use the usual git push <remote> -u <branch> command to push the branch to the remote, and set up a remote-tracking branch at the same time.
git push origin -u hiring
3 Verify that the branch has been pushed to the remote by visiting the fork on GitHub, and looking for the origin/hiring remote-tracking ref in the local repo.
done!
Branches in a remote can be replicated in the local repo, and maintained in sync with each other.
Sometimes we need to create a local copy of a branch from a remote repository, make further changes to it, and keep it synchronised with the remote branch. Let's explore how to handle this in a few common use cases:
Use case 1: Working with branches that already existed in the remote repo when you cloned it to your computer.
When you clone a repository,
- Git checks out the default branch. You can start working on this branch immediately. This branch is tracking the default branch in the remote, which means you can easily synchronise changes in this branch with the remote by pulling and pushing.
- Git also fetches all the other branches from the remote. These other branches are not immediately available as local branches, but they are visible as remote-tracking branches.
You can think of remote-tracking branches as read-only references to the state of those branches in the remote repository at the time of cloning. They allow you to see what work has been done on those branches without yet making local copies of them.
To work on one of these branches, you can create a new local branch based on the remote-tracking branch. Once you do this, your local branch will usually be configured to track the corresponding branch on the remote, so you can easily synchronise your work later.
Preparation Use the same samplerepo-company repo you used in Lesson T8L1. Pushing Branches to a Remote. Fork and clone it if you haven't done that already.
1 Verify that the remote-tracking branch origin/track-sales exists in the local repo, but there is no local copy of it.
You can use the git branch -a command to list all local and tracking branches.
git branch -a
* hiring
main
remotes/origin/HEAD -> origin/main
remotes/origin/hiring
remotes/origin/main
remotes/origin/track-sales
The * in the output above indicates the currently active branch.
Note how there is no track-sales in the list of branches (i.e., no local branch named track-sales), but there is a remotes/origin/track-sales (i.e., the remote-tracking branch)
Observe how the branch track-sales appear under REMOTES → origin but not under BRANCHES.
2 Create a local copy of the remote branch origin/track-sales.
You can use the git switch -c <branch> <remote-branch> command for this e.g.,
git switch -c track-sales origin/track-sales
Locate the track-sales remote-tracking branch (look under REMOTES → origin), right-click, and choose Checkout....
In the next dialog, choose as follows:
The above command/action does several things:
- Creates a new branch
track-sales. - Sets the new branch to track the remote branch
origin/track-sales, which means the local branch reftrack-saleswill also move to where theorigin/track-salesis. - Switch to the newly-created branch i.e., makes it the current branch.
3 Add a commit to the track-sales branch and push to the remote, to verify that the local branch is tracking the remote branch.
Commands to perform this step in one shot:
echo "5 reams of paper" >> sales.txt
git commit -am "Update sales.txt"
git push origin track-sales
done!
Use case 2: Working with branches that were added to the remote repository after you cloned it e.g., a branch someone else pushed to the remote after you cloned.
Simply fetch to update your local repository with information about the new branch. After that, you can create a local copy of it and work with it just as you did in Use Case 1.
Fetching was covered in Lesson T3L3. Downloading Data Into a Local Repo.
New commits can appear in a remote branch after you have set up a local branch to track it (as per use case 1 or 2 given above). e.g., a branch someone else pushed a commit to the remote branch after you pulled the previous version of it. Often you would want to update your local copy of the branch with those new commits.
Here is an example:
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
branch bug-fix
checkout main
commit id: "m2"
checkout bug-fix
commit id: "[origin/bug-fix][HEAD → bug-fix] b1"
checkout main
[local repo: bug-fix branch is unaware
of the commit b2 in the remote]
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
branch bug-fix
checkout main
commit id: "m2"
checkout bug-fix
commit id: "b1"
commit id: "[bug-fix] b2"
checkout main
[remote repo: has an extra commit
in the bug-fix branch]
To bring the missing commits to the local branch, simply pull the remote branch from your local branch.
If you fetch first (or if your Git GUI is set to auto-fetch periodically) the local repo will be as follows, before and after the pull.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
branch bug-fix
checkout main
commit id: "m2"
checkout bug-fix
commit id: "[HEAD → bug-fix] b1"
commit id: "[origin/bug-fix] b2"
checkout main
[local repo: bug-fix branch is aware
of the commit b2]
→
[pull, or just merge]
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
branch bug-fix
checkout main
commit id: "m2"
checkout bug-fix
commit id: "b1"
commit id: "[origin/bug-fix][HEAD → bug-fix] b2"
checkout main
[local repo: now has the commit b2]
If both the local branch and the remote-tracking branch have new commits that the other does not, Git will try to combine the two diverged histories when you do a pull. By default, this is done by creating a merge commit, although this behaviour can be changed (for example, to use rebasing instead).
In the example below, the local branch bug-fix has a new commit b3 while its remote tracking branch has a new commit b2. After pulling, Git has combined the two diverged branches with a merged commit.
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
branch bug-fix
checkout main
commit id: "m2"
checkout bug-fix
commit id: "[origin/bug-fix] b1"
commit id: "[HEAD → bug-fix] b3"
checkout main
[local repo: bug-fix has a new commit b3]
→
[pull]
gitGraph BT:
%%{init: { 'theme': 'default', 'gitGraph': {'mainBranchName': 'main'}} }%%
commit id: "m1"
branch bug-fix
checkout main
commit id: "m2"
checkout bug-fix
commit id: "b1"
branch _
checkout _
commit id: "b2"
checkout bug-fix
commit id: "b3"
merge _ id: "[HEAD → bug-fix] merge commit"
checkout main
[local repo: now has b2, and a merge commit]
Often, you'll need to delete a branch in a remote repo after it has served its purpose.
To delete a branch in a remote repository, you simply tell Git to remove the reference to that branch from the remote. This does not delete the branch from your local repository — it only removes it from the remote, so others won’t see it anymore. This is useful for cleaning up clutter in the remote repo e.g., delete old or merged branches that are no longer needed on the remote.
Preparation Fork the samplerepo-books to your GitHub account. When doing so, un-tick the Copy the main branch only option.
After forking, go to the fork and ensure all three branches are in there.
Clone the fork to your computer.
1 Create a local copy of the fantasy branch in your clone.
Follow instructions in Lesson T8L2. Pulling Branches from a Remote.
2 Delete the remote branch fantasy.
You can use the git push <remote> --delete <branch> command to delete a branch in a remote. This is like pushing changes in a branch to a remote, except we request the branch to be deleted instead, by adding the --delete switch.
git push origin --delete fantasy
Locate the remote branch under REMOTES → origin, right-click on the branch name, and choose Delete...:
3 Verify that the branch was deleted from the remote, by going to the fork on GitHub and checking the branches page https://github.com/{YOUR_USERNAME}/samplerepo-books/branches
e.g., https://github.com/johndoe/samplerepo-books/branches.
Also verify that the local copy has not been deleted.
4 Restore the remote branch from the local copy.
Push the local branch to the remote, while enabling the tracking option (as if pushing the branch to the remote for the first time), as covered in Lesson T8L1. Pushing Branches to a Remote.
In the above steps, we first created a local copy of the branch before deleting it in the remote repo. Doing so is optional. You can delete a remote branch without ever checking it out locally — you just need to know its name on the remote. Deleting the remote branch directly without creating a local copy is recommended if you simply want to clean up a remote branch you no longer need.
done!
At this point: You should now be able to work with branches in a remote repo, and keep them synchronised with branches in the local repo.
What's next: Tour 9: Working with Pull Requests
Target Usage: To contribute to a project using GitHub's pull request mechanism.
Motivation: Pull Request (PR) is the most common way to contribute to a project hosted on GitHub.
Lesson plan:
T9L1. Creating Pull Requests covers that part.
T9L2. Reviewing Pull Requests covers that part.
T9L3. Merging Pull Requests covers that part.
Another way to contribute to a GitHub project is by giving your inputs via a pull request review.
PR reviews are a collaborative process in which project members examine and provide feedback on PRs submitted to a remote repo. After an initial review, the reviewer may suggest improvements or identify issues, prompting the submitter to refine and update their code in the PR. This review-refine-update cycle can repeat several times, with reviewers reassessing each new iteration until all feedback is addressed and the code meets the team’s expectations. Once approved, the PR can be merged, making the changes an official part of the codebase.
Preparation If you do not have access to a PR that you can review, you can create one for yourself as follows:
- Create a branch in a repo that you have forked and cloned (e.g., samplerepo-pr-practice).
- Do some changes in the branch.
- Push the branch to the remote repo.
- Create a PR within your fork, from the new branch to the
mainbranch.
1 Locate the PR:
- Go to the GitHub page of the repo.
- Click on the Pull requests tab.
- Click on the PR you want to review.
2Read the PR description. It might contain information relevant to reviewing the PR.
3Click on the Files changed tab to see the diff view.
You can use the following setting to try the two different views available and pick the one you like.
4Add review comments:
- Hover over the line you want to comment on and click on the icon that appears on the left margin. That should create a text box for you to enter your comment.
- To give a comment related to multiple lines, click-and-drag the icon. The result will look like this:
- To give a comment related to multiple lines, click-and-drag the icon. The result will look like this:
- Enter your comment.
- This page @SE-EDU/guides has some best practices PR reviewers can follow.
- To suggest an in-line code change, click on this icon:
After that, you can proceed to edit thesuggestioncode block generated by GitHub (as seen in the screenshot above).
The comment will look like this to the viewers:
- After typing in the comment, click on the Start a review button (not the Add single comment button. This way, your comment is saved but not visible to others yet. It will be visible to others only when you have finished the entire review.
- Repeat the above steps to add more comments.
5Submit the review:
- When there are no more comments to add, click on the Review changes button (on the top right of the diff page).
- Type in an overall comment about the PR, if any. e.g.,
Overall, I found your code easy to read for the most part except a few places where the nesting was too deep. I noted a few minor coding standard violations too. Some of the classes are getting quite long. Consider splitting into smaller classes if that makes sense.LGTMis often used in such overall comments, to indicateLooks good to me(orLooks good to merge).
nit(as in nit-picking) is another such term, used to indicate minor flaws e.g.,LGTM. Just a few nits to fix.. - Choose
Approve,Comment, orRequest changesoption as appropriate and click on the Submit review button.
done!
If you have appropriate level of access to a GitHub repo, you can merge pull requests.
A project member with sufficient access to the remote repo can merge a PR, incorporating proposed changes into the main codebase. Merging a PR is similar to performing a Git merge in a local repo, except that it occurs in the remote repository.
Preparation If you would like to try merging a PR yourself, you can create a dummy PR in the following manner.
- Create a branch in a repo that you have forked and cloned (e.g., samplerepo-pr-practice).
- Do some changes in the branch.
- Push the branch to the remote repo.
- Create a PR within your fork, from the new branch to the
mainbranch.
1 Locate the PR to be merged in your repo's GitHub page.
2 Click on the Conversation tab and scroll to the bottom. You'll see a panel containing the PR status summary.
3 If the PR is not merge-able in the current state, the Merge pull request will not be green. Here are the possible reasons and remedies:
- Problem: The PR code is out-of-date, indicated by the message This branch is out-of-date with the base branch. That means the repo's
mainbranch has been updated since the PR code was last updated.- If the PR author has allowed you to update the PR and you have sufficient permissions, GitHub will allow you to update the PR simply by clicking the Update branch on the right side of the 'out-of-date' error message. If that option is not available, post a message in the PR requesting the PR author to update the PR.
- Problem: There are merge conflicts, indicated by the message This branch has conflicts that must be resolved. That means the repo's
mainbranch has been updated since the PR code was last updated, in a way that the PR code conflicts with the currentmainbranch. Those conflicts must be resolved before the PR can be merged.- If the conflicts are simple, GitHub might allow you to resolve them using the Web interface.
- If that option is not available, post a message in the PR requesting the PR author to update the PR.
4 Merge the PR by clicking on the Merge pull request button, followed by the Confirm merge button. You should see a Pull request successfully merged and closed message after the PR is merged.
- You can choose between three merging options by clicking on the down-arrow in the Merge pull request button. If you are new to Git and GitHub, the
Create merge commitoption is recommended.
done!
After a PR is merged, you need to sync other related repos. Merging a PR simply merges the code in the upstream remote repository in which it was merged. The PR author (and other members of the repo) needs to pull the merged code from the upstream repo to their local repos and push the new code to their respective forks to sync the fork with the upstream repo.
At this point: Now you can contribute to a GitHub project by creating reviewing, and even merging PRs in a GitHub repository.
What's next: Tour 10: Managing Git-Based Projects
Target Usage: To manage a multi-person project on GitHub
Motivation: To manage a multi-person project on GitHub, one needs to know some additional project management features GitHub offers.
Lesson plan:
T10L1. Git Workflows covers that part.
T10L2. Forking Workflow (with Branching) covers that part.
T10L3. Other Project Management Features covers that part.
A branch-based forking workflow is common in open-source projects and other large projects.
In a branch-based forking workflow, the official code lives in a designated 'main' repo, while each developer works in their own fork (hence, the name) and submits pull requests from separate branches (either long-lived branches or short term branches) back to the main repo. That is, it is a combination of the forking model and the feature-branch strategy. Not only this workflow is common for OSS projects and other large-team projects, it provides a good foundation for learning Git workflows (because other workflows are simpler than this, once you learn this workflow, it is easy to move to other workflows).
To illustrate how the workflow goes, let’s assume Jean wants to fix a bug in the code. Here are the steps:
- Jean creates a separate branch in her local repo and fixes the bug in that branch.
Common mistake: Doing the proposed changes in themainbranch -- if Jean does that, she will not be able to have more than one PR open at any time because any changes to themainbranch will be reflected in all open PRs. - Jean pushes the branch to her fork.
- Jean creates a pull request from that branch in her fork to the main repo.
- Other members review Jean’s pull request.
- If reviewers suggested any changes, Jean updates the PR accordingly.
- When reviewers are satisfied with the PR, one of the members (usually the team lead or a designated 'maintainer' of the main repo) merges the PR, which brings Jean’s code to the main repo.
- Other members (and Jean), realizing there is new code in the upstream repo, sync their forks with the new upstream repo (i.e., the main repo). This is done by pulling the new code to their own local repo and pushing the updated code to their own fork. If there are unmerged branches in the local repo, they can be updated too e.g., by merging the new
mainbranch to each of them.
Potential mistake: Creating another 'reverse' PR from the team repo to the team member's fork to sync the member's fork with the merged code. PRs are meant to go from downstream repos to upstream repos, not in the other direction.
One main benefit of this workflow is that it does not require most contributors to have write permissions to the main repository. Only those who are merging PRs need write permissions. The main drawback of this workflow is the extra overhead of sending everything through forks.
This practical is best done as a team.
Preparation One member: set up the team org and the team repo.
Create a GitHub organization for your team, if you don't have one already. The org name is up to you. We'll refer to this organization as team org from now on.
Add a team called
developersto your team org.Add team members to the
developersteam.Fork se-edu/samplerepo-workflow-practice to your team org. We'll refer to this as the team repo.
Add the forked repo to the
developersteam. Give write access.
1 Each team member: create PRs via own fork.
Fork that repo from your team org to your own GitHub account.
Create a branch named
add-{your name}-info(e.g.add-johnTan-info) in the local repo.Add a file
yourName.mdinto themembersdirectory (e.g.,members/johnTan.md) containing some info about you into that branch.Push that branch to your fork.
Create a PR from that branch to the
mainbranch of the team repo.
2 For each PR: review, update, and merge.
[A team member (not the PR author)] Review the PR by adding comments (can be just dummy comments).
[PR author] Update the PR by pushing more commits to it, to simulate updating the PR based on review comments.
[Another team member] Approve and merge the PR using the GitHub interface.
[All members] Sync your local repo (and your fork) with upstream repo. In this case, your upstream repo is the repo in your team org.
- The basic mechanism for this has two steps (which you can do using Git CLI or any Git GUI):
(1) First, pull from the upstream repo -- this will update your clone with the latest code from the upstream repo.
If there are any unmerged branches in your local repo, you can update them too e.g., you can merge the newmainbranch to each of them.
(2) Then, push the updated branches to your fork. This will also update any PRs from your fork to the upstream repo. - Some alternatives mechanisms to achieve the same can be found in this GitHub help page.
If you are new to Git, we recommend that you use the above two-step mechanism instead, so that you get a better view of what's actually happening behind the scene.
- The basic mechanism for this has two steps (which you can do using Git CLI or any Git GUI):
3 Create conflicting PRs.
[One member]: Update README: In the
mainbranch, remove John Doe and Jane Doe from theREADME.md, commit, and push to the main repo.[Each team member] Create a PR to add yourself under the
Team Memberssection in theREADME.md. Use a new branch for the PR e.g.,add-johnTan-name.
4 Merge conflicting PRs one at a time. Before merging a PR, you’ll have to resolve conflicts.
[Optional] A member can inform the PR author (by posting a comment) that there is a conflict in the PR.
[PR author] Resolve the conflict locally:
- Pull the
mainbranch from the repo in your team org. - Merge the pulled
mainbranch to your PR branch. - Resolve the merge conflict that crops up during the merge.
- Push the updated PR branch to your fork.
- Pull the
[Another member or the PR author]: Merge the de-conflicted PR: When GitHub does not indicate a conflict anymore, you can go ahead and merge the PR.
done!
At this point: You are now able to use an appropriate workflow for your project, and also, make use of other project management features offered by GitHub.
What's next: This is the last of the Git-Mastery tours!