Immutability Patterns

Create new objects instead of mutating existing ones. Immutability prevents shared-state bugs, makes code easier to reason about, and enables reliable change detection.

When to Use

• Writing any function that transforms data
• Updating React/Vue/Svelte state
• Working with Redux or other state management
• Reviewing code that modifies objects or arrays in place
• Any time .push(), .splice(), delete, or direct property assignment appears on shared data

Core Patterns

JavaScript/TypeScript: Object Updates

// WRONG: Mutation
function updateUser(user: User, name: string): User {
  user.name = name;  // Mutates the original!
  return user;
}

// CORRECT: Spread operator
function updateUser(user: User, name: string): User {
  return { ...user, name };
}

// CORRECT: Nested object update
function updateAddress(user: User, city: string): User {
  return {
    ...user,
    address: {
      ...user.address,
      city,
    },
  };
}

// CORRECT: Conditional field update
function toggleAdmin(user: User): User {
  return {
    ...user,
    role: user.role === "admin" ? "user" : "admin",
  };
}

JavaScript/TypeScript: Array Operations

// WRONG: Mutating arrays
function addItem(items: Item[], item: Item): Item[] {
  items.push(item);   // Mutates!
  return items;
}

// CORRECT: Immutable array operations
const added = [...items, newItem];                          // append
const prepended = [newItem, ...items];                      // prepend
const removed = items.filter((item) => item.id !== id);     // remove
const updated = items.map((item) =>                         // update
  item.id === id ? { ...item, name: "new" } : item
);
const inserted = [                                          // insert at index
  ...items.slice(0, index),
  newItem,
  ...items.slice(index),
];

TypeScript: Readonly Types

// Mark data as immutable at the type level
interface Config {
  readonly apiUrl: string;
  readonly timeout: number;
  readonly retries: number;
}

// Readonly arrays
function processItems(items: readonly Item[]): readonly Item[] {
  // items.push(x) -> TypeScript error!
  return items.filter((item) => item.active);
}

// Deep readonly utility
type DeepReadonly<T> = {
  readonly [K in keyof T]: T[K] extends object ? DeepReadonly<T[K]> : T[K];
};

// Readonly records
const routes: Readonly<Record<string, string>> = {
  home: "/",
  login: "/auth/login",
  dashboard: "/app/dashboard",
};

Deep Cloning and Complex Updates

// structuredClone for deep copies (no functions or symbols)
const copy = structuredClone(complexObject);

// Object.freeze for shallow immutability (runtime enforcement)
const config = Object.freeze({
  api: "https://api.example.com",
  timeout: 5000,
});
// config.timeout = 3000; -> throws in strict mode, silent in sloppy mode

// For complex nested updates, use a library like immer
import { produce } from "immer";

const nextState = produce(state, (draft) => {
  draft.users[0].address.city = "New York";  // Looks like mutation
  draft.items.push(newItem);                  // but produces a new object
});

Python: Immutable Patterns

from dataclasses import dataclass, replace
from typing import NamedTuple

# Use frozen dataclasses
@dataclass(frozen=True)
class User:
    name: str
    email: str
    role: str = "user"

# Create updated copy with replace()
user = User(name="Alice", email="[email protected]")
updated = replace(user, role="admin")  # New object, original unchanged

# Use tuples instead of lists for fixed collections
ALLOWED_ROLES = ("admin", "user", "viewer")

# Use frozenset for immutable sets
VALID_STATUSES = frozenset({"active", "inactive", "pending"})

# NamedTuple for immutable records
class Point(NamedTuple):
    x: float
    y: float

# Dict updates without mutation
original = {"a": 1, "b": 2}
updated = {**original, "b": 3, "c": 4}  # New dict

Go: Immutable Patterns

// Return new structs instead of modifying pointers
type User struct {
    Name  string
    Email string
    Role  string
}

// WRONG: Mutation via pointer
func (u *User) SetRole(role string) {
    u.Role = role
}

// CORRECT: Return new value
func (u User) WithRole(role string) User {
    u.Role = role  // Modifying the copy (value receiver)
    return u
}

// Immutable slice operations
func appendItem(items []Item, item Item) []Item {
    result := make([]Item, len(items)+1)
    copy(result, items)
    result[len(items)] = item
    return result
}

func removeItem(items []Item, index int) []Item {
    result := make([]Item, 0, len(items)-1)
    result = append(result, items[:index]...)
    result = append(result, items[index+1:]...)
    return result
}

Anti-Patterns

What NOT to Do

• Direct property assignment on shared objects: user.name = "new" — always spread into a new object.
• Array mutators on shared arrays: .push(), .pop(), .splice(), .sort(), .reverse() — use immutable alternatives or call on a copy.
• Assuming Object.freeze is deep: It only freezes the top level. Nested objects are still mutable.
• Over-cloning: Do not structuredClone on every operation — use spread for shallow updates, deep clone only when needed.

// BAD: Sorting mutates the original array
const sorted = items.sort((a, b) => a.name.localeCompare(b.name));

// GOOD: toSorted (ES2023) or spread + sort
const sorted = items.toSorted((a, b) => a.name.localeCompare(b.name));
const sorted = [...items].sort((a, b) => a.name.localeCompare(b.name));

// BAD: Reversing mutates
const reversed = items.reverse();

// GOOD: toReversed (ES2023) or spread + reverse
const reversed = items.toReversed();
const reversed = [...items].reverse();

Quick Reference

JavaScript/TypeScript Immutable Operations:
  Object update:    { ...obj, key: value }
  Array append:     [...arr, item]
  Array remove:     arr.filter(x => x.id !== id)
  Array update:     arr.map(x => x.id === id ? { ...x, ...changes } : x)
  Array sort:       arr.toSorted(fn)  or  [...arr].sort(fn)
  Array reverse:    arr.toReversed()  or  [...arr].reverse()
  Deep clone:       structuredClone(obj)
  Nested update:    immer produce()

TypeScript Types:
  readonly prop     Readonly<T>     ReadonlyArray<T>     as const

Python:
  @dataclass(frozen=True)   replace(obj, field=val)
  tuple()   frozenset()     {**dict, key: val}

Go:
  Value receivers   Return new structs   copy() for slices

Mutating Methods to AVOID (on shared data):
  .push() .pop() .shift() .unshift() .splice()
  .sort() .reverse() .fill() delete obj.key