krasimir/actml

ActML is a library that allows you to use React's JSX syntax for writing business logic.

💫 ActML is a library that allows you to use JSX syntax outside of React world. It aims to provide the same experience in terms of composability and patterns.

/** @jsx A */
import { A, run } from 'actml';

const Message = ({ user, children }) => {
  console.log(children(user));
}
const Greeting = (user) => {
  return `Hello ${ user.name }!`;
}

run(
  <Message user={ { name: 'Emma' } }>
    <Greeting />
  </Message>
);
// Hello Emma!

(Try it yourself here.)

---

• How to use it
• Introduction
  • Children
  • Asynchronous
  • Context
• Hooks
  • useState
  • useEffect
  • useReducer
  • usePubSub
  • useContext
• Examples

---

How to use it

• npm i actml or yarn install actml
• ActML uses JSX so you need to have some sort of Babel integration (or any other transpiler that understands JSX)
• ActML requires you to add /** @jsx A */ at the top of the file. Otherwise the ActML elements will be transpiled to React.createElement

Introduction

ActML looks like React but it's not about rendering UI. It's about executing your JavaScript.

const Foo = () => 'bar';
run(<Foo />); // bar

You'll probably wonder why using ActML and instead of writing Foo() we use <Foo />? The answer is same reason why we write <Component /> instead of React.createElement(Component, null). We are declarative instead of imperative. It's better to say what we want to happen instead of how it happens. Being declarative means having more options for composition. The declarative code is easier to read and extend.

Also when we write <Foo /> we are not executing our function Foo. We are just saying that this function should be executed at this place. We create a descriptor of that call which is passed to ActML runtime. This abstraction creates a layer where we can do bunch of things. We can for example process async operations behind the scenes or we can build a map the application's logic. We can deffer things or enhance them with additional arguments.

Children

Every function run by ActML receives a children prop. Similarly to React that prop represents the children of the element. Here we have two use case - we can call children as a function or we can return it as a result.

const X = ({ children }) => {
  children();
  children();
};
const Y = ({ children }) => {
  return children;
}
const Message = () => {
  console.log('Hello') 
};
run(<X><Message /></X>); // prints Hello twice
run(<Y><Message /></Y>); // prints Hello once

If we are calling children we are getting back an array containing the results of the nested elements. Or if the array contains one item we get directly that item.

const X = () => 'foo';
const Y = () => 'bar';
const Results = ({ children }) => {
  console.log(JSON.stringify(children()));
};
run(
  <Results>
    <X />
    <Y />
  </Results>
); // prints ["foo","bar"]

Calling manually children means running the children X and Y and receiving the result of them.

Asynchronous

ActML runtime supports both asynchronous and synchronous elements. You can mix them in a single expression. As soon as there is something asynchronous ActML marks the call as such and the result of it is a promise. For example:

const App = async ({ children }) => {
  const message = await children();
  
  return message.join(' ');
}
const Greeting = () => 'Hey';
const GetUserFirstName = async () => {
  const { data: { first_name }} = await (await fetch('https://reqres.in/api/users/2')).json();
  return first_name;
}
const FavoriteColor = () => 'your favorite color is';
const GetFavoriteColor = async () => {
  const { data: { color }} = await (await fetch('https://reqres.in/api/products/3')).json();
  return color;
}

run(
  <App>
    <Greeting />
    <GetUserFirstName />
    <FavoriteColor />
    <GetFavoriteColor />
  </App>
).then(message => console.log(message));

// outputs: Hey Janet your favorite color is #BF1932

(online demo here)

Notice that <Greeting> and <FavoriteColor> are synchronous. ActML waits for all the children to be processed and then resolves the promise returned by the children call. If all the elements were synchronous then we'll get an array straight away.

Context

I can't say it better than React docs:

Context provides a way to pass data through the component tree without having to pass props down manually at every level.

import { createContext } from 'actml';

const Context = createContext('<default value>');
const { Provider, Consumer } = Context;

const F = () => {
  return (
    <Consumer>
      { value => console.log(value) }
    </Consumer>
  );
};

run(
  <Provider value='foo'>
    <F />
  </Provider>
); 
// outputs: 'foo'

ActML searches for the <Provider> up the tree from the place where the <Consumer> is positioned. If it can't find the <Provider> then it shows a warning and delivers <default value> instead.

Hooks

useState

import { A, useState } from 'actml';

const E = () => {
  const [ getState, setState ] = useState(initialState);
}

Returns two functions for setting and retrieving a state value. In the original React docs the first item is the state value directly but here ActML diverges a little bit by providing a function. It is done to provide a mechanism for immediate retrieval of the update value.

useEffect

import { A, useEffect } from 'actml';

const E = () => {
  useEffect(function sideEffect() {
    // ...
    return function onElementRemoved() {
      // ...
    }
  }, [ dependencyA, dependencyB ]);
}

The function sideEffect is fired after the function E finishes. After that it gets fired only if some of the dependencyA or dependencyB are changed. If we pass an empty array we are creating a side effect that is fired only once no matter how many times E is executed. The function that we pass to useEffect may return another function that is invoked when the element is removed from the tree.

ActML's useEffect mimics React's useEffect

useReducer

import { A, useEffect } from 'actml';

function reducer(state, action) {
  switch (action.type) {
    case 'increment':
      return {count: state.count + 1};
    case 'decrement':
      return {count: state.count - 1};
    default:
      throw new Error();
  }
}
const E = function () {
  const [ getState, dispatch ] = useReducer(reducer, initialState);
  //...
  dispatch({ type: 'increment' })
}

Very similar to useState. In fact useReducer internally uses useState. The mechanism for updating the state is by using actions which are dispatched and then processed by the reducer` which returns the new version of the state.

ActML's useReducer mimics React's useReducer

usePubSub

import { A, Fragment, usePubSub } from 'actml';

const TYPE = 'TYPE';

const Publisher = function () {
  const { publish } = usePubSub();

  publish(TYPE, 42);
};
const Subscribe = function () {
  const { subscribe } = usePubSub();

  subscribe(TYPE, (answer) => {
    console.log(`The answer is ${ answer }`);
  });
};

run(
  <Fragment>
    <Subscribe />
    <Publisher />
  </Fragment>
);

The usePubSub hook can help you if you need to communicate between elements on different levels in the tree. It returns an object with two methods publish and subscribe:

• publish(<type>, <payload>)
• subscribe(<type>, <callback>)

useContext

Consume the value of a context.

const Context = createContext();
const F = () => {
  const value = useContext(Context);
  // here value is 'foo'
};

run(
  <Context.Provider value='foo'>
    <F />
  </Context.Provider>
);

Examples

• Playground
• ToDoMVC app