Configuration
While the intention of core.io is to adhere to the idea of convention over configuration, it still grants you, the developer, full control over most aspects of your application by letting you override default values.
core.io configuration process is purportedly simple, a core.io application takes an options object with configuration parameters and override values. core.io does not really care how you come up with that object.
However, the Application
class provides a helper static method to collect, merge, and resolve dependencies of configuration files that are located in the ./config
folder of a project.
The resulting configuration object will be made available at runtime on the application context as context.config
.
When you create a new Application
instance you can pass an options
object to it's constructor.
This options
object has two purposes. If you define a configuration
key, it's value will be added to context.config
.
All other keys in this object will extend the application instance, like a [mixin][mixin]. The application instance extends itself with this object in it's init
method which is called directly from the constructor.
You can use it to override methods before the instance makes use of any of them or to add new methods to your instance.
const App = require('core.io').Application;
/*
* Autoload and merge files inside
* `config/` directory.
*/
const config = App.loadConfig({
//...default values
}, true);
const app = new App({
myCustomMethod: function(e) {
this.emit('custom.event', e);
},
config: config
});
app.myCustomMethod({});
Configuration instance
For convenience core.io wraps the config
object with a get
and set
methods.
This is so that you can access a deep object without fear of some object in the path not being defined. It also enables you to provide a default value for such cases.
/*
* Get the value of "environment" defined in config/app.js
* return "production" if undefined.
*/
let environment = context.config.get('app.environment', 'production');
It's more useful when you need to access a deeply nested object:
//Get the value of "repl.options.prompt", return "poke-repl >" if undefined.
let prompt = context.config.get('repl.options.prompt', 'poke-repl >');
Module configuration
When core.io registers a module, first it will require
the module and then will look for a key in context.config
that matches the module's moduleid. It will then call module.init
with a reference to the value of this key.
Pseudo code to illustrate:
let moduleId = 'persistence';
let config = this.config.get(moduleId, {});
module.init(this, config);
Configuration Files
Application.loadConfig
will load all configuration files found inside the ./config
directory of your application.
It will then load the files, and merge them in a single object using the file name as a key.
If you have a configuration file that has the same name as a given module's moduleId
then the contents of that file will be passed to the module during the initialization phase.
In a configuration file you can reference values from the same configuration object or from other configuration objects. Using two different syntaxes you can reference strings or objects:
- Strings:
${app.name}
- Objects:
@{app.locals}
The configuration solve routine will solve all cross references between configuration files. It runs after merging all files into a single object.
As an example, ${app.name}
will be resolved to config.app.name
:
config/app.js
:
module.exports = {
name: 'MyApplication'
};
config/repl.js
:
module.exports = {
prompt: '${app.name}'
};
There is also the possibility of processing the contents of a configuration file after it has been merged and loaded.
If you export a function named afterSolver
it will be called after all dependencies have been resolved. The function will be called with the whole configuration object.
module.exports.afterSolver = function(config) {
config.set('amqp.amqp', require('amqp'));
};
Configuration files are regular JavaScript files, which means you can build different logic into them.
Under the hood core.io uses the [simple config loader][scl] package. You can read more in the packages repository.
core.io provides a convenience method to collect these configuration files.
var App = require('core.io').Application;
/*
* Autoload and merge files inside
* `config/`
*/
var config = App.loadConfig({
//...default values
}, true);
var app = new App({
//Top level attributes will extend the application
//instance.
myCustomMethod: function(){},
config: config
});
You can specify the path from where to look for the configuration files.
Supporting Different Environments
Another aspect in which core.io tries to simplify the configuration process is by how it supports different development environments, like staging, development, production, etc.
In short: it does not.
To be more precise, core.io takes a very pragmatic stance and does not provide any way to (directly) manage different environments but has some recommendations that make having configuration files per environment unnecessary.
A lot of the things that need to change on each environment are secrets like service tokens or user keys. We chose to manage those by using process.env
and environment variables.
If you use the provided Application.loadConfig
then your configuration files are javascript files which, obviously, can have logic in it. Meaning that you can check the value of process.env.NODE_ENV
and export different objects based on that value.
Another benefit of Application.loadConfig
is that you can reference other parts of your configuration files and solve them at runtime, making your configuration files modular.
You can also make use of the afterSolver
facility which gets access to the merged configuration object. In it you can access the environment
key which holds the value of the current environment and modify your configuration file at runtime.
Case in point, you have options.
Ideally your configuration files should be logic lightweight in order to reduce possible errors and keep things simple, but you are free to do as you please.
You can also use an environment manager like [envset][envset] to dynamically populate your process.env
variables. All you need is an .envset
file where you define your environments, environment variables and their values:
[production]
NODE_AWS_SECRET_ACCESS_KEY=FS40N0QY22p2bpciAh7wuAeHjJURgXIBQ2cGodpJD3FRjw2EyYGjyXpi73Ld8zWO
NODE_AWS_ACCESS_KEY_ID=LSLhv74Q1vH8auQKUt5pFwnix0FUl0Ml
NODE_HONEYBADGER_KEY=LCgZgqsxKfhO
NODE_POSTGRES_ENDPOINT=50.23.54.25
NODE_POSTGRES_DATABASE=myproject
NODE_POSTGRES_PSWD=Pa$sW03d
NODE_POSTGRES_USER=myproject
[development]
NODE_AWS_SECRET_ACCESS_KEY=HN5ITok3lDaA1ASHxtEpy1U9XCjZwThmfgoIYZk8bkOqc5yk6sT7AWd3ooNeRFV9
NODE_AWS_ACCESS_KEY_ID=m35W4PGGVZfj1gOxYvztFxBD5F2605B3
NODE_HONEYBADGER_KEY=f3MNPUhZoki6
NODE_POSTGRES_ENDPOINT=localhost
NODE_POSTGRES_DATABASE=postgres
NODE_POSTGRES_PSWD=postgres
NODE_POSTGRES_USER=postgres
NOTE: If you use .envset
remember to add it to your .gitignore
file.
Lastly but more importantly, you can BYOS- bring your own solution- and use whatever configuration system you prefer.
Configuration files located in the config/
folder of projects will be merged together in a single object, which will be available at runtime as a property of your application instance, i.e. context.config
.
The top-level keys on the context.config
(i.e. context.config.repl
) object correspond to a particular configuration file name under your config/
directory (i.e. config/repl.js
). Most individual configuration files are specific to a module, with the exception of config/app.js
which should hold options for the current application, like the application's name, it's base directory, environment in run under, etc.
The intention of these files is to provide modules with configuration options. When a module is loaded, it will be called with the application's instance and a config
top-level key that matches the module's name.
Configuration dependencies
Your configuration files can contain references to a value found in other configuration files using a simple syntax that will get resolved after merging all files into a single object.
As an example, ${moduleA.name}
will be resolved with config.moduleA.name
:
config/moduleA.js
:
module.exports = {
name: 'ModuleA'
};
config/moduleB.js
:
module.exports = {
prompt: '${moduleA.name}'
};
core.io provides a convenience method to collect these configuration files.
const Application = require('application-core').Application;
/*
* Autoload and merge files inside
* `config/`
*/
const config = Application.loadConfig({
//...default values
}, true);
const app = new Application({config});
Configuration Extra Properties
The configuration process attaches two properties to the config object:
- package: contents of
package.json
file minus the readme - environment: value of
process.env.NODE_ENV
Application configuration file
- banner: String|Function
The ./config/app.js
is different than other configuration files in that the application will extend itself with the object like if it was a mixin.
Module configuration
When core.io registers a module, first it will require
the module and then will look for a key in context.config
that matches the module's moduleid. It will then call module.init
with a reference to the value of this key.
Pseudo code to illustrate:
let moduleId = 'persistence';
let config = this.config.get(moduleId, {});
module.init(this, config);
Solving Configuration Dependencies
Configuration files can have interpolated values where you reference the value of any attribute of the configuration object using the attribute's keypath.
You reference objects or properties by their keypath. A keypath is a string representing the location of a piece of data.
var data = {
user: {
name: 'Peperone',
address: {
city: 'New York'
}
}
};
You can reference strings or objects using two different syntaxes:
- Object interpolation:
@{user}
or@{user.address}
- String interpolation:
${user.name}
or${user.address.city}