This is my take on an often necessary task, especially since the advent of so-called Web 2.0, AJAX, and now the HTML5 History API.
jQuery.getUrlVars differs from other implementations by interpreting
both the search and hash as members of the same object. If a key
is found in multiple places, the value is appended to an array
in the returned object.
One place for improvement I see is storing the requests in a cache
indexed by URL. If a request is made for the exact same string, we
could quickly return a compiled object instead of repeating the same
calculations. — completed as of 2013-02-21