lockandload AMD-loader
Lockandload is a minimalist AMD-loader-compatible boilerplate to kickstart
your website. It includes special support for single-page-apps.
Gzipped, the essential script content amounts to roughly 859 bytes of code.
Without compression it blows up to about 1520 bytes. Further minifying
this code does not result in any significant gains (773 vs 859 bytes), it
would just hinder readability.
Why
After reviewing the javascript-loader landscape, looking for a loader
that satisfies the following criteria:
- Small enough to be inlined (to avoid paying an extra request-tax to
get the loader onboard).
- Inlined config file (get rid of a request).
- Config file in pure javascript (no parsing overhead).
- Let the browser do the actual loading (browsers are smart these days)
of files.
- Connect all asynchronously loaded modules together.
- Support for single page apps that include legacy code that
uses sprinkled
$(function(){...})
constructs, yet I insist on loading
jQuery late and asynchronously to speed things up.
I came to the conclusion that (apparently?) none of the existing loaders
fit the bill. So I wrote lockandload
back in 2014 and used it in
various internal projects, but decided to open-source
it now in case others find it useful. As always, when open-sourcing something,
actually writing documentation takes most of the time. The documentation
can always be improved. Suggestions are welcome!
Features
- Less filling: 859 bytes of gzipped script content.
- Handminified to retain readable and maintainable code.
- It's so small, it can and should be inlined on your HTML page
(which is also one of the reasons to handminify it only).
- Because it is inlined, it is faster than all other loaders.
- Fully asynchronous script loader: AMD-loader compatible.
- Supports anonymous define() calls.
- Supports local require() calls (with one and two arguments, RequireJS-style).
- Supports require.undef() for hot-reloading scripts.
- Supports require.load() to load scripts on demand.
- Supports implicit and explicit ['require', 'exports', 'module'] dependencies.
- Explicit circular dependencies will silently hang in unresolved state
(or put differently: do not do that), use 'exports' or 'require' to resolve
those instead.
- No extra diagnostic code to minimise code weight and optimise loading speed.
- Does not support require.toUrl() nor simplified CommonJS wrapping.
- Fully event driven, no polling timers.
- Standard supported dependencies:
require
, exports
, module
and domready
. - Both high and low priority asynchronous loading of Javascript and CSS files.
- Leverages native browser speed for high priority loading (by getting out
of the way).
- Legacy support for $(...) jquery riddled synchronous code.
- Legacy support for loading synchronous Javascript.
- Single-page-app support using $$(...) page refresh callbacks.
- Supports IE10 and up and all other webbrowsers.
- No config file, means: no syntax to learn, no config file parser code.
- No module system: all needed functionality is included already because
it was/is so small, that writing a module system would take more code
than the source of all the added functionality.
- Integrated hooks for Google Tag Manager (GTM) support.
Requirements
It runs inside any webbrowser environment (starting at IE10 and up).
Usage
Using npm
Running npm install lockandload
in the webroot of your site,
should create the following file and directory structure:
node_modules
lockandload
lockandload_master.inc
: Placed right after the charset definition
on the page.lockandload_headready.inc
: The start of the headready-script that
lives at the end of the <head>
.lockandload_trailer.inc
: The end of the headready-script that lives
at the end of the <head>
.index.php
: PHP boilerplate.index_inlined.html
: High performance HTML only boilerplate.lockandload_master.js
: For lazy <script>
loading.lockandload_headready.js
: Boilerplate for lazy <script>
loading.index_extern.html
: Lazy HTML only boilerplate.main.js
: Example SPA (Single Page Application).
Using PHP
Copy the boilerplate node_modules/lockandload/index.php
file to your
webroot; then customise the copied file to taste.
Using other serverside scripting languages
Look at the PHP boilerplate node_modules/lockandload/index.php
, and
translate this to your own scripting language.
Without serverside scripting (lazy)
Copy the node_modules/lockandload/index_extern.html
boilerplate
file to your webroot; then customise the copied file to taste.
Copy the node_modules/lockandload/lockandload_headready.js
file
to your javascript directory and customise it taste. Do not forget
to change the path of the script
directive in index_extern.html
to point to the new location of the headready script.
Without serverside scripting (high performance)
Copy the node_modules/lockandload/index_inlined.html
boilerplate
file to your webroot; then customise the copied file to taste.
The index_inlined.html
contains two <script>
sections. The first section
should not be preceded by any other <script>
tags and should be left
verbatim.
The second section should be placed at or close to the end of the <head>
,
and should not precede any direct <link type="stylesheet">
tags.
Inside this second section there is a clearly marked section that is
your configuration area.
The basic structure of a page should be:
- html
- head
- Charset declaration.
- Inline
lockandload
master script. - High priority async external scripts.
- Viewport declaration.
- High priority CSS scripts.
<title>
.- All other tags that should go in the
<head>
. - Inline
lockandload
headready-script.
- CSS scripts fullfilling a custom applied-style dependency.
- Low priority CSS scripts.
- Medium priority async Javascript scripts.
- Low priority async Javascript scripts.
- Low priority synchronous Javascript scripts.
- body
- All other inline scripts (if you must).
The index_inlined.html
file is a production-stripped version
of annotated.html
. Look at annotated.html
to understand the code and
read additional inline documentation. These index.*
, lockandload_*.inc
and lockandload_*.js
files are not present in the git source repository, they can only be found in
the npm repository (or after running npm run prepublish
).
API
Module ids
Module ids are short strings that uniquely identify a module.
In lockandload
these ids typically do not contain parts of a path.
If a module id is derived from the javascript filename that is being
loaded, it will only refer to the final path component without
.js
or version extension (i.e. anything after the last /
and
before the first .
).
Globally
-
define(id?, dependencies?, factory)
The standard AMD global
entrypoint.
id
declares the module id we are defining. If omitted, we derive
a module id from the name of the javascript file we are loading.dependencies
is an array of strings of module ids this module depends on.
If the parameter is missing, a default dependency list of
["require", "exports", "module"]
is supplied.factory
is the callback function that gets called as soon as
all dependencies have been loaded. The factory function gets
references to all the exported symbols from its dependencies, and
should subsequently return its own symbols it wants to export to
other modules. The factory can be a function or a static object.
-
require(dependencies, callback)
Allows you to load dependencies (an array of strings of module ids)
asynchronously, the callback is called
as soon as all dependencies have loaded. Parameters to the callback
are references to the dependencies just as in the factory function
in define
.
Locally
In the secondary lockandload
headready-script; all url arguments
are used verbatim in <link href="url">
or <script src="url">
tags:
css(url, id?)
Loads low priority ordered css files asynchronously;
after the stylesheet has been applied, it fulfills the optional
id
dependency.js(url, "async"?)
Loads Javascript file, if the second optional argument "async"
is
provided, the load will be asynchronous.jsa(alias, path)
Define aliases for javascript file paths to be referenced through
require.load(alias)
to load the file on demand.
Standard require module
If you use "require"
in your dependencylist, you get a reference to the
internal require-module. It supports the following functionality:
require(id)
Returns a reference to the exports of module id
. Beware:
this will not trigger loading that module, if the module is not loaded
because it was already in your dependencylist, or if the module is
not being requested by other modules, this will return zero.
As such, it can be used to find out if a certain module is being loaded
at all.require.undef(id)
Clears the cached module id
exported symbols list. This allows the module
to be hot-reloaded by a subsequent require.load()
. Beware that existing
references to the old module are not overwritten. Any modules using
require(id)
before the require.load()
will return zero, any modules
using require(id)
after the require.load()
will return a reference
to the exports of the reloaded module.require.load(file)
Asynchronously loads the referenced javascript file. To centralise
file-location management, it is advisable to use jsa()
calls in the
headready section to declare aliases for javascript files which can
be used instead of actual file paths in the require.load()
calls.
Dealing with jQuery
In order to support legacy code that uses inline $(function(){...})
scattered
throughout pages, this loader allows you to use that construct even before
the jQuery library has been loaded,
and thus enables you to load jQuery in an asynchronous and non-blocking fashion.
The standard headready-script contains a dependency on domready
and
jquery
which finally runs domready(1)
which will run all the
registered delayed functions the first time.
SPA (Single Page App) support
To ease SPA development, the loader defines a
$$(function(jquery_document){...})
function which registers functions
for execution on every SPA-controlled page refresh. The registered functions
receive a convenience argument $(document)
when executed.
To run the registered functions, one needs to make a call to the entrypoint
of the AMD-dependency on domready
without parameters or with exactly
one parameter; if not provided, this single parameter will default to
$(document)
(the jquery object/scope referring to the whole document).
Convention states that if provided the argument should normally be the jquery
object referring to the element tree that contains the changes.
Ultimately you decide what your $$(function(argument){...})
scheduled
scripts will use the argument for. All domready()
calls before domready(1)
has been run will silently be ignored.
E.g. in your application, you could use code like this:
!function(){
define("main", ["domready"], function (domready) {
function refreshpage() {
var newdiv = $("#contentdiv");
newdiv.html("your glorious new page content");
domready(newdiv);
}
});
}();
References
Other loaders: