This workshop has been written with a focus on using Git Bash on Windows or bash shell, alternative setups will likely work without issue.
Before being able to commit files to any of the git repositories locally git requires two pieces of information, your name and email address. This is used to identify you as the person who commited the code. These settings can be set on a per repository basis or globally (for all repositories) using the following commands.
git config --global user.name "Your Name"
git config --global user.email "[email protected]"
When working with remote repositories there are two methods of connecting to the remote git server, using HTTPS and using SSH. When pushing to a git server when connecting with HTTPS you will be asked for your username and password for the server, this can be tedious to repeatedly enter when simply uploading your changes, by using SSH keys you no longer need to enter your credentials. This approach uses OpenSSH, it is also possible to use alternatives but this is not covered here.
cd ~/.ssh
ls -a
If you see the file id_rsa.pub then you already have an SSH key and can skip to Section 3.
We will be using the ssh-keygen tool to create a new SSH key with the RSA cryptosystem, your email is also required in this step.
ssh-keygen -t rsa -C "[email protected]"
The first prompt will ask where the SSH key should be stored, the default location is preferable so that Git will be able to find the key without issue, simply press enter.
The you will be prompted to enter a passphrase (and confirmation) required to use the SSH key, if you are working on a trusted system just press enter twice so that no passphrase is required.
You should now have two new files in your ~/.ssh directory: your private key id_rsa which must be kept safe, and public key id_rsa.pub which can be provided to a Git server to enable SSH authentication.
This step depends on who your chosen Git hosting provider is, for sites such as GitHub and Bitbucket and self hosted installations of GitLab has a settings section which allows you to add your public key to the server quickly and easily.
Solving merge conflicts can be a difficult and tedious task which can be made much easier with a visual merge tool such as meld. meld is available for Windows as well as Unix based systems. Other merge tools are available and are usually a personal preference.
In order to setup Git to take advantage of an installed merge tool on Windows we need to perform the following configuration. Take care to replace the path to your merge tool with the correct path to the exectuable.
git config --global mergetool.meld.path "C:\Program Files (x86)\Meld\bin\meld\meld.exe"
git config --global merge.tool meld
It is a common problem to type a Git command incorrectly and recieve a message such as the following.
git: 'stauts' is not a git command. See 'git --help'.
Did you mean this?
status
If this is a common problem for you then it is possible to enable a helper setting which will make an attempt to guess which command you were attempting to use.
git config --global help.autocorrect 1
When working on cross platforms projects, or using the same hard drive in combination with different operating systems, problems arise with the way text is stored on disk. Specifically Windows and Unix like systems use different characters to define the end of a line, Windows uses CRLF and Unix like systems use LR.
The following setting tells Git to automatically deal with line endings globally.
git config --global core.autocrlf true
If you checkout a commit before setting this option, you may be asked to commit changes to files you have not changed. In this situation the easist apporach is to remove the files which were checked out with the wrong line endings then perform a hard reset on the repository, this will checkout the files with the correct line endings. Be careful not to loss any changes you have made.
When working on projects with many branches it can be confusing to determine the order of commits and which branches were merges and where. With the following alias it is possible show this information on the command line.
First we need to create an alias called tree (you can choose your own alias name) which will be the shorthand for the command. Note that this is a long command which is why having an alias is preferable.
git config --global alias.tree "log --graph --abbrev-commit --decorate --date=relative --format=format:'%C(bold blue)%h%C(reset) - %C(bold green)(%ar)%C(reset) %C(white)%s%C(reset) %C(dim white)- %an%C(reset)%C(bold yellow)%d%C(reset)' --all"
Once this is done we can use the alias with the command git tree.
To start using Git first we need to be in a repository. We will initialize a new
one, to do this we will need to run git init. To follow this tutorial, you
will need to be in a terminal with Git installed or using the Windows Git Bash.
To get started run the following commands:
mkdir git-tutorial
cd git-tutorial
git init
You can also initialize projects using git init <directory>, this will create
the git repository in the specified repository. Running either of the two
commands should produce the following output from Git:
Initialized empty Git repository in .git/
Many open source projects already use git (github). To work
on any of them you need to clone the project, which creates a copy of the
repository. You can do this with git clone <url> but for now we will move on.
Now that we have initialized the repository, we can check the status of the
repository. If you ever wish to see the state of the working directory or
staging area just run git status, let's do this now.
On branch master
Initial commit
nothing to commit (create/copy files and use "git add" to track)
Now that we can get git to tell us the status of the repository, lets create
some content for git to keep track of.
echo "foo" > foo.txt
echo "bar" > bar.txt
By adding files to the working directory of the repository we have changes its
status. Lets check how the change the status of the repository, run git status again
On branch master
Initial commit
Untracked files:
(use "git add <file>..." to include in what will be committed)
bar.txt
foo.txt
nothing added to commit but untracked files present (use "git add" to track)
Adding these files changed the working directory of our repository, we can use
the git add <file> command to add these changes to the staging area. This
is an area which tracks which changes to the working directory will be in the
next commit. The staging area doesn't change the repository as these changes
won't actually be recorded until you commit them. This is one of Git's more
unique features, it allows you to craft your commits splitting changes into
relevant groups, instead of just saving all of your changes since the last
commit.
To add your changes run git add ., this command adds all changes to files in
the repository (but not deleted files). Now check the result by running
git status again. You should see the following output:
On branch master
Initial commit
Changes to be committed:
(use "git rm --cached <file>..." to unstage)
new file: bar.txt
new file: foo.txt
git add can also be used to add directories in the same manner as files, it
is simple git add <directory>.
Now that you have some changes in the staging area, we can go ahead and create
your first commit: git commit -m "Initial commit". The -m tells git commit
that you are giving it a message on the command line (without it this will open
the default text editor set in your .gitconfig, but we will ignore this
detail).
Now our changes that were in the staging area have been stored as a commit. If
you run git status again you will see the following message.
On branch master
nothing to commit, working directory clean
If you make a mistake when writing a commit message use git commit --amend, this
will allow you to edit the commit message.
All version control systems including Git show a projects commit history. The
git log command displays the commited snapshots, showing your commit history.
If you run git log you should see similar output:
Author: Sean Jones <[email protected]>
Date: Sat May 10 15:38:15 2014 +0100
Initial commit
Let's create some more commits so we can see what git log looks like when
populated with more than one commit.
echo "foobar" > foobar.txt
echo "foo is not bar" > foo.txt
echo "bar is not foo" > bar.txt
Instead of using git add ., use git add -p You should be prompted with output
similar to this:
diff --git a/bar.txt b/bar.txt
index 5716ca5..02626e7 100644
--- a/bar.txt
+++ b/bar.txt
@@ -1 +1,2 @@
bar
+ is not foo
Stage this hunk [y,n,q,a,d,/,e,?]?
Enter y and then git commit -m "Add changes to foo & bar", finally do the
following:
git add foobar.txt
git commit -m "Add foobar.txt"
git log
Hopefully, you should see similar output:
commit bff897c9d3d27cfee74516935500388b417f1c11
Author: Sean Jones <[email protected]>
Date: Sat May 10 22:50:14 2014 +0100
Add foobar.txt
commit 0c5275028bd818395de035aa1733a3f2889532e5
Author: Sean Jones <[email protected]>
Date: Sat May 10 22:45:13 2014 +0100
Add changes to foo & bar
commit 36a24afa1548fb767439f36e3e8bee8458f571ee
Author: Sean Jones <[email protected]>
Date: Sat May 10 22:23:27 2014 +0100
Initial commit
This output is fairly simple to understand, every commit has a 40 character SHA1
hash, which is a checksum for the commit and a unique ID. Running git log --decorate will show some extra output.
commit bff897c9d3d27cfee74516935500388b417f1c11 (HEAD, master)
Author: Sean Jones <[email protected]>
Date: Sat May 10 22:50:14 2014 +0100
Add foobar.txt
This extra output (HEAD, master) indicates that the latest commit bff897c
is the current commit and that it is the tip of the master branch. You can use
the HEAD reference and branch names with Git commands that require a commit
id.
Try some of the following commands, see the different output they produce:
git log -n <limit>
git log --oneline
git log --stat
git log -p
git log --author"<pattern>"
git log --grep="<pattern>"
git log <since>..<until>
git log <file>
git log --graph --decorate --oneline
git log --graph --stat -p --decorate
Using git log <since>..<until> command allows you to seen between the
specified range. For example git log HEAD~2..HEAD:
commit bff897c9d3d27cfee74516935500388b417f1c11
Author: Sean Jones <[email protected]>
Date: Sat May 10 22:50:14 2014 +0100
Add foobar.txt
commit 0c5275028bd818395de035aa1733a3f2889532e5
Author: Sean Jones <[email protected]>
Date: Sat May 10 22:45:13 2014 +0100
Add changes to foo & bar
The ~ character is used for referencing the relative parent of the specified
commit. HEAD1 being the commit previous to HEAD and bff897c1 being 0c52750,
the commit before bff897c. For more information about git log visit [here]
(http://git-scm.com/docs/git-log)
Now that we have some history in the repository, let's use it for what it is
mean to do, store file versions safely. Run the following command cat foo.txt
you should see:
foo is not bar
If you wanted to use the original file, or just inspect what it contained you
can run git checkout HEAD~2 foo.txt is now in original state. Running cat foo.txt should yeild the original contents:
foo
git status will show that foo.txt has been modified compared to HEAD (the
current commit).
To redo the changes made to foo.txt there are several options:
git checkout foo.txt
git reset HEAD foo.txt
git reset --hard HEAD
Try a command out then run git status to confirm it has been restored to the
latest version (the second command will not full restore the file, the third
command will). Use git checkout HEAD~2 to set foo.txt back to the first
version and try the next command. If foo.txt has been restored to the latest
version you should see the following:
On branch master
nothing to commit, working directory clean
The first command in the list, creates a foo.txt from the snapshot at HEAD,
replacing the contents of the file. The second command resets any changes that
were in the staging area for that file, all the changes are still present in
the file. The last command will restore all tracked files in the staging area
and the working directory to those in the HEAD commit.
When making changes to your files, occasionally you might want to remove the the change in the case where you added a mistake. Run the following commands:
echo "foobar is neither foo nor bar" > foobar.txt
git add foobar.txt
git status
Now the change to foobar.txt has been added to the staging area, check with
git status to confirm:
On branch master
Changes to be committed:
(use "git reset HEAD <file>..." to unstage)
modified: foobar.txt
Now to remove this change to foobar.txt from the staging area, you can use
git reset HEAD <file> to remove changes to files from the staging area.
To remove the change to foobar.txt run git reset HEAD foobar.txt.
echo "foobar is neither foo nor bar" > foobar.txt
git add foobar.txt
git commit -m "Add changes to foobar"
Say removing foobar.txt was a mistake, we can reset the index of the
repository, essentially rewinding history. Let's try it out:
git rm foobar.txt
git commit -m "Remove foobar"
To rewind this commit we can run the following command git reset HEAD~1,
this will reset the index to the previous commit.
WARNING: Only do this on commits to your local repository. DO NOT do this when a commit has come from a remote repository, it LITTERALLY rewrites history.
The correct way to undo a commit is to use git revert. This command creates
a commit which will undo the changes of a specified commit. Let's try it git revert HEAD, reverting HEAD (the lastest commit) should result in the
following git log:
commit 5b4c1b41f75673407c6965466b9e7dbe96d074d4
Author: Sean Jones <[email protected]>
Date: Sun May 11 09:28:22 2014 +0100
Revert "removed foobar"
This reverts commit d3840ad88286808062169c1a201316a00d85f880.
commit d3840ad88286808062169c1a201316a00d85f880
Author: Sean Jones <[email protected]>
Date: Sun May 11 03:04:09 2014 +0100
removed foobar
We can see better what has happened by using git log -p --oneline HEAD~2..HEAD
5b4c1b4 Revert "removed foobar"
diff --git a/foobar.txt b/foobar.txt
new file mode 100644
index 0000000..4768904
--- /dev/null
+++ b/foobar.txt
@@ -0,0 +1 @@
+foobar is neither foo nor bar
<inserted space for clarity>
d3840ad removed foobar
diff --git a/foobar.txt b/foobar.txt
deleted file mode 100644
index 4768904..0000000
--- a/foobar.txt
+++ /dev/null
@@ -1 +0,0 @@
-foobar is neither foo nor bar
We can see in these diffs that git revert HEAD has create a commit that does the
exact opposite of the previous commit, it has added foobar.txt back. This is the
correct way to undo commits if they have been pushed to public projects.
The git clone command was previously mentioned in [Creating Projects]
(#creating-a-new-project). The git clone <url> command can be used to create exact
copies of repositories, either on your machine or remote locations.
Let's try cloning the repository that has been used as an example throughout this tutorial.
cd ..
git clone https://github.com/NeuralSandwich/git-tutorial-basic.git
Following output:
Cloning into 'git-tutorial-basic'...
remote: Counting objects: 15, done.
remote: Compressing objects: 100% (6/6), done.
remote: Total 15 (delta 3), reused 15 (delta 3)
Unpacking objects: 100% (15/15), done.
Checking connectivity... done.
Congradulations, if you so the following output then you have successfully created a clone of the repository.
Before the next section, we need to do some house keeping. Run the following
command rm -rf git-tutorial:
Remote repositories are normally stored as bare repositories. Bare repositories
do not have working directories. These repositories just store all the
information from inside the .git directory. They are set out like this:
.
├── branches
├── config
├── description
├── HEAD
├── hooks
├── info
├── objects
├── packed-refs
└── refs
5 directories, 4 files
Since, you will not have access to push changes to the repository you just cloned, you need to create a local bare repository to use for the rest of the tutorial.
git clone --bare git-tutorial-basic
This should produce the following output:
Cloning into bare repository 'git-tutorial-basic.git'...
done.
Let's go back to our repository and get it working with our bare repository. All of the following commands will work the same even if the repository was an actual remote repository. The only differene would be the URL to access it.
cd git-tutorial-basic
git remote set-url origin ../git-tutorial-basic.git
When cloning the repository Git will also set a remote connection called origin
This was set to pull from GitHub.com. git remote set-url <name> <url> This
command tells Git to change the URL of the remote connection origin to
../git-tutorial-basic.git we need to do this so we can use the new bare
repository as a remote connection.
git push -u origin master
This should produce similar output:
Counting objects: 4, done.
Delta compression using up to 4 threads.
Compressing objects: 100% (2/2), done.
Writing objects: 100% (3/3), 359 bytes | 0 bytes/s, done.
Total 3 (delta 0), reused 0 (delta 0)
To ../git-tutorial-basic.git
7a2bf50..cd10233 master -> master
git push send your commits to the remote repository. To get changes from the
remote server you can use git fetch and getting the changes and applying them
to the working directory you can use git pull
+Git's more popular feature is its support of light weight branching. A branch is +an abstract representation of an independent line of development. It allows you +to make changes to your code, as if you were being given new working directory, +staging area and project history. However you can instantly switch to others +which may have other development in them.
So far in this tutorial, you have been working exclusively one master, the
default branch name. To create and switch to a new branch there is two methods:
git branch <name>
git checkout <name>
Or.
git checkout -b <name>
Both of these create a new branch and switch the working directory to it. Creating branches for specific tasks such as bug fixes, adding features and refactoring code are common practise. It allows several lines of development to be carried out in parrallel without conflict.
Once a line of development is complete, it should be merged into the appropriate
branch and then deleted. This can be done using git branch -d <name> if a
branch has not been merged but you still wish to delete it use git branch -D <name>
Switching between branch is as simple as calling git checkout <name> this will
switch HEAD to the relevant snapshot.
To learn more about branch visit the great interactive LearnGitBranching tutorial by Peter M Cottle.
When working on projects it is common that extra files will be generated by the development process, whether these are related to the build system, temporary files generated by your application running, or just something that you like having around, it is useful to be able to ignore these files. The git method for not adding files to the repository easily it the .gitignore file. This file lives in the root of your repository but is not provided by default, creating one is easy.
touch .gitignore
To ignore a specific file you simply add the file name to .gitignore, however it is usually easier to ignore directories or types of files. The * signifies a wildcard, meaning it will match any string.
# Ignore a directory
images
# Ignore all .txt files
*.txt
# Ignore all .txt files in images
images/*.txt
And that's it, but remember that files which have already been commited to the repository will not be ignored.
The stash is a temporary place where local changes can stored whilst other operations are performed on the repository, such as pulling the lastest commits or changing branch before a commit. To stash changes is simple.
git stash
You should see the following if the stash was successful.
Saved working directory and index state WIP on master: <commit> <message>
Otherwise if you have not made any changes you will see the following, please make a change to a file in order to continue.
No local changes to save
Now that we have stashed some changes we can do whichever task for which the stash was required. It is possible to list all the current stashes.
git stash list
Which will show output in the following form.
stash@{0}: WIP on master: <commit> <message>
... possibly more stashes ...
Stashes are stored in a stack, so the latest stash will have the stash@{0} identifier, any older stashes will have the form stash@{n} where n is the position in the stack.
When the time comes when you want to unstash the changes there are two options, apply the changes or pop the changes which does an apply then a drop. We will start with apply.
git stash apply stash@{0}
A successful apply will show the output of a call to git status and your changes will be applied to the repository. If we call git stash list you will see that your stash is still avaialble.
The alternative is to use git stash pop which will apply the changes then drop the stash from the stash list. First we need to perform a hard reset to avoid a merge conflict.
git reset --hard HEAD
git stash pop
git stash list
If all goes well the stash should be applied to the working directory and the displayed list will no longer show the stash. Lets add the stash once again, then explore how to view the contents of a stash.
git stash
git stash list
To view the contents of a stash we can use git stash show stash@{0} to see the insertions and deletions, however it is usually more useful to see the changes in patch form.
git stash show -p stash@{0}
Finally if we decide we no longer require the changes containted within it. The stash can be dropped as follows.
git stash drop stash@{0}
git stash list
The majoritory of the time git can automatically merge files without problems, this gives learning how to deal with merge conflicts a lower priority especially when working on your own. However when a merge conflict does arise it can be a slow and painful experience to fix it properly. This section relies on having a visual merge tool setup on your system, we will be using meld which you should have setup if you followed the Mergetool section above.
First of all lets setup a situation where a merge conflict will occur, for this we need to have a file commited in out repository.
echo "Some text" > conflict.txt
git add conflict.txt
git commit -m "Add conflict.txt"
Now that we have our file in the repository tree we require an additional branch, switch to this branch and make a change to contents of conflict.txt. Once this is done we can commit these changes and change back to the master branch.
git branch conflict
git checkout conflict
echo "Some changed text" > conflict.txt
git commit -am "Changed conflicts.txt"
git checkout master
Back on the master branch now and the final change to conflict.txt before we can see how to solve a merge conflict.
echo "Another change" > conflict.txt
git commit -am "The conflicting change"
We are now in a situation where a merge conflict will occur when we attempt to merge the conflict branch into master as follows.
git merge conflict
This will output the following message
Auto-merging conflict.txt
CONFLICT (content): Merge conflict in conflict.txt
Automatic merge failed; fix conflicts and then commit the result.
This is usually a good time to view the status of the repository in order to see the extent of the merge conflict.
git status
Which should look like this.
On branch master
You have unmerged paths.
(fix conflicts and run "git commit")
Unmerged paths:
(use "git add <file>..." to mark resolution)
both modified: conflict.txt
no changes added to commit (use "git add" and/or "git commit -a")
So now git has told us that we need to resolve the merge conflict in the file conflict.txt we do this by invoking the merge tool we set up earlier.
git mergetool conflict.txt
Will prompt you with the following, just hit enter and a meld window will appear.
Merging:
conflict.txt
Normal merge conflict for 'conflict.txt':
{local}: modified file
{remote}: modified file
Hit return to start merge resolution tool (meld):
meld will now appear looking similar to the following image, here we have three views of the file conflict.txt in different states. On the left is the state of the file on the master branch, on the right is the state of the file in the conflict branch, and in the middle in the view we want to contain the resolved state. The arrows in the middle view show where the merge conflict has occured and which file the changes have come from.
In this case the chosen change is not really important so we can just click on the arrow pointing from the left view to the middle view, if both views contain important changes then you can edit the middle view just like a regular text file.
This screenshot shows the result of the previous action, as you can see we have chosen the master branches change and ignored the conflict branched change.
Finally we can exit meld and save the conflict.txt file.
We have successfully resolve the merge conflict, the final task is to commit the changes.
git commit
The commit message will look like this, there is no need to change it.
Merge branch 'conflict'
Conflicts:
conflict.txt
And your done, merge conflicts are no longer your enemy.
A patch is a set of changes in a portable format, pathces are useful for distributing changes when you do not have write access to a repository and are the basis of code review systems. Generating a patch is pretty simple when in a repository which has local changes.
git diff > ../a.patch
This command produces a differential between the most recent commit and the local changes which have been made, the second part of the command is piping > the output from git diff to the file ../a.patch, if we did not pipe this output it would be printed on the command prompt.
Now that we have a patch which holds the changes we can clean up the repository and then apply the patch. If we attempt to apply the patch to the respository without cleaning it, the command will abort because the changes are already there.
git apply ../a.patch
The majority of projects are built upon existing libraries, distributing these projects can prove difficult because these external dependencies must be available on the target system which makes building a problem. One approach to solving this problem is to supply the source code for the libraries you rely on with your own project, however the naive solution of copying the source code into your repository is hard to update and can lead to files being accidentally changed.
Submodules are the git approach to this problem, a submodule is an existing git repository that you reference in your own git repository. To add a submodule to your repository you simply need to the URL of the repository you want to use. For this example we will create a new repository and add the GitHub GLFW repository as a submodule.
mkdir glfw-example
cd glfw-example
git init
It is good practice, but not essential, to keep your submodule in a directory within your repository. Here we will store GLFW in the external repository although it is also common to use ThirdParty or similar.
mkdir external
Now we will add our submodule
git submodule add https://github.com/glfw/glfw.git external/glfw
Which results in the following output
Cloning into 'external/glfw'...
remote: Reusing existing pack: 13159, done.
remote: Total 13159 (delta 0), reused 0 (delta 0)
Receiving objects: 100% (13159/13159), 6.41 MiB | 1.41 MiB/s, done.
Resolving deltas: 100% (8082/8082), done.
Checking connectivity... done.
warning: LF will be replaced by CRLF in .gitmodules.
The file will have its original line endings in your working directory.
Now we can see that the GLFW library source code has been added to the external/glfw directory, but we are not finished. If we check the status of our repository we will see the following output.
On branch master
Initial commit
Changes to be committed:
(use "git rm --cached <file>..." to unstage)
new file: .gitmodules
new file: external/glfw
There are two entries to be added, the first is the file .gitmodules which contains a list of all the submodules which have been added to the project. The second is the submodule its self. Lets commit our changes.
git commit -m "Add glfw submodule"
Removing a submodule from a repository can be tricky, so here is the process. First we need to edit the .gitmodules file and remove the following entry.
[submodule "external/glfw"]
path = external/glfw
url = https://github.com/glfw/glfw.git
And then stage .gitmodules
git add .gitmodules
Then we need to edit the file .git/config removing the following entry.
[submodule "external/glfw"]
url = https://github.com/glfw/glfw.git
Now we need to remove the submodule from the repository tree, and also remove the module reference.
git rm --cached external/glfw
rm -rf .git/modules/external/glfw
Now we can commit the changes made to remove the submodule
git commit -m "Remove submodule external/glfw"
Finally we can now remove the physical files from the file system and view the state of the repository.
rm -rf external/glfw
git status
Which results in the following
On branch master
nothing to commit, working directory clean
Git rebasing, it a very powerful tool. It can be be used to rewrite your
repositories history, for better or worse. git rebase does this by copying
commits from a branch and copying them to new base commit.
Let's git rebase <base> to understand how it works.
cd ../
git clone https://github.com/NeuralSandwich/git-tutorial-rebasing.git
cd git-tutorial-rebasing
If you run git log you will see that this repository has an extra commit on
the master branch compare to the last repository
commit f172eb1d904be0621d4152f2bdde0b5d89433152
Author: Sean Jones <[email protected]>
Date: Mon May 12 07:48:38 2014 +0100
Add barfoo.txt
commit 7a2bf50939c737a2bb5fa47d2a2763b692adda79
Author: Sean Jones <[email protected]>
Date: Sun May 11 09:48:50 2014 +0100
Remove foobar
By running git branch -v we can see there is more than one branch.
bring-back-foobar b24dcb0 Revert "Remove foobar"
* master f172eb1 Add barfoo.txt
It appears both have different tip commits. git log bring-back-foobar reveals
that the history is also different between these branchs:
commit b24dcb0b0da31ba7cfe84dde0298b35f0e0213d4
Author: Sean Jones <[email protected]>
Date: Mon May 12 08:02:25 2014 +0100
Revert "Remove foobar"
This reverts commit 7a2bf50939c737a2bb5fa47d2a2763b692adda79.
commit 7a2bf50939c737a2bb5fa47d2a2763b692adda79
Author: Sean Jones <[email protected]>
Date: Sun May 11 09:48:50 2014 +0100
Remove foobar
If we wanted to merge bring-back-foobar into master, we would end up with a
non-linear history. git log --graph --oneline
* fcfb589 Merge branch 'bring-back-foobar'
|\
| * b24dcb0 Revert "Remove foobar"
* | f172eb1 Add barfoo.txt
|/
* 7a2bf50 Remove foobar
* fff6840 Add changes to foobar
* 2042f6c Add foobar.txt
* 9085b36 Add changes to foo & bar
* 2f7907c Initial commit
While this in itself is not a bad thing, for the trivial change that this branch brings, it doesn't need to introduce a mess into the history. Let's fix this so when we do merge, we can get a nice linear history.
git co bring-back-foobar
git rebase master
Now if we look at the history of this branch we will see a nice linear series of
commits. git log --graph --oneline
* da75a4c Revert "Remove foobar"
* f172eb1 Add barfoo.txt
* 7a2bf50 Remove foobar
* fff6840 Add changes to foobar
* 2042f6c Add foobar.txt
* 9085b36 Add changes to foo & bar
* 2f7907c Initial commit
Now if we merge bring-back-foobar into master, it will fast-forward and create the nice linear history we can see above.
And that is rebasing, for more challenging uses try out the great interactive LearnGitBranching tutorial by Peter M Cottle. It contains exercises in rebasing, branching and merging that will help you better understand the process.