Git

A distributed version control system

19-Mar-13

Version control systems

Version control (or revision control, or source control) is all about managing multiple versions of documents, programs, web sites, etc.

Almost all real projects use some kind of version control Essential for team projects, but also very useful for individual projects

Some well-known version control systems are CVS, Subversion, Mercurial, and Git

CVS and Subversion use a central repository; users check out files, work on them, and check them in Mercurial and Git treat all repositories as equal

Distributed systems like Mercurial and Git are newer and are gradually replacing centralized systems like CVS and Subversion
2

Why version control?

For working by yourself:

Gives you a time machine for going back to earlier versions Gives you great support for different versions (standalone, web app, etc.) of the same basic project Greatly simplifies concurrent work, merging changes Any company with a clue uses some kind of version control Companies without a clue are bad places to work

For working with others:

For getting an internship or job:

Why Git?

Git has many advantages over earlier systems such as CVS and Subversion

More efficient, better workflow, etc. See the literature for an extensive list of reasons Of course, there are always those who disagree I like Mercurial better Same concepts, slightly simpler to use In my (very limited) experience, the Eclipse plugin is easier to install and use Much less popular than Git
4

Best competitor: Mercurial

Download and install Git

There are online materials that are better than any that I could provide Heres the standard one: http://git-scm.com/downloads Heres one from StackExchange: http://stackoverflow.com/questions/315911/git-for-beginners-thedefinitive-practical-guide#323764 Note: Git is primarily a command-line tool I prefer GUIs over command-line tools, but The GIT GUIs are more trouble than they are worth (YMMV)

Introduce yourself to Git

Enter these lines (with appropriate changes):

git config --global user.name "John Smith" git config --global user.email [email protected]

You only need to do this once

If you want to use a different name/email address for a particular project, you can change it for just that project

cd to the project directory Use the above commands, but leave out the --global

Create and fill a repository

3. Type in git add .

This adds all your current files to the repository

4. Type in git commit m "Initial commit"

Clone a repository from elsewhere

git clone URL git clone URL mypath

All repositories are equal

But you can treat some particular repository (such as one on Github) as the master directory

Typically, each team member works in his/her own repository, and merges with other repositories as appropriate

The repository

Your top-level working directory contains everything about your project

The working directory probably contains many subdirectoriessource code, binaries, documentation, data files, etc. One of these subdirectories, named .git, is your repository

At any time, you can take a snapshot of everything (or selected things) in your project directory, and put it in your repository

This snapshot is called a commit object The commit object contains (1) a set of files, (2) references to the parents of the commit object, and (3) a unique SHA1 name Commit objects do not require huge amounts of memory

You can work as much as you like in your working directory, but the repository isnt updated until you commit something
9

init and the .git repository

When you said git init in your project directory, or when you cloned an existing project, you created a repository

The repository is a subdirectory named .git containing various files The dot indicates a hidden directory You do not work directly with the contents of that directory; various git commands do that for you You do need a basic understanding of what is in the repository

10

Making commits

You do your work in your project directory, as usual If you create new files and/or folders, they are not tracked by Git unless you ask it to do so

git add newFile1 newFolder1 newFolder2 newFile2

Committing makes a snapshot of everything being tracked into your repository

A message telling what you have done is required git commit m Uncrevulated the conundrum bar git commit This version opens an editor for you the enter the message To finish, save and quit the editor One line containing the complete summary If more than one line, the second line must be blank
11

Format of the commit message

Commits and graphs

A commit is when you tell git that a change (or addition) you have made is ready to be included in the project When you commit your change to git, it creates a commit object

A commit object represents the complete state of the project, including all the files in the project The very first commit object has no parents Usually, you take some commit object, make some changes, and create a new commit object; the original commit object is the parent of the new commit object

Hence, most commit objects have a single parent The new commit object has two parents

You can also merge two commit objects to form a new one

Hence, commit objects form a directed graph

Git is all about using and manipulating this graph

Working with your own repository

A head is a reference to a commit object The current head is called HEAD (all caps) Usually, you will take HEAD (the current commit object), make some changes to it, and commit the changes, creating a new current commit object

This results in a linear graph: A B C HEAD

You can also take any previous commit object, make changes to it, and commit those changes

This creates a branch in the graph of commit objects

You can merge any previous commit objects

Commit messages

In git, Commits are cheap. Do them often. When you commit, you must provide a one-line message stating what you have done

Terrible message: Fixed a bunch of things Better message: Corrected the calculation of median scores

Commit messages can be very helpful, to yourself as well as to your team members You cant say much in one line, so commit often

14

Choose an editor

When you commit, git will require you to type in a commit message For longer commit messages, you will use an editor The default editor is probably vim To change the default editor:

git config --global core.editor /path/to/editor

You may also want to turn on colors:

git config --global color.ui auto

15

Working with others

All repositories are equal, but it is convenient to have one central repository in the cloud Heres what you normally do:

Download the current HEAD from the central repository Make your changes Commit your changes to your local repository Check to make sure someone else on your team hasnt updated the central repository since you got it Upload your changes to the central repository It is your responsibility to merge your two versions

If the central repository has changed since you got it:

This is a strong incentive to commit and upload often!

Git can often do this for you, if there arent incompatible changes
16

Typical workflow

git pull remote_repository

Get changes from a remote repository and merge them into your own repository See what Git thinks is going on Use this frequently!

git status

Work on your files (remember to add any new ones) git commit m What I did git push

17

Multiple versions
Initial commit Second commit Third commit Bob gets a copy

Fourth commit
Bobs commit Merge

18

Keeping it simple

If you:

Make sure you are current with the central repository Make some improvements to your code Update the central repository before anyone else does

Then you dont have to worry about resolving conflicts or working with multiple branches

All the complexity in git comes from dealing with these

Therefore:

Make sure you are up-to-date before starting to work Commit and update the central repository frequently

If you need help: https://help.github.com/

The End
When I say I hate CVS with a passion, I have to also say that if there are any SVN [Subversion] users in the audience, you might want to leave. Because my hatred of CVS has meant that I see Subversion as being the most pointless project ever started. The slogan of Subversion for a while was "CVS done right", or something like that, and if you start with that kind of slogan, there's nowhere you can go. There is no way to do CVS right.
--Linus Torvalds, as quoted in Wikipedia
20