Note: You need to know all but the last example (self-replicating programs) for exams.

char: a single byte, signed. It is treated exactly like an 8-bit signed integer.
int: a signed integer, system dependent, often 4 bytes, that is, 32 bits.
double: a floating point value, system dependent, often 8 bytes, that is, 64 bits.

Then the ones we will seldom if ever use. (The unsigned types are too specialized.  long is usually the same as int, so it is not needed.  short is too short to be of much use.  float has far too little precision to be recommended for use.)

short: a short signed integer, system dependent, often 2 bytes.
long: a long signed integer, system dependent, often also 4 bytes.
unsigned char: an unsigned version of char.
unsigned short: an unsigned version of short.
unsigned int: an unsigned version of integer.
unsigned long: an unsigned version of long.
float: a floating point value, system dependent, often 4 bytes.

% cat pi.c
#include <stdio.h> /* for printf */
#include <math.h>  /* required for atan */

int main( ) {
   double pi = 4.0*atan(1.0);
   printf("Pi = %20.16f\n", pi);
   return 0;
}
% cc -o pi pi.c -lm # -lm for math library
% pi                # now execute pi
Pi =   3.1415926535897931

The function printf must always have an initial argument that is a string (in double quotes, the control string). This is followed by zero or more variables or expressions, whose values are printed according to the control string: Ordinary characters in the string are printed as they are, but a conversion specifier triggers the output of the next value in the list following the initial control string. Each conversion specifier begins with a percent character (%), which is not printed itself. The conversion specifier above is %20.16f, which controls the output of the value of the variable pi. The % starts the conversion specifier. Then the 20 says to use 20 or more columns for output. The .16 specifies 16 digits to the right of a decimal point, and finally the f says to print a float or a double in a fixed decimal format. The last character in the string above is a newline (\n) which must be explicitly present to skip to a new line at the end.

Here are two tables: conversion specifiers and special excape sequences:

A minimum field width and decimal precision can used to create formatted output.  Left or right justification can be managed as well. Here are a few examples, where an underscore stands for a blank:

Don't worry about g and G in this course (they combine features of both f and e).

% cat pi2.c
#include <stdio.h> /* for printf */
#include <math.h>  /* required for atan */

int main( ) {
   int a = 314159, b = 271828;
   double pi = 4.0*atan(1.0);
   printf("%20.16i, %i, %i\n", pi, a, b);
   return 0;
}
% cc -o pi2 pi2.c -lm # compile on linux
% pi2                 # run on linux
    0000001413754136, 1074340347, 314159
% cc -o pi2 pi2.c -lm # compile on a Sun
% pi2                 # run on a Sun
    0000001074340347, 1413754136, 314159
% lint -err=warn -Ncheck=%all -Nlevel=4 -Xarch=v9 pi2.c

name used but not defined
    atan                pi2.c(6)

function returns value which is always ignored
    printf

function argument ( number ) type inconsistent with format
    printf (arg 2)      double  :: (format) int         pi2.c(7)

.c file has no corresponding .h file
    /home/wagner/public_html/CS2213/lectures/pi2.c

Notice what has happened here. We expected the variable pi to be printed incorrectly, since it had the wrong specifier, but we expected the next two variables a and b, to be printed as 314159 and 271828, repectively. We see that pi is indeed printed as garbage, but a is also printed as garbage, and b is printed as if it had a's value. Also, running on a Sun gives different garbage than running under Linux. (In this case, just the first half of the storage is used up for the first print, and the printing is out-of-synch after that.) Notice that lint running on a Sun did correctly diagnose the error.

% cat age.c
#include <stdio.h>

int main( ) {
   int age;
   double height;
   printf("Please enter age in years, and height in meters ---> ");
   scanf("%i  %lf", &age, &height);
   printf("Age: %i years, height: %5.2f meters\n", age, height);
   return 0;
}
% cc -o age age.c
% age
Please enter age in years, and height in meters ---> 23  1.68
Age: 23 years, height:  1.68 meters
% age
Please enter age in years, and height in meters ---> 23

1.697
Age: 23 years, height:  1.70 meters
% age
Please enter age in years, and height in meters ---> 23  2
Age: 23 years, height:  2.00 meters
% age
Please enter age in years, and height in meters ---> 23.5  2
Age: 23 years, height:  0.50 meters
% age
Please enter age in years, and height in meters ---> xyz  1.68
Age: -1073744696 years, height:  5.12 meters

Notice about the above program and its runs:

• A double requires a %lf as conversion specifier, rather than just %f, which is used on input.
• The function scanf keeps processing input characters until it finds the items it is supposed to look for.
• Even with erroneous input (last example), it doesn't give a segmentation violation, or other error, but just puts wrong values into the variables.
• In processing input, scanf will skip over whitespace, such as newlines and tabs.
• In the fourth input above, scanf was looking for an int followed by a double, and 23.5 2 appeared on input. The 23 worked for the int, and it was terminated by a dot. Then the .5 , terminated by a blank was taken as the double, not what was intended.
• On the previous input, a 2 terminated by a blank was converted to double.

A common mistake arises because printf uses %f for a double, but scanf uses %lf (percent-ell-eff) for a double Another common mistake is to leave the ampersand off the variable being stored into, usually producing a segmentation error when you execute. scanf ignores whitespace characters such as a newline in its format string.

Later the section on strings will present the functions sprintf and sscanf, which print to a string and scan from a string, respectively -- very useful when needed.

Open, study and execute fahrcels.c

A formal parameter that is meant to be the address of an int is declared with an extra *, that is, int *. To access the actual value at the address of this formal parameter, you use the dereference operator: another *. So if the formal parameter is declared int * mp (pointer to m), then *mp would refer to the actual value. Instead of scanf (which is a little complicated to show here), if we were trying to get values back for the variables above, say from a function myfetch, the program would look as follows:

% cat fetch.c
#include <stdio.h>

void myfetch(int *mp, double *xp);

int main( ) {
   int m; /* unititialized! */
   double x; /* unititialized! */
   myfetch(&m, &x);
   printf("m = %i, x = %10.6f\n", m, x);
}

void myfetch(int *mp, double *xp) {
   *mp = 666;
   *xp = 2.718281828;
   return;
}
% cc -o fetch fetch.c
% fetch
m = 666, x =   2.718282

Self-reproducing C program.