How to exclude urlArgs from build using r.js

Question

I use r.js optimizer to combine js files based on build profile as it's suggested in documentation. Here's my build-config.js:

({
    baseUrl: ".",
    paths: {
        jquery: '//ajax.googleapis.com/ajax/libs/jquery/1.9.1/jquery.min',
    },
    name: "main",
    out: "main-built.2013-07-30.js"
})

As you can see it's based upon main.js file, here's a code of it:

requirejs.config({
    baseUrl: 'scripts',
    urlArgs: "bust=" + (new Date()).getTime(),
    paths: {
        jquery: [
            '//ajax.googleapis.com/ajax/libs/jquery/1.9.1/jquery.min',
            'lib/jquery-1.9.1.min',
        ],
    },
});

require([
    'layout',
    'cue',
], function() {

});

If I preserve urlArgs: "bust=" + (new Date()).getTime() in main.js all external files (here jquery which is loaded from CDN) look like .../jquery.js?bust=1377412213

So it's PITA to comment out this line every time I make a build. I've read through all the documentation and googled for a solution but everything in vain. Maybe I do it wrong?

Gnu

Answer 1

The following solution would work in your case, where you're renaming the main.js file in the r.js build:

urlArgs: require.specified('main') ? "bust="+(new Date()).getTime() : null

The above snippet will check for the module named 'main', which will match in development, but not in production, where the module is named 'main-built.2013-07-30'.

I've tested in development and production builds and it works! :-)

On the require.specified() function: With requirejs is it possible to check if a module is defined without attempting to load it?

Multinuclear

Answer 2

Way late to the party on this, but here the solution I used: append the urlArgs param to the config with a subsequent call.

HTML:

<script src="js/libs/require.js"></script>
<script src="js/config.js"></script>
<script>require(['main-app']);</script>

Config File:

requirejs.config({
  paths: {...},
  shim: {...}
}); 

// Apply the cache bust only for the browser. 
if (window) {
  requirejs.config({
    urlArgs: REQUIRE_NOCACHE ? "bust="+(new Date()).getTime() : null
  });
}

The optimizer only takes the first requirejs.config declaration and it ignores the subsequent code. The second requirejs.config declaration extends rather than overrides the first, so urlArgs is still successfully applied to modules in the browser. Hope that helps.

Pharyngo

What is REQUIRE_NOCACHE? – Gnu

REQUIRE_NOCACHE is just a global boolean variable I declare in my dev environments. When it exists, it makes sure that the cache is busted whenever the page is refreshed. – Pharyngo

How do you organize your dev and prod environments? – Gnu

That's kind of a broad question... what specifically are you asking about? – Pharyngo

Well, your example doesn't work for me because r.js generates result config like window && urlArgs: bust... which a browser then perfectly understands and thus adds bust to each loaded file. So it's kinda useless. What could work is the parameter REQUIRE_NOCACHE which should be false during r.js work and true in dev environment. – Gnu

Answer 3

As of version 2.2.0, urlArgs now accepts functions.

It sends the moduleId and the url, so you can segment depending on the path if it should have extra args or not. See https://requirejs.org/docs/api.html#config-urlArgs

Bonaventura