Git is a file system. We need a database for the code
Summary
The article argues that Git functions as a content-addressable filesystem rather than a true database for code, and advocates building a CRDT-based revision control system that operates on AST-like trees. It outlines the monorepo, overlay, merge, and data-model challenges in current SCM, surveys CRDT-based approaches like DeltaDB and RDX, and sketches a roadmap for a future database-centric code revision system.