Examples:
- $("#test").load("test.html #something");
- $("#test").load("test.html p.user");

Caveats:
- No scripts are injected when a selector is used.
- The selector is rooted inside the head and body - it's equivalent to doing:
  $("body,head").find(selector)

Marked all the appropriate methods as being deprecated for the 1.1.4 release (in preparation for 1.2).

Brought back a husk method to represent evalScripts. I didn't realize so many plugins relied on it (it's deprecated in this release).

Only bind .ready() once per instance of jQuery - and only bind if the ready() method is actually called (nothing is bound if window/load is used).

Made 'jQuery' unrollable via .noConflict(true) and instances of new Function() have been converted to eval(function(){}) in order to handle jQuery name changes. (Bug #1393)

Using some of the ideas presented by rformato, I've significantly sped up $("#id") selection. It's now just 10% slower than doing: $(document.getElementById("test")), which seems quite acceptable. (Bug #1316)

Fix for a selector speed regression (calling a simple selector many times resulted in a significant speed down). This has been fixed by breaking the RegExps out into the global scope. This required that a closure be implemented around the full jQuery script (which is now the case). Some simple changes were made in addition to the RegExp one, allowing for some greater flexibility on our part - and hopefully better compression.

Speed results:
http://dev.jquery.com/~john/ticket/1351/ vs.
http://dev.jquery.com/~john/ticket/1351/113.html vs.
http://dev.jquery.com/~john/ticket/1351/112.html

Two quick fixes in relation to #1028. ajaxSetup shouldn't be deep and the boolean arg should be used, not overwritten with 'true'.

Make deep .extend() an optional argument - it will go deep if you pass in an boolean as the first argument (fixed bug #1028).

Just pushed in my changes for making jQuery.each on objects faster, sample results: http://dev.jquery.com/~john/ticket/each/obj.html

Just pushed some major speed improvements through for $.each() - we're now seeing a 2x speed improvement over 1.1.3. Some crude results can be found here: http://dev.jquery.com/~john/ticket/each/

Complete overhaul of the Ajax test suite, it's now passing in all browsers. In order to achieve this I had to fix a numbe
r of bugs in the suite itself, along with other random bugs that popped up. The following bugs were resolved along the wa
y: #1236 (.extend() keeps processing when it hits nulls), #1028 (.extend() now works recursively), #1080 ($.get no longer
 overwrites the data parameter), #1210 (Creating script and link tags now work), and #1463 (jQuery.global has been re-too
led to no longer leak memory and slow things down).

Added a fix for IE returning comment nodes in * queries. I put the logic in $.merge() but added a conditional such that the speed hit only effects IE users. (Bug #1155)

Applied arrix's fix for getting style values on elements that aren't in the DOM, in Safari. (Bug #1482)

We were catching exceptions within the success callback of an Ajax request, then causing an error callback to be called (which is incorrect). (Bug #1441)

reverted my "fixes", seems like they work only with the latest (dev) rhino version, which isn't that helpful

When we disabled triggering of clicks, we also killed off blur and focus, this fixes that (bug #1440).

The test for underscore selectors broke under Opera, so the element is now loaded via Ajax and tested. (bug #1084)