Testing#

A test "name": block runs when you invoke ms test. A block passes when it finishes normally; panic from @std/faults.ms fails it with a message. There is no assert statement or built-in expect function. Write a condition that calls panic, or define a helper for the values you compare.

Save this complete file as test.ms:

import panic from "@std/faults.ms"

fn add(a: Int, b: Int) -> Int:
    return a + b

fn expect_equal(actual: Int, expected: Int) -> Unit:
    if actual != expected:
        return panic("Expected {expected}, got {actual}")

test "addition":
    expect_equal(add(1, 2), 3)

test "zero is unchanged":
    let actual = add(0, 7)
    if actual != 7:
        return panic("Expected 7, got {actual}")

ms test test.ms
ms test test.ms --filter addition
ms test test.ms --timeout 10s

The helper is an ordinary function specialized for Int. Define a helper with the appropriate parameter types when testing text or another type. It evaluates the actual and expected values once and includes both in its failure message. return panic(...) stops the test; panic never returns normally.

To see a failure, change the first test’s expected value from 3 to 4. The CLI prints a line containing FAIL test.ms::addition: Expected 4, got 3, reports one failed test, and returns a nonzero exit status. Restore 3 to make both tests pass. There is no need for main in a tests file, and ms test does not execute an application’s main.

Run tests in a project#

ms test ms.toml
ms test ms.toml --filter addition --timeout 10s
ms test ms.toml --json

A manifest run discovers test.ms and *_test.ms files recursively below the project root, as well as tests in imported modules. A direct ms test file.ms runs that module’s tests and the tests in its imports; it does not discover unrelated sibling files. The exact traversal and excluded-directory rules live in project configuration .

Tests run from the manifest’s directory, or from the selected source file’s directory for a single-file run. Relative fixture paths therefore resolve consistently when you invoke ms test from another directory; see build and test .

Put reusable application functions in an exported module and import them into a test file with a relative path, for example import add from "./src/math.ms". Keep the import consistent with your project’s actual files; see modules .

--filter selects tests whose module::name contains the supplied text. --timeout bounds each test (60 seconds by default; 0 disables it). A programming fault or timeout fails the test; a function returning error(...) is still an ordinary value that you must check. Do not discard a Result when testing success or failure.

For automation, --json emits a versioned result with per-test outcomes, counts, source diagnostics and captured stdout. Stderr remains stderr; failures keep their exit status. It supports one run and cannot be combined with --watch. See the JSON contract .

For example, this fragment belongs inside a test after calling a fallible function:

match result:
    ok(value):
        expect_equal(value, 3)
    error(cause):
        return panic("Expected success, got {cause.message()}")

The fragment assumes Result[Int, Error] and the Int helper above. Handle domain-specific error cases explicitly when a test is meant to check a failure. See Results and programming faults .