git ready

learn git one commit at a time
by Nick Quaranto

what's inside your .git directory

committed 23 Mar 2009

UPDATE: I’ve recieved some very helpful comments regarding corrections to the various files and what they do. Thanks for letting me know and keeping me on my toes.

One of the things I like best about Git is that it keeps all of its information in one place: your .git directory in your project’s root. If you haven’t been digging around it yet, don’t fret! There’s plenty of goodies to be had within it. Let’s take a look into the important files and folders and try to get a better sense of what’s going on under the hood.

The basic structure looks like this:

.
|-- COMMIT_EDITMSG
|-- FETCH_HEAD
|-- HEAD
|-- ORIG_HEAD
|-- branches
|-- config
|-- description
|-- hooks
|   |-- applypatch-msg
|   |-- commit-msg
|   |-- post-commit
|   |-- post-receive
|   |-- post-update
|   |-- pre-applypatch
|   |-- pre-commit
|   |-- pre-rebase
|   |-- prepare-commit-msg
|   `-- update
|-- index
|-- info
|   `-- exclude
|-- logs
|   |-- HEAD
|   `-- refs
|-- objects
`-- refs
    |-- heads
    |-- remotes
    |-- stash
    `-- tags

Let’s go over some of the normal files that you may see living in the base directory:

  • COMMIT_EDITMSG: This is the last commit’s message. It’s not actually used by Git at all, but it’s there mostly for your reference after you made a commit.
  • config: Contains settings for this repository. Specific configuration variables can be dumped in here (and even aliases!) What this file is most used for is defining where remotes live and some core settings, such as if your repository is bare or not.
  • description: If you’re using gitweb or firing up git instaweb, this will show up when you view your repository or the list of all versioned repositories.
  • FETCH_HEAD: The SHAs of branch/remote heads that were updated during the last git fetch
  • HEAD: The current ref that you’re looking at. In most cases it’s probably refs/heads/master
  • index: The staging area with meta-data such as timestamps, file names and also SHAs of the files that are already wrapped up by Git.
  • packed-refs: Packs away dormant refs, this is not the definitive list of refs in your repository (the refs folder has the real ones!) Take a look at gitster’s comment to see more information on this.
  • ORIG_HEAD: When doing a merge, this is the SHA of the branch you’re merging into.
  • MERGE_HEAD: When doing a merge, this is the SHA of the branch you’re merging from.
  • MERGE_MODE: Used to communicate constraints that were originally given to git merge to git commit when a merge conflicts, and a separate git commit is needed to conclude it. Currently --no-ff is the only constraints passed this way.
  • MERGE_MSG: Enumerates conflicts that happen during your current merge.
  • RENAMED-REF: Still trying to track this one down. From a basic grep through the source, it seems like this file is related to errors when saving refs.

There’s plenty of directories as well:

  • hooks: A directory that will fast become your best friend: this contains scripts that are executed at certain times when working with Git, such as after a commit or before a rebase. An entire series of articles will be coming about hooks.
  • info: Relatively uninteresting except for the exclude file that lives inside of it. We’ve seen this before in the ignoring files article, but as a reminder, you can use this file to ignore files for this project, but beware! It’s not versioned like a .gitignore file would be.
  • logs: Contains history for different branches. Seems to be used mostly with the reflog command.
  • objects: Git’s internal warehouse of blobs, all indexed by SHAs.
  • rebase-apply: The workbench for rebasing and for git am. You can dig into its patch file when it does not apply cleanly if you’re brave.
  • refs: The master copy of all refs that live in your repository, be they for stashes, tags, remote tracking branches, or local branches.

Just one word of wisdom when messing around with Git internals: make sure you know what you’re doing, and if not, have a backup! Messing around with the config files or hooks is pretty simple but I wouldn’t go spelunking in the datastore if I didn’t have to. If for some reason you are as part of your normal workflow, you might be doing it wrong.

There’s plenty more about the internals of Git that we haven’t covered yet. One of the best guides is the Git Community Book, and of course, you can just download the source for yourself and take a look. If you have more information about what’s going on in the .git folder, let us know in the comments!