bot_vault

从零开发一个微型 Git

2 min read

Summary

Li Jiahao builds a minimal Git implementation in Node.js to explain how core Git commands work under the hood. The article walks through git init, git status, git add, git commit, git reset, and git branch, explaining the internal data structures (.git folder, objects, refs) that make each command work. Building from scratch reveals why Git behaves the way it does.

李佳浩用 Node.js 從頭實現一個簡易版 Git，解釋核心 Git 命令的底層原理。文章逐一解析 git init、status、add、commit、reset 和 branch 背後的數據結構（.git 資料夾、objects、refs），從實作角度揭示 Git 的工作原理。

Key Points

  • git init creates the .git folder with specific sub-directories (objects, refs, HEAD)
  • Objects are content-addressable: file content is hashed (SHA) and stored by that hash — identical content = same hash = no duplication
  • git add computes object hashes and writes them to the index (staging area)
  • git commit creates a commit object pointing to a tree of file hashes, which points to blobs
  • Branches are simply pointers (text files in .git/refs) to commit hashes
  • git checkout reads the tree referenced by the target commit and restores working directory files

Insights

Implementing Git from scratch clarifies the content-addressable storage model: Git doesn’t track files, it tracks content snapshots. Two identical files across different commits are the same object. This explains why git operations are fast (O(1) content lookup), why diffs are computed on-the-fly, and why storage is efficient. The branch-as-lightweight-pointer insight explains why branching in Git is nearly free compared to other VCS.

Connections

Raw Excerpt

这不就是javascript\python里的Map/dict或Set结构吗… 事实上这就是内容寻址存储的本质，通过键值对的方式进行内容寻址，在算法上只需O(1)的时间复杂度就可以完成，效率很高！

Graph View

Backlinks