Comp 210 Lab 11: From Scheme to C

For an introduction to the C language, be sure to see the Extended C Primer (html or ps), available through the "Readings" link of the main class web page.

A Sample Program

;; (f n)
;; given an integer n,
;; return 3n2 + 4n + 2
;;
(define f
  (lambda (n)
     (+ (* 3 n n)
        (* 4 n)
        2)))

(f 7)	

#include <iostream.h>         // Include the input/output library header,
                              //   to be able to use "cin", "cout"

/* f( n )
 * return 3n2 + 4n + 2
 */
int f( int n ) {             // A function which takes int n, returns an int.
  int result = 3*n*n + 4*n + 2;
  return result;             // You have to "return" explicitly.
  }


int main() {                 // Every program needs one function named "main"
  cout << "f at 7 is " << f(7) << "." << endl;
                             // Nothing prints by iteslf; endl is a newline.

  return 0;                  /* Return a value...to where?  Who called main?
                                Always return 0 from main, to UNIX. */
  }

You've seen most of these constructs in lecture already, but ask about anything you don't recognize. A few things to note:

• Yes, it is possible to shorten some of the above if you wanted.
• The word int can be viewed as a data definition, declaring that the value of result is always an integer.
• The line int result = ... creates a placeholder result and gives it an initial value, just like Scheme's define. But in C, it is possible to do these two tasks separately. If we write int result;, the variable is declared but beware--it initially has some garbage value.
• There are two types of comments. They don't nest (Ian's pet peeve; if you ever specify the syntax for a new computer language, please allow nested comments.)
• The assignment operator = (pronounced ``gets'') is like Scheme's set!. Do not confuse = with the comparison function == (pronounced ``equals''), which is the analog of Scheme's eq?.
• For input/output, such lines always start with cin, cout, cerr (c for character?), and the arrows point in the direction you want the data to go.

The Process of Creating a C Program

1. Type your program into a text file. Important: The filename should end with .cc, for example silly.cc. You can use whatever text editor you like to create this file; we'll use emacs since it has some nice features for using C.

   Select emacs from your window manager menu, and then inside emacs be sure to select ``Open File'' and open a new file silly.cc (alternately, type emacs silly.cc & from a UNIX shell).

   Recall that you can run an emacs tutorial from the "Help" menu. For further commands, see the emacs reference card, or review some basic emacs commands from a previous lab. Also, check out the new menu labelled ``C++'' with some handy functions.

2. Compile your file, that is translate it from a text file to a binary format the machine can understand directly.

   Translating your program to machine code would be tedious by hand, but fortunately there is a program which will do it automatically.

      owl%  g++ silly.cc
   

   If you are using emacs, you can compile the file with M-c (meta-c), or select "compile" under the C menu. You'll see that emacs invokes g++ for you, but with some useful extra arguments, which your lab instructor will go over.

   If emacs doesn't invoke g++ when you type M-c, or there isn't a C++ menu, make sure that you are editing a file whose name ends in .cc. If that still doesn't fix the problem, type "M-x c++-mode" and that should set everything aright.

3. If your program had any syntax errors (i.e., it wasn't legal C), then g++ will inform you. You must go back to Step 1, and fix your source file.

4. If you didn't have any syntax errors, g++ will create a new executable file, named silly if you compiled with those extra arguments, or a.out (booo) if you didn't. To run your program, realize it's the same as any other UNIX program: just type in the name of the executable file.

   Test your program on different inputs. If you have any mistakes...you guessed it: back to Step 1.

5. Celebrate!

#include <iostream.h>

bool biggern( int a, int b ) {      /* "bool" is C's Boolean type */
  return (a > b);
  }

int main() {
  int income = 7;
  int outgo  = 9;

  if (biggern( income, outgo ))
    cout << "Woo-hoo!  Made a profit!" << endl;
  else {
    cout << "Uh-oh, in the red." << endl;
    cout << "Must raise $" << outgo-income << "quickly!" << endl;
    }
    
  return 0;
  }

Practice Makes Perfect

Remember that in C, 5.0/9.0 is different thatn 5/9. Dividing int 5 by int 9 insists that the answer must be an int, regardless of what is done with the result (such as assigning it to a double).

Group Exercise (thanks to Dan): Go back and change one single character in your program (at random, or cunningly). Re-compile, and see what error message you get. Is it similar to the error message your neighbor got?

Warning: Programs are intended to be read by humans as well as machines. Emacs does some indenting for you, but also notice how spaces and blank lines are used judiciously (in the original sense of the word) in the above sample programs. Your exact C coding-style is something you'll settle into, but some issues are non-negotiable.

• Functions should be separated by a blank line or two (at least). Also, leave a bit of vertical space after the #includes at the start.
• In assignment statements, include spaces around =, for example x = 5;.
• In calling functions with more than one argument, leave a space after the comma in foo( a, b ).

Exercise: Write the function factorial in Scheme. Transliterate this to C, compile and run. (You are not allowed to forget what we've taught you using Scheme, about having recursive programs which match the recursive nature of the data definition (natural numbers, in this case)).

Write a main function which prints 0 through 30 factorial (remember C's while loop from lecture; write the corresponding Scheme while-fn first).

Hey, this starts giving weird answers starting around 14! There are even some negative results?! What gives? (Hint: Remember that ints can overflow, if they only have a fixed number of digits.) How can you fix this problem? How good a fix is it?

One common error to beware of: Just as in Scheme you defined a variable only once but may set! it many times, in C you want to only specify a variable's type once even if you assign to it repeatedly. For example:

int r = 3;
r = r + 1;    // right

int w = 0;
int w = w + 1; // wrong: this declares a second variable w

Optional Exercise: Write the function Stirling, which is similar to factorial except that it doesn't use recursion but instead uses Stirling's approximation for factorial: