Compilers

OCaml Transition Guide

This document is a transition guide for students in the Swarthmore CS curriculum. It contains examples of code written in Python, C++, and OCaml.

Variable Declaration
String Concatenation
Standard User Interaction
Function Declaration
Recursive Functions
Loops
Using Libraries
String Formatting and Printing
Exceptions
Tuples
Absent Values
Polymorphic Dictionaries

Variable Declaration

Python
```
x = 5
y = "Hello"
```
C++
```
int x = 5;
string y = "Hello";
```

OCaml

let x : int = 5;;
let y : string = "Hello";;

Note: the OCaml compiler is capable of some type inference, which allows it to guess which types you meant. The above can be written

let x = 5;;
let y = "Hello";;

You are free to use type inference as much as you want, but instruction in this course will avoid it (at least early on) to reduce confusion.

String Concatenation

Python

s1 = "Hello,"
s2 = " world!"
result = s1 + s2

C++

string s1 = "Hello,";
string s2 = " world!";
result = s1 + s2;

OCaml

let s1 : string = "Hello,";;
let s2 : string = " world!";;
let result : string = s1 ^ s2;;

Standard User Interaction

Python 3

print("Please enter a line of text.")
text = input()
print("I like this text: " + text)

C++11

cout << "Please enter a line of text." << endl;
string text;
getline(cin, text);
cout << "I like this text: " << text << endl;

OCaml

print_endline "Please enter a line of text.";;
let text : string = read_line ();;
print_endline ("I like this text: " ^ text);;

Function Declaration

Python
```
def sum_two(x,y):
  return x + y
```

C++

int sum_two(int x, int y) {
  return x + y;
}

OCaml

let sum_two (x : int) (y : int) : int =
  x + y
;;

Again, type inference can be used to write the above as

let sum_two x y = x + y;;

This course avoids relying on type inference especially in the case of functions, as the inference rules can be quite confusing to newcomers at first.

Recursive Functions

Python

def count_down(n):
  if n == 0:
    print("Liftoff!")
  else:
    print(n)
    count_down(n-1)

C++

void count_down(int n);
void count_down(int n) {
  if (n == 0) {
    cout << "Liftoff!" << endl;
  } else {
    cout << n << endl;
    count_down(n-1);
  }
}

OCaml

let rec count_down (n : int) : unit =
  if n = 0 then
    print_endline "Liftoff!"
  else
    print_endline (string_of_int n);
    count_down (n-1)
;;

Loops

Python

n = 0
while n < 10:
  print(n)
  n += 1

C++

int n = 0;
while (n < 10) {
  cout << n << endl;
  n += 1;
}

OCaml

let rec loop (n : int) =
  if n < 10 then
    print_endline (string_of_int n);
    loop (n+1)
  else
    ()
in
loop 0

Technically, OCaml does have mutable variables and while loops. For this course, however, we will rely heavily upon our ability to think recursively and we will avoid those language features whenever possible. Much of the code we write will be much better expressed recursively!

Using Libraries

Python
```
import random
x = random.randint(0,4)
```

C++

#include <cstdlib>
#include <ctime>
using namespace std;
...
srand((int)time(0));
int x = rand() % 4;

OCaml

Random.self_init ();;
let x : int = Random.int 4;;

String Formatting and Printing

Python Python has two forms of built-in string formatting. The first we discuss in CS21:
```
print("I am %d years old." % age)
```
That form is somewhat dated, however, and we are recommended to use the new form:
```
print("I am {} years old.".format(age))
```
C++ C++ string formatting is done largely through the iostream library. However, the printf library from C is also available:
```
#include <cstdio>
...
printf("I am %d years old.", age);
```
OCaml OCaml provides a Printf module similar to that of C:
```
open Printf;;
...
printf "I am %d years old." age;;
```

Exceptions

Python

try:
  do_some_stuff()
except SomeError as e:
  handle_error()
except OtherError as e:
  handle_other_error()

C++

try {
  doSomeStuff();
} catch (SomeError& e) {
  handleError();
} catch (OtherError& e) {
  handleOtherError();
}

OCaml

try
  do_some_stuff ()
with
| SomeError _ ->
  handle_error()
| SomeOtherError _ ->
  handle_other_error()

Tuples

Python Tuples are a built-in feature of Python, but we avoid discussing them in CS21.
```
my_pair = (4,"yes")
my_triple = ("stuff",True,1)
```
C++ C++ does not have tuples as a language feature. In CS35, we use a standard library class called pair:
```
pair<int,string> myPair(4,"yes"); // statically allocated
```
The C++ STL also has a more general class called tuple:
```
tuple<string,bool,int> myTriple("stuff",true,1); // statically allocated
```

OCaml OCaml has a built-in syntax for tuple similar to Python’s.

let my_pair : int * string = (4,"yes");;
let my_triple : string * bool * int = ("stuff",true,1);;

Absent Values

Python Python’s built-in None value typically serves as a way of signaling no value (without raising an exception).
```
def first_element_of_list(the_list):
  if len(the_list) == 0:
    return None
  else:
    return the_list[0]
```

C++ In C++, it is not uncommon (though not very tidy) to use NULL to refer to the absence of a value.

template <typename T>
T* firstElementOfList(List<T*>* theList) {
  if (theList->getSize() == 0) {
    return NULL;
  } else {
    return theList->get(0);
  }
}

Note, however, that this only works if the value is a pointer.

OCaml In OCaml, a built-in type declaration type 'a option = None | Some of 'a allows (and requires) you to be explicit about when “no value” is a possible answer.
```
let first_element_of_list (the_list : 'a list) : 'a option =
  match the_list with
  | [] -> None
  | first :: rest -> Some(first)
;;
```

Polymorphic Dictionaries

Python In Python, there is no static type system. As a result, you can store any key-value mappings you like in any dictionary. If you mess up, it’s your problem.
```
my_dictionary = {}
my_dictionary[1] = "one"
my_dictionary[2] = "two"
print(my_dictionary[1])
```
C++ C++ uses templates for polymorphic data types. C++ templates are usually instantiated implicitly, meaning that you simply need to name the data type you want to use.
```
std::map<int, string> my_dictionary;
my_dictionary[1] = "one";
my_dictionary[2] = "two";
cout << my_dictionary[1] << endl;
```
OCaml OCaml uses functors, which are similar to C++ templates but must be invoked explicitly. In the case of dictionaries, this means that we need a module which defines the type of the key and how it is compared. We only need to define our dictionary (e.g. IntMap) once and we may then use it anywhere, but there is no implicit instantiation similar to C++.
```
module IntKey = struct
  type t = int
  let compare = Pervasives.compare
end;;
module IntMap = Map.Make(IntKey);;
...
let empty_dictionary = IntMap.empty in
let singleton_dictionary =
  IntMap.add 1 "one" empty_dictionary in
let doubleton_dictionary =
  IntMap.add 2 "two" singleton_dictionary in
print_endline (IntMap.find 1 doubleton_dictionary);;
```