Why does 'git commit' not save my changes?

问题:

I did a git commit -m "message" like this:

> git commit -m "save arezzo files"
# On branch master
# Changes not staged for commit:
#   (use "git add <file>..." to update what will be committed)
#   (use "git checkout -- <file>..." to discard changes in working directory)
#
#       modified:   arezzo.txt
#       modified:   arezzo.jsp
#
no changes added to commit (use "git add" and/or "git commit -a")

But afterwards, when I do git status it shows the same modified files:

> git status
# On branch master
# Changes not staged for commit:
#   (use "git add <file>..." to update what will be committed)
#   (use "git checkout -- <file>..." to discard changes in working directory)
#
#       modified:   arezzo.txt
#       modified:   arezzo.jsp
#
no changes added to commit (use "git add" and/or "git commit -a")

What am I doing wrong?

回答1:

As the message says:

no changes added to commit (use "git add" and/or "git commit -a")

Git has a "staging area" where files need to be added before being committed, you can read an explanation of it here.

For your specific example, you can use:

git commit -am "save arezzo files"

(note the extra a in the flags, can also be written as git commit -a -m "message" - both do the same thing)

Alternatively, if you want to be more selective about what you add to the commit, you use the git add command to add the appropriate files to the staging area, and git status to preview what is about to be added (remembering to pay attention to the wording used).

You can also find general documentation and tutorials for how to use git on the git documentation page which will give more detail about the concept of staging/adding files.

One other thing worth knowing about is interactive staging - this allows you to add parts of a file to the staging area, so if you've made three distinct code changes (for related but different functionality), you can use interactive mode to split the changes and add/commit each part in turn. Having smaller specific commits like this can be helpful.

回答2:

You didn't add the changes. Either specifically add them via

git add filename1 filename2

or add all changes (from root path of the project)

git add .

or use the shorthand -a while commiting:

git commit -a -m "message".

回答3:

You should do:

git commit . -m "save arezzo files"

回答4:

You could have done a:

git add -u -n

To check which files you modified and are going to be added (dry run: -n option), and then

git add -u

To add just modified files

回答5:

I copied a small sub project I had that was under Git source control into another project and forgot to delete the .git folder. When I went to commit I got the same message as above and couldn't clear it until I deleted the .git folder.

It is a bit silly, but it is worth checking you don't have a .git folder under the folder that doesn't commit.

回答6:

The reason why this is happening is because you have a folder that is already being tracked by Git inside another folder that is also tracked by Git. For example, I had a project and I added a subfolder to it. Both of them were being tracked by Git before I put one inside the other. In order to stop tracking the one inside, find it and remove the Git file with:

rm -rf .git

In my case I had a WordPress application and the folder I added inside was a theme. So I had to go to the theme root, and remove the Git file, so that the whole project would now be tracked by the parent, the WordPress application.

回答7:

I find this problem appearing when I've done a git add . in a subdirectory below where my .gitignore file lives (the home directory of my repository, so to speak). Try changing directories to your uppermost directory and running git add . followed by git commit -m "my commit message".

回答8:

Maybe an obvious thing, but...

If you have problem with the index, use git-gui. You get a very good view how the index (staging area) actually works.

Another source of information that helped me understand the index was Scott Chacons "Getting Git" page 259 and forward.

I started off using the command line because most documentation only showed that...

I think git-gui and gitk actually make me work faster, and I got rid of bad habits like "git pull" for example... Now I always fetch first... See what the new changes really are before I merge.

回答9:

I had an issue where I was doing commit --amend even after issuing a git add . and it still wasn't working. Turns out I made some .vimrc customizations and my editor wasn't working correctly. Fixing these errors so that vim returns the correct code resolved the issue.

回答10:

I had a very similar issue with the same error message. "Changes not staged for commit", yet when I do a diff it shows differences. I finally figured out that a while back I had changed a directories case. ex. "PostgeSQL" to "postgresql". As I remember now sometimes git will leave a file or two behind in the old case directory. Then you will commit a new version to the new case.

Thus git doesn't know which one to rely on. So to resolve it, I had to go onto the github's website. Then you're able to view both cases. And you must delete all the files in the incorrect cased directory. Be sure that you have the correct version saved off or in the correct cased directory.

Once you have deleted all the files in the old case directory, that whole directory will disappear. Then do a commit.

At this point you should be able to do a Pull on your local computer and not see the conflicts any more. Thus being able to commit again. :)

回答11:

if you have a subfolder, which was cloned from other git-Repository, first you have to remove the $.git$ file from the child-Repository: rm -rf .git after that you can change to parent folder and use git add -A.