Zeroact is a custom React-like framework written from scratch with the primary goal of learning React internals by implementing them, instead of treating React as a black box.

This project focuses on how React works, not on feature completeness or production use.

The main goal of Zeroact is to:

• Understand React internals deeply
• Learn how rendering, reconciliation, and hooks work under the hood
• Gain confidence reading and reasoning about React source code
• Build strong mental models instead of relying on abstractions

Zeroact is actively used in SquidPong, my final school project, to prove that the concepts are understood and usable in a real application.

• ⚛️ Function components only (no class components)
• 🧵 Fiber-based architecture
• 🔁 Incremental rendering with a scheduler
• 🪝 Custom hooks implementation:
  • useState
  • useEffect
  • useRef
  • useCallback
  • useContext
• 🔄 Localized re-renders (subtree updates, not full root re-renders)
• 🧠 Custom reconciliation logic
• 🧩 JSX support

Zeroact re-implements core React ideas, including:

• Virtual element creation
• Fiber tree creation and traversal
• Work loop and scheduling
• Hook state storage and ordering
• Effect queues and execution
• Update propagation through the fiber tree

Every part exists to be read, debugged, and understood.

Zeroact powers the UI layer of SquidPong, a multiplayer game built as a final school project.

Using Zeroact in a real project validates:

• The correctness of the architecture
• Hook behavior under real interaction
• State updates, effects, and rendering performance

• Production readiness
• Full React API compatibility
• Performance optimizations beyond learning needs

• React source code
• Build Your Own React by Paul O’Shannessy
• Redux and React architecture patterns

Zeroact is a learning project.