Convert Erlang to Elm

Convert Erlang code to idiomatic Elm. This skill extends meta-convert-dev with Erlang-to-Elm specific type mappings, idiom translations, and architectural patterns for moving from distributed backend systems to type-safe frontend applications.

This Skill Extends

• meta-convert-dev - Foundational conversion patterns (APTV workflow, testing strategies)

For general concepts like the Analyze → Plan → Transform → Validate workflow, testing strategies, and common pitfalls, see the meta-skill first.

This Skill Adds

• Type mappings: Erlang types → Elm types
• Idiom translations: Erlang patterns → idiomatic Elm
• Architecture patterns: OTP behaviors → The Elm Architecture (TEA)
• Message passing: Process mailboxes → Elm commands/subscriptions
• Error handling: let-it-crash → Maybe/Result types
• Concurrency: Processes/gen_server → Elm runtime effects

This Skill Does NOT Cover

• General conversion methodology - see meta-convert-dev
• Erlang language fundamentals - see lang-erlang-dev
• Elm language fundamentals - see lang-elm-dev
• Reverse conversion (Elm → Erlang) - see convert-elm-erlang
• Backend-to-backend conversions - see other conversion skills

---

Quick Reference

Erlang	Elm	Notes
atom()	String or custom type	Atoms become string literals or union types
binary()	String	UTF-8 encoded strings
integer()	Int	Arbitrary precision → fixed size
float()	Float	Direct mapping
boolean()	Bool	true/false mapping
list()	List a	Homogeneous typed lists
tuple()	Custom type or record	Named fields preferred
map()	Dict k v	Key-value storage
pid()	N/A	No direct equivalent (use Cmd/Sub)
undefined	Nothing in Maybe a	Explicit nullability
{ok, Value}	Just Value or Ok Value	Success wrapper
{error, Reason}	Err Reason in Result e a	Error wrapper

---

Architectural Paradigm Shift

From OTP to The Elm Architecture (TEA)

Aspect	Erlang OTP	Elm TEA
Purpose	Distributed, fault-tolerant backend	Type-safe, reactive frontend
Concurrency	Millions of processes	Single-threaded event loop
State	Process-local mutable state	Immutable application state
Communication	Message passing between processes	Commands/Subscriptions to runtime
Error handling	Let-it-crash + supervision trees	Compiler-enforced exhaustive handling

Mapping OTP Behaviors to TEA Components

Erlang gen_server:

-module(counter_server).
-behaviour(gen_server).

-record(state, {count = 0}).

init([]) -> {ok, #state{}}.

handle_call(get, _From, State) ->
    {reply, State#state.count, State};
handle_call({increment, N}, _From, State) ->
    NewCount = State#state.count + N,
    {reply, NewCount, State#state{count = NewCount}}.

Elm equivalent using TEA:

module Counter exposing (Model, Msg, init, update, view)

-- MODEL
type alias Model =
    { count : Int }

init : Model
init =
    { count = 0 }

-- UPDATE
type Msg
    = Increment Int
    | Get

update : Msg -> Model -> (Model, Cmd Msg)
update msg model =
    case msg of
        Increment n ->
            ( { model | count = model.count + n }, Cmd.none )

        Get ->
            ( model, Cmd.none )

-- VIEW
view : Model -> Html Msg
view model =
    div []
        [ text ("Count: " ++ String.fromInt model.count)
        , button [ onClick (Increment 1) ] [ text "Increment" ]
        ]

---

Type System Mapping

Primitive Types

Erlang	Elm	Notes
true / false	True / False	Capitalized in Elm
42	42	Integer literals
3.14	3.14	Float literals
<<"binary">>	"String"	UTF-8 strings
'atom'	"string" or custom type	Context-dependent

Collection Types

Erlang	Elm	Example
[1, 2, 3]	[1, 2, 3]	Homogeneous lists
#{key => value}	Dict.fromList [("key", value)]	Requires Dict module
{ok, 42}	Ok 42	Result type
{error, "failed"}	Err "failed"	Result type
undefined	Nothing	Maybe type

Structured Types

Erlang Records → Elm Type Aliases

%% Erlang
-record(user, {
    id :: integer(),
    name :: binary(),
    age :: integer() | undefined
}).

-- Elm
type alias User =
    { id : Int
    , name : String
    , age : Maybe Int
    }

---

Idiom Translation

1. Pattern Matching

Erlang:

classify(N) when N > 0 -> positive;
classify(N) when N < 0 -> negative;
classify(0) -> zero.

Elm:

classify : Int -> String
classify n =
    case compare n 0 of
        GT -> "positive"
        LT -> "negative"
        EQ -> "zero"

2. List Processing

Erlang:

Squares = [X * X || X <- [1, 2, 3, 4, 5], X rem 2 == 0].

Elm:

squares : List Int
squares =
    [1, 2, 3, 4, 5]
        |> List.filter (\x -> modBy 2 x == 0)
        |> List.map (\x -> x * x)

3. Error Handling

Erlang:

parse_int(Str) ->
    try binary_to_integer(Str) of
        Int -> {ok, Int}
    catch
        error:badarg -> {error, invalid_integer}
    end.

Elm:

parseInt : String -> Result String Int
parseInt str =
    String.toInt str
        |> Result.fromMaybe "Invalid integer"

4. Optional Values

Erlang:

get_timeout(#config{timeout = undefined}) -> 5000;
get_timeout(#config{timeout = T}) -> T.

Elm:

getTimeout : Config -> Int
getTimeout config =
    Maybe.withDefault 5000 config.timeout

5. HTTP Requests (Message Passing Replacement)

Erlang:

fetch_data(Url) ->
    Pid = self(),
    spawn(fun() ->
        Response = httpc:request(get, {Url, []}, [], []),
        Pid ! {http_response, Response}
    end).

Elm:

type Msg
    = GotData (Result Http.Error String)

fetchData : String -> Cmd Msg
fetchData url =
    Http.get
        { url = url
        , expect = Http.expectString GotData
        }

6. State Machine

Erlang:

locked(cast, {button, Code}, #{code := Code} = Data) ->
    {next_state, unlocked, Data};
locked(cast, {button, _}, Data) ->
    {keep_state, Data}.

Elm:

type DoorState
    = Locked
    | Unlocked

type Msg
    = ButtonPressed String

update : Msg -> Model -> (Model, Cmd Msg)
update msg model =
    case (msg, model.state) of
        (ButtonPressed code, Locked) ->
            if code == model.correctCode then
                ( { model | state = Unlocked }, Cmd.none )
            else
                ( model, Cmd.none )
        _ ->
            ( model, Cmd.none )

---

Error Handling Philosophy

Philosophy Shift

Erlang	Elm
Let-it-crash: Supervisors restart failed processes	Prevent-all-crashes: Compiler enforces handling all cases
Runtime errors are acceptable	Compile-time guarantees eliminate runtime errors

Practical Translation

Erlang:

safe_divide(_, 0) -> {error, division_by_zero};
safe_divide(X, Y) -> {ok, X / Y}.

Elm:

type DivisionError = DivisionByZero

safeDivide : Float -> Float -> Result DivisionError Float
safeDivide a b =
    if b == 0 then
        Err DivisionByZero
    else
        Ok (a / b)

---

Migration Strategy

What CAN be converted:

• Business logic (calculations, validations)
• Data transformations
• State machines
• Request/response patterns

What CANNOT be converted:

• Process supervision (no equivalent)
• Distributed systems (Elm is frontend-only)
• Hot code reloading
• Low-level concurrency

Architecture Mapping

Erlang OTP Application
│
├── Supervision Tree ──────────> [Remains in Erlang backend]
├── gen_server (State) ────────> Elm Model + Update
├── handle_call/cast ──────────> Msg variants + update cases
├── State transitions ─────────> Model updates
└── API endpoints ─────────────> Elm HTTP commands

Result: Hybrid architecture
- Backend: Erlang OTP (supervision, distributed state)
- Frontend: Elm (UI, client state, type-safe interactions)
- Communication: HTTP/WebSocket APIs

---

Common Pitfalls

1. Trying to Port Process Concurrency

Problem: Erlang's concurrency model doesn't translate to Elm. Solution: Re-architect around TEA with commands/subscriptions.

2. Expecting Mutable State

Problem: Erlang processes have mutable state; Elm is purely functional. Solution: Embrace immutability. Return new model versions from update.

3. Over-relying on Dynamic Types

Problem: Erlang's dynamic typing has no direct Elm equivalent. Solution: Use custom types (union types) to model all possibilities.

4. Ignoring JSON Boundaries

Problem: Assuming Erlang terms can be directly used in Elm. Solution: Always create explicit JSON encoders/decoders for API contracts.

---

Tooling Translation

Erlang	Elm	Purpose
rebar3 compile	elm make	Build project
rebar3 eunit	elm-test	Run tests
rebar3 shell	elm repl	Interactive shell
dialyzer	elm compiler	Type checking
observer	Elm debugger	Runtime inspection

---

Example: Counter with Backend

Elm Frontend:

module Counter exposing (main)

import Browser
import Html exposing (..)
import Html.Events exposing (onClick)
import Http

type alias Model =
    { count : Int, loading : Bool }

type Msg
    = Increment Int
    | GotCount (Result Http.Error Int)

update : Msg -> Model -> (Model, Cmd Msg)
update msg model =
    case msg of
        Increment n ->
            ( { model | loading = True }
            , incrementCount n
            )

        GotCount (Ok count) ->
            ( { model | count = count, loading = False }
            , Cmd.none
            )

        GotCount (Err _) ->
            ( { model | loading = False }
            , Cmd.none
            )

incrementCount : Int -> Cmd Msg
incrementCount n =
    Http.post
        { url = "/api/counter"
        , body = Http.jsonBody (Encode.object [("increment", Encode.int n)])
        , expect = Http.expectJson GotCount countDecoder
        }

---

See Also

• lang-erlang-dev - Erlang language fundamentals
• lang-elm-dev - Elm language fundamentals
• meta-convert-dev - General conversion methodology
• convert-elm-erlang - Reverse conversion