kosha

Kosha

⚡Kosha: A Modern, Lightweight, and High-Performance State Management Library for React

Kosha is a production-ready, minimalistic global state management solution for modern React applications. At just ~450 bytes minzipped, it's optimized for performance-critical applications, full React 18+ support, and clean developer ergonomics.

Live demo: https://kosha-six.vercel.app

📚 Table of Contents

• 🚀 Features
• 📦 Installation
• 🧑‍💻 Usage
• 🧰 Immer Middleware
• ⚖️ Zustand Comparison
• 📁 Examples
• ❓ FAQ
• 🚧 Known Limitations
• 🤝 Contributing
• 🧪 Testing and Mocking
• 🧠 Internals & Caveats
• 📜 License

🚀 Features

• Ultra-Lightweight

  • Minzipped size: ~420–460 bytes, ideal for bundle-sensitive projects.

• Optimized by Default: Zero Unnecessary Re-renders

  • Uses useSyncExternalStore for subscription handling and memoization.
  • No need to manually optimize with shallow equality — Kosha uses JSON.stringify internally to compare selector outputs.

• Partial State Updates

  • Update only what you need without spreading old state:

    set({ count });
    set(state => ({ count: state.count + 1 }));
    

• Concurrent Rendering Ready

  • Fully supports React 18+ and concurrent features.

• Flexible Consumption API

  • Consume whole store or optimized slices via selectors.

    const count = useKosha(state => state.count); // optimized - re-renders only when count changes
    const { count, setCount } = useKosha(); // re-renders for every state change
    

• Middleware Architecture (BYO Middleware)

  • Middleware support is built-in. While Kosha doesn’t yet include a plugin ecosystem like Zustand, you can define and compose custom middlewares easily.
  • Includes working persist middleware and immer middleware example out-of-the-box

📦 Installation

Install using your preferred package manager:

pnpm add kosha

or

npm install kosha

or

yarn add kosha

🧑‍💻 Usage

1. Create a Store

import { create } from "kosha";

const useKosha = create(set => ({
  count: 0,
  increment: () => set(state => ({ count: state.count + 1 })),
}));

2. Consume the Store (Full Access)

const { count, increment } = useKosha();

This will trigger re-render even when count or increment are not changed but other states [part of the same store] change.

3. Select Specific State

const count = useKosha(state => state.count);
const increment = useKosha(state => state.increment);

// or use shorthand
const { count, increment } = useKosha(({ count, increment }) => ({ count, increment }));

4. Update Outside React

useKosha.getState().increment();

⚠️ Avoid In-Place Mutations

Kosha detects changes using shallow comparison (JSON.stringify) between the previous and new result of the selector. If you mutate state in-place and return the same object reference, listeners will not be triggered, and your UI will not re-render — even though the state has technically changed.

🔴 Problem Example

create(set => ({
  todos: [],
  addTodo: item =>
    set(state => {
      state.todos.push(item); // ❌ in-place mutation
      return state; // same reference!
    }),
}));

✅ Correct Way – Return a New Object

create(set => ({
  todos: [],
  addTodo: item =>
    set(state => ({
      ...state,
      todos: [...state.todos, item], // ✅ returns a new object
    })),
}));

✅ Even Better – Use Immer

import { produce } from "immer";

create(set => ({
  todos: [],
  addTodo: item =>
    set(
      produce(state => {
        state.todos.push(item);
      }),
    ),
}));

ℹ️ Kosha does not bundle immer by default, but you can use it safely with your own setup or use immer middleware.

🧰 Immer Middleware

Kosha provides a convenient immer middleware to simplify immutable state updates by enabling you to write mutative logic inside the store setter. It internally applies immer's produce function automatically.

Example usage:

import { create } from "kosha";
import { immer } from "kosha/middleware";

const useKosha = create(
  immer(set => ({
    todos: [],
    addTodo: (item: string) =>
      set(state => {
        state.todos.push(item); // safe mutable update with immer middleware
      }),
  })),
);

This middleware allows you to write concise, mutable-looking update code while keeping the state immutable under the hood.

You can combine immer middleware with other middlewares like persist for powerful state management.

🧩 Modular Slice Composition

For large or modular applications, it's often useful to split your store logic into isolated, reusable slices. Kosha supports this pattern with a simple and intuitive API that lets you compose your global store from smaller, focused units of state and behavior.

✅ Why Use Slices?

• Promote separation of concerns and cleaner organization.
• Encapsulate logic around a specific domain (e.g., auth, UI, settings).
• Enable independent testing and reusability of slices.
• Access the full store inside each slice via get() — helpful when cross-slice coordination is needed.

📦 Example

interface CounterSlice {
  count: number;
  setCount: (count: number) => void;
}

interface ThemeSlice {
  theme: string;
  setTheme: (theme: string) => void;
}

type StoreType = CounterSlice & ThemeSlice;

const createCounterSlice: SliceCreator<StoreType, CounterSlice> = set => ({
  count: 0,
  setCount: count => set({ count }),
});

const createThemeSlice: SliceCreator<StoreType, ThemeSlice> = set => ({
  theme: "light",
  setTheme: theme => set({ theme }),
});

const useStore = create<StoreType>((...a) => ({
  ...createCounterSlice(...a),
  ...createThemeSlice(...a),
}));

🔍 Notes

• Each slice gets access to set() and get() functions.
• Kosha allows full-store access within any slice, offering more flexibility than Zustand’s default slice typings, where type constraints may prevent access to the full state.
• You can still strongly type your slices using generics — though we're continuing to improve TypeScript ergonomics for advanced use cases.

Feature / Concern	Kosha	Zustand
Slice Access to Store	Full access to entire store via get()	Often restricted to own slice unless manually cast or configured
DX with TypeScript	More transparent	Auto inferred slice typing, but can obscure full-store access
Slice Reusability	Easily extractable and composable	Extractable, but full-store interop requires extra typing
Cross-Slice Communication	Direct via get()	Indirect or requires casting

⚖️ Why Choose Kosha Over Zustand?

Feature	Kosha	Zustand
Size (minzipped)	~450 bytes	~0.6–2.5kB+ (depending on usage)
Optimized Selectors	✅ Built-in via stringify	⚠️ Manual / shallow equality
Partial Updates	✅	✅
Middleware Support	✅ (custom)	✅ (rich plugin ecosystem)
Devtools	❌ (custom only)	✅
Plugin Ecosystem	🚧 (in development)	✅

Kosha is perfect for apps that prioritize bundle size, simplicity, and predictable updates without the need for extensive tooling.

📁 Examples

Explore the examples/ directory for working codebases, including component and selector examples.

❓ FAQ

1. Is Kosha production-ready?

Yes. Kosha is small, stable, and well-tested.

2. Does Kosha support async logic?

Absolutely. Define async functions inside the store:

const useStore = create(set => ({
  fetchUser: async () => {
    const user = await fetch("/api/user").then(res => res.json());
    set({ user });
  },
}));

3. Does Kosha support middleware like Zustand?

Yes. Middleware support is built-in. A working persist middleware is included. You can easily build your own or extend with logging, devtools, etc.

4. My state updates, but components don’t re-render. Why?

You might be mutating the existing state object in-place instead of returning a new one. Kosha relies on reference comparison (JSON.stringify) to detect changes. Always return a new object, or use libraries like immer or the immer middleware from kosha/middleware to handle immutable updates safely.

5. Can I use it with Set, Map, or Date objects?

While Date serializes fine, avoid storing Set or Map directly in global state, since Kosha uses JSON.stringify to diff selector outputs. Use arrays or plain objects instead for best results.

6. Isn’t JSON.stringify unreliable because key order might change?

No — in Kosha, you’re comparing outputs of the same selector function across renders. Since the order of keys in JavaScript objects is preserved in deterministic function outputs, JSON.stringify remains stable and reliable in this context.

🚧 Known Limitations

Kosha is intentionally minimal by design — built to offer just what most React apps need, without bloat. That comes with a few tradeoffs:

🧠 Selector Comparison via JSON.stringify

Kosha uses JSON.stringify internally to compare selector outputs for change detection. This works extremely well for the majority of cases — even with moderately large or deeply nested selectors.

However, there are a few caveats:

• Avoid non-serializable values in selectors like Set, Map, WeakMap, or circular objects.
• Very large selector outputs may incur performance costs during diffing.

To clarify: ✅ It’s perfectly fine to have a large store or global state. ⚠️ What matters is the output of your selector. If you’re selecting a large slice like state => ({ a: state.a, ..., z: state.z }), it’s more efficient to either:

• Access the store directly without a selector (useKosha()), or
• Extract only the minimal fields you actually need.

🔌 Plugin Ecosystem Still Growing

Kosha includes full support for custom middleware and already ships with a working persist middleware. However:

• Built-in plugins like devtools are not yet included.
• A community-driven plugin ecosystem is still in its early stages.

That said, the underlying architecture is solid and middleware-ready — you're free to build and compose your own middleware as needed.

🤝 Contributing

We welcome contributions!

• 💬 Start a discussion
• 🐛 Report bugs on the issue tracker
• 🧪 Submit PRs to improve or extend Kosha

Got an idea for a middleware plugin or improvement? We'd love to collaborate.

Here’s a polished and clarified version of your 🧪 Testing and Mocking section, optimized for developer clarity, best SEO, and practical DX. It clearly introduces the utilities, explains the pattern, and emphasizes what's already available via utils/test.

🧪 Testing and Mocking

Kosha provides built-in support for test-friendly patterns that allow you to inject and reset store state during testing.

🔧 Test Store Creation with Reset Support

For advanced test isolation, Kosha allows you to create test stores that can be reset to their initial state after each test.

We recommend using the utilities exported from utils/test:

import { create, resetAllStores } from "kosha/utils/test";

Here’s what they do:

// utils/test.ts
import { create as actualCreate, BaseType, StoreCreator } from "kosha";

export const storeResetFns = new Set<() => void>();

export const create = <T extends BaseType>(creator: StoreCreator<T>) => {
  const useStore = actualCreate(creator);
  const initial = useStore.getState();
  storeResetFns.add(() => useStore.setState(initial!, true));
  return useStore;
};

export const resetAllStores = () => storeResetFns.forEach(fn => fn());

This utility enables the following DX pattern in your test setup:

import { act } from "react-dom/test-utils";
import { afterEach } from "vitest"; // or jest
import { resetAllStores } from "kosha/utils/test";

afterEach(() => {
  act(() => resetAllStores());
});

🧪 Why It Matters

• ✅ Stable Testing: Reset store state between tests to prevent bleed-through.
• ✅ Mock with Confidence: Easily inject specific state values for targeted testing.
• ✅ No Manual Boilerplate: Reuse create() from utils/test for clean and repeatable test setups.

This approach ensures clean state isolation and minimal friction when writing tests for components relying on global state. It's also ideal for mocking feature flags, user sessions, themes, etc.

🧠 Internals & Caveats

🔍 Why use JSON.stringify for selector comparison?

Kosha compares previous and next selector outputs using JSON.stringify to prevent unnecessary re-renders. This approach has several benefits:

• It's fast and works for most serializable primitives and plain objects.
• It ensures deep comparison out-of-the-box without manual equality functions.

⚠️ What about key order in objects?

This is a common concern, but not an issue in Kosha:

Kosha always compares results of the same selector function, so the key order is preserved unless your selector behaves non-deterministically (e.g., relies on Object.keys() or mutation). As long as your selector returns a consistent structure, JSON.stringify comparison will be reliable.

📜 License

Kosha is licensed under the MPL-2.0 license.

Explore my courses or support development.

Built with 💖 by Mayank Kumar Chaudhari