I’m working on a project that I’ve built a very comprehensive Javascript test suite for with QUnit. I run tests a few times in a few different ways before subjecting the world to my code.

First I load the QUnit test page in multiple browsers to ensure I’m not going to be bitten by a cross-browser compatibility bug (IE). 

Afterward, when I run my Grunt build script to check, concatenate, and minify my JS, it re-runs those tests before concatenating, after concatenating, and finally on the minified version as well.

Occasionally I run into some false positives (failed tests) that are tied to some browser quirk like Chrome not allowing file cookies (unable to set cookies when the URL is local: file:///…) or the fact that PhantomJS isn’t a true browser so it isn’t able to handle cookies gracefully. Don’t skip writing tests for troublesome bits, instead just conditionally skip those tests based on user agents.

For example, I have two test helpers in a file test_helpers.js:

var helpers = {
	// Determine if tests are running in Chrome
	is_chrome: navigator.userAgent.toLowerCase().indexOf('chrome') > -1,

	// Determine if tests are running in PhantomJS.
	is_phantom: navigator.userAgent.toLowerCase().indexOf('phantom') > -1,
};

I include test_helpers.js file into my QUnit test page(s). Then I wrap any tests that succumb to environment issues in an if statement:

test('Session creates a session & cookie upon instantiation (TEST SKIPPED IN CHROME! WILL NOT PASS).', function(){
	if(!helpers.is_chrome && !helpers.is_phantom){
		// Test that runs only outside of Chrome and Phantom.js
	}
});

My troublesome QUnit tests will run in all clients except those I’ve explicitly avoided. My tests run green in both Chrome and my Grunt build script.