Context
The problem
The best way I know to really understand Git is to build it: storing files, comparing versions and merging changes, without any libraries.
Approach
How I built it
Stores files, folders and commits by their SHA-1 hash, the same way Git does.
Uses the Myers diff algorithm and three-way merges, and adds conflict markers when two changes clash.
Comes with a small built-in web server and dashboard, plus a static version that runs on Vercel.
Result
What came out of it
No outside libraries
Real diff and merge algorithms
A live dashboard