Git is all about composing and saving snapshots of your project and then working with and comparing those snapshots. This section will explain the commands needed to compose and commit snapshots of your project.
An important concept here is that Git has an 'index', which acts as sort of a staging area for your snapshot. This allows you to build up a series of well composed snapshots from changed files in your working directory, rather than having to commit all of the file changes at once.
In a nutshell, you will use git add to start tracking new
files and also to stage changes to already tracked files, then
git status and git diff to see what has been
modified and staged and finally git commit to record your
snapshot into your history. This will be the basic workflow that you use
most of the time.
In Git, you have to add file contents to your staging area before you
can commit them. If the file is new, you can run git add
to initially add the file to your staging area, but even if the file
is already "tracked" - ie, it was in your last commit - you still need
to call git add to add new modifications to your staging
area. Let's see a few examples of this.
Going back to our Hello World example, once we've initiated the project,
we would now start adding our files to it and we would do that with
git add. We can use git status to see what the
state of our project is.
$ git status -s ?? README ?? hello.rbSo right now we have two untracked files. We can now add them.
$ git add README hello.rbNow if we run
git status again, we'll see that they've been
added.
$ git status -s A README A hello.rb
It is also common to recursively add all files in a new project by specifying
the current working directory like this: git add .. Since Git
will recursively add all files under a directory you give it, if you give it
the current working directory, it will simply start tracking every file
there. In this case, a git add . would have done the same
thing as a git add README hello.rb, or for that matter so would
git add *, but that's only because we don't have subdirectories
which the * would not recurse into.
OK, so now if we edit one of these files and run git status
again, we will see something odd.
$ vim README $ git status -s AM README A hello.rb
The 'AM' status means that the file has been modified on disk since we
last added it. This means that if we commit our snapshot right now, we will
be recording the version of the file when we last ran git add,
not the version that is on our disk. Git does not assume that what the file
looks like on disk is necessarily what you want to snapshot - you have to
tell Git with the git add command.
In a nutshell,
you run git add on a file when you want to
include whatever changes you've made to it in your next commit snapshot.
Anything you've changed that is not added will not be included - this means
you can craft your snapshots with a bit more precision than most other SCM
systems.
For a very interesting example of using this flexibility to stage only
parts of modified files at a time, see the '-p' option to
git add in the Pro Git book.
As you saw in the git add section, in order to see what the
status of your staging area is compared to the code in your working
directory, you can run the git status command. I demonstrated
using it with the -s option, which gives you short output.
Without that flag, the git status command will give you more
context and hints. Here is the same status output with and without the
-s. The short output looks like this:
$ git status -s AM README A hello.rbWhere the same status with the long output looks like this:
$ git status # On branch master # # Initial commit # # Changes to be committed: # (use "git rm --cached..." to unstage) # # new file: README # new file: hello.rb # # Changed but not updated: # (use "git add ..." to update what will be committed) # (use "git checkout -- ..." to discard changes in working directory) # # modified: README #
You can easily see how much more compact the short output is, but the long output has useful tips and hints as to what commands you may want to use next.
Git will also tell you about files that were deleted since your last commit or files that were modified or staged since your last commit.
$ git status -s M README D hello.rbYou can see there are two columns in the short status output. The first column is for the staging area, the second is for the working directory. So for example, if you have the README file staged and then you modify it again without running
git add a second time, you'll see
this:
$ git status -s MM README D hello.rb
In a nutshell,
you run git status to see if anything has been modified
and/or staged since your last commit so you can decide if you want to
commit a new snapshot and what will be recorded in it.
There are two main uses of the git diff command. One use we
will describe here, the other we will describe later in the
"Inspection and Comparison"
section. The way we're going to use it here is to describe the
changes that are staged or modified on disk but unstaged.
Without any extra arguments, a simple git diff will display
in unified diff format (a patch) what code or content you've changed in your
project since the last commit that are not yet staged for the next commit
snapshot.
$ vim hello.rb $ git status -s M hello.rb $ git diff diff --git a/hello.rb b/hello.rb index d62ac43..8d15d50 100644 --- a/hello.rb +++ b/hello.rb @@ -1,7 +1,7 @@ class HelloWorld def self.hello - puts "hello world" + puts "hola mundo" end end
So where git status will show you what files have changed
and/or been staged since your last commit, git diff will
show you what those changes actually are, line by line. It's generally
a good follow-up command to git status
The git diff --cached command will show you what contents
have been staged. That is, this will show you the changes that will
currently go into the next commit snapshot. So, if you were to stage
the change to hello.rb in the example above,
git diff by itself won't show you any output because it will
only show you what is not yet staged.
$ git status -s M hello.rb $ git add hello.rb $ git status -s M hello.rb $ git diff $
If you want to see the staged changes, you can run
git diff --cached instead.
$ git status -s M hello.rb $ git diff $ $ git diff --cached diff --git a/hello.rb b/hello.rb index d62ac43..8d15d50 100644 --- a/hello.rb +++ b/hello.rb @@ -1,7 +1,7 @@ class HelloWorld def self.hello - puts "hello world" + puts "hola mundo" end end
If you want to see both staged and unstaged changes together, you
can run git diff HEAD - this basically means you want to
see the difference between your working directory and the last commit,
ignoring the staging area. If we make another change to our
hello.rb file then we'll have some changes staged and some
changes unstaged. Here are what all three diff commands
will show you:
$ vim hello.rb $ git diff diff --git a/hello.rb b/hello.rb index 4f40006..2ae9ba4 100644 --- a/hello.rb +++ b/hello.rb @@ -1,7 +1,7 @@ class HelloWorld + # says hello def self.hello puts "hola mundo" end end $ git diff --cached diff --git a/hello.rb b/hello.rb index 2aabb6e..4f40006 100644 --- a/hello.rb +++ b/hello.rb @@ -1,7 +1,7 @@ class HelloWorld def self.hello - puts "hello world" + puts "hola mundo" end end $ git diff HEAD diff --git a/hello.rb b/hello.rb index 2aabb6e..2ae9ba4 100644 --- a/hello.rb +++ b/hello.rb @@ -1,7 +1,8 @@ class HelloWorld + # says hello def self.hello - puts "hello world" + puts "hola mundo" end end
If we don't want the full diff output, but we want more than the
git status output, we can use the --stat
option, which will give us a summary of changes instead. Here is the
same example as above, but using the --stat option instead.
$ git status -s MM hello.rb $ git diff --stat hello.rb | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) $ git diff --cached --stat hello.rb | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) $ git diff HEAD --stat hello.rb | 3 ++- 1 files changed, 2 insertions(+), 1 deletions(-)
You can also provide a file path at the end of any of these options
to limit the diff output to a specific file or subdirectory.
In a nutshell,
you run git diff to see details of the git status
command - how files have been modified or staged on a line by line
basis.
Now that you have staged the content you want to snapshot with the
git add command, you run git commit to actually
record the snapshot.
Git records your name and email address with every commit you make,
so the first step is to tell Git what these are.
$ git config --global user.name 'Your Name' $ git config --global user.email you@somedomain.com
Let's stage and commit all the changes to our
hello.rb file. In this first example, we'll use the
-m option to provide the commit message on the command line.
$ git add hello.rb
$ git status -s
M hello.rb
$ git commit -m 'my hola mundo changes'
[master 68aa034] my hola mundo changes
1 files changed, 2 insertions(+), 1 deletions(-)
Now we have recorded the snapshot. If we run git status
again, we will see that we have a "clean working directory", which means
that we have not made any changes since our last commit - there is no
un-snapshotted work in our checkout.
$ git status # On branch master nothing to commit (working directory clean)
If you leave off the -m option, Git will try to open a
text editor for you to write your commit message. In vim,
which it will default to if it can find nothing else in your settings,
the screen might look something like this:
# Please enter the commit message for your changes. Lines starting # with '#' will be ignored, and an empty message aborts the commit. # On branch master # Changes to be committed: # (use "git reset HEAD..." to unstage) # # modified: hello.rb # ~ ~ ".git/COMMIT_EDITMSG" 9L, 257C
At this point you add your actual commit message at the top of the
document. Any lines starting with '#' will be ignored - Git will put
the output of the git status command in there for you as
a reminder of what you have modified and staged.
In general, it's very important to write a good commit message. For open source projects, it's generally a rule to write your message more or less in this format:
Short (50 chars or less) summary of changes More detailed explanatory text, if necessary. Wrap it to about 72 characters or so. In some contexts, the first line is treated as the subject of an email and the rest of the text as the body. The blank line separating the summary from the body is critical (unless you omit the body entirely); some git tools can get confused if you run the two together. Further paragraphs come after blank lines. - Bullet points are okay, too - Typically a hyphen or asterisk is used for the bullet, preceded by a single space, with blank lines in between, but conventions vary here # Please enter the commit message for your changes. Lines starting # with '#' will be ignored, and an empty message aborts the commit. # On branch master # Changes to be committed: # (use "git reset HEAD..." to unstage) # # modified: hello.rb # ~ ~ ~ ".git/COMMIT_EDITMSG" 25L, 884C written
The commit message is very important. Since much of the power of Git is this flexibility in carefully crafting commits locally and then sharing them later, it is very powerful to be able to write three or four commits of logically separate changes so that your work may be more easily peer reviewed. Since there is a separation between committing and pushing those changes, do take the time to make it easier for the people you are working with to see what you've done by putting each logically separate change in a separate commit with a nice commit message so it is easier for them to see what you are doing and why.
If you think the git add stage of the workflow is too
cumbersome, Git allows you to skip that part with the -a
option. This basically tells Git to run git add on any file
that is "tracked" - that is, any file that was in your last commit and
has been modified. This allows you to do a more Subversion style workflow
if you want, simply editing files and then running git commit -a
when you want to snapshot everything that has been changed. You still need
to run git add to start tracking new files, though, just like
Subversion.
$ vim hello.rb $ git status -s M hello.rb $ git commit -m 'changes to hello file' # On branch master # Changed but not updated: # (use "git add..." to update what will be committed) # (use "git checkout -- ..." to discard changes in working directory) # # modified: hello.rb # no changes added to commit (use "git add" and/or "git commit -a") $ git commit -am 'changes to hello file' [master 78b2670] changes to hello file 1 files changed, 2 insertions(+), 1 deletions(-)
Notice how if you don't stage any changes and then run
git commit, Git will simply give you the output of the
git status command, reminding you that nothing is staged.
I've highlighted the important part of that message, saying that nothing
is added to be committed. If you use -a, it will add and
commit everything at once.
This now lets you complete the entire snapshotting workflow - you
make changes to your files, then use git add to stage
files you want to change, git status and git diff
to see what you've changed, and then finally git commit
to actually record the snapshot forever.
In a nutshell,
you run git commit to record the snapshot of your staged
content. This snapshot can then be compared, shared and reverted to
if you need to.
git reset is probably the most confusing command written
by humans. I've been using Git for years, even wrote a book on it and I
still get confused by what it is going to do at times. So, I'll just
tell you the three specific invocations of it that are generally
helpful and ask you to blindly use it as I do - because it can be
very useful.
In the first case, we can use it to unstage something that you have
accidentally staged. Let's say that you have modified two files and want
to record them into two different commits. You should stage and commit
one, then stage and commit the other. If you accidentally stage both of
them, how do you un-stage one? You do it with
git reset HEAD -- file. Technically here you don't have to
add the -- - it is used to tell Git when you have stopped
listing options and are now listing file paths, but it's probably good to
get into the habit of using it to separate options from paths even if you
don't need to.
So, let's see what it looks like to unstage something. Here we have two files that have been modified since our last commit. We will stage both, then unstage one of them.
$ git status -s M README M hello.rb $ git add . $ git status -s M README M hello.rb $ git reset HEAD -- hello.rb Unstaged changes after reset: M hello.rb $ git status -s M README M hello.rb
Now you can run a git commit which will just record
the changes to the README file, not the now unstaged
hello.rb.
In case you're curious, what it's actually doing here is it is resetting
the checksum of the entry for that file in the "index" to be what it was
in the last commit. Since git add checksums a file and adds
it to the "index", git reset HEAD overwrites that with what
it was before, thereby effectively unstaging it.
If you want to be able to just run git unstage, you can easily
setup an alias in Git. Just run
git config --global alias.unstage "reset HEAD".
Once you have run that, you can then just run
git unstage [file] instead.
If you forget the command to unstage something, Git is helpful in
reminding you in the output of the normal git status
command. For example, if you run git status without
the -s when you have staged files, it will tell you
how to unstage them:
$ git status # On branch master # Changes to be committed: # (use "git reset HEAD..." to unstage) # # modified: README # modified: hello.rb #
When you run git reset without specifying a flag
it defaults to --mixed. The other options are
--soft and --hard.
The first thing git reset does is undo the last
commit and put the files back onto the stage. If you include the
--soft flag this is where it stops. For example,
if you run git reset --soft HEAD~ (the parent of the
HEAD) the last commit will be undone and the files touched
will be back on the stage again.
$ git status -s M hello.rb $ git commit -am 'hello with a flower' [master 5857ac1] hello with a flower 1 files changed, 3 insertions(+), 1 deletions(-) $ git status # On branch master nothing to commit (working directory clean) $ git reset --soft HEAD~ $ git status -s M hello.rb
This is basically doing the same thing as
git commit --amend, allowing you to do more work
before you roll in the file changes into the same commit.
The third option is to go --hard and make your working
directory look like the index, unstage files and undo the last commit.
This is the most dangerous option and not working directory safe. Any
changes not in the index or have not been commited will be lost.
$ git status # On branch master # Changes to be committed: # (use "git reset HEAD..." to unstage) # # modified: README # # Changes not staged for commit: # (use "git add ..." to update what will be committed) # (use "git checkout -- ..." to discard changes in working directory) # # modified: README # $ git reset --hard HEAD HEAD is now at 5857ac1 hello with a flower $ git status # On branch master nothing to commit (working directory clean)
In the above example, while we had both changes ready to commit and
ready to stage, a git reset --hard wiped them out.
On top of that, the last commit has been undone.
You can replace HEAD with a commit SHA-1 or another
parent reference to reset to that specific point.
In a nutshell,
you run git reset HEAD to undo the last commit, unstage
files that you previously ran git add on and wish to not
include in the next commit snapshot
git rm will remove entries from the staging area.
This is a bit different from git reset HEAD which "unstages"
files. By "unstage" I mean it reverts the staging area to what was
there before we started modifying things. git rm on the
other hand just kicks the file off the stage entirely, so that it's not
included in the next commit snapshot, thereby effectively deleting it.
By default, a git rm file will remove the file from the
staging area entirely and also off your disk (the working directory). To
leave the file in the working directory, you can use git rm --cached
.
Unlike most other version control systems, Git does not track file renames.
Instead, it just tracks the snapshots and then figures out what files were
likely renamed by comparing snapshots. If a file was removed from one
snapshot and another file was added to the next one and the contents are
similar, Git figures it was most likely a rename. So, although the
git mv command exists, it is superfluous - all it does is a
git rm --cached, moves the file on disk, then runs a
git add on the new file. You don't really need to use it, but
if it's easier, feel free.
I personally don't use this command that much in it's normal form - to
delete files. It's often easier to just remove the files off your disk and
then run a git commit -a, which will automatically remove them
from your index, too.
In a nutshell,
you run git rm to remove files from being tracked in Git. It
will also remove them from your working directory.