A Git Workflow is a recipe or recommendation for how to use Git to accomplish work in a consistent and productive manner. Git workflows encourage users to leverage Git effectively and consistently. Git offers a lot of flexibility in how users manage changes. Given Git's focus on flexibility, there is no standardized process on how to interact with Git. When working with a team on a Git managed project, it’s important to make sure the team is all in agreement on how the flow of changes will be applied. To ensure the team is on the same page, an agreed upon Git workflow should be developed or selected. There are several publicized Git workflows that may be a good fit for your team. Here, we’ll be discussing some of these workflow options.
The array of possible workflows can make it hard to know where to begin when implementing Git in the workplace. This page provides a starting point by surveying the most common Git workflows for software teams.
What is a successful Git workflow?
When evaluating a workflow for your team, it's most important that you consider your team’s culture. You want the workflow to enhance the effectiveness of your team and not be a burden that limits productivity. Some things to consider when evaluating a Git workflow are:
- Does this workflow scale with team size?
- Is it easy to undo mistakes and errors with this workflow?
- Does this workflow impose any new unnecessary cognitive overhead to the team?
The Centralized Workflow is a great Git workflow for teams transitioning from SVN. Like Subversion, the Centralized Workflow uses a central repository to serve as the single point-of-entry for all changes to the project. Instead of
trunk, the default development branch is called
master and all changes are committed into this branch. This workflow doesn’t require any other branches besides
Transitioning to a distributed version control system may seem like a daunting task, but you don’t have to change your existing workflow to take advantage of Git. Your team can develop projects in the exact same way as they do with Subversion.
However, using Git to power your development workflow presents a few advantages over SVN. First, it gives every developer their own local copy of the entire project. This isolated environment lets each developer work independently of all other changes to a project - they can add commits to their local repository and completely forget about upstream developments until it's convenient for them.
Second, it gives you access to Git’s robust branching and merging model. Unlike SVN, Git branches are designed to be a fail-safe mechanism for integrating code and sharing changes between repositories. The Centralized Workflow is similar to other workflows in its utilization of a remote server-side hosted repository that developers push and pull form. Compared to other workflows, the Centralized Workflow has no defined pull request or forking patterns. A Centralized Workflow is generally better suited for teams migrating from SVN to Git and smaller size teams.
Developers start by cloning the central repository. In their own local copies of the project, they edit files and commit changes as they would with SVN; however, these new commits are stored locally - they’re completely isolated from the central repository. This lets developers defer synchronizing upstream until they’re at a convenient break point.
To publish changes to the official project, developers "push" their local
master branch to the central repository. This is the equivalent of
svn commit, except that it adds all of the local commits that aren’t already in the central
Initialize the central repository
最初に、誰かがサーバー上に中央リポジトリを作成しなければなりません。新規プロジェクトの場合は誰がそれを実行しても構いません。既存プロジェクトの場合は、既存の Git または SVN リポジトリをインポートする必要があります。
ssh user@host git init --bare /path/to/repo.git
Be sure to use a valid SSH username for
user, the domain or IP address of your server for
host, and the location where you'd like to store your repo for
/path/to/repo.git. Note that the
.git extension is conventionally appended to the repository name to indicate that it’s a bare repository.
Hosted central repositories
Central repositories are often created through 3rd party Git hosting services like Bitbucket Cloud or Bitbucket Server. The process of initializing a bare repository discussed above is handled for you by the hosting service. The hosting service will then provide an address for the central repository to access from your local repository.
Clone the central repository
Next, each developer creates a local copy of the entire project. This is accomplished via the
git clone command:
git clone ssh://user@host/path/to/repo.git
When you clone a repository, Git automatically adds a shortcut called
origin that points back to the “parent” repository, under the assumption that you'll want to interact with it further on down the road.
Make changes and commit
Once the repository is cloned locally, a developer can make changes using the standard Git commit process: edit, stage, and commit. If you’re not familiar with the staging area, it’s a way to prepare a commit without having to include every change in the working directory. This lets you create highly focused commits, even if you’ve made a lot of local changes.
git status # View the state of the repo git add <some-file> # Stage a file git commit # Commit a file</some-file>
既に説明したように、これらのコマンドはローカルなコミットを行うものであり、従って John は中央リポジトリで起こっていることを気にすることなくコミットを何度でも繰り返すことができます。これは、作業を単純で小規模な部分に分割する必要のある大規模フィーチャーの場合に特に有用な機能です。
Push new commits to central repository
Once the local repository has new changes committed. These change will need to be pushed to share with other developers on the project.
git push origin master
This command will push the new committed changes to the central repository. When pushing changes to the central repository, it is possible that updates from another developer have been previously pushed that contain code which conflict with the intended push updates. Git will output a message indicating this conflict. In this situation,
git pull will first need to be executed. This conflict scenario will be expanded on in the following section.
中央リポジトリはプロジェクトを公式に代表するものであり、そのコミット履歴は大切に扱わなければならず、また安易に改変してはなりません。開発者のローカルなコミット履歴が中央リポジトリと分岐状態にある場合は、中央リポジトリに誤書き込みを起こす可能性があるため、変更のプッシュは Git によって拒否されます。
この場合開発者は、フィーチャーを公開する前に最新の中央リポジトリをフェッチしてその先端にローカルな変更をリベースする必要があります。この操作は、「私の変更作業は皆が変更を完了したものをベースとして行いたい」と言うに等しいものです。その結果、履歴は従来の SVN ワークフローの場合と同様に完全に直線的になります。
If local changes directly conflict with upstream commits, Git will pause the rebasing process and give you a chance to manually resolve the conflicts. The nice thing about Git is that it uses the same
git status and
git add commands for both generating commits and resolving merge conflicts. This makes it easy for new developers to manage their own merges. Plus, if they get themselves into trouble, Git makes it very easy to abort the entire rebase and try again (or go find help).
Let’s take a general example at how a typical small team would collaborate using this workflow. We’ll see how two developers, John and Mary, can work on separate features and share their contributions via a centralized repository.
In his local repository, John can develop features using the standard Git commit process: edit, stage, and commit.
Remember that since these commands create local commits, John can repeat this process as many times as he wants without worrying about what’s going on in the central repository.
Meanwhile, Mary is working on her own feature in her own local repository using the same edit/stage/commit process. Like John, she doesn’t care what’s going on in the central repository, and she really doesn’t care what John is doing in his local repository, since all local repositories are private.
Once John finishes his feature, he should publish his local commits to the central repository so other team members can access it. He can do this with the
git push command, like so:
git push origin master
origin is the remote connection to the central repository that Git created when John cloned it. The
master argument tells Git to try to make the
master branch look like his local
master branch. Since the central repository hasn’t been updated since John cloned it, this won’t result in any conflicts and the push will work as expected.
John がローカルな変更を首尾よく中央リポジトリに公開した後で Mary がフィーチャーのプッシュを試みるとどうなるでしょうか。全く同様にプッシュコマンドを使用することはできます:
git push origin master
error: failed to push some refs to '/path/to/repo.git' hint: Updates were rejected because the tip of your current branch is behind hint: its remote counterpart. Merge the remote changes (e.g. 'git pull') hint: before pushing again. hint: See the 'Note about fast-forwards' in 'git push --help' for details.
この機能により、Mary が中央リポジトリに対して誤書き込みを起こすことが防止できます。Mary は、John による変更をローカルリポジトリにプルしてそれにローカルな変更を統合し、再度プッシュを試みなければなりません。
Mary が John のコミットの先端にリベースします
Mary can use
git pull to incorporate upstream changes into her repository. This command is sort of like
svn update—it pulls the entire upstream commit history into Mary’s local repository and tries to integrate it with her local commits:
git pull --rebase origin master
--rebase option tells Git to move all of Mary’s commits to the tip of the
master branch after synchronising it with the changes from the central repository, as shown below:
Rebasing works by transferring each local commit to the updated
master branch one at a time. This means that you catch merge conflicts on a commit-by-commit basis rather than resolving all of them in one massive merge commit. This keeps your commits as focused as possible and makes for a clean project history. In turn, this makes it much easier to figure out where bugs were introduced and, if necessary, to roll back changes with minimal impact on the project.
Mary と John が作業しているフィーチャーが無関係なものであった場合は、リベースの際に競合が起こる可能性は小さいものと考えられます。それでも競合が起こった場合、リベース動作は処理中のコミットで停止され、次のようなメッセージと関連情報を表示します
CONFLICT (content): Merge conflict in <some-file>
The great thing about Git is that anyone can resolve their own merge conflicts. In our example, Mary would simply run a
git status to see where the problem is. Conflicted files will appear in the Unmerged paths section:
# マージされていないパス: # ("git reset HEAD <some-file>..." を使用してアンステージします) # (必要に応じて "git add/rm <some-file>..." を使用し解決状況にマークを付けます) # # 両方とも修正済み: <some-file>
Then, she’ll edit the file(s) to her liking. Once she’s happy with the result, she can stage the file(s) in the usual fashion and let
git rebase do the rest:
git add <some-file> git rebase --continue
If you get to this point and realize and you have no idea what’s going on, don’t panic. Just execute the following command and you’ll be right back to where you started:
git rebase --abort
git push origin master
ここまで説明してきたように、少数の Git コマンドのみを使用して Subversion を利用した従来型の開発環境を踏襲することが可能です。これは、SVN からの移行過程にあるチームにとっては非常に有用ですが、これではGit の分散的特質を活用できません。
The Centralized Workflow is great for small teams. The conflict resolution process detailed above can form a bottleneck as your team scales in size. If your team is comfortable with the Centralized Workflow but wants to streamline its collaboration efforts, it's definitely worth exploring the benefits of the Feature Branch Workflow. By dedicating an isolated branch to each feature, it’s possible to initiate in-depth discussions around new additions before integrating them into the official project.
Other common workflows
The Centralized Workflow is essentially a building block for other Git workflows. Most popular Git workflows will have some sort of centralized repo that individual developers will push and pull from. Below we will briefly discuss some other popular Git workflows. These extended workflows offer more specialized patterns in regard to managing branches for feature development, hot fixes, and eventual release.
Feature Branching is a logical extension of Centralized Workflow. The core idea behind the Feature Branch Workflow is that all feature development should take place in a dedicated branch instead of the
master branch. This encapsulation makes it easy for multiple developers to work on a particular feature without disturbing the main codebase. It also means the
master branch should never contain broken code, which is a huge advantage for continuous integration environments.
The Gitflow Workflow was first published in a highly regarded 2010 blog post from Vincent Driessen at nvie. The Gitflow Workflow defines a strict branching model designed around the project release. This workflow doesn’t add any new concepts or commands beyond what’s required for the Feature Branch Workflow. Instead, it assigns very specific roles to different branches and defines how and when they should interact.
The Forking Workflow is fundamentally different than the other workflows discussed in this tutorial. Instead of using a single server-side repository to act as the “central” codebase, it gives every developer a server-side repository. This means that each contributor has not one, but two Git repositories: a private local one and a public server-side one.
There is no one size fits all Git workflow. As previously stated, it’s important to develop a Git workflow that is a productivity enhancement for your team. In addition to team culture, a workflow should also complement business culture. Git features like branches and tags should complement your business’s release schedule. If your team is using task tracking project management software you may want to use branches that correspond with tasks in progress. In addition, some guidelines to consider when deciding on a workflow are:
The longer a branch lives separate from the production branch, the higher the risk for merge conflicts and deployment challenges. Short-lived branches promote cleaner merges and deploys.
Minimize and simplify reverts
It’s important to have a workflow that helps proactively prevent merges that will have to be reverted. A workflow that tests a branch before allowing it to be merged into the
master branch is an example. However, accidents do happen. That being said, it’s beneficial to have a workflow that allows for easy reverts that will not disrupt the flow for other team members.
Match a release schedule
A workflow should complement your business’s software development release cycle. If you plan to release multiple times a day, you will want to keep your
master branch stable. If your release schedule is less frequent, you may want to consider using Git tags to tag a branch to a version.
In this document we discussed Git workflows. We took an in-depth look at a Centralized Workflow with practical examples. Expanding on the Centralized Workflow we discussed additional specialized workflows. Some key takeaways from this document are:
- There is no one-size-fits-all Git workflow
- A workflow should be simple and enhance the productivity of your team
- Your business requirements should help shape your Git workflow
To read about the next Git workflow check out our comprehensive breakdown of the Feature Branch Workflow.