The Plumbing#

So far we’ve been using Git in a “porcelain” manner. But to sate our insatiable curiosity, let’s take a look at some of the “plumbing”.

It’s Just Files All the Way Down#

All the data in a Git repository is stored directly in the (hidden) .git directory. That includes all the commits, branches, tags, and other objects we’ll learn about later.

Git is made up of objects that are stored in the .git/objects directory. A commit is just a type of object.

Assignment#

  1. Use git log -n 10 to find your commit hash again.
  2. Use ls -al to list the contents of the .git/objects directory.
  3. Look for a directory that suspiciously matches the first two characters of your commit hash.
  4. Use ls -al to list the contents of that directory and answer the question.

Cat File#

Luckily, Git has a built-in plumbing command, cat-file, that allows us to see the contents of a commit without needing to futz around with the object files directly.

git cat-file -p <hash>

Assignment#

  1. Use the cat-file command to view the contents of your last commit.
  2. For the CLI to be able to check your output, do it again, but redirect the output to a temporary file:
[catfile command here] > /tmp/catfileout.txt

Trees and Blobs#

Now that we understand some of our plumbing equipment, let’s get into the pipes. Here are some terms to know:

  • tree: git’s way of storing a directory
  • blob: git’s way of storing a file

Here’s what I got when I inspected my last commit:

> git cat-file -p 5ba786fcc93e8092831c01e71444b9baa2228a4f

tree 4e507fdc6d9044ccd8a4a3061324c9f711c4667d
author alperen <the.alperen@aol.com> 1705891256 -0700
committer alperen <the.alperen@aol.com> 1705891256 -0700

A: add contents.md

Notice that we can see:

  • The tree object
  • The author
  • The committer
  • The commit message

However, we cannot see the contents of the contents.md file itself! That’s because the blob object stores it.

Storing Data#

Git stores an entire snapshot of files on a per-commit level. This was a surprise to me! I always assumed each commit only stored the changes made in that commit.

Optimization#

While it’s true that Git stores entire snapshots, it does have some performance optimizations so that your .git directory doesn’t get too unbearably large.

  • Git compresses and packs files to store them more efficiently.
  • Git deduplicates files that are the same across different commits. If a file doesn’t change between commits, Git will only store it once.
Backward gitignore