Freeloader.js is a declarative behavior system for server-rendered webapps. It consists of two main pieces:
Install via npm:
$ npm install freeloader-js --save
A freeloader app looks like this:
var freeloader =var app =
Once you have your app, you can declaratively add behaviors, in the form of controllers.
Here's an example of a controller that keeps its element's aspect ratio constant, both when it's first created and when the window changes size:
As evident above, controllers are simply event routers for parts of your page. "Events" brings to mind DOM events, but those are just one kind of event a controller can respond to. Types of events include:
Ancestor and descendant controllers just means controller instances attached to elements higher or lower in DOM tree. There are no rules or limits about what the hierarchy can look like. You can bind one controller to many elements, or multiple controllers to a single element. A controller has a regular structure:
Here are the event names, along with how it's triggered.
life- Lifecycle events. Triggered internally by freeloader. Event names are limited to
events- DOM events. Triggered by the user. Event names take the form
<type> <selector>(for delegated events) or
<type>(for non-delegated events).
above- Ancestor messages. Sent by ancestor controllers, using
below- Descendant messages. Sent by descendant controllers, using
subs- Subscription events. Sent by any controller, using
this.publish(name). Alternatively, sent from non-controller contexts as long as you have a reference to the app, via
All reference-keeping work is offloaded to the browser via the DOM, so you don't need to worry about memory leaks or zombie handlers that often arise when naively tying lots of things together using events.
No controller that isn't alive and awake in the document will ever receive an event, or be retained in memory after its DOM node is removed.
document.getElementById('foo').innerHTML = '' are perfectly acceptable ways of removing content, as far as freeloader is concerned.
Freeloader provides two primary ways to add new content to the DOM.
this.inject()from within controllers.
To use navigation, call
Whether your browser supports the history API or not, a call to
app.navigate('/foo?bar') always results in the URL bar being updated to
At no point will
/?bar#foo ever appear in your URL.
In the former case, the contents of
/foo?bar are fetched via XHR and injected into the page, css/js fetched, and the URL updated using the history API.
In the latter case, the browser simply navigates to the new page:
location.href = '/foo?bar'.
The end result is the same, except modern browsers will be lots faster.
A page's scripts should therefore in no way depend on events such as
onload, or otherwise make synchronous assumptions with respect to initial page-load orchestration.
This is one of the few things freeloader is opinionated about.
Fortunately freeloader's declarative paradigm encourages no such assumptions to begin with, so things should be copacetic.
Besides the speed boost in modern browsers, freeloader's navigation automatically rebinds declared behaviors. This includes existing ones and any ones declared in newly-loaded scripts. This allows you to break apart your scripts on a per-page basis, thus minimizing initial download time.
From within a controller, you may
The inection API is flexible.
options argument is an argument with a source directive and a target directive.
The source is either a string of HTML code, or a URL to retrieve HTML from.
The target dictates where to put that HTML.
Suppose you have a URL called
/posts that returns a DOM tree like this:
Then suppose you have a controller, with a DOM tree like this:
div|--ul.posts| |--li.post#post1| |--li.post#post2| `--li.post#post3`--button.load-more
You can inect like this:
Source directives can be one of these forms:
url: 'url'(contents of the URL)
url: 'url selector'(a selection from within the URL)
html: 'string'(arbitrary chunk of HTML code)
html: 'element'(one or more DOM or jQuery objects)
Target directives can be one of these forms:
replace: true(replace this element)
replace: 'selector'(replace a sub-element)
before: true(insert before this element)
before: 'selector'(insert before a sub-element)
after: true(insert after this element)
after: 'selector'(insert after a sub-element)
prepend: true(prepend to this element)
prepend: 'selector'(prepend to a sub-element)
append: true(append to this element)
append: 'selector'(append to a sub-element)
Like the navigation API, the injection API binds behaviors to newly-added content. Thus the second thing freeloader is opionated about: always add new content using freeloader.
Or don't. The only consequence of bypassing freeloader when adding new content is that the new content won't get behavior. But it won't actually break anything.