@fynjs/run

npm run enhanced - A powerful task runner and build tool for modern JavaScript projects.

• Compatible with npm run for npm scripts
• Concurrent & Serial execution of tasks
• JavaScript extensibility with functions and promises
• Provider packages for reusable task libraries
• TypeScript support - node strips types natively; tsx is picked up automatically for syntax it cannot strip
• Advanced CLI with argument parsing and remaining args support
• Namespace organization for better task management
• and more

Running npm scripts

This module provides a command xrun to run all your npm scripts in package.json.

And you can run multiple of them concurrently or serially.

Some examples below:

Alias for the options:

• -s: --serial

npm scripts run through xrun get the same environment variables npm run sets, so scripts that branch on them work unchanged: npm_lifecycle_event, npm_lifecycle_script, npm_package_name, npm_package_version, npm_package_json, npm_execpath, npm_node_execpath, INIT_CWD.

Running JavaScript tasks

You can write your tasks in JavaScript and run them with xrun.

This is useful when a shell script is too long to fit in a JSON string, or when it's not easy to do something with shell script.

These APIs are provided: concurrent, serial, exec, env, and load.

Put your tasks in a file xrun-tasks.js and xrun will load it automatically.

An example xrun-tasks.js:

const { load, exec, concurrent, serial } = require("@fynjs/run");
load({
  //
  // define a task hello, with a string definition
  // because a string is the task's direct value, it will be executed as a shell command.
  //
  hello: "echo hello",
  //
  // define a task world, using a JavaScript function to print something
  //
  world: () => console.log("world"),
  //
  // define a task serialTask, that will execute the three tasks serially, first two are
  // the hello and world tasks defined above, and 3rd one is a shell command defined with exec.
  // because the 3rd one is not a direct value of a task, it has to use exec to define a shell command.
  //
  serialTask: serial("hello", "world", exec("echo hi from exec")),
  //
  // define a task concurrentTask, that will execute the three tasks concurrently
  //
  concurrentTask: concurrent("hello", "world", exec("echo hi from exec")),
  //
  // define a task nesting, that does complex nesting of concurrent/serial constructs
  //
  nesting: concurrent(serial("hello", "world"), serial("serialTask", concurrent("hello", "world")))
});
Copy

To run the tasks defined above from the command prompt, below are some examples:

exec and shell scripts

Use exec to invoke a shell command from JavaScript.

Here are some examples:

• exec supports options that can set a few things. Some examples below:

Function tasks

A task in JavaScript can be just a function.

load({
  hello: () => console.log("hello")
});
Copy

A function task can do a few things:

• Return a promise or be an async function, and xrun will wait for the Promise.
• Return a stream and xrun will wait for the stream to end.
• Return another task for xrun to execute further.
• Access parsed options with context.argOpts.

Example:

load({
  // A function task named hello that access parsed options with `context.argOpts`
  async hello(context) {
    console.log("hello argOpts:", context.argOpts);
    return ["foo"];
  },
  h2: ["hello world"],
  foo: "echo bar"
});
Copy

Running tasks with concurrent and serial

Use concurrent and serial to define a task that run multiple other tasks concurrently or serially.

Some examples:

• To do the same thing as the shell script echo hello && echo world:

serial(exec("echo hello"), exec("echo world"));
Copy

• or concurrently:

concurrent(exec("echo hello"), exec("echo world"));
Copy

• You can specify any valid tasks:

serial(
  exec("echo hello"),
  () => console.log("world"),
  "name-of-a-task",
  concurrent("task1", "task2")
);
Copy

Tasks to set process.env

env allows you to create a task to set variables in process.env.

You use it by passing an object of env vars, like env({VAR_NAME: "var-value"})

Examples:

load({
  setEnv: serial(env({ FOO: "bar" }), () => console.log(process.env.FOO))
});
Copy

And to put it all together

A popular CI/CD use case is to start servers and then run tests, which can be achieved using xrun JavaScript tasks:

const { concurrent, serial, load, stop } = require("@fynjs/run");
const waitOn = require("wait-on");

const waitUrl = url => waitOn({ resources: [url] });

load({
  "start-server-and-test": concurrent(
    // start the servers concurrently
    concurrent("start-mock-server", "start-app-server"),
    serial(
      // wait for servers concurrently, and then run tests
      concurrent("wait-mock-server", "wait-app-server"),
      "run-tests",
      // Finally stop servers and exit.
      // This is only needed because there are long running servers.
      () => stop()
    )
  ),
  "start-mock-server": "mock-server",
  "start-app-server": "node lib/server",
  "wait-mock-server": () => waitUrl("http://localhost:8000"),
  "wait-app-server": () => waitUrl("http://localhost:3000"),
  "run-tests": "cypress run --headless -b chrome"
});
Copy

xrun adds node_modules/.bin to PATH. That's why npx is not needed to run commands like cypress that's installed in node_modules.

Provider Packages

@fynjs/run supports provider packages - reusable task libraries that can be shared across projects. This allows teams to standardize common build tasks and workflows.

What makes a provider package?

A provider package is identified by either:

1. Having xrunProvider config in its package.json
2. Having @fynjs/run as a dependency

Creating a provider package

// In your provider package's package.json
{
  "name": "my-build-tasks",
  "xrunProvider": {
    "module": "tasks.js"  // optional: specify which module exports loadTasks
  }
}
Copy

// In your provider's tasks.js (or main module)
module.exports = {
  loadTasks(xrun) {
    // can pass in optional namespace with xrun.load("namespace", {...})
    return xrun.load({
      build: "webpack --mode=production",
      test: "jest",
      lint: "eslint src/",
      ci: ["lint", "test", "build"]
    });
  }
};
Copy

Using provider packages

Provider packages are automatically loaded when:

1. You have no tasks loaded (automatic discovery)
2. You explicitly enable them by setting loadProviderModules: true in your @fynjs/run config

Provider tasks are loaded from:

• dependencies
• devDependencies
• optionalDependencies

Example package.json:

{
  "name": "my-app",
  "dependencies": {
    "my-build-tasks": "^1.0.0"
  },
  "@fynjs/run": {
    "loadProviderModules": true
  }
}
Copy

Now you can run provider tasks directly:

xrun build    # runs the build task from my-build-tasks
xrun ci       # runs the ci task which executes lint, test, build serially
Copy

Shorthands

Not a fan of full API names like concurrent, serial, exec? You can skip them.

• concurrent: Any array of tasks are concurrent, except when they are specified at the top level.
• exec: Any string starting with ~$ are treated as shell script.
• serial: An array of tasks specified at the top level is executed serially.

Example:

load({
  executeSerially: ["task1", "task2"], // top level array serially
  concurrentArray: [["task1", "task2"]], // Any other array (the one within) are concurrent
  topLevelShell: "echo hello", // top level string is a shell script
  shellScripts: [
    "~$echo hello", // any string started with ~$ is shell script
    "~(tty,spawn)$echo hello" // also possible to specify tty and spawn flag between ~ and $
  ]
});
Copy

Full List of Features

• Core Execution Engine

  • Serial and concurrent task execution with proper nesting hierarchy
  • Promise, node.js stream, or callback support for JavaScript tasks
  • Run time flow control - return further tasks to execute from JS task functions
  • Tasks can have a finally hook that always runs after task finish or fail

• Developer Experience

  • Compatible with and loads npm scripts from package.json
  • Auto completion for bash and zsh
  • TypeScript support with automatic tsx/ts-node loading (tsx preferred)
  • Advanced CLI with comprehensive options (see CLI reference)
  • Argument parsing with -- remaining args support
  • Specify complex task execution patterns from command line

• Extensibility & Organization

  • Provider packages - reusable task libraries for sharing common workflows
  • Namespaces for organizing tasks across modules
  • Define tasks in JavaScript files with full programmatic control
  • Support flexible function tasks that can return more tasks to run
  • Custom task execution reporters

• Advanced Features

  • TTY control for interactive commands
  • Environment variable management with env() tasks
  • Shell command execution with exec() and spawn options
  • Task dependency resolution and execution planning

Getting Started

Still reading? Maybe you want to take it for a test drive?

A Simple Example

Here is a simple sample.

1. First setup the directory and project:

mkdir xrun-test
cd xrun-test
npm init --yes
npm install rimraf @fynjs/run
Copy

2. Save the following code to xrun-tasks.js:

"use strict";
const { load } = require("@fynjs/run");

const tasks = {
  hello: "echo hello world", // a shell command to be exec'ed
  jsFunc() {
    console.log("JS hello world");
  },
  both: ["hello", "jsFunc"] // execute the two tasks serially
};

// Load the tasks into @fynjs/run
load(tasks);
Copy

3. And try one of these commands:

A More Complex Example

Here is a more complex example to showcase a few more features:

"use strict";

const util = require("util");
const { exec, concurrent, serial, env, load } = require("@fynjs/run");
const rimraf = util.promisify(require("rimraf"));

const tasks = {
  hello: "echo hello world",
  jsFunc() {
    console.log("JS hello world");
  },
  both: {
    desc: "invoke tasks hello and jsFunc in serial order",
    // only array at top level like this is default to serial, other times
    // they are default to concurrent, or they can be marked explicitly
    // with the serial and concurrent APIs (below).
    task: ["hello", "jsFunc"]
  },
  // invoke tasks hello and jsFunc concurrently as a simple concurrent array
  both2: concurrent("hello", "jsFunc"),
  shell: {
    desc: "Run a shell command with TTY control and set an env",
    task: exec({ cmd: "echo test", flags: "tty", env: { foo: "bar" } })
  },
  babel: exec("babel src -D lib"),
  // serial array of two tasks, first one to set env, second to invoke the babel task.
  compile: serial(env({ BABEL_ENV: "production" }), "babel"),
  // more complex nesting serial/concurrent tasks.
  build: {
    desc: "Run production build",
    task: serial(
      () => rimraf("dist"), // cleanup, (returning a promise will be awaited)
      env({ NODE_ENV: "production" }), // set env
      concurrent("babel", exec("webpack")) // invoke babel task and run webpack concurrently
    )
  }
};

load(tasks);
Copy

Install Globally

If you'd like to get the command xrun globally, you can install this module globally.

$ npm install -g @fynjs/run
Copy

However, it will still try to require and use the copy from your node_modules if you installed it.

Load and Run Tasks Programmatically

If you don't want to use the CLI, you can load and invoke tasks in your JavaScript code using the run API.

Example:

const { run, load, concurrent } = require("@fynjs/run");
const myTasks = require("./tools/tasks");

load(myTasks);
// assume task1 and task2 are defined, below will run them concurrently
run(concurrent("task1", "task2"), err => {
  if (err) {
    console.log("run tasks failed", err);
  } else {
    console.log("tasks completed");
  }
});
Copy

Promise version of run is asyncRun

Task file formats

Your task file can be any of these, and xrun searches for them in this order:

A CommonJS task file exports with module.exports, an ES module with export default. Either can export the tasks object directly, or a function that receives the xrun instance:

export default xrun => {
  xrun.load({ hello: "echo hello" });
};
Copy

Top-level await

An ES module task file may use top-level await:

const config = await loadConfigFromSomewhere();

export default xrun => {
  xrun.load({ build: `build --target ${config.target}` });
};
Copy

Task files load through import(), which is what makes this possible - require refuses a module graph containing top-level await and always will.

TypeScript

Name your task file xrun-tasks.ts, .mts, or .cts.

No TypeScript runtime is needed for ordinary task files. Node strips type annotations itself, unflagged since 22.18, so a task file with types just works.

Syntax that must be transformed rather than erased - enums, namespaces, parameter properties - is beyond what stripping does. For those, install tsx:

npm install -D tsx typescript
Copy

xrun loads it automatically when it finds a TypeScript task file. Without it, node reports ERR_UNSUPPORTED_TYPESCRIPT_SYNTAX and xrun tells you which limit you hit. Running node with --experimental-transform-types is the other way out.

Command Line Usage

Any task can be invoked with the command xrun:

$ xrun task1 [task1 options] [<task2> ... <taskN>]
Copy

ie:

$ xrun build
Copy

Passing Arguments to Tasks

You can pass arguments after -- to shell commands. These arguments are automatically appended to the last shell task:

$ xrun build -- --watch --verbose
$ xrun test -- --grep "specific test"
Copy

For JavaScript function tasks, parsed options are available via the context param:

It's also pass as the this context for the function.

load({
  myTask(context) {
    console.log("Parsed options:", context.argOpts);
  }
});
Copy

CLI Options

Common CLI options include:

• --serial, -s - Execute tasks serially instead of concurrently
• --cwd <path>, -w - Set working directory
• --list, -l - List available tasks
• --npm, -n - Load npm scripts (default: true)
• --quiet, -q - Suppress output
• --soe <mode>, -e - Stop on error mode: no, soft, full

For complete CLI reference:

$ xrun -h
Copy

See CLI Options for full details.

To load npm scripts into the npm namespace, use the --npm option:

This is enabled by default. To turn it off use --no-npm option.

$ xrun --npm test
Copy

You can also specify command line options under @fynjs/run in your package.json.

Specifying Complex Tasks from command line

• You can specify your tasks as an array from the command line. For example, to have xrun execute the tasks [task_a, task_b] concurrently:

$ xrun [task_a, task_b]
$ xrun --concurrent [task_a, task_b]
Copy

• You can also execute them serially with:

$ xrun [--serial, task_a, task_b]
$ xrun --serial [task_a, task_b]
Copy

• You can execute tasks serially, and then an inner array with concurrent tasks. The following will execute task_a, then task_b, and finally task_c1 and task_c2 concurrently

$ xrun --serial [task_a, task_b, [task_c1, task_c2]]
Copy

• You can also make inner arrays serial using --serial as the first element. Other shortcuts for "--serial" are: . and -s.

$ xrun [task_a, task_b, [--serial, task_c1, task_c2]]
Copy

• You can pass the whole array in as a single string, which will be parsed as an array with string elements only.

$ xrun "[task_a, task_b, [task_c1, task_c2]]"
Copy

Task Name

Task name is any alphanumeric string that does not contain /, or starts with ? or ~$.

Tasks can be invoked from command line:

• xrun foo/task1 indicates to execute task1 in namespace foo
• xrun ?task1 or xrun ?foo/task1 indicates that executing task1 is optional.

xrun treats these characters as special:

• / as namespace separator
• prefix ? to let you indicate that the execution of a task is optional so it won't fail if the task is not found.
• prefix ~$ to indicate the task to be a string as a shell command

Optional Task Execution

By prefixing the task name with ? when invoking, you can indicate the execution of a task as optional so it won't fail in case the task is not found.

For example:

• xrun ?foo/task1 or xrun ?task1 won't fail if task1 is not found.

Task Definition

A task can be string, array, function, or object. See reference for details.

package.json

You can define @fynjs/run tasks and options in your package.json.

Tasks

You can also define xrun tasks without JavaScript capability in your package.json.

They will be loaded into a namespace pkg.

For example:

{
  "name": "my-app",
  "@fynjs/run": {
    "tasks": {
      "task1": "echo hello from package.json",
      "task2": "echo hello from package.json",
      "foo": ["task1", "task2"]
    }
  }
}
Copy

And you can invoke them with xrun pkg/foo, or xrun foo if there are no other namespace with a task named foo.

Options

Command line options can also be specified under @fynjs/run or xrun inside your package.json.

For example:

{
  "name": "my-app",
  "@fynjs/run": {
    "npm": true
  }
}
Copy

Async Tasks

You can provide a JS function for a task that executes asynchronously. Your function just need to take a callback or return a Promise or a node.js stream.

ie:

const tasks = {
  cb_async: (cb) => {
    setTimeout(cb, 10);
  },
  promise_async: () => {
    return new Promise(resolve => {
      setTimeout(resolve, 10);
    }
  }
}
Copy

Detailed Reference

See reference for more detailed information on features such as load tasks into namespace, and setup auto complete with namespace for your shell.

License

Licensed under the Apache License, Version 2.0