Create a test runner worker. Some specific ports(ish) are required as arguments, as well as the "master" test concatenating all tests (CF SeededRunners.fromTest).

The main Elm module calling this one will typically look like the example below. In that code, {{ user_imports }} and {{ tests }} are to be replaced by the list of needed imports from user code and the list of tests to run.

port module Runner exposing (main)


{{ user_imports }}
import Test
import ElmTestRunner.Runner exposing (Flags, Model, Msg)
import Json.Encode exposing (Value)


port askNbTests : (Value -> msg) -> Sub msg
port sendNbTests : { type_ : String, nbTests : Int } -> Cmd msg
port receiveRunTest : (Int -> msg) -> Sub msg
port sendResult : { type_ : String, id: Int, result : Value } -> Cmd msg


main : Program Flags Model Msg
main =
    [ {{ tests }} ]
        |> Test.concat
        |> ElmTestRunner.Runner.worker
            { askNbTests = askNbTests
            , sendNbTests = \nb -> sendNbTests { type_ = "nbTests", nbTests = nb }
            , receiveRunTest = receiveRunTest
            , sendResult = \id res -> sendResult { type_ = "result", id = id, result = res }
            }

It can later be spawned as a Node worker with a tiny bit of JS similar to:

const { parentPort } = require("worker_threads");
const { Elm } = require("./Runner.elm.js");

// Start the Elm app
const flags = { initialSeed: ..., fuzzRuns: ... };
const app = Elm.Runner.init({ flags: flags });

// Communication between Elm runner worker and Supervisor via port
app.ports.outgoing.subscribe((msg) => parentPort.postMessage(msg));
parentPort.on("message", (msg) => app.ports.incoming.send(msg));

Ports(ish) required by the worker program to function. They aren't exactly ports (sendResult isn't a valid port for example) but will basically be wrapped by an actual port in the main Elm caller module.

The initial random seed and the number of fuzz runs are passed as flags.

Main model. Exposed for usage in type definitions.

Internal messages.