type dot = {mutable x:int; mutable y:int}
let f (dot1:dot) (dot2:dot) : int =
dot1.x <- 12;
dot2.x <- 26;
dot1.x
I'm not sure what this would evaluate to. I assume it would evaluate to 12 always but the answer key says that it only evaluates to 12 "sometimes". I guess this has to do with whether or not dot1 and dot2 are aliases but I still don't understand why this would matter considering we are only modifying dot1 and accessing the x field of dot1.
Here is a simple example, when the returned value would be 26 not 12:
# let p = {x=0; y=0};;
val p : dot = {x = 0; y = 0}
# f p p;;
- : int = 26
Here p is an object that has two mutable fields x and y. When applied, function f gets the same object as dot1 and dot2, so, both names dot1 and dot2 refer to the same object and, no matter which name you're using, it will lead to the modification of the p object.
Another example, that doesn't involve any functions, but just let bound names,
# let p1 = p;;
val p1 : dot = {x = 26; y = 0}
# let p2 = p;;
val p2 : dot = {x = 26; y = 0}
# p1.y <- 1;;
- : unit = ()
# p;;
- : dot = {x = 26; y = 1}
# p2.x <- 11;;
- : unit = ()
# p;;
- : dot = {x = 11; y = 1
As you can see name binding is just a name biding, so it is not creating a new object or anything like this.