The consensus from mailing list questions was that people did not expect jq1 to be touched in situations like this:

 var jq1 = $("p");
 var jq2 = jq1.find("a");

The 1.1 behavior ensures they won't be surprised. In 1.1 you can do this to get the equivalent of the 1.0 behavior:

 var jq1 = $("p");
 var jq2 = jq1 = jq1.find("a");

For the .add case, just do this:

 jq = jq.add([element]);

jQuery now takes the fully functional approach, that is, no method modifies the object on which the method is called (though it may modify the underlying objects, eg. DOM elements).

That could be added as a "general contract" to the reference documentation.

There is a fundamental difference between the find method and the add method: when I use find, it is implicit that I don't want to perform an operation on the method owner, but I want to select a different element set from that of the called object, so it is natural that the method returns a new object. But when I call the add method, I want to add an element to *that* object on which I call the method, not to another object. This is a context like that in which I discovered the problem:

var jq = $('#someDiv');

for (var i = 0; i < someArray.length; i++) {

if (anything) {

jq.add(someElement);

}

}

Obviously, there are easy workarounds for that, but the point is that this behaviour is completely counterintuitive, and therefore highly error prone. It is against any logic and any object oriented principle, that calling an add method on object A, the passed objects are added to object B and not to object A.

In my opinion is very important to clearly distinguish the different types of operations, on the basis of its semantic meaning and its expectable behaviour, rather than standardize everything on the basis of general principles like 'no method modifies the object on which the method is called'.

P.S. If you have time, take also a look to the message I posted to the forum in date Jan 16 "Method return values in jQuery 1.1", for further observations on the subject

Same bug report:

http://jquery.com/dev/bugs/bug/837/

What you're looking for is an equivalent to .push() - .add() is not .push(). We've actively removed all destructive operations from jQuery and it would be counter-intuitive to reintroduce some at the point. For the time being, .add() should remain as it is, with some consideration given to implementing a traditional-style .push(), .slice(), etc.

You can achieve what you're looking for with the following code:

var jq = $('#someDiv');

for (var i = 0; i < someArray.length; i++) {
        if (anything) {
                jq = jq.add(someElement);
        }
}

Or with the following extra method:

jQuery.fn.push = function( elem ) {
    [].push.apply( this, elem.constructor == Array ? elem : [ elem ] );
    return this;
};

What about performance?

When you need to collect a large number of elements in a selector, it's hardly efficient to keep constructing and throwing away the previous object...