What is Git?

Git is a content-addressable, distributed version control system created by Linus Torvalds in 2005 to manage Linux kernel development. Unlike centralized VCS (SVN, CVS), every clone contains the full repository — its complete history, all branches, all tags.

Storage model: Git stores snapshots, not diffs. Each commit references a tree object (the directory state at that point), which references blob objects (file contents). Identical content is deduplicated by SHA-1 (transitioning to SHA-256). Diffs are computed on demand at display time.

Distributed semantics: a 'remote' is just another Git repository, typically reachable over SSH/HTTPS. There is no privileged 'central' server in the protocol — that role is a convention. You fetch (download) and push (upload) commits between repositories. Conflicts are detected at merge time, not at the file lock level.

Object model (4 types): blob (file content), tree (directory listing), commit (snapshot pointer + metadata + parent pointers), tag (named reference, optionally signed). Refs (refs/heads/main, refs/tags/v1.0) are pointers from human-readable names to commits. HEAD is a symbolic ref pointing to the current branch.

Why distributed beats centralized: offline work, branch-cheap experimentation (a branch is one ref pointer, not a copy), full local history (instant git log/git blame), redundancy (every clone is a backup). Trade-off: more concepts to learn upfront, no implicit locking.

Git vs GitHub vs GitLab: Git is the open-source CLI/library. Hosting providers add web UI, CI/CD, issues, pull/merge requests, access control. The Git protocol (the wire format) is identical across hosts — git push doesn't care whether the remote is GitHub, GitLab, or a bare repo on your own server.