AI Hints
Git & GitHub · Lesson 107 of 144

Following the Pointers

Trees and blobs.

⏱ 2 min read▶ Video at 1:57:38🎓 Free
Watch this lesson in the Git & GitHub courseWatch on YouTube ↗Full Git & GitHub course

Let's follow the pointers from a commit, all the way down to a file's contents. Every object is named by its hash.

Following the Pointers
From the video: Following the Pointers
  • From commit
  • To tree
  • To blob

From tree to blob

Cat file on the commit's tree, using head, a caret, and curly braces with the word tree. It lists our three files, each with the hash of its blob. Now cat file on the index dot HTML blob. There's the file's content, exactly as committed.

git
git cat-file -p "HEAD^{tree}"
100644 blob 058b39baaca23d0c9c5cf8c9f2e0ef8c351e460c    about.html
100644 blob 93dd595b334c23932a183a8a4b3e98bdad2c9494    index.html
100644 blob 693ce9b84057d37872139150e32284abbbe76754    style.css
git cat-file -p 93dd595
<h1>My Site</h1>
<p>Welcome!</p>
<a href="about.html">About</a>

Why this is clever

Because objects are named by their content, two identical files are stored only once. A commit that changes one file reuses all the other blobs from before. That's why commits are fast, and history takes so little space, even after thousands of commits.

Named by contentthe hash
Identical filesstored once
New commitreuses old blobs
Resultfast and small

Summary

Commits point to trees, trees point to blobs, and identical content is stored once. Next, what a branch really is.

Key takeaways
  • git cat-file -p
  • Objects named by hash
  • Shared, not copied
Next up: What a Branch Really Is — a file with a hash