I just looked at the performance improvements in Google V8 Javascript since I want to include it in one of my own projects.
My interest at the moment is related to hidden classes. The basic idea is that V8 creates hidden classes as properties are added to the object in order to efficiently search for this property, avoiding dictionary searches.
For example, when you create a new object Pointwith a name p, it creates a hidden class C0, which is a class with no properties, and attaches this class to the object:

The operator p.x = 0will modify the object to add the property, and then create a new class C1, indicating that the property xcan be found with a specific offset within the object.

Thus, obtaining p.xafter this is a relatively effective operation.
Finally, execution p.y = 0will perform a similar operation, the result is:

Now this is actually quite elegant, because if you create another object Point p2, a new hidden class should not be created, it just gets "assigned" C0. Similarly, the addition of properties x, and yin this manner will also be effective in terms of the need to create new hidden class.
. , :
Point p3 = new Point();
p3.y = 3141592653589;
p3.x = 2718281828459;
, : y y x .
-, x y, C2, , x y .
, , ? , x, y y, x, .
, .
, , , ?
, , , , :
For x, see offset 0
For y, see offset 1
?
, , .
- ?