blob: 9c4c41df5af03e53f140cab6f9e680e0b3b2a91f [file] [log] [blame]
Git User's Manual
This manual is designed to be readable by someone with basic unix
command-line skills, but no previous knowledge of git.
Chapter 1 gives a brief overview of git commands, without any
explanation; you may prefer to skip to chapter 2 on a first reading.
Chapters 2 and 3 explain how to fetch and study a project using
git--the tools you'd need to build and test a particular version of a
software project, to search for regressions, and so on.
Chapter 4 explains how to do development with git, and chapter 5 how
to share that development with others.
Further chapters cover more specialized topics.
Comprehensive reference documentation is available through the man
pages. For a command such as "git clone", just use
$ man git-clone
Git Quick Start
This is a quick summary of the major commands; the following chapters
will explain how these work in more detail.
Creating a new repository
From a tarball:
$ tar xzf project.tar.gz
$ cd project
$ git init
Initialized empty Git repository in .git/
$ git add .
$ git commit
From a remote repository:
$ git clone git://
$ cd project
Managing branches
$ git branch # list all branches in this repo
$ git checkout test # switch working directory to branch "test"
$ git branch new # create branch "new" starting at current HEAD
$ git branch -d new # delete branch "new"
Instead of basing new branch on current HEAD (the default), use:
$ git branch new test # branch named "test"
$ git branch new v2.6.15 # tag named v2.6.15
$ git branch new HEAD^ # commit before the most recent
$ git branch new HEAD^^ # commit before that
$ git branch new test~10 # ten commits before tip of branch "test"
Create and switch to a new branch at the same time:
$ git checkout -b new v2.6.15
Update and examine branches from the repository you cloned from:
$ git fetch # update
$ git branch -r # list
$ git checkout -b masterwork origin/master
Fetch a branch from a different repository, and give it a new
name in your repository:
$ git fetch git:// theirbranch:mybranch
$ git fetch git:// v2.6.15:mybranch
Keep a list of repositories you work with regularly:
$ git remote add example git://
$ git remote # list remote repositories
$ git remote show example # get details
* remote example
URL: git://
Tracked remote branches
master next ...
$ git fetch example # update branches from example
$ git branch -r # list all remote branches
Exploring history
$ gitk # visualize and browse history
$ git log # list all commits
$ git log src/ # ...modifying src/
$ git log v2.6.15..v2.6.16 # v2.6.16, not in v2.6.15
$ git log master..test # branch test, not in branch master
$ git log test..master # branch master, but not in test
$ git log test...master # one branch, not in both
$ git log -S'foo()' # ...where difference contain "foo()"
$ git log --since="2 weeks ago"
$ git log -p # show patches as well
$ git show # most recent commit
$ git diff v2.6.15..v2.6.16 # diff between two tagged versions
$ git diff v2.6.15..HEAD # diff with current head
$ git grep "foo()" # search working directory for "foo()"
$ git grep v2.6.15 "foo()" # search old tree for "foo()"
$ git show v2.6.15:a.txt # look at old version of a.txt
Search for regressions:
$ git bisect start
$ git bisect bad # current version is bad
$ git bisect good v2.6.13-rc2 # last known good revision
Bisecting: 675 revisions left to test after this
# test here, then:
$ git bisect good # if this revision is good, or
$ git bisect bad # if this revision is bad.
# repeat until done.
Making changes
Make sure git knows who to blame:
$ cat >~/.gitconfig <<\EOF
name = Your Name Comes Here
email =
Select file contents to include in the next commit, then make the
$ git add a.txt # updated file
$ git add b.txt # new file
$ git rm c.txt # old file
$ git commit
Or, prepare and create the commit in one step:
$ git commit d.txt # use latest content only of d.txt
$ git commit -a # use latest content of all tracked files
$ git merge test # merge branch "test" into the current branch
$ git pull git:// master
# fetch and merge in remote branch
$ git pull . test # equivalent to git merge test
Sharing your changes
Importing or exporting patches:
$ git format-patch origin..HEAD # format a patch for each commit
# in HEAD but not in origin
$ git am mbox # import patches from the mailbox "mbox"
Fetch a branch in a different git repository, then merge into the
current branch:
$ git pull git:// theirbranch
Store the fetched branch into a local branch before merging into the
current branch:
$ git pull git:// theirbranch:mybranch
After creating commits on a local branch, update the remote
branch with your commits:
$ git push ssh:// mybranch:theirbranch
When remote and local branch are both named "test":
$ git push ssh:// test
Shortcut version for a frequently used remote repository:
$ git remote add example ssh://
$ git push example test
Repository maintenance
Check for corruption:
$ git fsck
Recompress, remove unused cruft:
$ git gc
Repositories and Branches
How to get a git repository
It will be useful to have a git repository to experiment with as you
read this manual.
The best way to get one is by using the gitlink:git-clone[1] command
to download a copy of an existing repository for a project that you
are interested in. If you don't already have a project in mind, here
are some interesting examples:
# git itself (approx. 10MB download):
$ git clone git://
# the linux kernel (approx. 150MB download):
$ git clone git://
The initial clone may be time-consuming for a large project, but you
will only need to clone once.
The clone command creates a new directory named after the project
("git" or "linux-2.6" in the examples above). After you cd into this
directory, you will see that it contains a copy of the project files,
together with a special top-level directory named ".git", which
contains all the information about the history of the project.
In most of the following, examples will be taken from one of the two
repositories above.
How to check out a different version of a project
Git is best thought of as a tool for storing the history of a
collection of files. It stores the history as a compressed
collection of interrelated snapshots (versions) of the project's
A single git repository may contain multiple branches. It keeps track
of them by keeping a list of <<def_head,heads>> which reference the
latest version on each branch; the gitlink:git-branch[1] command shows
you the list of branch heads:
$ git branch
* master
A freshly cloned repository contains a single branch head, by default
named "master", with the working directory initialized to the state of
the project referred to by that branch head.
Most projects also use <<def_tag,tags>>. Tags, like heads, are
references into the project's history, and can be listed using the
gitlink:git-tag[1] command:
$ git tag -l
Tags are expected to always point at the same version of a project,
while heads are expected to advance as development progresses.
Create a new branch head pointing to one of these versions and check it
out using gitlink:git-checkout[1]:
$ git checkout -b new v2.6.13
The working directory then reflects the contents that the project had
when it was tagged v2.6.13, and gitlink:git-branch[1] shows two
branches, with an asterisk marking the currently checked-out branch:
$ git branch
* new
If you decide that you'd rather see version 2.6.17, you can modify
the current branch to point at v2.6.17 instead, with
$ git reset --hard v2.6.17
Note that if the current branch head was your only reference to a
particular point in history, then resetting that branch may leave you
with no way to find the history it used to point to; so use this command
Understanding History: Commits
Every change in the history of a project is represented by a commit.
The gitlink:git-show[1] command shows the most recent commit on the
current branch:
$ git show
commit 2b5f6dcce5bf94b9b119e9ed8d537098ec61c3d2
Author: Jamal Hadi Salim <>
Date: Sat Dec 2 22:22:25 2006 -0800
[XFRM]: Fix aevent structuring to be more complete.
aevents can not uniquely identify an SA. We break the ABI with this
patch, but consensus is that since it is not yet utilized by any
(known) application then it is fine (better do it now than later).
Signed-off-by: Jamal Hadi Salim <>
Signed-off-by: David S. Miller <>
diff --git a/Documentation/networking/xfrm_sync.txt b/Documentation/networking/xfrm_sync.txt
index 8be626f..d7aac9d 100644
--- a/Documentation/networking/xfrm_sync.txt
+++ b/Documentation/networking/xfrm_sync.txt
@@ -47,10 +47,13 @@ aevent_id structure looks like:
struct xfrm_aevent_id {
struct xfrm_usersa_id sa_id;
+ xfrm_address_t saddr;
__u32 flags;
+ __u32 reqid;
As you can see, a commit shows who made the latest change, what they
did, and why.
Every commit has a 40-hexdigit id, sometimes called the "object name" or the
"SHA1 id", shown on the first line of the "git show" output. You can usually
refer to a commit by a shorter name, such as a tag or a branch name, but this
longer name can also be useful. Most importantly, it is a globally unique
name for this commit: so if you tell somebody else the object name (for
example in email), then you are guaranteed that name will refer to the same
commit in their repository that it does in yours (assuming their repository
has that commit at all). Since the object name is computed as a hash over the
contents of the commit, you are guaranteed that the commit can never change
without its name also changing.
In fact, in <<git-internals>> we shall see that everything stored in git
history, including file data and directory contents, is stored in an object
with a name that is a hash of its contents.
Understanding history: commits, parents, and reachability
Every commit (except the very first commit in a project) also has a
parent commit which shows what happened before this commit.
Following the chain of parents will eventually take you back to the
beginning of the project.
However, the commits do not form a simple list; git allows lines of
development to diverge and then reconverge, and the point where two
lines of development reconverge is called a "merge". The commit
representing a merge can therefore have more than one parent, with
each parent representing the most recent commit on one of the lines
of development leading to that point.
The best way to see how this works is using the gitlink:gitk[1]
command; running gitk now on a git repository and looking for merge
commits will help understand how the git organizes history.
In the following, we say that commit X is "reachable" from commit Y
if commit X is an ancestor of commit Y. Equivalently, you could say
that Y is a descendent of X, or that there is a chain of parents
leading from commit Y to commit X.
Understanding history: History diagrams
We will sometimes represent git history using diagrams like the one
below. Commits are shown as "o", and the links between them with
lines drawn with - / and \. Time goes left to right:
o--o--o <-- Branch A
o--o--o <-- master
o--o--o <-- Branch B
If we need to talk about a particular commit, the character "o" may
be replaced with another letter or number.
Understanding history: What is a branch?
When we need to be precise, we will use the word "branch" to mean a line
of development, and "branch head" (or just "head") to mean a reference
to the most recent commit on a branch. In the example above, the branch
head named "A" is a pointer to one particular commit, but we refer to
the line of three commits leading up to that point as all being part of
"branch A".
However, when no confusion will result, we often just use the term
"branch" both for branches and for branch heads.
Manipulating branches
Creating, deleting, and modifying branches is quick and easy; here's
a summary of the commands:
git branch::
list all branches
git branch <branch>::
create a new branch named <branch>, referencing the same
point in history as the current branch
git branch <branch> <start-point>::
create a new branch named <branch>, referencing
<start-point>, which may be specified any way you like,
including using a branch name or a tag name
git branch -d <branch>::
delete the branch <branch>; if the branch you are deleting
points to a commit which is not reachable from this branch,
this command will fail with a warning.
git branch -D <branch>::
even if the branch points to a commit not reachable
from the current branch, you may know that that commit
is still reachable from some other branch or tag. In that
case it is safe to use this command to force git to delete
the branch.
git checkout <branch>::
make the current branch <branch>, updating the working
directory to reflect the version referenced by <branch>
git checkout -b <new> <start-point>::
create a new branch <new> referencing <start-point>, and
check it out.
The special symbol "HEAD" can always be used to refer to the current
branch. In fact, git uses a file named "HEAD" in the .git directory to
remember which branch is current:
$ cat .git/HEAD
ref: refs/heads/master
Examining an old version without creating a new branch
The git-checkout command normally expects a branch head, but will also
accept an arbitrary commit; for example, you can check out the commit
referenced by a tag:
$ git checkout v2.6.17
Note: moving to "v2.6.17" which isn't a local branch
If you want to create a new branch from this checkout, you may do so
(now or later) by using -b with the checkout command again. Example:
git checkout -b <new_branch_name>
HEAD is now at 427abfa... Linux v2.6.17
The HEAD then refers to the SHA1 of the commit instead of to a branch,
and git branch shows that you are no longer on a branch:
$ cat .git/HEAD
git branch
* (no branch)
In this case we say that the HEAD is "detached".
This can be an easy way to check out a particular version without having
to make up a name for a new branch. However, keep in mind that when you
switch away from the (for example, by checking out something else), you
can lose track of what the HEAD used to point to.
Examining branches from a remote repository
The "master" branch that was created at the time you cloned is a copy
of the HEAD in the repository that you cloned from. That repository
may also have had other branches, though, and your local repository
keeps branches which track each of those remote branches, which you
can view using the "-r" option to gitlink:git-branch[1]:
$ git branch -r
You cannot check out these remote-tracking branches, but you can
examine them on a branch of your own, just as you would a tag:
$ git checkout -b my-todo-copy origin/todo
Note that the name "origin" is just the name that git uses by default
to refer to the repository that you cloned from.
Naming branches, tags, and other references
Branches, remote-tracking branches, and tags are all references to
commits. All references are named with a slash-separated path name
starting with "refs"; the names we've been using so far are actually
- The branch "test" is short for "refs/heads/test".
- The tag "v2.6.18" is short for "refs/tags/v2.6.18".
- "origin/master" is short for "refs/remotes/origin/master".
The full name is occasionally useful if, for example, there ever
exists a tag and a branch with the same name.
As another useful shortcut, if the repository "origin" posesses only
a single branch, you can refer to that branch as just "origin".
More generally, if you have defined a remote repository named
"example", you can refer to the branch in that repository as
"example". And for a repository with multiple branches, this will
refer to the branch designated as the "HEAD" branch.
For the complete list of paths which git checks for references, and
the order it uses to decide which to choose when there are multiple
references with the same shorthand name, see the "SPECIFYING
REVISIONS" section of gitlink:git-rev-parse[1].
Updating a repository with git fetch
Eventually the developer cloned from will do additional work in her
repository, creating new commits and advancing the branches to point
at the new commits.
The command "git fetch", with no arguments, will update all of the
remote-tracking branches to the latest version found in her
repository. It will not touch any of your own branches--not even the
"master" branch that was created for you on clone.
Fetching branches from other repositories
You can also track branches from repositories other than the one you
cloned from, using gitlink:git-remote[1]:
$ git remote add linux-nfs git://
$ git fetch linux-nfs
* refs/remotes/linux-nfs/master: storing branch 'master' ...
commit: bf81b46
New remote-tracking branches will be stored under the shorthand name
that you gave "git remote add", in this case linux-nfs:
$ git branch -r
If you run "git fetch <remote>" later, the tracking branches for the
named <remote> will be updated.
If you examine the file .git/config, you will see that git has added
a new stanza:
$ cat .git/config
[remote "linux-nfs"]
url = git://
fetch = +refs/heads/*:refs/remotes/linux-nfs/*
This is what causes git to track the remote's branches; you may modify
or delete these configuration options by editing .git/config with a
text editor. (See the "CONFIGURATION FILE" section of
gitlink:git-config[1] for details.)
Exploring git history
Git is best thought of as a tool for storing the history of a
collection of files. It does this by storing compressed snapshots of
the contents of a file heirarchy, together with "commits" which show
the relationships between these snapshots.
Git provides extremely flexible and fast tools for exploring the
history of a project.
We start with one specialized tool that is useful for finding the
commit that introduced a bug into a project.
How to use bisect to find a regression
Suppose version 2.6.18 of your project worked, but the version at
"master" crashes. Sometimes the best way to find the cause of such a
regression is to perform a brute-force search through the project's
history to find the particular commit that caused the problem. The
gitlink:git-bisect[1] command can help you do this:
$ git bisect start
$ git bisect good v2.6.18
$ git bisect bad master
Bisecting: 3537 revisions left to test after this
[65934a9a028b88e83e2b0f8b36618fe503349f8e] BLOCK: Make USB storage depend on SCSI rather than selecting it [try #6]
If you run "git branch" at this point, you'll see that git has
temporarily moved you to a new branch named "bisect". This branch
points to a commit (with commit id 65934...) that is reachable from
v2.6.19 but not from v2.6.18. Compile and test it, and see whether
it crashes. Assume it does crash. Then:
$ git bisect bad
Bisecting: 1769 revisions left to test after this
[7eff82c8b1511017ae605f0c99ac275a7e21b867] i2c-core: Drop useless bitmaskings
checks out an older version. Continue like this, telling git at each
stage whether the version it gives you is good or bad, and notice
that the number of revisions left to test is cut approximately in
half each time.
After about 13 tests (in this case), it will output the commit id of
the guilty commit. You can then examine the commit with
gitlink:git-show[1], find out who wrote it, and mail them your bug
report with the commit id. Finally, run
$ git bisect reset
to return you to the branch you were on before and delete the
temporary "bisect" branch.
Note that the version which git-bisect checks out for you at each
point is just a suggestion, and you're free to try a different
version if you think it would be a good idea. For example,
occasionally you may land on a commit that broke something unrelated;
$ git bisect visualize
which will run gitk and label the commit it chose with a marker that
says "bisect". Chose a safe-looking commit nearby, note its commit
id, and check it out with:
$ git reset --hard fb47ddb2db...
then test, run "bisect good" or "bisect bad" as appropriate, and
Naming commits
We have seen several ways of naming commits already:
- 40-hexdigit object name
- branch name: refers to the commit at the head of the given
- tag name: refers to the commit pointed to by the given tag
(we've seen branches and tags are special cases of
- HEAD: refers to the head of the current branch
There are many more; see the "SPECIFYING REVISIONS" section of the
gitlink:git-rev-parse[1] man page for the complete list of ways to
name revisions. Some examples:
$ git show fb47ddb2 # the first few characters of the object name
# are usually enough to specify it uniquely
$ git show HEAD^ # the parent of the HEAD commit
$ git show HEAD^^ # the grandparent
$ git show HEAD~4 # the great-great-grandparent
Recall that merge commits may have more than one parent; by default,
^ and ~ follow the first parent listed in the commit, but you can
also choose:
$ git show HEAD^1 # show the first parent of HEAD
$ git show HEAD^2 # show the second parent of HEAD
In addition to HEAD, there are several other special names for
Merges (to be discussed later), as well as operations such as
git-reset, which change the currently checked-out commit, generally
set ORIG_HEAD to the value HEAD had before the current operation.
The git-fetch operation always stores the head of the last fetched
branch in FETCH_HEAD. For example, if you run git fetch without
specifying a local branch as the target of the operation
$ git fetch git:// theirbranch
the fetched commits will still be available from FETCH_HEAD.
When we discuss merges we'll also see the special name MERGE_HEAD,
which refers to the other branch that we're merging in to the current
The gitlink:git-rev-parse[1] command is a low-level command that is
occasionally useful for translating some name for a commit to the object
name for that commit:
$ git rev-parse origin
Creating tags
We can also create a tag to refer to a particular commit; after
$ git tag stable-1 1b2e1d63ff
You can use stable-1 to refer to the commit 1b2e1d63ff.
This creates a "lightweight" tag. If the tag is a tag you wish to
share with others, and possibly sign cryptographically, then you
should create a tag object instead; see the gitlink:git-tag[1] man
page for details.
Browsing revisions
The gitlink:git-log[1] command can show lists of commits. On its
own, it shows all commits reachable from the parent commit; but you
can also make more specific requests:
$ git log v2.5.. # commits since (not reachable from) v2.5
$ git log test..master # commits reachable from master but not test
$ git log master..test # ...reachable from test but not master
$ git log master...test # ...reachable from either test or master,
# but not both
$ git log --since="2 weeks ago" # commits from the last 2 weeks
$ git log Makefile # commits which modify Makefile
$ git log fs/ # ... which modify any file under fs/
$ git log -S'foo()' # commits which add or remove any file data
# matching the string 'foo()'
And of course you can combine all of these; the following finds
commits since v2.5 which touch the Makefile or any file under fs:
$ git log v2.5.. Makefile fs/
You can also ask git log to show patches:
$ git log -p
See the "--pretty" option in the gitlink:git-log[1] man page for more
display options.
Note that git log starts with the most recent commit and works
backwards through the parents; however, since git history can contain
multiple independent lines of development, the particular order that
commits are listed in may be somewhat arbitrary.
Generating diffs
You can generate diffs between any two versions using
$ git diff master..test
Sometimes what you want instead is a set of patches:
$ git format-patch master..test
will generate a file with a patch for each commit reachable from test
but not from master. Note that if master also has commits which are
not reachable from test, then the combined result of these patches
will not be the same as the diff produced by the git-diff example.
Viewing old file versions
You can always view an old version of a file by just checking out the
correct revision first. But sometimes it is more convenient to be
able to view an old version of a single file without checking
anything out; this command does that:
$ git show v2.5:fs/locks.c
Before the colon may be anything that names a commit, and after it
may be any path to a file tracked by git.
Check whether two branches point at the same history
Suppose you want to check whether two branches point at the same point
in history.
$ git diff origin..master
will tell you whether the contents of the project are the same at the
two branches; in theory, however, it's possible that the same project
contents could have been arrived at by two different historical
routes. You could compare the object names:
$ git rev-list origin
$ git rev-list master
Or you could recall that the ... operator selects all commits
contained reachable from either one reference or the other but not
both: so
$ git log origin...master
will return no commits when the two branches are equal.
Find first tagged version including a given fix
Suppose you know that the commit e05db0fd fixed a certain problem.
You'd like to find the earliest tagged release that contains that
Of course, there may be more than one answer--if the history branched
after commit e05db0fd, then there could be multiple "earliest" tagged
You could just visually inspect the commits since e05db0fd:
$ gitk e05db0fd..
Or you can use gitlink:git-name-rev[1], which will give the commit a
name based on any tag it finds pointing to one of the commit's
$ git name-rev --tags e05db0fd
e05db0fd tags/v1.5.0-rc1^0~23
The gitlink:git-describe[1] command does the opposite, naming the
revision using a tag on which the given commit is based:
$ git describe e05db0fd
but that may sometimes help you guess which tags might come after the
given commit.
If you just want to verify whether a given tagged version contains a
given commit, you could use gitlink:git-merge-base[1]:
$ git merge-base e05db0fd v1.5.0-rc1
The merge-base command finds a common ancestor of the given commits,
and always returns one or the other in the case where one is a
descendant of the other; so the above output shows that e05db0fd
actually is an ancestor of v1.5.0-rc1.
Alternatively, note that
$ git log v1.5.0-rc1..e05db0fd
will produce empty output if and only if v1.5.0-rc1 includes e05db0fd,
because it outputs only commits that are not reachable from v1.5.0-rc1.
As yet another alternative, the gitlink:git-show-branch[1] command lists
the commits reachable from its arguments with a display on the left-hand
side that indicates which arguments that commit is reachable from. So,
you can run something like
$ git show-branch e05db0fd v1.5.0-rc0 v1.5.0-rc1 v1.5.0-rc2
! [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
! [v1.5.0-rc0] GIT v1.5.0 preview
! [v1.5.0-rc1] GIT v1.5.0-rc1
! [v1.5.0-rc2] GIT v1.5.0-rc2
then search for a line that looks like
+ ++ [e05db0fd] Fix warnings in sha1_file.c - use C99 printf format if
Which shows that e05db0fd is reachable from itself, from v1.5.0-rc1, and
from v1.5.0-rc2, but not from v1.5.0-rc0.
Developing with git
Telling git your name
Before creating any commits, you should introduce yourself to git. The
easiest way to do so is:
$ cat >~/.gitconfig <<\EOF
name = Your Name Comes Here
email =
(See the "CONFIGURATION FILE" section of gitlink:git-config[1] for
details on the configuration file.)
Creating a new repository
Creating a new repository from scratch is very easy:
$ mkdir project
$ cd project
$ git init
If you have some initial content (say, a tarball):
$ tar -xzvf project.tar.gz
$ cd project
$ git init
$ git add . # include everything below ./ in the first commit:
$ git commit
How to make a commit
Creating a new commit takes three steps:
1. Making some changes to the working directory using your
favorite editor.
2. Telling git about your changes.
3. Creating the commit using the content you told git about
in step 2.
In practice, you can interleave and repeat steps 1 and 2 as many
times as you want: in order to keep track of what you want committed
at step 3, git maintains a snapshot of the tree's contents in a
special staging area called "the index."
At the beginning, the content of the index will be identical to
that of the HEAD. The command "git diff --cached", which shows
the difference between the HEAD and the index, should therefore
produce no output at that point.
Modifying the index is easy:
To update the index with the new contents of a modified file, use
$ git add path/to/file
To add the contents of a new file to the index, use
$ git add path/to/file
To remove a file from the index and from the working tree,
$ git rm path/to/file
After each step you can verify that
$ git diff --cached
always shows the difference between the HEAD and the index file--this
is what you'd commit if you created the commit now--and that
$ git diff
shows the difference between the working tree and the index file.
Note that "git add" always adds just the current contents of a file
to the index; further changes to the same file will be ignored unless
you run git-add on the file again.
When you're ready, just run
$ git commit
and git will prompt you for a commit message and then create the new
commit. Check to make sure it looks like what you expected with
$ git show
As a special shortcut,
$ git commit -a
will update the index with any files that you've modified or removed
and create a commit, all in one step.
A number of commands are useful for keeping track of what you're
about to commit:
$ git diff --cached # difference between HEAD and the index; what
# would be commited if you ran "commit" now.
$ git diff # difference between the index file and your
# working directory; changes that would not
# be included if you ran "commit" now.
$ git status # a brief per-file summary of the above.
Creating good commit messages
Though not required, it's a good idea to begin the commit message
with a single short (less than 50 character) line summarizing the
change, followed by a blank line and then a more thorough
description. Tools that turn commits into email, for example, use
the first line on the Subject line and the rest of the commit in the
How to merge
You can rejoin two diverging branches of development using
$ git merge branchname
merges the development in the branch "branchname" into the current
branch. If there are conflicts--for example, if the same file is
modified in two different ways in the remote branch and the local
branch--then you are warned; the output may look something like this:
$ git merge next
100% (4/4) done
Auto-merged file.txt
CONFLICT (content): Merge conflict in file.txt
Automatic merge failed; fix conflicts and then commit the result.
Conflict markers are left in the problematic files, and after
you resolve the conflicts manually, you can update the index
with the contents and run git commit, as you normally would when
creating a new file.
If you examine the resulting commit using gitk, you will see that it
has two parents, one pointing to the top of the current branch, and
one to the top of the other branch.
In more detail:
Resolving a merge
When a merge isn't resolved automatically, git leaves the index and
the working tree in a special state that gives you all the
information you need to help resolve the merge.
Files with conflicts are marked specially in the index, so until you
resolve the problem and update the index, gitlink:git-commit[1] will
$ git commit
file.txt: needs merge
Also, gitlink:git-status[1] will list those files as "unmerged", and the
files with conflicts will have conflict markers added, like this:
<<<<<<< HEAD:file.txt
Hello world
>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
All you need to do is edit the files to resolve the conflicts, and then
$ git add file.txt
$ git commit
Note that the commit message will already be filled in for you with
some information about the merge. Normally you can just use this
default message unchanged, but you may add additional commentary of
your own if desired.
The above is all you need to know to resolve a simple merge. But git
also provides more information to help resolve conflicts:
Getting conflict-resolution help during a merge
All of the changes that git was able to merge automatically are
already added to the index file, so gitlink:git-diff[1] shows only
the conflicts. It uses an unusual syntax:
$ git diff
diff --cc file.txt
index 802992c,2b60207..0000000
--- a/file.txt
+++ b/file.txt
@@@ -1,1 -1,1 +1,5 @@@
++<<<<<<< HEAD:file.txt
+Hello world
+ Goodbye
++>>>>>>> 77976da35a11db4580b80ae27e8d65caf5208086:file.txt
Recall that the commit which will be commited after we resolve this
conflict will have two parents instead of the usual one: one parent
will be HEAD, the tip of the current branch; the other will be the
tip of the other branch, which is stored temporarily in MERGE_HEAD.
During the merge, the index holds three versions of each file. Each of
these three "file stages" represents a different version of the file:
$ git show :1:file.txt # the file in a common ancestor of both branches
$ git show :2:file.txt # the version from HEAD, but including any
# nonconflicting changes from MERGE_HEAD
$ git show :3:file.txt # the version from MERGE_HEAD, but including any
# nonconflicting changes from HEAD.
Since the stage 2 and stage 3 versions have already been updated with
nonconflicting changes, the only remaining differences between them are
the important ones; thus gitlink:git-diff[1] can use the information in
the index to show only those conflicts.
The diff above shows the differences between the working-tree version of
file.txt and the stage 2 and stage 3 versions. So instead of preceding
each line by a single "+" or "-", it now uses two columns: the first
column is used for differences between the first parent and the working
directory copy, and the second for differences between the second parent
and the working directory copy. (See the "COMBINED DIFF FORMAT" section
of gitlink:git-diff-files[1] for a details of the format.)
After resolving the conflict in the obvious way (but before updating the
index), the diff will look like:
$ git diff
diff --cc file.txt
index 802992c,2b60207..0000000
--- a/file.txt
+++ b/file.txt
@@@ -1,1 -1,1 +1,1 @@@
- Hello world
++Goodbye world
This shows that our resolved version deleted "Hello world" from the
first parent, deleted "Goodbye" from the second parent, and added
"Goodbye world", which was previously absent from both.
Some special diff options allow diffing the working directory against
any of these stages:
$ git diff -1 file.txt # diff against stage 1
$ git diff --base file.txt # same as the above
$ git diff -2 file.txt # diff against stage 2
$ git diff --ours file.txt # same as the above
$ git diff -3 file.txt # diff against stage 3
$ git diff --theirs file.txt # same as the above.
The gitlink:git-log[1] and gitk[1] commands also provide special help
for merges:
$ git log --merge
$ gitk --merge
These will display all commits which exist only on HEAD or on
MERGE_HEAD, and which touch an unmerged file.
Each time you resolve the conflicts in a file and update the index:
$ git add file.txt
the different stages of that file will be "collapsed", after which
git-diff will (by default) no longer show diffs for that file.
Undoing a merge
If you get stuck and decide to just give up and throw the whole mess
away, you can always return to the pre-merge state with
$ git reset --hard HEAD
Or, if you've already commited the merge that you want to throw away,
$ git reset --hard ORIG_HEAD
However, this last command can be dangerous in some cases--never
throw away a commit you have already committed if that commit may
itself have been merged into another branch, as doing so may confuse
further merges.
Fast-forward merges
There is one special case not mentioned above, which is treated
differently. Normally, a merge results in a merge commit, with two
parents, one pointing at each of the two lines of development that
were merged.
However, if one of the two lines of development is completely
contained within the other--so every commit present in the one is
already contained in the other--then git just performs a
<<fast-forwards,fast forward>>; the head of the current branch is
moved forward to point at the head of the merged-in branch, without
any new commits being created.
Fixing mistakes
If you've messed up the working tree, but haven't yet committed your
mistake, you can return the entire working tree to the last committed
state with
$ git reset --hard HEAD
If you make a commit that you later wish you hadn't, there are two
fundamentally different ways to fix the problem:
1. You can create a new commit that undoes whatever was done
by the previous commit. This is the correct thing if your
mistake has already been made public.
2. You can go back and modify the old commit. You should
never do this if you have already made the history public;
git does not normally expect the "history" of a project to
change, and cannot correctly perform repeated merges from
a branch that has had its history changed.
Fixing a mistake with a new commit
Creating a new commit that reverts an earlier change is very easy;
just pass the gitlink:git-revert[1] command a reference to the bad
commit; for example, to revert the most recent commit:
$ git revert HEAD
This will create a new commit which undoes the change in HEAD. You
will be given a chance to edit the commit message for the new commit.
You can also revert an earlier change, for example, the next-to-last:
$ git revert HEAD^
In this case git will attempt to undo the old change while leaving
intact any changes made since then. If more recent changes overlap
with the changes to be reverted, then you will be asked to fix
conflicts manually, just as in the case of <<resolving-a-merge,
resolving a merge>>.
Fixing a mistake by editing history
If the problematic commit is the most recent commit, and you have not
yet made that commit public, then you may just
<<undoing-a-merge,destroy it using git-reset>>.
Alternatively, you
can edit the working directory and update the index to fix your
mistake, just as if you were going to <<how-to-make-a-commit,create a
new commit>>, then run
$ git commit --amend
which will replace the old commit by a new commit incorporating your
changes, giving you a chance to edit the old commit message first.
Again, you should never do this to a commit that may already have
been merged into another branch; use gitlink:git-revert[1] instead in
that case.
It is also possible to edit commits further back in the history, but
this is an advanced topic to be left for
<<cleaning-up-history,another chapter>>.
Checking out an old version of a file
In the process of undoing a previous bad change, you may find it
useful to check out an older version of a particular file using
gitlink:git-checkout[1]. We've used git checkout before to switch
branches, but it has quite different behavior if it is given a path
name: the command
$ git checkout HEAD^ path/to/file
replaces path/to/file by the contents it had in the commit HEAD^, and
also updates the index to match. It does not change branches.
If you just want to look at an old version of the file, without
modifying the working directory, you can do that with
$ git show HEAD^:path/to/file
which will display the given version of the file.
Ensuring good performance
On large repositories, git depends on compression to keep the history
information from taking up to much space on disk or in memory.
This compression is not performed automatically. Therefore you
should occasionally run gitlink:git-gc[1]:
$ git gc
to recompress the archive. This can be very time-consuming, so
you may prefer to run git-gc when you are not doing other work.
Ensuring reliability
Checking the repository for corruption
The gitlink:git-fsck[1] command runs a number of self-consistency checks
on the repository, and reports on any problems. This may take some
time. The most common warning by far is about "dangling" objects:
$ git fsck
dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
dangling blob 218761f9d90712d37a9c5e36f406f92202db07eb
dangling commit bf093535a34a4d35731aa2bd90fe6b176302f14f
dangling commit 8e4bec7f2ddaa268bef999853c25755452100f8e
dangling tree d50bb86186bf27b681d25af89d3b5b68382e4085
dangling tree b24c2473f1fd3d91352a624795be026d64c8841f
Dangling objects are objects that are harmless, but also unnecessary;
you can remove them at any time with gitlink:git-prune[1] or the --prune
option to gitlink:git-gc[1]:
$ git gc --prune
This may be time-consuming. Unlike most other git operations (including
git-gc when run without any options), it is not safe to prune while
other git operations are in progress in the same repository.
For more about dangling objects, see <<dangling-objects>>.
Recovering lost changes
Say you modify a branch with gitlink:git-reset[1] --hard, and then
realize that the branch was the only reference you had to that point in
Fortunately, git also keeps a log, called a "reflog", of all the
previous values of each branch. So in this case you can still find the
old history using, for example,
$ git log master@{1}
This lists the commits reachable from the previous version of the head.
This syntax can be used to with any git command that accepts a commit,
not just with git log. Some other examples:
$ git show master@{2} # See where the branch pointed 2,
$ git show master@{3} # 3, ... changes ago.
$ gitk master@{yesterday} # See where it pointed yesterday,
$ gitk master@{"1 week ago"} # ... or last week
The reflogs are kept by default for 30 days, after which they may be
pruned. See gitlink:git-reflog[1] and gitlink:git-gc[1] to learn
how to control this pruning, and see the "SPECIFYING REVISIONS"
section of gitlink:git-rev-parse[1] for details.
Note that the reflog history is very different from normal git history.
While normal history is shared by every repository that works on the
same project, the reflog history is not shared: it tells you only about
how the branches in your local repository have changed over time.
Examining dangling objects
In some situations the reflog may not be able to save you. For
example, suppose you delete a branch, then realize you need the history
it contained. The reflog is also deleted; however, if you have not
yet pruned the repository, then you may still be able to find
the lost commits; run git-fsck and watch for output that mentions
"dangling commits":
$ git fsck
dangling commit 7281251ddd2a61e38657c827739c57015671a6b3
dangling commit 2706a059f258c6b245f298dc4ff2ccd30ec21a63
dangling commit 13472b7c4b80851a1bc551779171dcb03655e9b5
You can examine
one of those dangling commits with, for example,
$ gitk 7281251ddd --not --all
which does what it sounds like: it says that you want to see the commit
history that is described by the dangling commit(s), but not the
history that is described by all your existing branches and tags. Thus
you get exactly the history reachable from that commit that is lost.
(And notice that it might not be just one commit: we only report the
"tip of the line" as being dangling, but there might be a whole deep
and complex commit history that was dropped.)
If you decide you want the history back, you can always create a new
reference pointing to it, for example, a new branch:
$ git branch recovered-branch 7281251ddd
Sharing development with others
Getting updates with git pull
After you clone a repository and make a few changes of your own, you
may wish to check the original repository for updates and merge them
into your own work.
We have already seen <<Updating-a-repository-with-git-fetch,how to
keep remote tracking branches up to date>> with gitlink:git-fetch[1],
and how to merge two branches. So you can merge in changes from the
original repository's master branch with:
$ git fetch
$ git merge origin/master
However, the gitlink:git-pull[1] command provides a way to do this in
one step:
$ git pull origin master
In fact, "origin" is normally the default repository to pull from,
and the default branch is normally the HEAD of the remote repository,
so often you can accomplish the above with just
$ git pull
See the descriptions of the branch.<name>.remote and
branch.<name>.merge options in gitlink:git-config[1] to learn
how to control these defaults depending on the current branch.
In addition to saving you keystrokes, "git pull" also helps you by
producing a default commit message documenting the branch and
repository that you pulled from.
(But note that no such commit will be created in the case of a
<<fast-forwards,fast forward>>; instead, your branch will just be
updated to point to the latest commit from the upstream branch.)
The git-pull command can also be given "." as the "remote" repository,
in which case it just merges in a branch from the current repository; so
the commands
$ git pull . branch
$ git merge branch
are roughly equivalent. The former is actually very commonly used.
Submitting patches to a project
If you just have a few changes, the simplest way to submit them may
just be to send them as patches in email:
First, use gitlink:git-format-patch[1]; for example:
$ git format-patch origin
will produce a numbered series of files in the current directory, one
for each patch in the current branch but not in origin/HEAD.
You can then import these into your mail client and send them by
hand. However, if you have a lot to send at once, you may prefer to
use the gitlink:git-send-email[1] script to automate the process.
Consult the mailing list for your project first to determine how they
prefer such patches be handled.
Importing patches to a project
Git also provides a tool called gitlink:git-am[1] (am stands for
"apply mailbox"), for importing such an emailed series of patches.
Just save all of the patch-containing messages, in order, into a
single mailbox file, say "patches.mbox", then run
$ git am -3 patches.mbox
Git will apply each patch in order; if any conflicts are found, it
will stop, and you can fix the conflicts as described in
"<<resolving-a-merge,Resolving a merge>>". (The "-3" option tells
git to perform a merge; if you would prefer it just to abort and
leave your tree and index untouched, you may omit that option.)
Once the index is updated with the results of the conflict
resolution, instead of creating a new commit, just run
$ git am --resolved
and git will create the commit for you and continue applying the
remaining patches from the mailbox.
The final result will be a series of commits, one for each patch in
the original mailbox, with authorship and commit log message each
taken from the message containing each patch.
Setting up a public repository
Another way to submit changes to a project is to simply tell the
maintainer of that project to pull from your repository, exactly as
you did in the section "<<getting-updates-with-git-pull, Getting
updates with git pull>>".
If you and maintainer both have accounts on the same machine, then
then you can just pull changes from each other's repositories
directly; note that all of the commands (gitlink:git-clone[1],
git-fetch[1], git-pull[1], etc.) that accept a URL as an argument
will also accept a local directory name; so, for example, you can
$ git clone /path/to/repository
$ git pull /path/to/other/repository
If this sort of setup is inconvenient or impossible, another (more
common) option is to set up a public repository on a public server.
This also allows you to cleanly separate private work in progress
from publicly visible work.
You will continue to do your day-to-day work in your personal
repository, but periodically "push" changes from your personal
repository into your public repository, allowing other developers to
pull from that repository. So the flow of changes, in a situation
where there is one other developer with a public repository, looks
like this:
you push
your personal repo ------------------> your public repo
^ |
| |
| you pull | they pull
| |
| |
| they push V
their public repo <------------------- their repo
Now, assume your personal repository is in the directory ~/proj. We
first create a new clone of the repository:
$ git clone --bare proj-clone.git
The resulting directory proj-clone.git will contains a "bare" git
repository--it is just the contents of the ".git" directory, without
a checked-out copy of a working directory.
Next, copy proj-clone.git to the server where you plan to host the
public repository. You can use scp, rsync, or whatever is most
If somebody else maintains the public server, they may already have
set up a git service for you, and you may skip to the section
"<<pushing-changes-to-a-public-repository,Pushing changes to a public
repository>>", below.
Otherwise, the following sections explain how to export your newly
created public repository:
Exporting a git repository via http
The git protocol gives better performance and reliability, but on a
host with a web server set up, http exports may be simpler to set up.
All you need to do is place the newly created bare git repository in
a directory that is exported by the web server, and make some
adjustments to give web clients some extra information they need:
$ mv proj.git /home/you/public_html/proj.git
$ cd proj.git
$ git update-server-info
$ chmod a+x hooks/post-update
(For an explanation of the last two lines, see
gitlink:git-update-server-info[1], and the documentation
link:hooks.txt[Hooks used by git].)
Advertise the url of proj.git. Anybody else should then be able to
clone or pull from that url, for example with a commandline like:
$ git clone
(See also
for a slightly more sophisticated setup using WebDAV which also
allows pushing over http.)
Exporting a git repository via the git protocol
This is the preferred method.
For now, we refer you to the gitlink:git-daemon[1] man page for
instructions. (See especially the examples section.)
Pushing changes to a public repository
Note that the two techniques outline above (exporting via
<<exporting-via-http,http>> or <<exporting-via-git,git>>) allow other
maintainers to fetch your latest changes, but they do not allow write
access, which you will need to update the public repository with the
latest changes created in your private repository.
The simplest way to do this is using gitlink:git-push[1] and ssh; to
update the remote branch named "master" with the latest state of your
branch named "master", run
$ git push ssh:// master:master
or just
$ git push ssh:// master
As with git-fetch, git-push will complain if this does not result in
a <<fast-forwards,fast forward>>. Normally this is a sign of
something wrong. However, if you are sure you know what you're
doing, you may force git-push to perform the update anyway by
proceeding the branch name by a plus sign:
$ git push ssh:// +master
As with git-fetch, you may also set up configuration options to
save typing; so, for example, after
$ cat >.git/config <<EOF
[remote "public-repo"]
url = ssh://
you should be able to perform the above push with just
$ git push public-repo master
See the explanations of the remote.<name>.url, branch.<name>.remote,
and remote.<name>.push options in gitlink:git-config[1] for
Setting up a shared repository
Another way to collaborate is by using a model similar to that
commonly used in CVS, where several developers with special rights
all push to and pull from a single shared repository. See
link:cvs-migration.txt[git for CVS users] for instructions on how to
set this up.
Allow web browsing of a repository
The gitweb cgi script provides users an easy way to browse your
project's files and history without having to install git; see the file
gitweb/INSTALL in the git source tree for instructions on setting it up.
TODO: topic branches, typical roles as in everyday.txt, ?
Rewriting history and maintaining patch series
Normally commits are only added to a project, never taken away or
replaced. Git is designed with this assumption, and violating it will
cause git's merge machinery (for example) to do the wrong thing.
However, there is a situation in which it can be useful to violate this
Creating the perfect patch series
Suppose you are a contributor to a large project, and you want to add a
complicated feature, and to present it to the other developers in a way
that makes it easy for them to read your changes, verify that they are
correct, and understand why you made each change.
If you present all of your changes as a single patch (or commit), they
may find that it is too much to digest all at once.
If you present them with the entire history of your work, complete with
mistakes, corrections, and dead ends, they may be overwhelmed.
So the ideal is usually to produce a series of patches such that:
1. Each patch can be applied in order.
2. Each patch includes a single logical change, together with a
message explaining the change.
3. No patch introduces a regression: after applying any initial
part of the series, the resulting project still compiles and
works, and has no bugs that it didn't have before.
4. The complete series produces the same end result as your own
(probably much messier!) development process did.
We will introduce some tools that can help you do this, explain how to
use them, and then explain some of the problems that can arise because
you are rewriting history.
Keeping a patch series up to date using git-rebase
Suppose that you create a branch "mywork" on a remote-tracking branch
"origin", and create some commits on top of it:
$ git checkout -b mywork origin
$ vi file.txt
$ git commit
$ vi otherfile.txt
$ git commit
You have performed no merges into mywork, so it is just a simple linear
sequence of patches on top of "origin":
o--o--o <-- origin
o--o--o <-- mywork
Some more interesting work has been done in the upstream project, and
"origin" has advanced:
o--o--O--o--o--o <-- origin
a--b--c <-- mywork
At this point, you could use "pull" to merge your changes back in;
the result would create a new merge commit, like this:
o--o--O--o--o--o <-- origin
\ \
a--b--c--m <-- mywork
However, if you prefer to keep the history in mywork a simple series of
commits without any merges, you may instead choose to use
$ git checkout mywork
$ git rebase origin
This will remove each of your commits from mywork, temporarily saving
them as patches (in a directory named ".dotest"), update mywork to
point at the latest version of origin, then apply each of the saved
patches to the new mywork. The result will look like:
o--o--O--o--o--o <-- origin
a'--b'--c' <-- mywork
In the process, it may discover conflicts. In that case it will stop
and allow you to fix the conflicts; after fixing conflicts, use "git
add" to update the index with those contents, and then, instead of
running git-commit, just run
$ git rebase --continue
and git will continue applying the rest of the patches.
At any point you may use the --abort option to abort this process and
return mywork to the state it had before you started the rebase:
$ git rebase --abort
Modifying a single commit
We saw in <<fixing-a-mistake-by-editing-history>> that you can replace the
most recent commit using
$ git commit --amend
which will replace the old commit by a new commit incorporating your
changes, giving you a chance to edit the old commit message first.
You can also use a combination of this and gitlink:git-rebase[1] to edit
commits further back in your history. First, tag the problematic commit with
$ git tag bad mywork~5
(Either gitk or git-log may be useful for finding the commit.)
Then check out that commit, edit it, and rebase the rest of the series
on top of it (note that we could check out the commit on a temporary
branch, but instead we're using a <<detached-head,detached head>>):
$ git checkout bad
$ # make changes here and update the index
$ git commit --amend
$ git rebase --onto HEAD bad mywork
When you're done, you'll be left with mywork checked out, with the top
patches on mywork reapplied on top of your modified commit. You can
then clean up with
$ git tag -d bad
Note that the immutable nature of git history means that you haven't really
"modified" existing commits; instead, you have replaced the old commits with
new commits having new object names.
Reordering or selecting from a patch series
Given one existing commit, the gitlink:git-cherry-pick[1] command
allows you to apply the change introduced by that commit and create a
new commit that records it. So, for example, if "mywork" points to a
series of patches on top of "origin", you might do something like:
$ git checkout -b mywork-new origin
$ gitk origin..mywork &
And browse through the list of patches in the mywork branch using gitk,
applying them (possibly in a different order) to mywork-new using
cherry-pick, and possibly modifying them as you go using commit
Another technique is to use git-format-patch to create a series of
patches, then reset the state to before the patches:
$ git format-patch origin
$ git reset --hard origin
Then modify, reorder, or eliminate patches as preferred before applying
them again with gitlink:git-am[1].
Other tools
There are numerous other tools, such as stgit, which exist for the
purpose of maintaining a patch series. These are outside of the scope of
this manual.
Problems with rewriting history
The primary problem with rewriting the history of a branch has to do
with merging. Suppose somebody fetches your branch and merges it into
their branch, with a result something like this:
o--o--O--o--o--o <-- origin
\ \
t--t--t--m <-- their branch:
Then suppose you modify the last three commits:
o--o--o <-- new head of origin
o--o--O--o--o--o <-- old head of origin
If we examined all this history together in one repository, it will
look like:
o--o--o <-- new head of origin
o--o--O--o--o--o <-- old head of origin
\ \
t--t--t--m <-- their branch:
Git has no way of knowing that the new head is an updated version of
the old head; it treats this situation exactly the same as it would if
two developers had independently done the work on the old and new heads
in parallel. At this point, if someone attempts to merge the new head
in to their branch, git will attempt to merge together the two (old and
new) lines of development, instead of trying to replace the old by the
new. The results are likely to be unexpected.
You may still choose to publish branches whose history is rewritten,
and it may be useful for others to be able to fetch those branches in
order to examine or test them, but they should not attempt to pull such
branches into their own work.
For true distributed development that supports proper merging,
published branches should never be rewritten.
Advanced branch management
Fetching individual branches
Instead of using gitlink:git-remote[1], you can also choose just
to update one branch at a time, and to store it locally under an
arbitrary name:
$ git fetch origin todo:my-todo-work
The first argument, "origin", just tells git to fetch from the
repository you originally cloned from. The second argument tells git
to fetch the branch named "todo" from the remote repository, and to
store it locally under the name refs/heads/my-todo-work.
You can also fetch branches from other repositories; so
$ git fetch git:// master:example-master
will create a new branch named "example-master" and store in it the
branch named "master" from the repository at the given URL. If you
already have a branch named example-master, it will attempt to
"fast-forward" to the commit given by's master branch. So
next we explain what a fast-forward is:
Understanding git history: fast-forwards
In the previous example, when updating an existing branch, "git
fetch" checks to make sure that the most recent commit on the remote
branch is a descendant of the most recent commit on your copy of the
branch before updating your copy of the branch to point at the new
commit. Git calls this process a "fast forward".
A fast forward looks something like this:
o--o--o--o <-- old head of the branch
o--o--o <-- new head of the branch
In some cases it is possible that the new head will *not* actually be
a descendant of the old head. For example, the developer may have
realized she made a serious mistake, and decided to backtrack,
resulting in a situation like:
o--o--o--o--a--b <-- old head of the branch
o--o--o <-- new head of the branch
In this case, "git fetch" will fail, and print out a warning.
In that case, you can still force git to update to the new head, as
described in the following section. However, note that in the
situation above this may mean losing the commits labeled "a" and "b",
unless you've already created a reference of your own pointing to
Forcing git fetch to do non-fast-forward updates
If git fetch fails because the new head of a branch is not a
descendant of the old head, you may force the update with:
$ git fetch git:// +master:refs/remotes/example/master
Note the addition of the "+" sign. Be aware that commits that the
old version of example/master pointed at may be lost, as we saw in
the previous section.
Configuring remote branches
We saw above that "origin" is just a shortcut to refer to the
repository that you originally cloned from. This information is
stored in git configuration variables, which you can see using
$ git config -l
If there are other repositories that you also use frequently, you can
create similar configuration options to save typing; for example,
$ git config remote.example.url git://
then the following two commands will do the same thing:
$ git fetch git:// master:refs/remotes/example/master
$ git fetch example master:refs/remotes/example/master
Even better, if you add one more option:
$ git config remote.example.fetch master:refs/remotes/example/master
then the following commands will all do the same thing:
$ git fetch git:// master:ref/remotes/example/master
$ git fetch example master:ref/remotes/example/master
$ git fetch example example/master
$ git fetch example
You can also add a "+" to force the update each time:
$ git config remote.example.fetch +master:ref/remotes/example/master
Don't do this unless you're sure you won't mind "git fetch" possibly
throwing away commits on mybranch.
Also note that all of the above configuration can be performed by
directly editing the file .git/config instead of using
See gitlink:git-config[1] for more details on the configuration
options mentioned above.
Git internals
Git depends on two fundamental abstractions: the "object database", and
the "current directory cache" aka "index".
The Object Database
The object database is literally just a content-addressable collection
of objects. All objects are named by their content, which is
approximated by the SHA1 hash of the object itself. Objects may refer
to other objects (by referencing their SHA1 hash), and so you can
build up a hierarchy of objects.
All objects have a statically determined "type" aka "tag", which is
determined at object creation time, and which identifies the format of
the object (i.e. how it is used, and how it can refer to other
objects). There are currently four different object types: "blob",
"tree", "commit", and "tag".
A <<def_blob_object,"blob" object>> cannot refer to any other object,
and is, as the name implies, a pure storage object containing some
user data. It is used to actually store the file data, i.e. a blob
object is associated with some particular version of some file.
A <<def_tree_object,"tree" object>> is an object that ties one or more
"blob" objects into a directory structure. In addition, a tree object
can refer to other tree objects, thus creating a directory hierarchy.
A <<def_commit_object,"commit" object>> ties such directory hierarchies
together into a <<def_DAG,directed acyclic graph>> of revisions - each
"commit" is associated with exactly one tree (the directory hierarchy at
the time of the commit). In addition, a "commit" refers to one or more
"parent" commit objects that describe the history of how we arrived at
that directory hierarchy.
As a special case, a commit object with no parents is called the "root"
object, and is the point of an initial project commit. Each project
must have at least one root, and while you can tie several different
root objects together into one project by creating a commit object which
has two or more separate roots as its ultimate parents, that's probably
just going to confuse people. So aim for the notion of "one root object
per project", even if git itself does not enforce that.
A <<def_tag_object,"tag" object>> symbolically identifies and can be
used to sign other objects. It contains the identifier and type of
another object, a symbolic name (of course!) and, optionally, a
Regardless of object type, all objects share the following
characteristics: they are all deflated with zlib, and have a header
that not only specifies their type, but also provides size information
about the data in the object. It's worth noting that the SHA1 hash
that is used to name the object is the hash of the original data
plus this header, so `sha1sum` 'file' does not match the object name
for 'file'.
(Historical note: in the dawn of the age of git the hash
was the sha1 of the 'compressed' object.)
As a result, the general consistency of an object can always be tested
independently of the contents or the type of the object: all objects can
be validated by verifying that (a) their hashes match the content of the
file and (b) the object successfully inflates to a stream of bytes that
forms a sequence of <ascii type without space> + <space> + <ascii decimal
size> + <byte\0> + <binary object data>.
The structured objects can further have their structure and
connectivity to other objects verified. This is generally done with
the `git-fsck` program, which generates a full dependency graph
of all objects, and verifies their internal consistency (in addition
to just verifying their superficial consistency through the hash).
The object types in some more detail:
Blob Object
A "blob" object is nothing but a binary blob of data, and doesn't
refer to anything else. There is no signature or any other
verification of the data, so while the object is consistent (it 'is'
indexed by its sha1 hash, so the data itself is certainly correct), it
has absolutely no other attributes. No name associations, no
permissions. It is purely a blob of data (i.e. normally "file
In particular, since the blob is entirely defined by its data, if two
files in a directory tree (or in multiple different versions of the
repository) have the same contents, they will share the same blob
object. The object is totally independent of its location in the
directory tree, and renaming a file does not change the object that
file is associated with in any way.
A blob is typically created when gitlink:git-update-index[1]
is run, and its data can be accessed by gitlink:git-cat-file[1].
Tree Object
The next hierarchical object type is the "tree" object. A tree object
is a list of mode/name/blob data, sorted by name. Alternatively, the
mode data may specify a directory mode, in which case instead of
naming a blob, that name is associated with another TREE object.
Like the "blob" object, a tree object is uniquely determined by the
set contents, and so two separate but identical trees will always
share the exact same object. This is true at all levels, i.e. it's
true for a "leaf" tree (which does not refer to any other trees, only
blobs) as well as for a whole subdirectory.
For that reason a "tree" object is just a pure data abstraction: it
has no history, no signatures, no verification of validity, except
that since the contents are again protected by the hash itself, we can
trust that the tree is immutable and its contents never change.
So you can trust the contents of a tree to be valid, the same way you
can trust the contents of a blob, but you don't know where those
contents 'came' from.
Side note on trees: since a "tree" object is a sorted list of
"filename+content", you can create a diff between two trees without
actually having to unpack two trees. Just ignore all common parts,
and your diff will look right. In other words, you can effectively
(and efficiently) tell the difference between any two random trees by
O(n) where "n" is the size of the difference, rather than the size of
the tree.
Side note 2 on trees: since the name of a "blob" depends entirely and
exclusively on its contents (i.e. there are no names or permissions
involved), you can see trivial renames or permission changes by
noticing that the blob stayed the same. However, renames with data
changes need a smarter "diff" implementation.
A tree is created with gitlink:git-write-tree[1] and
its data can be accessed by gitlink:git-ls-tree[1].
Two trees can be compared with gitlink:git-diff-tree[1].
Commit Object
The "commit" object is an object that introduces the notion of
history into the picture. In contrast to the other objects, it
doesn't just describe the physical state of a tree, it describes how
we got there, and why.
A "commit" is defined by the tree-object that it results in, the
parent commits (zero, one or more) that led up to that point, and a
comment on what happened. Again, a commit is not trusted per se:
the contents are well-defined and "safe" due to the cryptographically
strong signatures at all levels, but there is no reason to believe
that the tree is "good" or that the merge information makes sense.
The parents do not have to actually have any relationship with the
result, for example.
Note on commits: unlike real SCM's, commits do not contain
rename information or file mode change information. All of that is
implicit in the trees involved (the result tree, and the result trees
of the parents), and describing that makes no sense in this idiotic
file manager.
A commit is created with gitlink:git-commit-tree[1] and
its data can be accessed by gitlink:git-cat-file[1].
An aside on the notion of "trust". Trust is really outside the scope
of "git", but it's worth noting a few things. First off, since
everything is hashed with SHA1, you 'can' trust that an object is
intact and has not been messed with by external sources. So the name
of an object uniquely identifies a known state - just not a state that
you may want to trust.
Furthermore, since the SHA1 signature of a commit refers to the
SHA1 signatures of the tree it is associated with and the signatures
of the parent, a single named commit specifies uniquely a whole set
of history, with full contents. You can't later fake any step of the
way once you have the name of a commit.
So to introduce some real trust in the system, the only thing you need
to do is to digitally sign just 'one' special note, which includes the
name of a top-level commit. Your digital signature shows others
that you trust that commit, and the immutability of the history of
commits tells others that they can trust the whole history.
In other words, you can easily validate a whole archive by just
sending out a single email that tells the people the name (SHA1 hash)
of the top commit, and digitally sign that email using something
like GPG/PGP.
To assist in this, git also provides the tag object...
Tag Object
Git provides the "tag" object to simplify creating, managing and
exchanging symbolic and signed tokens. The "tag" object at its
simplest simply symbolically identifies another object by containing
the sha1, type and symbolic name.
However it can optionally contain additional signature information
(which git doesn't care about as long as there's less than 8k of
it). This can then be verified externally to git.
Note that despite the tag features, "git" itself only handles content
integrity; the trust framework (and signature provision and
verification) has to come from outside.
A tag is created with gitlink:git-mktag[1],
its data can be accessed by gitlink:git-cat-file[1],
and the signature can be verified by
The "index" aka "Current Directory Cache"
The index is a simple binary file, which contains an efficient
representation of a virtual directory content at some random time. It
does so by a simple array that associates a set of names, dates,
permissions and content (aka "blob") objects together. The cache is
always kept ordered by name, and names are unique (with a few very
specific rules) at any point in time, but the cache has no long-term
meaning, and can be partially updated at any time.
In particular, the index certainly does not need to be consistent with
the current directory contents (in fact, most operations will depend on
different ways to make the index 'not' be consistent with the directory
hierarchy), but it has three very important attributes:
'(a) it can re-generate the full state it caches (not just the
directory structure: it contains pointers to the "blob" objects so
that it can regenerate the data too)'
As a special case, there is a clear and unambiguous one-way mapping
from a current directory cache to a "tree object", which can be
efficiently created from just the current directory cache without
actually looking at any other data. So a directory cache at any one
time uniquely specifies one and only one "tree" object (but has
additional data to make it easy to match up that tree object with what
has happened in the directory)
'(b) it has efficient methods for finding inconsistencies between that
cached state ("tree object waiting to be instantiated") and the
current state.'
'(c) it can additionally efficiently represent information about merge
conflicts between different tree objects, allowing each pathname to be
associated with sufficient information about the trees involved that
you can create a three-way merge between them.'
Those are the ONLY three things that the directory cache does. It's a
cache, and the normal operation is to re-generate it completely from a
known tree object, or update/compare it with a live tree that is being
developed. If you blow the directory cache away entirely, you generally
haven't lost any information as long as you have the name of the tree
that it described.
At the same time, the index is at the same time also the
staging area for creating new trees, and creating a new tree always
involves a controlled modification of the index file. In particular,
the index file can have the representation of an intermediate tree that
has not yet been instantiated. So the index can be thought of as a
write-back cache, which can contain dirty information that has not yet
been written back to the backing store.
The Workflow
Generally, all "git" operations work on the index file. Some operations
work *purely* on the index file (showing the current state of the
index), but most operations move data to and from the index file. Either
from the database or from the working directory. Thus there are four
main combinations:
working directory -> index
You update the index with information from the working directory with
the gitlink:git-update-index[1] command. You
generally update the index information by just specifying the filename
you want to update, like so:
$ git-update-index filename
but to avoid common mistakes with filename globbing etc, the command
will not normally add totally new entries or remove old entries,
i.e. it will normally just update existing cache entries.
To tell git that yes, you really do realize that certain files no
longer exist, or that new files should be added, you
should use the `--remove` and `--add` flags respectively.
NOTE! A `--remove` flag does 'not' mean that subsequent filenames will
necessarily be removed: if the files still exist in your directory
structure, the index will be updated with their new status, not
removed. The only thing `--remove` means is that update-cache will be
considering a removed file to be a valid thing, and if the file really
does not exist any more, it will update the index accordingly.
As a special case, you can also do `git-update-index --refresh`, which
will refresh the "stat" information of each index to match the current
stat information. It will 'not' update the object status itself, and
it will only update the fields that are used to quickly test whether
an object still matches its old backing store object.
index -> object database
You write your current index file to a "tree" object with the program
$ git-write-tree
that doesn't come with any options - it will just write out the
current index into the set of tree objects that describe that state,
and it will return the name of the resulting top-level tree. You can
use that tree to re-generate the index at any time by going in the
other direction:
object database -> index
You read a "tree" file from the object database, and use that to
populate (and overwrite - don't do this if your index contains any
unsaved state that you might want to restore later!) your current
index. Normal operation is just
$ git-read-tree <sha1 of tree>
and your index file will now be equivalent to the tree that you saved
earlier. However, that is only your 'index' file: your working
directory contents have not been modified.
index -> working directory
You update your working directory from the index by "checking out"
files. This is not a very common operation, since normally you'd just
keep your files updated, and rather than write to your working
directory, you'd tell the index files about the changes in your
working directory (i.e. `git-update-index`).
However, if you decide to jump to a new version, or check out somebody
else's version, or just restore a previous tree, you'd populate your
index file with read-tree, and then you need to check out the result
$ git-checkout-index filename
or, if you want to check out all of the index, use `-a`.
NOTE! git-checkout-index normally refuses to overwrite old files, so
if you have an old version of the tree already checked out, you will
need to use the "-f" flag ('before' the "-a" flag or the filename) to
'force' the checkout.
Finally, there are a few odds and ends which are not purely moving
from one representation to the other:
Tying it all together
To commit a tree you have instantiated with "git-write-tree", you'd
create a "commit" object that refers to that tree and the history
behind it - most notably the "parent" commits that preceded it in
Normally a "commit" has one parent: the previous state of the tree
before a certain change was made. However, sometimes it can have two
or more parent commits, in which case we call it a "merge", due to the
fact that such a commit brings together ("merges") two or more
previous states represented by other commits.
In other words, while a "tree" represents a particular directory state
of a working directory, a "commit" represents that state in "time",
and explains how we got there.
You create a commit object by giving it the tree that describes the
state at the time of the commit, and a list of parents:
$ git-commit-tree <tree> -p <parent> [-p <parent2> ..]
and then giving the reason for the commit on stdin (either through
redirection from a pipe or file, or by just typing it at the tty).
git-commit-tree will return the name of the object that represents
that commit, and you should save it away for later use. Normally,
you'd commit a new `HEAD` state, and while git doesn't care where you
save the note about that state, in practice we tend to just write the
result to the file pointed at by `.git/HEAD`, so that we can always see
what the last committed state was.
Here is an ASCII art by Jon Loeliger that illustrates how
various pieces fit together.
commit obj
| |
| |
| Object DB |
| Backing |
| Store |
write-tree | |
tree obj | |
| | read-tree
| | tree obj
| Index |
| "cache" |
update-index ^
blob obj | |
| |
checkout-index -u | | checkout-index
stat | | blob obj
| Working |
| Directory |
Examining the data
You can examine the data represented in the object database and the
index with various helper tools. For every object, you can use
gitlink:git-cat-file[1] to examine details about the
$ git-cat-file -t <objectname>
shows the type of the object, and once you have the type (which is
usually implicit in where you find the object), you can use
$ git-cat-file blob|tree|commit|tag <objectname>
to show its contents. NOTE! Trees have binary content, and as a result
there is a special helper for showing that content, called
`git-ls-tree`, which turns the binary content into a more easily
readable form.
It's especially instructive to look at "commit" objects, since those
tend to be small and fairly self-explanatory. In particular, if you
follow the convention of having the top commit name in `.git/HEAD`,
you can do
$ git-cat-file commit HEAD
to see what the top commit was.
Merging multiple trees
Git helps you do a three-way merge, which you can expand to n-way by
repeating the merge procedure arbitrary times until you finally
"commit" the state. The normal situation is that you'd only do one
three-way merge (two parents), and commit it, but if you like to, you
can do multiple parents in one go.
To do a three-way merge, you need the two sets of "commit" objects
that you want to merge, use those to find the closest common parent (a
third "commit" object), and then use those commit objects to find the
state of the directory ("tree" object) at these points.
To get the "base" for the merge, you first look up the common parent
of two commits with
$ git-merge-base <commit1> <commit2>
which will return you the commit they are both based on. You should
now look up the "tree" objects of those commits, which you can easily
do with (for example)
$ git-cat-file commit <commitname> | head -1
since the tree object information is always the first line in a commit
Once you know the three trees you are going to merge (the one "original"
tree, aka the common case, and the two "result" trees, aka the branches
you want to merge), you do a "merge" read into the index. This will
complain if it has to throw away your old index contents, so you should
make sure that you've committed those - in fact you would normally
always do a merge against your last commit (which should thus match what
you have in your current index anyway).
To do the merge, do
$ git-read-tree -m -u <origtree> <yourtree> <targettree>
which will do all trivial merge operations for you directly in the
index file, and you can just write the result out with
Merging multiple trees, continued
Sadly, many merges aren't trivial. If there are files that have
been added.moved or removed, or if both branches have modified the
same file, you will be left with an index tree that contains "merge
entries" in it. Such an index tree can 'NOT' be written out to a tree
object, and you will have to resolve any such merge clashes using
other tools before you can write out the result.
You can examine such index state with `git-ls-files --unmerged`
command. An example:
$ git-read-tree -m $orig HEAD $target
$ git-ls-files --unmerged
100644 263414f423d0e4d70dae8fe53fa34614ff3e2860 1 hello.c
100644 06fa6a24256dc7e560efa5687fa84b51f0263c3a 2 hello.c
100644 cc44c73eb783565da5831b4d820c962954019b69 3 hello.c
Each line of the `git-ls-files --unmerged` output begins with
the blob mode bits, blob SHA1, 'stage number', and the
filename. The 'stage number' is git's way to say which tree it
came from: stage 1 corresponds to `$orig` tree, stage 2 `HEAD`
tree, and stage3 `$target` tree.
Earlier we said that trivial merges are done inside
`git-read-tree -m`. For example, if the file did not change
from `$orig` to `HEAD` nor `$target`, or if the file changed
from `$orig` to `HEAD` and `$orig` to `$target` the same way,
obviously the final outcome is what is in `HEAD`. What the
above example shows is that file `hello.c` was changed from
`$orig` to `HEAD` and `$orig` to `$target` in a different way.
You could resolve this by running your favorite 3-way merge
program, e.g. `diff3` or `merge`, on the blob objects from
these three stages yourself, like this:
$ git-cat-file blob 263414f... >hello.c~1
$ git-cat-file blob 06fa6a2... >hello.c~2
$ git-cat-file blob cc44c73... >hello.c~3
$ merge hello.c~2 hello.c~1 hello.c~3
This would leave the merge result in `hello.c~2` file, along
with conflict markers if there are conflicts. After verifying
the merge result makes sense, you can tell git what the final
merge result for this file is by:
$ mv -f hello.c~2 hello.c
$ git-update-index hello.c
When a path is in unmerged state, running `git-update-index` for
that path tells git to mark the path resolved.
The above is the description of a git merge at the lowest level,
to help you understand what conceptually happens under the hood.
In practice, nobody, not even git itself, uses three `git-cat-file`
for this. There is `git-merge-index` program that extracts the
stages to temporary files and calls a "merge" script on it:
$ git-merge-index git-merge-one-file hello.c
and that is what higher level `git merge -s resolve` is implemented with.
How git stores objects efficiently: pack files
We've seen how git stores each object in a file named after the
object's SHA1 hash.
Unfortunately this system becomes inefficient once a project has a
lot of objects. Try this on an old project:
$ git count-objects
6930 objects, 47620 kilobytes
The first number is the number of objects which are kept in
individual files. The second is the amount of space taken up by
those "loose" objects.
You can save space and make git faster by moving these loose objects in
to a "pack file", which stores a group of objects in an efficient
compressed format; the details of how pack files are formatted can be
found in link:technical/pack-format.txt[technical/pack-format.txt].
To put the loose objects into a pack, just run git repack:
$ git repack
Generating pack...
Done counting 6020 objects.
Deltifying 6020 objects.
100% (6020/6020) done
Writing 6020 objects.
100% (6020/6020) done
Total 6020, written 6020 (delta 4070), reused 0 (delta 0)
Pack pack-3e54ad29d5b2e05838c75df582c65257b8d08e1c created.
You can then run
$ git prune
to remove any of the "loose" objects that are now contained in the
pack. This will also remove any unreferenced objects (which may be
created when, for example, you use "git reset" to remove a commit).
You can verify that the loose objects are gone by looking at the
.git/objects directory or by running
$ git count-objects
0 objects, 0 kilobytes
Although the object files are gone, any commands that refer to those
objects will work exactly as they did before.
The gitlink:git-gc[1] command performs packing, pruning, and more for
you, so is normally the only high-level command you need.
Dangling objects
The gitlink:git-fsck[1] command will sometimes complain about dangling
objects. They are not a problem.
The most common cause of dangling objects is that you've rebased a
branch, or you have pulled from somebody else who rebased a branch--see
<<cleaning-up-history>>. In that case, the old head of the original
branch still exists, as does obviously everything it pointed to. The
branch pointer itself just doesn't, since you replaced it with another
There are also other situations too that cause dangling objects. For
example, a "dangling blob" may arise because you did a "git add" of a
file, but then, before you actually committed it and made it part of the
bigger picture, you changed something else in that file and committed
that *updated* thing - the old state that you added originally ends up
not being pointed to by any commit or tree, so it's now a dangling blob
Similarly, when the "recursive" merge strategy runs, and finds that
there are criss-cross merges and thus more than one merge base (which is
fairly unusual, but it does happen), it will generate one temporary
midway tree (or possibly even more, if you had lots of criss-crossing
merges and more than two merge bases) as a temporary internal merge
base, and again, those are real objects, but the end result will not end
up pointing to them, so they end up "dangling" in your repository.
Generally, dangling objects aren't anything to worry about. They can
even be very useful: if you screw something up, the dangling objects can
be how you recover your old tree (say, you did a rebase, and realized
that you really didn't want to - you can look at what dangling objects
you have, and decide to reset your head to some old dangling state).
For commits, the most useful thing to do with dangling objects tends to
be to do a simple
$ gitk <dangling-commit-sha-goes-here> --not --all
For blobs and trees, you can't do the same, but you can examine them.
You can just do
$ git show <dangling-blob/tree-sha-goes-here>
to show what the contents of the blob were (or, for a tree, basically
what the "ls" for that directory was), and that may give you some idea
of what the operation was that left that dangling object.
Usually, dangling blobs and trees aren't very interesting. They're
almost always the result of either being a half-way mergebase (the blob
will often even have the conflict markers from a merge in it, if you
have had conflicting merges that you fixed up by hand), or simply
because you interrupted a "git fetch" with ^C or something like that,
leaving _some_ of the new objects in the object database, but just
dangling and useless.
Anyway, once you are sure that you're not interested in any dangling
state, you can just prune all unreachable objects:
$ git prune
and they'll be gone. But you should only run "git prune" on a quiescent
repository - it's kind of like doing a filesystem fsck recovery: you
don't want to do that while the filesystem is mounted.
(The same is true of "git-fsck" itself, btw - but since
git-fsck never actually *changes* the repository, it just reports
on what it found, git-fsck itself is never "dangerous" to run.
Running it while somebody is actually changing the repository can cause
confusing and scary messages, but it won't actually do anything bad. In
contrast, running "git prune" while somebody is actively changing the
repository is a *BAD* idea).
Notes and todo list for this manual
This is a work in progress.
The basic requirements:
- It must be readable in order, from beginning to end, by
someone intelligent with a basic grasp of the unix
commandline, but without any special knowledge of git. If
necessary, any other prerequisites should be specifically
mentioned as they arise.
- Whenever possible, section headings should clearly describe
the task they explain how to do, in language that requires
no more knowledge than necessary: for example, "importing
patches into a project" rather than "the git-am command"
Think about how to create a clear chapter dependency graph that will
allow people to get to important topics without necessarily reading
everything in between.
Say something about .gitignore.
Scan Documentation/ for other stuff left out; in particular:
some of technical/?
list of commands in gitlink:git[1]
Scan email archives for other stuff left out
Scan man pages to see if any assume more background than this manual
Simplify beginning by suggesting disconnected head instead of
temporary branch creation?
Add more good examples. Entire sections of just cookbook examples
might be a good idea; maybe make an "advanced examples" section a
standard end-of-chapter section?
Include cross-references to the glossary, where appropriate.
Document shallow clones? See draft 1.5.0 release notes for some
Add a section on working with other version control systems, including
CVS, Subversion, and just imports of series of release tarballs.
More details on gitweb?
Write a chapter on using plumbing and writing scripts.