Booter DI (sub-project)

Main idea of Booter project.

I khnow that for now we have a lot of "boot" react, redux, reflax projects. That is just experiment).
I really like spring boot (It's java:)) project. Out of the box you will get logging, running, testing, configuraion and more features.
So aim of this project is to be as close as possible to spring boot mission.

Modern applications need some common tools:

• Logging.
• Configurations.
• Running & building tools.
• App security solutions.
• Test environment and runners.
• So on.

Booter dependency injection

Greate feature of Spring project is dependency injection & dependency resolving.

Let's look on universal application.
It's simple to understand, we really have 2 applications: Server and Client. And that applications have different entry poins(main functions or classes). Also, sometimes, they depends on different libraries (evironment is different). And we have to resolve this issue. For me is ugly way to try solve this by next approach:

import superagent from 'superagent';
import config from '../config';

const methods = ['get', 'post', 'put', 'patch', 'del'];

function formatUrl(path) {
  const adjustedPath = path[0] !== '/' ? '/' + path : path;
  if (__SERVER__) {
    // Prepend host and port of the API server to the path.
    return 'http://' + config.apiHost + ':' + config.apiPort + adjustedPath;
  }
  // Prepend `/api` to relative URL, to proxy to API server.
  return '/api' + adjustedPath;
}

Take attention if (__SERVER__) {. Sometimes can be if (__SERVER__ && __PRODUCTION__) {.

May we solve this problem in another way? And yes, we can!

For example http library can be configured for client and server application in different ways, but in the same time have one interface. So for client it will be new Http(), for server new Http({origin: config.apiOrigin}). That is! We have two beans(instances). And in client app we will use the first, in server the second.

Documentation

So, how it works?.
Termin "bean" means object instance.

So our application will have one main class(entry point). Lets give name Application. And for this application will have set of beans. This beans must be provided in next way:

@ApplicationBeanProvider(BeanProvider)
class MyApp {
  run() {
  }
}

Where BeanProvider is class with method provide:

class BeanProvider {
  provide() {
    // set of beans
    return {
      'http': () => {
      }
    }
  }
}

Okey, now we have declared set of beans and Application class. We can create MyApp in next way:

let app = new MyApp()
app.run();

Yep, app instance is created. Let's inject some bean into class:

@Inject()
class SomeClass {

  @InjectBean('apiClient')
  setHttp(client) {
    this.apiClient = client;
  }
  
  doSomething() {
    this.apiClient.fetchUser();
  }
}

or just:

@Inject([
  'apiClient'
])
class SomeClass {

  doSomething() {
    this.apiClient.fetchUser();
  }
}

And beans will be injected successfuly.
Under the hood injector will create bean and "save" it into cashe. Every beans are singletons.

Reference. Top API.

@ApplicationBeanProvider

Decorator for setting application bean provider.

Usage:

@ApplicationBeanProvider(BeanProvider)
class MyClass {}

BeanProvider is a class with at least one method provide.
Example:

class BeanProvider {
  provide() {
    // set of beans
    return {
      'http': () => {
      }
    }
  }
}

@Inject

Decorator mark class. In that class will be injected required beans.
Example:

@Inject()
class BeanProvider {
}

or:

@Inject(['http'])
class BeanProvider {
}

@InjectBean

Decorator mark class method as bean setter.
Example:

@Inject()
class BeanProvider {

  @InjectBean('http')
  setHttp(http) {
    this.http = http;
  }
}

Build booter-di

npm install
npm build

Changelog:

30.06.2016 - Init commit