Why do we use Git?
- Git is a first-class distributed version control system, so we use it to keep a record of changes to avoid loss-of-work and to appropriately explain/document changes as projects develop.
- Git history, which is composed of “snapshots” of the source code, can be used to go back at any point in time, which leads to reproducible science.
This web site has lots of tutorials on Git; see the left sidebar.
One of the most powerful things Git can do is rewrite a series of patches after the fact. This is a powerful technique worth learning. There are many guides available here and elsewhere:
- Modifying your commits section of the Contributing to a plugin page
- Advanced topic branch editing section of the Git topic branches tutorial
- Rewriting history tutorial from Atlassian
- 7.6 Git Tools - Rewriting History from the Git book
- About Git rebase article from GitHub
- Using Git rebase article from GitHub
Git for dummies
What is Git?
Git is a distributed Version Control System (“VCS”), or Source Code Management tool (“SCM”). The purpose is to track the development of source code through time, along with useful side information: Who did it? When? Why?
Principles of Git
The source code lives in a working directory, where you have tracked files (i.e. files which are version-controlled) and untracked files (for example, .class files; they are generated from the source code, so they are uninteresting).
The timeline of the development is made up of revisions (commits in Git speak) of the source code. Each commit knows about its predecessors. In the common case, commits form something like a perl of revisions. By comparing a commit to its parent commit (i.e. comparing a revision to the previous one), one can see the changes introduced by the “child” commit.
Commits contain the names and contents of the tracked files, information about the author (and the committer, who does not need to be the author), the time of the commit, and a commit message where the author should say what the reasons for the changes are.
To make a new commit, you first tell Git what changed contents you want to be part of the new commit by adding (or staging) them, and then performing the commit itself, which will open an editor in which you can type the commit message.
All Git-specific things, such as revisions, are stored in the repository, which lives in the subdirectory .git/ in your project’s root directory.
To collaborate with others, there are also remote repositories from where you can clone, and with which you can synchronize by fetching and pushing.
Why do I need Git?
Git makes it not only to record what you did, it makes it easy to follow the development of other developers, and integrate those changes (merge in Git speak).
Even if we are all brilliant developers, from time to time the program appears to be completely broken by our work. Git can tell you which changes you made relative to the current commit—quickly!
- Study the history of a project to understand not only what the developers did when, but also why by reading their commit messages (i.e., the log of changes).
- Recover any past revision, such as when the latest version suffers from a bug (i.e., a regression) not present in an older version.
- Easily find the commit which introduced such a regression.
- Ensure the code and history is never lost, even if your machine fails—anyone with a clone of the repository has a copy of the entire history.
- Work on multiple new features or bug-fixes simultaneously, easily organizing and switching between them using branches.
The only negative to Git is the activation barrier of learning it. Once you become proficient, Git is a huge asset to any development project.
If you feel comfortable in using the commandline but you are still not familiar with Git the Git workshop page might be a perfect entry point for you.
Essential Git commands
Except for the clone command, this follows the common workflow:
git diff HEADwill show you the changes of your working directory relative to the current commit.
git add <file>...tells Git that you want the next commit to contain the current version of that file/these files.
git committells Git that you want to commit a new revision with all staged changes.
git fetch originsynchronizes the local copy of the default remote repository.
git merge origin/masterintegrates the default remote repository’s changes into your local repository and working directory (never do this when you have uncommitted changes!).
git push origin HEADpublish your changes in the remote repository.
Initially, you need to start from somewhere:
git clone <URL>will make a new local repository and initialize it from a remote one.
See also the more verbose Git mini howto.
Substantial differences to other version control systems
If you know CVS or Subversion, you are in for some surprises:
- In Git, each repository is local. To publish your changes, you need to have a remote repository, too, and push your work there.
- In Git, branches are easy and fast.
- In Git, you
git addcontent, not files. In other words, when the file README is already tracked,
git add READMEwill tell Git that you want the changes in said file to be part of the next commit.
- In Git, you never, ever try to integrate remote changes into an uncommitted state. In other words, if you have uncommitted changes, you always commit them before calling
git fetch origin; git merge origin/master.