Proposal

Tried all these cases in IE6, FF2, Opera 9.25 and Safari 3 and they all worked well.

var keys = $.map( {a:1,b:2}, function( v, k ){
	return k;
}, [ ] );
alert(keys);
var values = $.map( {a:1,b:2}, function( v, k ){
	return v;
}, [ ] );
alert(values);
var mapped = $.map( document.getElementsByTagName('script'), function( v, k ){
	return v;
}, {length:0} );
alert(mapped);
var flat = $.map( Array(8), function( v, k ){
	return k % 2 ? k : [k,k,k];//try mixing array and regular returns
});
alert(flat);

Could you convert that to the testsuite, too?

The tests for the test suite

Commited test in [5342].

Your patch must be tested against the benchmark before applying it, as map is a core function of jQuery. Its raw speed has higher priority then being useful for other cases. If we can do both, even better.

Shouldn't you/we accept the change before adding the tests to the test suite ? Or it will fail for others that haven't applied the proposed diff.

I meant you/we add the test, not accept. I didn't mean to say it's on me to accept :)

Updated to match against [5753]