Git: Difference between revisions

From David's Wiki
 
(18 intermediate revisions by the same user not shown)
Line 3: Line 3:
==Installation==
==Installation==
===Windows===
===Windows===
Download [https://git-scm.com/downloads Git for Windows]
Download [https://git-scm.com/downloads Git for Windows] or use [https://community.chocolatey.org/packages/git chocolatey].
<pre>
choco install git --params "/WindowsTerminalProfile"
</pre>
 
===Linux===
===Linux===
[https://git-scm.com/download/linux git-scm install]
<syntaxhighlight lang="bash">
<syntaxhighlight lang="bash">
# PPA for the latest stable version.
sudo add-apt-repository ppa:git-core/ppa
sudo apt update
sudo apt install git git-lfs
sudo apt install git git-lfs
</syntaxhighlight>
</syntaxhighlight>
Line 12: Line 20:
<syntaxhighlight lang="bash">
<syntaxhighlight lang="bash">
# Clone or download a repository
# Clone or download a repository
git clone <url>
git clone <url> [<foldername>]
# cd into the cloned folder
cd <foldername>
# Note that some repos have submodules which will also need to be cloned
git submodule update --init --recursive


# Stage your changes
# Stage your changes
Line 31: Line 43:


==Branches==
==Branches==
Branches allow you to work on the repository without creating new forks and using more disk space.
You can use them to create and test changes from a static copy of the code repo. 
You can also discard a branch if you decide the feature is no longer needed. 
Once the feature has been fully developed, you should rebase the branch onto the head of the main branch. 
Then you can merge the branch to add the feature to main/master.
<syntaxhighlight lang="bash">
<syntaxhighlight lang="bash">
# Make a branch
# Make a branch
Line 41: Line 59:
# Push
# Push
git push --set-upstream origin my_feature
git push --set-upstream origin my_feature
# Checkout a branch
git checkout my_branch
# Checkout a single file from another branch
git checkout my_branch -- other_file.txt
</syntaxhighlight>
</syntaxhighlight>
==Patches==
Patches are a way to represent and share changes without actually making commits or new branches.
<syntaxhighlight lang="bash">
# For unstaged changes
git diff > changes.patch
# For staged changes
git diff --cached > changes.patch
# You can also add --binary to include binary files
# Apply the patch
git apply changes.patch
</syntaxhighlight>
See [https://stackoverflow.com/questions/5159185/create-a-git-patch-from-the-uncommitted-changes-in-the-current-working-directory create git patch]


==Pull Requests==
==Pull Requests==
See [https://stackoverflow.com/questions/20289599/are-pull-requests-part-of-git-or-a-feature-of-tools-like-github-gerrit-and-atl Are pull requests are part of Git?]
Pull requests are a way to submit changes to repositories you do not manage. 
First, you fork the repository to create a working-copy for yourself to develop on. 
Once your changes are finished, you create a pull request asking the original repo to incorporate your changes. 
It's easiest to just use the web interfaces of GitHub or GitLab for this.
To do this manually, see [https://git-scm.com/book/en/v2/Distributed-Git-Contributing-to-a-Project Git Book: Contributing to a project].


==Migrating Repositories==
==Migrating Repositories==
Line 65: Line 111:
</syntaxhighlight>
</syntaxhighlight>


==gitignore==
==Special Files==
.gitignore is used to ignore certain files to make sure they do not get accidentally pushed to the remote git repo.<br>
===<code>.gitignore</code>===
<code>.gitignore</code> is used to ignore certain files to make sure they do not get accidentally pushed to the remote git repo.<br>
You can find a collection of .gitignore files on [https://github.com/github/gitignore Github's gitignore repo].<br>
You can find a collection of .gitignore files on [https://github.com/github/gitignore Github's gitignore repo].<br>
Or you can create a custom .gitignore file at [https://www.gitignore.io gitignore.io].
Or you can create a custom .gitignore file at [https://www.gitignore.io gitignore.io].


==gitattributes==
===<code>.gitattributes</code>===
.gitattributes are used to give attributes to paths.<br>
<code>.gitattributes</code> are used to give attributes to paths.<br>
They can be used to assign programming languages to filenames.<br>
They can be used to assign programming languages to filenames.<br>
They are also used to identify which files should be stored in the lfs.<br>
They are also used to identify which files should be stored in the lfs.<br>
Here is an example to store video and images in the lfs.<br>
[https://gitlab.davidl.me/-/snippets/1/raw/master/lfs.gitattributes Here] is an example to store video and images in the lfs.<br>
<pre>
## Video
*.mp4 filter=lfs diff=lfs merge=lfs -text
*.mov filter=lfs diff=lfs merge=lfs -text
*.flv filter=lfs diff=lfs merge=lfs -text
 
## Pictures
*.png filter=lfs diff=lfs merge=lfs -text
*.jpg filter=lfs diff=lfs merge=lfs -text
*.jpeg filter=lfs diff=lfs merge=lfs -text
 
## Binary files
*.bin filter=lfs diff=lfs merge=lfs -text
*.7z filter=lfs diff=lfs merge=lfs -text
</pre>


==Clean==
==Clean==
Line 102: Line 134:
==Git Large File Storage (LFS)==
==Git Large File Storage (LFS)==
[https://github.com/git-lfs/git-lfs/wiki/Tutorial Reference]
[https://github.com/git-lfs/git-lfs/wiki/Tutorial Reference]
* Note that Github charges for LFS so it's best to keep binary files outside of GitHub.
==Sparse checkout==
If you're interacting with a monorepo like [https://github.com/google-research/google-research google-research/google-research], you may be interested in git sparse checkout.<br>
See [https://git-scm.com/docs/git-sparse-checkout Git - git-sparse-checkout Documentation] and [https://github.blog/2020-01-17-bring-your-monorepo-down-to-size-with-sparse-checkout/ Bring your monorepo down to size with sparse-checkout | The GitHub Blog].<br>
This is a relatively new feature so be sure to use an updated version of git.
<syntaxhighlight lang="bash">
git clone --no-checkout $url
cd $project
git sparse-checkout init
git sparse-checkout add $folder
</syntaxhighlight>

Latest revision as of 13:14, 31 May 2023

Git is a popular version control system made by Linus Torvalds.

Installation

Windows

Download Git for Windows or use chocolatey.

choco install git --params "/WindowsTerminalProfile"

Linux

git-scm install

# PPA for the latest stable version.
sudo add-apt-repository ppa:git-core/ppa
sudo apt update
sudo apt install git git-lfs

Basic Usage

# Clone or download a repository
git clone <url> [<foldername>]
# cd into the cloned folder
cd <foldername>
# Note that some repos have submodules which will also need to be cloned
git submodule update --init --recursive

# Stage your changes
git add <my files>
# Or stage all files
# git add .

# Make a commit
git commit -m "My commit message"

# Push your changes to the repository
git push

# Pull the latest changes from the repository
# Equivalent to git fetch && git merge FETCH_HEAD
git pull

Branches

Branches allow you to work on the repository without creating new forks and using more disk space. You can use them to create and test changes from a static copy of the code repo.
You can also discard a branch if you decide the feature is no longer needed.
Once the feature has been fully developed, you should rebase the branch onto the head of the main branch.
Then you can merge the branch to add the feature to main/master.

# Make a branch
git checkout -b my_feature

# Add and commit to this branch
git add .
git commit -m "My commit message"

# Push
git push --set-upstream origin my_feature

# Checkout a branch
git checkout my_branch

# Checkout a single file from another branch
git checkout my_branch -- other_file.txt

Patches

Patches are a way to represent and share changes without actually making commits or new branches.

# For unstaged changes
git diff > changes.patch
# For staged changes
git diff --cached > changes.patch
# You can also add --binary to include binary files

# Apply the patch
git apply changes.patch

See create git patch

Pull Requests

See Are pull requests are part of Git?

Pull requests are a way to submit changes to repositories you do not manage.
First, you fork the repository to create a working-copy for yourself to develop on.
Once your changes are finished, you create a pull request asking the original repo to incorporate your changes.
It's easiest to just use the web interfaces of GitHub or GitLab for this.

To do this manually, see Git Book: Contributing to a project.

Migrating Repositories

How to migrate repositories to another Git server

git clone --mirror <original_repo>
cd <repo>

# Run the following lines only if you're using lfs
git lfs fetch --all
git lfs push --all <new_repo>

git push --all <new_repo>

Changing Remote URL

Reference
You will need to change the remote url if it is changed on the server. E.g. if you change the project name.

git remote set-url origin <new-url>

Special Files

.gitignore

.gitignore is used to ignore certain files to make sure they do not get accidentally pushed to the remote git repo.
You can find a collection of .gitignore files on Github's gitignore repo.
Or you can create a custom .gitignore file at gitignore.io.

.gitattributes

.gitattributes are used to give attributes to paths.
They can be used to assign programming languages to filenames.
They are also used to identify which files should be stored in the lfs.
Here is an example to store video and images in the lfs.

Clean

Atlassian Reference
Use git clean to delete untracked files.
Add -x to also delete gitignore'd files.

# Force delete any untracked files and directories
git clean -fd

Git Large File Storage (LFS)

Reference

  • Note that Github charges for LFS so it's best to keep binary files outside of GitHub.

Sparse checkout

If you're interacting with a monorepo like google-research/google-research, you may be interested in git sparse checkout.
See Git - git-sparse-checkout Documentation and Bring your monorepo down to size with sparse-checkout | The GitHub Blog.
This is a relatively new feature so be sure to use an updated version of git.

git clone --no-checkout $url
cd $project

git sparse-checkout init
git sparse-checkout add $folder