Basic inter-process routing with callback handling.
Shared in case others may find it useful.
- Table of Contents
- Installation
- Features
- Introduction
- Connections
- Routing
- Tests
- Contributing
- Changelog
- License
- See also
npm install --save cluster-callback-routing- callback driven communication between connections
- basic error handling (through IPC)
- easy management of routes
- routers
(Needs a proper introduction)
Run init() to set any options in a chain, or options() to get the options object when additional options is sat. Any configuration must be done before start() is invoked.
Note: Remember to always start the work through start().
const connection = require('cluster-callback-routing')
.init({
'env': <String>, // Runtime environment. Defaults to `process.env.NODE_ENV` if set or 'development'.
'module root': <String>, // Module root. Defaults to project root (cwd).
'count': <Number>, // Amount of workers to spawn. Defaults to amount of CPU-cores on system.
'respawn': <Boolean>, // Determine if child-processes throws or restarts on error. Defaults to true.
'routes': <Object or String>, // See #connections for details.
'case sensitive routing': <Boolean>, // Whether or not to respect case sensitivity while routing. Does not apply to route names. Defaults to true.
// Any other fields are ignored and can be used freely.
'foo': 'bar',
...
})
let all_options = conneciton.options({ ... })
...If you're okay with defaults, then you can just invoke start().
require('cluster-callback-routing').start()project
+-- index.js
`-- routes
+-- route.js
+-- route1.js
+-- route2.js
`-- worker.js
A connection route is a child-process communicator used to manage routes or send requests. A connection must have an unique name and will only spawn one at a time. The exception to this is the special connection worker, as explained below.
You can choose to use any valid string as a name, except names associated with special connection routes.
You can either declare your routes (1) in an object or (2) as a path. It is recommended to use the latter.
Both special routes are optional to use.
route: The default (request) routes handler for all requests, if present. If no default handler is set, then any requests not specifying a (connection) receiver will rebounce.
worker: As opposed to a normal route has no routing capabilities, but in return can spawn more than one at a time.
There are three ways to initialize routes in an object; (1) As a path, (2) directly, or (3) within an object with the field route.
...
.init({
...
routes: {
route: 'path/to/route', // It is possible to specify a path to require,
worker: connection => { // directly setting the route here,
...
},
route1: { // Or if you need variables initialized for a single route,
var1: 'only accessible for route1', // you can set them enclosed in an object,
route: connection => { // where 'route' is the handler for the route.
...
}
}
},
...
})
...It is possible to specify the path to a folder with scripts. Javascript files and folders directly after the specified path will then be used as routes. Route name according to filename or folder name.
...
.init({
...
routes: 'path/to/routes', // You can choose any path you want, but it must lead to a folder.
...
})
...
.init({
...
routes: './path/to/routes', // This is equal to the above statement.
...
})
...
.init({
...
routes: '/absolute/path/to/routes', // It is also possible to use an absolute path.
...
})
...Below are some brief explanations about the basic routing methods.
See the API Reference on the Wiki for more in-depth details of the API.
A route is a path with one or more callbacks tied to it. Requests will try to match against each route you define.
For more details, see the wiki page. (page needed)
...
// Defining a route
connection.route('path/to/callback', (req, next) => {
...
})
...When we want to do some work on another connection, we make a request. It is possible to attach a JSON-friendly load with the request.
It is recommended, but not needed, to reserve the first argument in all callbacks for error handling.
For more details, see the wiki page. (page needed)
...
// We can send plain requests without any load,
connection.request('path/to/callback/without/load', (err, result) => {
...
})
...
// wrap it in an array,
connection.request('path/to/callback/with/load', ['any', 'var', 0], (err, result) => {
...
})
...
// or add it directly.
connection.request('path/to/callback', 'any', 'var', 0, (err, result) => {
...
})
...To choose a different connection then the default, use the connection name as the first part of the path.
...
connection.request('/route1/path/to/callback', (err, result) => {
...
})
...An extension of the routing interface. Has a variety of uses. An instance can be acquired on any connection capable of routing.
Router is based on Express Router.
For more details, see the wiki page. (page needed)
...
console.log('Can %s spawn new router: ', connection.key, typeof connection.Router === 'function')
if (connection.key !== 'worker') {
let router = connection.router({
caseSensitive: <Boolean> // Same as global 'case sensitive routing' or sat here.
})
...
}
...Using a router is as simple as binding it to a route path.
...
// Get an instance
let router = connection.router({ ... })
...
// Bind it to a path.
connection.route('/some-router', router)
...Until some real tests are added will the file test.js have a small test for humans to run.
If someone is interested in contributing, feel free to create a PR or Issue on GitHub.
- Express - used their routing logic as a starting point.