World map

 

BM

World 10: Deep Git

Objects and hashes

You will learn: what Git really stores, and how each piece gets its name.

The problem

Your shop has 3 commits. style.css never changed. Does Git store it 3 times?

Watch

Git keeps everything as object: One thing Git stores: a blob, a tree, a commit or a tag. Its name is a hash of its content.. git cat-file -p HEAD opens the newest commit. It points to a tree: An object that lists one folder: names that point to blobs (files) and trees (folders). (the files). Its parent is the commit before.

Try it

A blob: An object that holds the content of one file. No name, just the content. is the content of one file. Open the commit, then its tree, then css. Find the blob of css/style.css.

Find the blob that holds css/style.css.

Pick a commit to look inside.

Check

Three kinds of object. Which one holds each thing?

Drag a card into a box. Or tap the card, then tap the box.

        Watch

        Each object’s name is a SHA-1: The recipe Git uses today to make a 40-character id from content. hash (commit id): A unique code that names one commit, like a1b2c3d.. It has 40 characters, made from the content. The short id is just the start.

        Predict

        You copy about.html to about-copy.html and commit. How many new blobs does Git store?

        Try it

        The fourth kind is a tag object: a label with a note. Open v1.0 and see where it points.

        c02eb601a02b3e9e8ddf8

        You did it!

        You can now open Git’s objects and name all four kinds.

        Go deeper (optional)

        Annotated tags (git tag -a) are objects. Lightweight tags are only refs. Git compresses each object and keeps it in .git/objects. git cat-file -t prints an object’s type. git cat-file -s prints its size. Git 3.0 plans SHA-256 ids for new repos.