Functype User Guide

Overview

Transform TypeScript code to use functype - a Scala-inspired functional programming library providing type-safe alternatives to null checks, exceptions, and imperative patterns. This skill helps integrate Option, Either, Try, List, IO, Task, and other functional types into projects.

When to Use This Skill

Trigger this skill when users:

• Convert imperative code to functional patterns
• Look up functype APIs or methods
• Handle nullable values or optional chaining
• Replace try-catch with functional error handling
• Work with immutable collections
• Compose side effects, async operations, or dependency injection with IO
• Run async operations with cancellation or progress tracking (Task)
• Debug functype code or understand error messages

Quick Start

Installation

npm install functype
# or
pnpm add functype

Core Imports

// Import from main bundle
import { Option, Either, Left, Right, Try, List, IO, Tag, Task, Layer } from "functype"

Constructor vs Companion Methods

Functype collections provide multiple ways to create instances:

Method	Use When	Example
List([...])	Creating from existing array	List(existingArray)
List.of(...)	Inline literal values	List.of(1, 2, 3)
List.empty()	Empty collections (typed)	List.empty<number>()

Decision Guide

Use Constructor List([...]) when:

• Converting existing arrays: List(data.items)
• Spreading iterables: List([...set])
• Variables holding arrays: List(myArray)

Use List.of(...) when:

• Inline literal values: List.of(1, 2, 3)
• Cleaner for small fixed lists: List.of("a", "b", "c")
• No need to wrap in array brackets

Use List.empty() when:

• Starting with empty collection: List.empty<User>()
• Type parameter needed but no initial values
• Returns singleton (efficient for repeated calls)

Examples

// Constructor - wrapping existing data
const users = List(fetchedUsers)
const items = List([...existingSet])

// .of() - inline literals
const colors = List.of("red", "green", "blue")
const primes = Set.of(2, 3, 5, 7, 11)

// .empty() - typed empty collections
const errors = List.empty<string>()
const cache = Map.empty<string, User>()

Pattern Conversion Guide

Null/Undefined Checks → Option

Before (Imperative):

if (value !== null && value !== undefined) {
  return value.toUpperCase()
}
return ""

After (Functype):

Option(value)
  .map((v) => v.toUpperCase())
  .orElse("")

Optional Chaining → Option Chain

Before:

const url = user?.profile?.avatar?.url

After:

const url = Option(user)
  .flatMap((u) => Option(u.profile))
  .flatMap((p) => Option(p.avatar))
  .map((a) => a.url)
  .orElse("/default-avatar.png")

Try-Catch → Try or Either

Before:

try {
  return JSON.parse(str)
} catch (e) {
  return null
}

After (with Try):

Try(() => JSON.parse(str))
  .toOption()
  .orElse(null)

After (with Either):

Try(() => JSON.parse(str))
  .toEither()
  .fold(
    (error) => `Parse failed: ${error.message}`,
    (data) => data,
  )

Array Operations → List

Before:

array.filter((x) => x > 0).map((x) => x * 2)

After:

List(array)
  .filter((x) => x > 0)
  .map((x) => x * 2)
  .toArray()

If-Else Chains → Cond

Before:

if (x > 10) {
  return "big"
} else if (x > 5) {
  return "medium"
} else {
  return "small"
}

After:

import { Cond } from "functype"

Cond.start<string>()
  .case(x > 10, "big")
  .case(x > 5, "medium")
  .otherwise("small")

Switch Statements → Match

Before:

switch (status) {
  case "success":
    return data
  case "error":
    return null
  default:
    return undefined
}

After:

import { Match } from "functype"

Match(status)
  .case("success", () => data)
  .case("error", () => null)
  .exhaustive()

Common Use Cases

Validation with Either

import { Either, Left, Right } from "functype"

function validateEmail(email: string): Either<string, string> {
  return email.includes("@") ? Right(email) : Left("Invalid email format")
}

function validateUser(user: any): Either<string, User> {
  return validateEmail(user.email)
    .map((email) => ({ ...user, email }))
    .flatMap((u) => (u.age >= 18 ? Right(u) : Left("Must be 18 or older")))
}

const result = validateUser({ email: "test@example.com", age: 20 }).fold(
  (error) => console.error(error),
  (user) => console.log("Valid user:", user),
)

Safe API Calls with Option

import { Option } from "functype"

interface User {
  id: string
  name: string
  email?: string
}

function getUserEmail(userId: string): Option<string> {
  return Option(fetchUser(userId))
    .flatMap((user) => Option(user.email))
    .filter((email) => email.includes("@"))
}

const email = getUserEmail("123").orElse("no-reply@example.com")

Error Recovery with Try

import { Try } from "functype"

const parseConfig = Try(() => JSON.parse(configStr))
  .recover((error) => {
    console.warn("Using default config:", error)
    return defaultConfig
  })
  .map((config) => validateConfig(config))

Collection Pipeline with List

import { List } from "functype"

const users = List([
  { name: "Alice", hobbies: ["reading", "coding"] },
  { name: "Bob", hobbies: ["gaming", "music"] },
])

const allHobbies = users
  .flatMap((user) => List(user.hobbies))
  .toSet() // Remove duplicates
  .toArray()

Additional Data Structures

IO<R,E,A> - Effect Type

IO is a lazy, composable effect type with typed errors and dependency injection:

• R = Requirements (environment/dependencies needed)
• E = Error type (typed failures)
• A = Success type (value produced on success)

import { IO, Tag, Layer } from "functype"

// Creation
IO.sync(() => computation()) // Sync operation
IO.succeed(value) // Pure success
IO.fail(error) // Pure failure
IO.async(() => promise) // Async operation
IO.tryPromise({
  // Promise with error mapping
  try: () => fetch(url),
  catch: (e) => new NetworkError(e),
})

// Dependency injection
const Database = Tag<DatabaseService>("Database")
const dbEffect = IO.service(Database) // Access a service
const program = dbEffect.flatMap((db) => IO.sync(() => db.query()))
program.provide(Layer.fromValue(Database, myDb)) // Provide deps

// Generator do-notation (cleaner syntax)
const program = IO.gen(function* () {
  const db = yield* IO.service(Database)
  const user = yield* IO.tryPromise(() => db.findUser(id))
  return user
})

// Execution
await effect.run() // Returns Promise<A>
effect.runSync() // Returns A (throws if async)
await effect.runEither() // Returns Promise<Either<E,A>>
await effect.runExit() // Returns Promise<Exit<E,A>>

// run() vs runExit(): Either has two branches, Exit has four. Use run() when all you
// need is success-or-not; use runExit() to tell a typed failure from a *defect* (a
// value that is not an E — a throwing IO.sync thunk or map/mapError callback, or
// IO.die) or from an interruption. run() puts both in the Left, raw.
const exit = await effect.runExit()
exit.isSuccess() // completed with a value
exit.isFailure() // a value from the declared E channel
exit.isDie() // a defect — NOT an E
exit.isInterrupted() // cancelled

// Both extra handlers are optional and fall back to the failure branch, so existing
// three-argument fold / three-key match calls keep working.
exit.fold(onFailure, onSuccess, onInterrupted?, onDie?)
exit.match({ Success, Failure, Interrupted, Die? })

// Defects stay recoverable: recover / recoverWith / fold / mapError treat a Die exactly
// as a Failure. Exit.Die records what the outcome was when nothing recovered it.

Tuple - Type-safe Fixed-length Array

import { Tuple } from "functype"

const pair = Tuple(42, "hello")
pair.first() // 42
pair.second() // "hello"
pair.mapFirst((x) => x * 2) // Tuple(84, "hello")
pair.swap() // Tuple("hello", 42)
pair.apply((a, b) => a + b.length) // 47
pair.concat(Tuple(true)) // Tuple(42, "hello", true)

Stack - Last-In-First-Out Collection

import { Stack } from "functype"

Stack.empty<number>()
const stack = Stack.of(1, 2, 3)
stack.push(value) // Returns new Stack
stack.pop() // Returns [Option<T>, Stack<T>]
stack.peek() // Returns Option<T>
stack.match({
  Empty: () => "empty stack",
  NonEmpty: (top, rest) => `top: ${top}`,
})

LazyList - Lazy-evaluated List

import { LazyList } from "functype"

// Creation
LazyList([1, 2, 3])
LazyList.of(1, 2, 3)
LazyList.empty<number>()

// Infinite sequences
const naturals = LazyList.from(0, (n) => n + 1)
const evens = naturals.filter((n) => n % 2 === 0).take(10)

// Operations are deferred until needed
const result = LazyList(hugeArray)
  .filter((x) => x > 0)
  .map((x) => x * 2)
  .take(5)
  .toArray() // Only processes first 5 matching elements

Do-Notation (Generator Comprehensions)

Scala-like for-comprehensions using JavaScript generators:

import { Do, DoAsync, $ } from "functype"

// Synchronous comprehensions
const result = Do(function* () {
  const x = yield* $(Option(5))
  const y = yield* $(Option(10))
  return x + y
}) // Option(15)

// Async comprehensions
const asyncResult = await DoAsync(async function* () {
  const user = yield* $(await fetchUserAsync(userId))
  const profile = yield* $(await fetchProfileAsync(user.id))
  return { user, profile }
})

// Cartesian products with List (2.5x-12x faster than nested flatMap)
const pairs = Do(function* () {
  const x = yield* $(List([1, 2, 3]))
  const y = yield* $(List([10, 20]))
  return { x, y }
}) // List([{x:1,y:10}, {x:1,y:20}, {x:2,y:10}, ...])

// Mixed monad types with automatic conversion
const mixed = Do(function* () {
  const a = yield* $(Option(5))
  const b = yield* $(Right<string, number>(10))
  return a + b
})

Note: First monad determines return type. Uses Reshapeable for automatic type conversion.

Companion Pattern & Type Guards

All types provide static type guards for narrowing:

import { Option, Either, Try } from "functype"

// Option type guards
if (Option.isSome(option)) {
  option.value // TypeScript knows value exists
}
if (Option.isNone(option)) {
  // Handle empty case
}

// Either type guards
if (Either.isRight(either)) {
  either.value // TypeScript knows it's Right
}
if (Either.isLeft(either)) {
  either.value // TypeScript knows it's error value
}

// Try type guards
if (Try.isSuccess(tryVal)) {
  tryVal.value // TypeScript knows it succeeded
}
if (Try.isFailure(tryVal)) {
  tryVal.error // TypeScript knows it failed
}

Serialization

All Serializable types provide JSON, YAML, and binary serialization:

// Serialization methods
option.serialize().toJSON()
option.serialize().toYAML()
option.serialize().toBinary() // Uint8Array

// Deserialization
Option.fromJSON<string>(jsonString)
Option.fromYAML<string>(yamlString)
Option.fromBinary<string>(binaryData)

Lint Conventions (eslint-plugin-functype)

recommended errors on no-let, no-imperative-loops, prefer-map, prefer-fold, prefer-functype-map and prefer-functype-set. prefer-option, prefer-either and prefer-try warn. Write code that passes, and when code is correct because it isn't FP-shaped, declare the boundary instead of disabling the rule:

Situation	Do this	Not this
DB row / HTTP body / JSONB with null fields or arrays	type UserRow = Wire<{ email: string | null; tags: ReadonlyArray<string> }>	eslint-disable on each field
A collection at a boundary	Wire<ReadonlyArray<Row>> (not Wire<Row>, which is one row)	Row[]
A bridge whose host requires a throw / rejection / nullable (React use(), React Query, a nullable→Option converter)	JSDoc @interop <reason> on the function	file-level eslint-disable
A throw for a programmer error, not an expected failure	invariant(cond, "msg") — narrows cond after it	throw (reported), Either for an impossible case, or a disable
A host needs a throw built from the failure	either.orThrow((l) => new StepError(l.message))	throw inside a fold
React state that may be empty	useState<User | null>(null) — allowed as-is	Option inside hook state
A cache or registry that is mutated on purpose	native Map/Set with .set/.add — allowed as-is	functype Map (immutable)
Loop that awaits sequentially and stops on failure	IO.forEach(items, f)	.forEach with an async callback

@interop needs a reason on the same line (@interop React Query needs a rejection.); a bare tag exempts nothing. It covers the tagged declaration and what is nested inside it. Rule of thumb: host needs the throw → @interop or orThrow(builder); only a bug can trip it → invariant(); can fail in normal operation → Either.

Looking Up Functype APIs

Feature Matrix Reference

For a complete overview of which methods are available on each data structure, consult the Feature Matrix at:

• references/feature-matrix.md (included with this skill)
• Or in the functype repo: docs/FUNCTYPE_FEATURE_MATRIX.md

The matrix shows which interfaces (Functor, Monad, Foldable, etc.) each type implements and what methods are available.

Common Methods by Type

Option

• map, flatMap, filter, fold
• orElse, or, orNull, orUndefined, orThrow
• isSome, isNone, contains

Either<L, R>

• map, flatMap, fold
• orElse, or, swap
• isLeft, isRight

Try

• map, flatMap, fold
• recover, recoverWith
• toOption, toEither
• isSuccess, isFailure

List

• map, flatMap, filter, reduce
• foldLeft, foldRight
• append, prepend, concat
• head, tail, isEmpty
• toArray, toSet

Additional Resources

For pattern conversion help, examples, and API reference:

• Feature Matrix: See references/feature-matrix.md for complete interface/method reference
• API Documentation: https://jordanburke.github.io/functype/
• GitHub Repository: https://github.com/jordanburke/functype
• CLI Documentation: Run npx functype for LLM-optimized API reference

Debugging Tips

Understanding Error Messages

"Type 'X' is not assignable to type 'Y'"

• Ensure proper type parameters: Option<string> not Option<any>
• Check that chains maintain type consistency

"Cannot read property 'map' of undefined"

• Remember to construct the type: Option(value) not just value
• Some types require explicit constructors: List([...]), Right(value)

Common Pitfalls

1. Forgetting to extract values

   // Wrong - returns Option<string>
   const name = Option(user).map((u) => u.name)
   
   // Correct - returns string
   const name = Option(user)
     .map((u) => u.name)
     .orElse("Unknown")
   

2. Using map instead of flatMap

   // Wrong - returns Option<Option<string>>
   Option(user).map((u) => Option(u.email))
   
   // Correct - returns Option<string>
   Option(user).flatMap((u) => Option(u.email))
   

3. Mutating instead of transforming

   // Wrong - mutates original array
   const list = List([1, 2, 3])
   list.toArray().push(4)
   
   // Correct - creates new List
   const newList = list.append(4)
   

Resources

references/

• feature-matrix.md - Complete interface and method reference
• common-patterns.md - Additional pattern examples and recipes
• quick-reference.md - Cheat sheet for functype APIs

For more examples and detailed documentation, visit:

• GitHub: https://github.com/jordanburke/functype
• Docs: https://jordanburke.github.io/functype/