commonjs require.js

(function($G)
{
  var cache = {};
  var XHR = $G.XMLHttpRequest || function(){ return new $G.ActiveXObject('Microsoft.XMLHTTP') };
  
  $G.require = function(url)
  {
    if(!(url in cache))
    {  
      var xhr = new XHR();
      
      xhr.open('GET', url, false);
      xhr.send();
      
      if(200 <= xhr.status && xhr.status < 300)
        cache[url] = (new Function('var export = {};' + xhr.responseText + '; return export;')).call($G); // may be ES5 strict mode
      else
        cache[url] = null;
    }
    
    if(cache[url] == null)
      throw url + ' not found';
    else
      return cache[url];
  };
})(this);

/*
  Why its bad.
  1) non cross domain
  2) block js while file is loading
  3) debugging eval (or new Function) is hell. You dont know nor script file name nor line number. And modern js engines hasnt support of something like eval(code, fileName, lineNumber) 
  
  Result
  Use normal asynch module load via script tag. CommonJS is not for client-side
*/

Oh I see the "Why its bad." code comment. Yes, blocking using sync requests is very bad. That was a big mistake of Dojo and they kept with it, incredulously, for many years. And so in light of that, it is surprisingly disturbing to see that strategy reemerge. Incredulous, and there it is.