Little Loader

A lightweight, IE8+ JavaScript loader that is actually tested...

... with a very narrow set of objectives:

• Tested all the way down to IE8
• Reliably calls back after script loads
• Captures script load errors down to IE8
• Really, really small (clocking in at ~519 minified + gzipped bytes)
• ... and that's it!

We currently test:

• Karma - Travis: PhantomJS, Firefox
• Selenium - Travis: PhantomJS, Firefox
• Selenium - Sauce Labs:
  • Windows: Firefox, Chrome, IE8-11
  • Mac: Safari

Usage

Integration

Alone, little loader attaches to window._lload for loading your Javascript:

<script>
  window._lload("http://example.com/foo.js", function (err) {
    // `err` is script load error.
    // otherwise, foo.js is loaded!
  }/*, [optional context (`this`) variable here] */);
</script>

If you use an AMD bundling tool (like RequireJS):

define(["little-loader"], function (load) {
  load("http://example.com/foo.js", function (err) {
    // ... your code ...
  });
});

If you use a CommonJS bundling tool (like Webpack):

var load = require("little-loader");

load("http://example.com/foo.js", function (err) {
  // ... your code ...
});

Calling

Little loader can be called in a number of ways:

// Load a script and don't worry about a callback
load("http://foo.com/foo.js");

// Load, then callback (and optionally with context.)
load("http://foo.com/foo.js", callback);
load("http://foo.com/foo.js", callback, this);

// Load, call `setup(script)` on the script tag before insertion, no callback
load("http://foo.com/foo.js", {
  setup: setup,         // setup(script)
  context: this         // (optional)
});

// Load, call `setup(script)` on the script tag before insertion, then
// callback with context (two ways)
load("http://foo.com/foo.js", {
  setup: setup,         // setup(script)
  callback: callback,   // callback(err)
  context: this
});
load("http://foo.com/foo.js", {
  setup: setup,         // setup(script)
  callback: callback    // callback(err)
}, this);

Installation

CDN

For the ready-to-use version from CDN, use

<!-- Minified, production version -->
<script src="https://unpkg.com/little-loader@VERSION/dist/little-loader.min.js"></script>
<!-- Development version -->
<script src="https://unpkg.com/little-loader@VERSION/lib/little-loader.js"></script>

NPM

To include little-loader as part of your own build, first install from npm:

$ npm install --save little-loader

The library has a UMD wrapper and should work like any other AMD or CommonJS module with your favorite bundling tool (Webpack, RequireJS, etc.).

If you do not use a CommonJS or AMD loader tool, then little loader will be exposed as the window._lload variable.

Development

Development requires two installation steps:

$ npm install
$ npm run install-dev

After that, run the full lint + tests:

$ npm run check

You can try out the live functional tests fixtures with our static server:

$ npm run server

and navigate to: http://127.0.0.1:3001/test/func/fixtures/

Tests

We run both Karma (client-side) and Selenium (functional) tests.

The Karma tests are faster and more flexible, but slightly "off" from real-world use because of their execution environment. We use Karma tests to kick the tires on our AMD and CommonJS abstractions and little, one-off use case scenarios.

The Selenium tests are slower and klunky, but they are the "real deal" executing little-loader in exactly the same manner as would be used on a real web page. We use Selenium to test a core set of fundamental use cases across all browsers in our matrix.

Parallel Local Tests

Our CI is setup with a specific optimized parallel workflow. To run parallel functional tests in development, here are some helper tasks...

Local Browsers

$ TEST_PARALLEL=true \
  builder envs test-func-local \
  --setup=setup-local \
  --buffer \
  '[ { "TEST_FUNC_PORT": 3030, "ROWDY_SETTINGS":"local.phantomjs" },
     { "TEST_FUNC_PORT": 3040, "ROWDY_SETTINGS":"local.firefox" },
     { "TEST_FUNC_PORT": 3050, "ROWDY_SETTINGS":"local.chrome" }
   ]'

The TEST_PARALLEL flag indicates to not do in-test setup which would conflict with other test processes. We also rely on setting TEST_FUNC_PORT specifically to non-conflicting ports with at least 3 ports total from the starting number for the two separate static servers we run during tests.

Sauce Labs

To run Sauce Labs tests in parallel from a local machine, you'll need the sc binary, which can be force installed with:

$ SAUCE_CONNECT_DOWNLOAD_ON_INSTALL=true npm install sauce-connect-launcher

After this, the module is available at: node_modules/sauce-connect-launcher/sc/*/bin/sc

From there, you can invoke our helper local commands:

$ TEST_PARALLEL=true \
  SAUCE_USERNAME=<INSERT_USERNAME> \
  SAUCE_ACCESS_KEY=<INSERT_ACCESS_KEY> \
  builder envs test-func-sauce \
  --setup=setup-sauce \
  --buffer \
  '[ { "TEST_FUNC_PORT": 3030, "ROWDY_SETTINGS":"sauceLabs.IE_8_Windows_2008_Desktop" },
     { "TEST_FUNC_PORT": 3040, "ROWDY_SETTINGS":"sauceLabs.IE_9_Windows_2008_Desktop" },
     { "TEST_FUNC_PORT": 3050, "ROWDY_SETTINGS":"sauceLabs.IE_10_Windows_2012_Desktop" }
   ]'

Releases

IMPORTANT - NPM: To correctly run preversion your first step is to make sure that you have a very modern npm binary:

$ npm install -g npm

First, you can optionally edit and commit the project history.

$ vim HISTORY.md
$ git add HISTORY.md
$ git commit -m "Update history for VERSION"

Now we're ready to publish. Choose a semantic update for the new version. If you're unsure, read about semantic versioning at http://semver.org/

$ npm version VERSION|major|minor|patch -m "Version %s - INSERT_REASONS"

Now postversion will push to git and publish to NPM.