testing.md

run_loop Testing Guide

There are two kinds of test infrastructure for run_loop: the standard async test helper and the GUI test harness.

test_helper (for ordinary async tests)

Use this when testing async code. It automatically handles RunLoop initialization, execution, and teardown.

Usage

use novonotes_run_loop::test_helper as test;
use serial_test::serial;

#[test]
#[serial]  // Required for serialization when running multiple tests.
fn test_example() {
    test::run_async(async {
        // Write async test code here
        RunLoop::current().delay(Duration::from_millis(10)).await;
        42  // Any type can be returned
    });
}

• Always attach #[serial]: RunLoop does not support concurrent execution by multiple threads.
• Return type is flexible: You can return Result<T, E> or any other type.
• Panics are handled: Panics inside the test are caught and reported as test failures.

test_harness (for GUI integration tests)

Tests that require GUI operations on macOS/iOS must run on the main thread. In those cases, disable the standard test harness and use the dedicated one instead.

Setup

The standard Rust harness does not run tests on the main thread, so it must be disabled.

# Cargo.toml
[[test]]
name = "gui_test"
path = "tests/gui_test.rs"
harness = false  # Disable the standard harness

Usage

use novonotes_run_loop::test_harness::run_gui_tests;

fn main() {
    run_gui_tests(vec![
        ("test_name", test_function),
        // Add more tests as needed
    ]);
}

fn test_function() -> Result<(), String> {
    RunLoop::current().schedule(Duration::ZERO, move || {
        // Some GUI test code
        assert_eq!(1 + 1, 2);

        RunLoop::current().stop_app();
    })
    .detach();

    // run_app blocks until stop_app is called.
    RunLoop::current().run_app();
    Ok(())
}