C Language Data Types: An Understandable Guide for Beginners

This chapter is from the book 

This chapter is from the book

C Primer Plus, 5th Edition

Learn More Buy

This chapter is from the book

This chapter is from the book 

C Primer Plus, 5th Edition

Learn More Buy

Basic C Data Types

Now let's look at the specifics of the basic data types used by C. For each type, we describe how to declare a variable, how to represent a constant, and what a typical use would be. Some older C compilers do not support all these types, so check your documentation to see which ones you have available.

The int Type

C offers many integer types, and you might wonder why one type isn't enough. The answer is that C gives the programmer the option of matching a type to a particular use. In particular, the C integer types vary in the range of values offered and in whether negative numbers can be used. The int type is the basic choice, but should you need other choices to meet the requirements of a particular task or machine, they are available.

The int type is a signed integer. That means it must be an integer and it can be positive, negative, or zero. The range in possible values depends on the computer system. Typically, an int uses one machine word for storage. Therefore, older IBM PC compatibles, which have a 16-bit word, use 16 bits to store an int. This allows a range in values from –32768 to 32767. Current personal computers typically have 32-bit integers and fit an int to that size. See Table 3.3 near the end of this chapter for examples. Now the personal computer industry is moving toward 64-bit processors that naturally will use even larger integers. ISO/ANSI C specifies that the minimum range for type int should be from –32767 to 32767. Typically, systems represent signed integers by using the value of a particular bit to indicate the sign. Chapter 15 discusses common methods.

Declaring an int Variable

As you saw in Chapter 2, "Introducing C," the keyword int is used to declare the basic integer variable. First comes int, and then the chosen name of the variable, and then a semicolon. To declare more than one variable, you can declare each variable separately, or you can follow the int with a list of names in which each name is separated from the next by a comma. The following are valid declarations:

int erns;
int hogs, cows, goats;

You could have used a separate declaration for each variable, or you could have declared all four variables in the same statement. The effect is the same: Associate names and arrange storage space for four int-sized variables.

These declarations create variables but don't supply values for them. How do variables get values? You've seen two ways that they can pick up values in the program. First, there is assignment:

cows = 112;

Second, a variable can pick up a value from a function—from scanf(), for example. Now let's look at a third way.

Initializing a Variable

To initialize a variable means to assign it a starting, or initial, value. In C, this can be done as part of the declaration. Just follow the variable name with the assignment operator (=) and the value you want the variable to have. Here are some examples:

int hogs = 21;
int cows = 32, goats = 14;
int dogs, cats = 94;    /* valid, but poor, form */

In the last line, only cats is initialized. A quick reading might lead you to think that dogs is also initialized to 94, so it is best to avoid putting initialized and noninitialized variables in the same declaration statement.

In short, these declarations create and label the storage for the variables and assign starting values to each (see Figure 3.4).

Figure 3.4 Defining and initializing a variable.

Type int Constants

The various integers (21, 32, 14, and 94) in the last example are integer constants. When you write a number without a decimal point and without an exponent, C recognizes it as an integer. Therefore, 22 and –44 are integer constants, but 22.0 and 2.2E1 are not. C treats most integer constants as type int. Very large integers can be treated differently; see the later discussion of the long int type in the section "long Constants and long long Constants."

Printing int Values

You can use the printf() function to print int types. As you saw in Chapter 2, the %d notation is used to indicate just where in a line the integer is to be printed. The %d is called a format specifier because it indicates the form that printf() uses to display a value. Each %d in the format string must be matched by a corresponding int value in the list of items to be printed. That value can be an int variable, an int constant, or any other expression having an int value. It's your job to make sure the number of format specifiers matches the number of values; the compiler won't catch mistakes of that kind. Listing 3.2 presents a simple program that initializes a variable and prints the value of the variable, the value of a constant, and the value of a simple expression. It also shows what can happen if you are not careful.

Listing 3.2 The print1.c Program

/* print1.c-displays some properties of printf() */
#include <stdio.h>
int main(void)
{
  int ten = 10;
  int two = 2;

  printf("Doing it right: ");
  printf("%d minus %d is %d\n", ten, 2, ten - two );
  printf("Doing it wrong: ");
  printf("%d minus %d is %d\n", ten ); // forgot 2 arguments

  return 0;
}

Compiling and running the program produced this output on one system:

Doing it right: 10 minus 2 is 8
Doing it wrong: 10 minus 10 is 2

For the first line of output, the first %d represents the int variable ten, the second %d represents the int constant 2, and the third %d represents the value of the int expression ten - two. The second time, however, the program used ten to provide a value for the first %d and used whatever values happened to be lying around in memory for the next two! (The numbers you get could very well be different from those shown here. Not only might the memory contents be different, but different compilers will manage memory locations differently.)

You might be annoyed that the compiler doesn't catch such an obvious error. Blame the unusual design of printf(). Most functions take a specific number of arguments, and the compiler can check to see whether you've used the correct number. However, printf() can have one, two, three, or more arguments, and that keeps the compiler from using its usual methods for error checking. Remember, check to see that the number of format specifiers you give to printf() matches the number of values to be displayed.

Octal and Hexadecimal

Normally, C assumes that integer constants are decimal, or base 10, numbers. However, octal (base 8) and hexadecimal (base 16) numbers are popular with many programmers. Because 8 and 16 are powers of 2, and 10 is not, these number systems occasionally offer a more convenient way for expressing computer-related values. For example, the number 65536, which often pops up in 16-bit machines, is just 10000 in hexadecimal. Also, each digit in a hexadecimal number corresponds to exactly 4 bits. For example, the hexadecimal digit 3 is 0011 and the hexadecimal digit 5 is 0101. So the hexadecimal value 35 is the bit pattern 0011 0101, and the hexadecimal value 53 is 0101 0011. This correspondence makes it easy to go back and forth between hexadecimal and binary (base 2) notation. But how can the computer tell whether 10000 is meant to be a decimal, hexadecimal, or octal value? In C, special prefixes indicate which number base you are using. A prefix of 0x or 0X (zero-ex) means that you are specifying a hexadecimal value, so 16 is written as 0x10, or 0X10, in hexadecimal. Similarly, a 0 (zero) prefix means that you are writing in octal. For example, the decimal value 16 is written as 020 in octal. Chapter 15 discusses these alternative number bases more fully.

Be aware that this option of using different number systems is provided as a service for your convenience. It doesn't affect how the number is stored. That is, you can write 16 or 020 or 0x10, and the number is stored exactly the same way in each case—in the binary code used internally by computers.

Displaying Octal and Hexadecimal

Just as C enables you write a number in any one of three number systems, it also enables you to display a number in any of these three systems. To display an integer in octal notation instead of decimal, use %o instead of %d. To display an integer in hexadecimal, use %x. If you want to display the C prefixes, you can use specifiers %#o, %#x, and %#X to generate the 0, 0x, and 0X prefixes, respectively. Listing 3.3 shows a short example. (Recall that you may have to insert a getchar(); statement in the code for some IDEs to keep the program execution window from closing immediately.)

Listing 3.3 The bases.c Program

/* bases.c--prints 100 in decimal, octal, and hex */
#include <stdio.h>
int main(void)
{
  int x = 100;

  printf("dec = %d; octal = %o; hex = %x\n", x, x, x);
  printf("dec = %d; octal = %#o; hex = %#x\n", x, x, x);

  return 0;
}

Compiling and running this program produces this output:

dec = 100; octal = 144; hex = 64
dec = 100; octal = 0144; hex = 0x64

You see the same value displayed in three different number systems. The printf() function makes the conversions. Note that the 0 and the 0x prefixes are not displayed in the output unless you include the # as part of the specifier.

Other Integer Types

When you are just learning the language, the int type will probably meet most of your integer needs. To be complete, however, we'll cover the other forms now. If you like, you can skim this section and jump to the discussion of the char type in the "Using Characters: Type char" section, returning here when you have a need.

C offers three adjective keywords to modify the basic integer type: short, long, and unsigned. Here are some points to keep in mind:

• The type short int or, more briefly, short may use less storage than int, thus saving space when only small numbers are needed. Like int, short is a signed type.

• The type long int, or long, may use more storage than int, thus enabling you to express larger integer values. Like int, long is a signed type.

• The type long long int, or long long (both introduced in the C99 standard), may use more storage than long, thus enabling you to express even larger integer values. Like int, long long is a signed type.

• The type unsigned int, or unsigned, is used for variables that have only nonnegative values. This type shifts the range of numbers that can be stored. For example, a 16-bit unsigned int allows a range from 0 to 65535 in value instead of from –32768 to 32767. The bit used to indicate the sign of signed numbers now becomes another binary digit, allowing the larger number.

• The types unsigned long int, or unsigned long, and unsigned short int, or unsigned short, are recognized as valid by the C90 standard. To this list, C99 adds unsigned long long int, or unsigned long long.

• The keyword signed can be used with any of the signed types to make your intent explicit. For example, short, short int, signed short, and signed short int are all names for the same type.

Declaring Other Integer Types

Other integer types are declared in the same manner as the int type. The following list shows several examples. Not all older C compilers recognize the last three, and the final example is new with the C99 standard.

long int estine;
long johns;
short int erns;
short ribs;
unsigned int s_count;
unsigned players;
unsigned long headcount;
unsigned short yesvotes;
long long ago;

Why Multiple Integer Types?

Why do we say that long and short types "may" use more or less storage than int? Because C guarantees only that short is no longer than int and that long is no shorter than int. The idea is to fit the types to the machine. On an IBM PC running Windows 3.1, for example, an int and a short are both 16 bits, and a long is 32 bits. On a Windows XP machine or a Macintosh PowerPC, however, a short is 16 bits, and both int and long are 32 bits. The natural word size on a Pentium chip or a PowerPC G3 or G4 chip is 32 bits. Because this allows integers in excess of 2 billion (see Table 3.3), the implementers of C on these processor/operating system combinations did not see a necessity for anything larger; therefore, long is the same as int. For many uses, integers of that size are not needed, so a space-saving short was created. The original IBM PC, on the other hand, has only a 16-bit word, which means that a larger long was needed.

Now that 64-bit processors, such as the IBM Itanium, AMD Opteron, and PowerPC G5, are beginning to become more common, there's a need for 64-bit integers, and that's the motivation for the long long type.

The most common practice today is to set up long long as 64 bits, long as 32 bits, short as 16 bits, and int to either 16 bits or 32 bits, depending on the machine's natural word size. In principle, however, these four types could represent four distinct sizes.

The C standard provides guidelines specifying the minimum allowable size for each basic data type. The minimum range for both short and int is –32,767 to 32,767, corresponding to a 16-bit unit, and the minimum range for long is –2,147,483,647 to 2,147,483,647, corresponding to a 32-bit unit. (Note: For legibility, we've used commas, but C code doesn't allow that option.) For unsigned short and unsigned int, the minimum range is 0 to 65,535, and for unsigned long, the minimum range is 0 to 4,294,967,295. The long long type is intended to support 64-bit needs. Its minimum range is a substantial –9,223,372,036,854,775,807 to 9,223,372,036,854,775,807, and the minimum range for unsigned long long is 0 to 18,446,744,073,709,551,615. (For those of you writing checks, that's eighteen quintillion, four hundred and forty-six quadrillion, seven hundred forty-four trillion, seventy-three billion, seven hundred nine million, five hundred fifty-one thousand, six hundred fifteen in U.S. notation, but who's counting?)

When do you use the various int types? First, consider unsigned types. It is natural to use them for counting because you don't need negative numbers, and the unsigned types enable you to reach higher positive numbers than the signed types.

Use the long type if you need to use numbers that long can handle and that int cannot. However, on systems for which long is bigger than int, using long can slow down calculations, so don't use long if it is not essential. One further point: If you are writing code on a machine for which int and long are the same size, and you do need 32-bit integers, you should use long instead of int so that the program will function correctly if transferred to a 16-bit machine.

Similarly, use long long if you need 64-bit integer values. Some computers already use 64-bit processors, and 64-bit processing in servers, workstations, and even desktops may soon become common.

Use short to save storage space if, say, you need a 16-bit value on a system where int is 32-bit. Usually, saving storage space is important only if your program uses arrays of integers that are large in relation to a system's available memory. Another reason to use short is that it may correspond in size to hardware registers used by particular components in a computer.

Integer Overflow

What happens if an integer tries to get too big for its type? Let's set an integer to its largest possible value, add to it, and see what happens. Try both signed and unsigned types. (The printf() function uses the %u specifier to display unsigned int values.)

/* toobig.c-exceeds maximum int size on our system */
#include <stdio.h>
int main(void)
{
  int i = 2147483647;
  unsigned int j = 4294967295;

  printf("%d %d %d\n", i, i+1, i+2);
  printf("%u %u %u\n", j, j+1, j+2);

  return 0;
}
Here is the result for our system:
2147483647 -2147483648 -2147483647
4294967295 0 1

The unsigned integer j is acting like a car's odometer. When it reaches its maximum value, it starts over at the beginning. The integer i acts similarly. The main difference is that the unsigned int variable j, like an odometer, begins at 0, but the int variable i begins at –2147483648. Notice that you are not informed that i has exceeded (overflowed) its maximum value. You would have to include your own programming to keep tabs on that.

The behavior described here is mandated by the rules of C for unsigned types. The standard doesn't define how signed types should behave. The behavior shown here is typical, but you could encounter something different

long Constants and long long Constants

Normally, when you use a number such as 2345 in your program code, it is stored as an int type. What if you use a number such as 1000000 on a system in which int will not hold such a large number? Then the compiler treats it as a long int, assuming that type is large enough. If the number is larger than the long maximum, C treats it as unsigned long. If that is still insufficient, C treats the value as long long or unsigned long long, if those types are available.

Octal and hexadecimal constants are treated as type int unless the value is too large. Then the compiler tries unsigned int. If that doesn't work, it tries, in order, long, unsigned long, long long, and unsigned long long.

Sometimes you might want the compiler to store a small number as a long integer. Programming that involves explicit use of memory addresses on an IBM PC, for instance, can create such a need. Also, some standard C functions require type long values. To cause a small constant to be treated as type long, you can append an l (lowercase L) or L as a suffix. The second form is better because it looks less like the digit 1. Therefore, a system with a 16-bit int and a 32-bit long treats the integer 7 as 16 bits and the integer 7L as 32 bits. The l and L suffixes can also be used with octal and hex integers, as in 020L and 0x10L.

Similarly, on those systems supporting the long long type, you can use an ll or LL suffix to indicate a long long value, as in 3LL. Add a u or U to the suffix for unsigned long long, as in 5ull or 10LLU or 6LLU or 9Ull.

Printing short, long, long long, and unsigned Types

To print an unsigned int number, use the %u notation. To print a long value, use the %ld format specifier. If int and long are the same size on your system, just %d will suffice, but your program will not work properly when transferred to a system on which the two types are different, so use the %ld specifier for long. You can use the l prefix for x and o, too. Therefore, you would use %lx to print a long integer in hexadecimal format and %lo to print in octal format. Note that although C allows both uppercase and lowercase letters for constant suffixes, these format specifiers use just lowercase.

C has several additional printf() formats. First, you can use an h prefix for short types. Therefore, %hd displays a short integer in decimal form, and %ho displays a short integer in octal form. Both the h and l prefixes can be used with u for unsigned types. For instance, you would use the %lu notation for printing unsigned long types. Listing 3.4 provides an example. Systems supporting the long long types use %lld and %llu for the signed and unsigned versions. Chapter 4 provides a fuller discussion of format specifiers.

Listing 3.4 The print2.c Program

/* print2.c-more printf() properties */
#include <stdio.h>
int main(void)
{
  unsigned int un = 3000000000; /* system with 32-bit int */
  short end = 200;       /* and 16-bit short    */
  long big = 65537;
  long long verybig = 12345678908642;

  printf("un = %u and not %d\n", un, un);
  printf("end = %hd and %d\n", end, end);
  printf("big = %ld and not %hd\n", big, big);
  printf("verybig= %lld and not %ld\n", verybig, verybig);

  return 0;
}

Here is the output on one system:

un = 3000000000 and not -1294967296
end = 200 and 200
big = 65537 and not 1
verybig= 12345678908642 and not 1942899938

This example points out that using the wrong specification can produce unexpected results. First, note that using the %d specifier for the unsigned variable un produces a negative number! The reason for this is that the unsigned value 3000000000 and the signed value –129496296 have exactly the same internal representation in memory on our system. (Chapter 15 explains this property in more detail.) So if you tell printf() that the number is unsigned, it prints one value, and if you tell it that the same number is signed, it prints the other value. This behavior shows up with values larger than the maximum signed value. Smaller positive values, such as 96, are stored and displayed the same for both signed and unsigned types.

Next, note that the short variable end is displayed the same whether you tell printf() that end is a short (the %hd specifier) or an int (the %d specifier). That's because C automatically expands a type short value to a type int value when it's passed as an argument to a function. This may raise two questions in your mind: Why does this conversion take place, and what's the use of the h modifier? The answer to the first question is that the int type is intended to be the integer size that the computer handles most efficiently. So, on a computer for which short and int are different sizes, it may be faster to pass the value as an int. The answer to the second question is that you can use the h modifier to show how a longer integer would look if truncated to the size of short. The third line of output illustrates this point. When the value 65537 is written in binary format as a 32-bit number, it looks like 00000000000000010000000000000001. Using the %hd specifier persuaded printf() to look at just the last 16 bits; therefore, it displayed the value as 1. Similarly, the final output line shows the full value of verybig and then the value stored in the last 32 bits, as viewed through the %ld specifier.

Earlier you saw that it is your responsibility to make sure the number of specifiers matches the number of values to be displayed. Here you see that it is also your responsibility to use the correct specifier for the type of value to be displayed.

Using Characters: Type char

The char type is used for storing characters such as letters and punctuation marks, but technically it is an integer type. Why? Because the char type actually stores integers, not characters. To handle characters, the computer uses a numerical code in which certain integers represent certain characters. The most commonly used code in the U.S. is the ASCII code given in the table on the inside front cover. It is the code this book assumes. In it, for example, the integer value 65 represents an uppercase A. So to store the letter A, you actually need to store the integer 65. (Many IBM mainframes use a different code, called EBCDIC, but the principle is the same. Computer systems outside the U.S. may use entirely different codes.)

The standard ASCII code runs numerically from 0 to 127. This range is small enough that 7 bits can hold it. The char type is typically defined as an 8-bit unit of memory, so it is more than large enough to encompass the standard ASCII code. Many systems, such as the IBM PC and the Apple Macintosh, offer extended ASCII codes (different for the two systems) that still stay within an 8-bit limit. More generally, C guarantees that the char type is large enough to store the basic character set for the system on which C is implemented.

Many character sets have many more than 127 or even 255 values. For example, there is the Japanese kanji character set. The commercial Unicode initiative has created a system to represent a variety of characters sets worldwide and currently has over 96,000 characters. The International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC) has developed a standard called ISO/IEC 10646 for character sets. Fortunately, the Unicode standard has been kept compatible with the more extensive ISO/IEC 10646 standard.

A platform that uses one of these sets as its basic character set could use a 16-bit or even a 32-bit char representation. The C language defines a byte to be the number of bits used by type char, so as far as C documentation goes, a byte would be 16 or 32 bits, rather than 8 bits on such systems.

Declaring Type char Variables

As you might expect, char variables are declared in the same manner as other variables. Here are some examples:

char response;
char itable, latan;

This code would create three char variables: response, itable, and latan.

Character Constants and Initialization

Suppose you want to initialize a character constant to the letter A. Computer languages are supposed to make things easy, so you shouldn't have to memorize the ASCII code, and you don't. You can assign the character A to grade with the following initialization:

char grade = 'A';

A single letter contained between single quotes is a C character constant. When the compiler sees 'A', it converts the 'A' to the proper code value. The single quotes are essential. Here's another example:

char broiled;     /* declare a char variable      */
broiled = 'T';    /* OK                           */
broiled = T;      /* NO! Thinks T is a variable   */
broiled = "T";    /* NO! Thinks "T" is a string   */

If you omit the quotes, the compiler thinks that T is the name of a variable. If you use double quotes, it thinks you are using a string. We'll discuss strings in Chapter 4.

Because characters are really stored as numeric values, you can also use the numerical code to assign values:

char grade = 65; /* ok for ASCII, but poor style */

In this example, 65 is type int, but, because the value is smaller than the maximum char size, it can be assigned to grade without any problems. Because 65 is the ASCII code for the letter A, this example assigns the value A to grade. Note, however, that this example assumes that the system is using ASCII code. Using 'A' instead of 65 produces code that works on any system. Therefore, it's much better to use character constants than numeric code values.

Somewhat oddly, C treats character constants as type int rather than type char. For example, on an ASCII system with a 32-bit int and an 8-bit char, the code

char grade = 'B';

represents 'B' as the numerical value 66 stored in a 32-bit unit, but grade winds up with 66 stored in an 8-bit unit. This characteristic of character constants makes it possible to define a character constant such as 'FATE', with four separate 8-bit ASCII codes stored in a 32-bit unit. However, attempting to assign such a character constant to a char variable results in only the last 8 bits being used, so the variable gets the value 'E'.

Nonprinting Characters

The single-quote technique is fine for characters, digits, and punctuation marks, but if you look through the table on the inside front cover of this book, you'll see that some of the ASCII characters are nonprinting. For example, some represent actions such as backspacing or going to the next line or making the terminal bell ring (or speaker beep). How can these be represented? C offers three ways.

The first way we have already mentioned—just use the ASCII code. For example, the ASCII value for the beep character is 7, so you can do this:

char beep = 7;

The second way to represent certain awkward characters in C is to use special symbol sequences. These are called escape sequences. Table 3.2 shows the escape sequences and their meanings.

Table 3.2 Escape Sequences

Escape sequences must be enclosed in single quotes when assigned to a character variable. For example, you could make the statement

char nerf = '\n';

and then print the variable nerf to advance the printer or screen one line.

Now take a closer look at what each escape sequence does. The alert character (\a), added by C90, produces an audible or visible alert. The nature of the alert depends on the hardware, with the beep being the most common. (With some systems, the alert character has no effect.) The ANSI standard states that the alert character shall not change the active position. By active position, the standard means the location on the display device (screen, teletype, printer, and so on) at which the next character would otherwise appear. In short, the active position is a generalization of the screen cursor with which you are probably accustomed. Using the alert character in a program displayed on a screen should produce a beep without moving the screen cursor.

Next, the \b, \f, \n, \r, \t, and \v escape sequences are common output device control characters. They are best described in terms of how they affect the active position. A backspace (\b) moves the active position back one space on the current line. A form feed character (\f) advances the active position to the start of the next page. A newline character (\n) sets the active position to the beginning of the next line. A carriage return (\r) moves the active position to the beginning of the current line. A horizontal tab character (\t) moves the active position to the next horizontal tab stop (typically, these are found at character positions 1, 9, 17, 25, and so on). A vertical tab (\v) moves the active position to the next vertical tab position.

These escape sequence characters do not necessarily work with all display devices. For example, the form feed and vertical tab characters produce odd symbols on a PC screen instead of any cursor movement, but they work as described if sent to a printer instead of to the screen.

The next three escape sequences (\\, \', and \") enable you to use \, ', and " as character constants. (Because these symbols are used to define character constants as part of a printf() command, the situation could get confusing if you use them literally.) Suppose you want to print the following line:

Gramps sez, "a \ is a backslash."

Then use this code:

printf("Gramps sez, \"a \\ is a backslash.\"\n");

The final two forms (\0oo and \xhh) are special representations of the ASCII code. To represent a character by its octal ASCII code, precede it with a backslash (\) and enclose the whole thing in single quotes. For example, if your compiler doesn't recognize the alert character (\a), you could use the ASCII code instead:

beep = '\007';

You can omit the leading zeros, so '\07' or even '\7' will do. This notation causes numbers to be interpreted as octal, even if there is no initial 0.

Beginning with C90, C provides a third option—using a hexadecimal form for character constants. In this case, the backslash is followed by an x or X and one to three hexadecimal digits. For example, the Ctrl+P character has an ASCII hex code of 10 (16, in decimal), so it can be expressed as '\x10' or '\X010'. Figure 3.5 shows some representative integer types.

Figure 3.5 Writing constants with the int family.

When you use ASCII code, note the difference between numbers and number characters. For example, the character 4 is represented by ASCII code value 52. The notation '4' represents the symbol 4, not the numerical value 4.

At this point, you may have three questions:

• Why aren't the escape sequences enclosed in single quotes in the last example (printf("Gramps sez, \"a \\ is a backslash\"\"n");)? When a character, be it an escape sequence or not, is part of a string of characters enclosed in double quotes, don't enclose it in single quotes. Notice that none of the other characters in this example (G, r, a, m, p, s, and so on) are marked off by single quotes. A string of characters enclosed in double quotes is called a character string. (Chapter 4 explores strings.) Similarly, printf("Hello!\007\n"); will print Hello! and beep, but printf("Hello!7\n"); will print Hello!7. Digits that are not part of an escape sequence are treated as ordinary characters to be printed.

• When should I use the ASCII code, and when should I use the escape sequences? If you have a choice between using one of the special escape sequences, say '\f', or an equivalent ASCII code, say '\014', use the '\f'. First, the representation is more mnemonic. Second, it is more portable. If you have a system that doesn't use ASCII code, the '\f' will still work.

• If I need to use numeric code, why use, say, '\032' instead of 032?—First, using '\032' instead of 032 makes it clear to someone reading the code that you intend to represent a character code. Second, an escape sequence such as \032 can be embedded in part of a C string, the way \007 was in the first point.

Printing Characters

The printf() function uses %c to indicate that a character should be printed. Recall that a character variable is stored as a 1-byte integer value. Therefore, if you print the value of a char variable with the usual %d specifier, you get an integer. The %c format specifier tells printf() to display the character that has that integer as its code value. Listing 3.5 shows a char variable both ways.

Listing 3.5 The charcode.c Program

/* charcode.c-displays code number for a character */
#include <stdio.h>
int main(void)
{
  char ch;

  printf("Please enter a character.\n");
  scanf("%c", &ch);  /* user inputs character */
  printf("The code for %c is %d.\n", ch, ch);

  return 0;
}

Here is a sample run:

Please enter a character.
C
The code for C is 67.

When you use the program, remember to press the Enter or Return key after typing the character. The scanf() function then fetches the character you typed, and the ampersand (&) causes the character to be assigned to the variable ch. The printf() function then prints the value of ch twice, first as a character (prompted by the %c code) and then as a decimal integer (prompted by the %d code). Note that the printf() specifiers determine how data is displayed, not how it is stored (see Figure 3.6).

Figure 3.6 Data display versus data storage.

Signed or Unsigned?

Some C implementations make char a signed type. This means a char can hold values typically in the range –128 through 127. Other implementations make char an unsigned type, which provides a range of 0 through 255. Your compiler manual should tell you which type your char is, or you can check the limits.h header file, discussed in the next chapter.

With C90, C enables you to use the keywords signed and unsigned with char. Then, regardless of what your default char is, signed char would be signed, and unsigned char would be unsigned. These versions of char are useful if you're using the type to handle small integers. For character use, just use the standard char type without modifiers.

The _Bool Type

The _Bool type is a C99 addition that's used to represent Boolean values—that is, the logical values true and false. Because C uses the value 1 for true and 0 for false, the _Bool type really is just an integer type, but one that, in principle, only requires 1 bit of memory, because that is enough to cover the full range from 0 to 1.

Programs use Boolean values to choose which code to execute next. Code execution is covered more fully in Chapter 6, "C Control Statements: Looping," and Chapter 7, "C Control Statements: Branching and Jumps," so let's defer further discussion until then.

Portable Types: inttypes.h

Are there even more integer types? No, but there are more names that you can use for the existing types. You might think you've seen more than an adequate number of names, but the primary names do have a problem. Knowing that a variable is an int doesn't tell you how many bits it is unless you check the documentation for your system. To get around this problem, C99 provides an alternative set of names that describes exactly what you get. For example, the name int16_t indicates a 16-bit signed integer type and the name uint32_t indicates a 32-bit unsigned integer type.

To make these names available to a program, include the inttypes.h header file. (Note that at the time this edition was prepared, some compilers don't yet support this feature.) That file uses the typedef facility (first described briefly in Chapter 5, "Operators, Expressions, and Statements") to create new type names. For example, it will make uint32_t a synonym or alias for a standard type with the desired characteristics—perhaps unsigned int on one system and unsigned long on another. Your compiler will provide a header file consistent with the computer system you are using. These new designations are called exact width types. Note that, unlike int, uint32_t is not a keyword, so the compiler won't recognize it unless you include the inttypes.h header file.

One possible problem with attempting to provide exact width types is that a particular system might not support some of the choices, so there is no guarantee that there will be, say, an int8_t type (8-bit signed). To get around that problem, the C99 standard defines a second set of names that promises the type is at least big enough to meet the specification and that no other type that can do the job is smaller. These types are called minimum width types. For example, int_least8_t will be an alias for the smallest available type that can hold an 8-bit signed integer value. If the smallest type on a particular system were 8 bits, the int8_t type would not be defined. However, the int_least8_t type would be available, perhaps implemented as a 16-bit integer.

Of course, some programmers are more concerned with speed than with space. For them, C99 defines a set of types that will allow the fastest computations. These are called the fastest minimum width types. For example, the int_fast8_t will be defined as an alternative name for the integer type on your system that allows the fastest calculations for 8-bit signed values.

Finally, for some programmers, only the biggest possible integer type on a system will do; intmax_t stands for that type, a type that can hold any valid signed integer value. Similarly, uintmax_t stands for the largest available unsigned type. Incidentally, these types could be bigger than long long and unsigned long because C implementations are permitted to define types beyond the required ones.

C99 not only provides these new, portable type names, it also has to assist with input and output. For example, printf() requires specific specifiers for particular types. So what do you do to display an int32_t value when it might require a %d specifier for one definition and an %ld for another? The C99 standard provides some string macros (introduced in Chapter 4) to be used to display the portable types. For example, the inttypes.h header file will define PRId16 as a string representing the appropriate specifier (hd or d, for instance) for a 16-bit signed value. Listing 3.6 shows a brief example illustrating how to use a portable type and its associated specifier.

Listing 3.6 The altnames.c Program

/* altnames.c -- portable names for integer types */
#include <stdio.h>
#include <inttypes.h> // supports portable types
int main(void)
{
  int16_t me16;   // me16 a 16-bit signed variable

  me16 = 4593;
  printf("First, assume int16_t is short: ");
  printf("me16 = %hd\n", me16);
  printf("Next, let's not make any assumptions.\n");
  printf("Instead, use a \"macro\" from inttypes.h: ");
  printf("me16 = %" PRId16 "\n", me16);

  return 0;
}

In the final printf() argument, the PRId16 is replaced by its inttypes.h definition of "hd", making the line this:

printf("me16 = %" "hd" "\n", me16);

But C combines consecutive quoted strings into a single quoted string, making the line this:

printf("me16 = %hd\n", me16);

Here's the output; note that the example also uses the \" escape sequence to display double quotation marks:

First, assume int16_t is short: me16 = 4593
Next, let's not make any assumptions.
Instead, use a "macro" from inttypes.h: me16 = 4593

Reference Section VI, "Expanded Integer Types," provides a complete rundown of the inttypes.h header file additions and also lists all the specifier macros.

Types float, double, and long double

The various integer types serve well for most software development projects. However, financial and mathematically oriented programs often make use of floating-point numbers. In C, such numbers are called type float, double, or long double. They correspond to the real types of FORTRAN and Pascal. The floating-point approach, as already mentioned, enables you to represent a much greater range of numbers, including decimal fractions. Floating-point number representation is similar to scientific notation, a system used by scientists to express very large and very small numbers. Let's take a look.

In scientific notation, numbers are represented as decimal numbers times powers of 10. Here are some examples.

The first column shows the usual notation, the second column scientific notation, and the third column exponential notation, or e-notation, which is the way scientific notation is usually written for and by computers, with the e followed by the power of 10. Figure 3.7shows more floating-point representations.

The C standard provides that a float has to be able to represent at least six significant figures and allow a range of at least 10–37 to 10+37. The first requirement means, for example, that a float has to represent accurately at least the first six digits in a number such as 33.333333. The second requirement is handy if you like to use numbers such as the mass of the sun (2.0e30 kilograms), the charge of a proton (1.6e–19 coulombs), or the national debt. Often, systems use 32 bits to store a floating-point number. Eight bits are used to give the exponent its value and sign, and 24 bits are used to represent the nonexponent part, called the mantissa or significand, and its sign.

Figure 3.7 Some floating-point numbers.

C also has a double (for double precision) floating-point type. The double type has the same minimum range requirements as float, but it extends the minimum number of significant figures that can be represented to 10. Typical double representations use 64 bits instead of 32. Some systems use all 32 additional bits for the nonexponent part. This increases the number of significant figures and reduces round-off errors. Other systems use some of the bits to accommodate a larger exponent; this increases the range of numbers that can be accommodated. Either approach leads to at least 13 significant figures, more than meeting the minimum standard.

C allows for a third floating-point type: long double. The intent is to provide for even more precision than double. However, C guarantees only that long double is at least as precise as double.

Declaring Floating-Point Variables

Floating-point variables are declared and initialized in the same manner as their integer cousins. Here are some examples:

float noah, jonah;
double trouble;
float planck = 6.63e-34;
long double gnp;

Floating-Point Constants

There are many choices open to you when you write a floating-point constant. The basic form of a floating-point constant is a signed series of digits, including a decimal point, followed by an e or E, followed by a signed exponent indicating the power of 10 used. Here are two valid floating-point constants:

-1.56E+12

2.87e-3

You can leave out positive signs. You can do without a decimal point (2E5) or an exponential part (19.28), but not both simultaneously. You can omit a fractional part (3.E16) or an integer part (.45E–6), but not both (that wouldn't leave much!). Here are some more valid floating-point constants:

3.14159

.2

4e16

.8E-5

100.

Don't use spaces in a floating-point constant.

Wrong: 1.56 E+12

By default, the compiler assumes floating-point constants are double precision. Suppose, for example, that some is a float variable and that you have the following statement:

some = 4.0 * 2.0;

Then 4.0 and 2.0 are stored as double, using (typically) 64 bits for each. The product is calculated using double precision arithmetic, and only then is the answer trimmed to regular float size. This ensures greater precision for your calculations, but it can slow down a program.

C enables you to override this default by using an f or F suffix to make the compiler treat a floating-point constant as type float; examples are 2.3f and 9.11E9F. An l or L suffix makes a number type long double; examples are 54.3l and 4.32e4L. Note that L is less likely to be mistaken for 1 (one) than is l. If the floating-point number has no suffix, it is type double.

C99 has added a new format for expressing floating-point constants. It uses a hexadecimal prefix (0x or 0X) with hexadecimal digits, a p or P instead of e or E, and an exponent that is a power of 2 instead of a power of 10. Here's what such a number might look like:

0xa.1fp10

The a is 10, the .1f is 1/16th plus 15/256th, and the p10 is 210, or 1024, making the complete value 10364.0 in base 10 notation.

Not all C compilers have added support for this C99 feature.

Printing Floating-Point Values

The printf() function uses the %f format specifier to print type float and double numbers using decimal notation, and it uses %e to print them in exponential notation. If your system supports the C99 hexadecimal format for floating-point numbers, you can use a or A instead of e or E. The long double type requires the %Lf, %Le, and %La specifiers to print that type. Note that both float and double use the %f, %e, or %a specifier for output. That's because C automatically expands type float values to type double when they are passed as arguments to any function, such as printf(), that doesn't explicitly prototype the argument type. Listing 3.7 illustrates these behaviors.

Listing 3.7 The showf_pt.c Program

/* showf_pt.c -- displays float value in two ways */
#include <stdio.h>
int main(void)
{
  float aboat = 32000.0;
  double abet = 2.14e9;
  long double dip = 5.32e-5;

  printf("%f can be written %e\n", aboat, aboat);
  printf("%f can be written %e\n", abet, abet);
  printf("%f can be written %e\n", dip, dip);

  return 0;
}

This is the output:

32000.000000 can be written 3.200000e+04
2140000000.000000 can be written 2.140000e+09
0.000053 can be written 5.320000e-05

This example illustrates the default output. The next chapter discusses how to control the appearance of this output by setting field widths and the number of places to the right of the decimal.

Floating-Point Overflow and Underflow

Suppose the biggest possible float value on your system is about 3.4E38 and you do this:

float toobig = 3.4E38 * 100.0f;
printf("%e\n", toobig);

What happens? This is an example of overflow—when a calculation leads to a number too large to be expressed. The behavior for this case used to be undefined, but now C specifies that toobig gets assigned a special value that stands for infinity and that printf() displays either inf or infinity (or some variation on that theme) for the value.

What about dividing very small numbers? Here the situation is more involved. Recall that a float number is stored as an exponent and as a value part, or mantissa. There will be a number that has the smallest possible exponent and also the smallest value that still uses all the bits available to represent the mantissa. This will be the smallest number that still is represented to the full precision available to a float value. Now divide it by 2. Normally, this reduces the exponent, but the exponent already is as small as it can get. So, instead, the computer moves the bits in the mantissa over, vacating the first position and losing the last binary digit. An analogy would be taking a base 10 value with four significant digits, such as 0.1234E-10, dividing by 10, and getting 0.0123E-10. You get an answer, but you've lost a digit in the process. This situation is called underflow, and C refers to floating-point values that have lost the full precision of the type as subnormal. So dividing the smallest positive normal floating-point value by 2 results in a subnormal value. If you divide by a large enough value, you lose all the digits and are left with 0. The C library now provides functions that let you check whether your computations are producing subnormal values.

There's another special floating-point value that can show up: NaN, or not-a-number. For example, you give the asin() function a value, and it returns the angle that has that value as its sine. But the value of a sine can't be greater than 1, so the function is undefined for values in excess of 1. In such cases, the function returns the NaN value, which printf() displays as nan, NaN, or something similar.

/* floaterr.c--demonstrates round-off error */
#include <stdio.h>
int main(void)
{
  float a,b;

  b = 2.0e20 + 1.0;
  a = b - 2.0e20;
  printf("%f \n", a);

  return 0;
}

0.000000  _older gcc on Linux
-13584010575872.000000  _Turbo C 1.5
4008175468544.000000  _CodeWarrior 9.0, MSVC++ 7.1

Complex and Imaginary Types

Many computations in science and engineering use complex and imaginary numbers. C99 supports these numbers, with some reservations. A free-standing implementation, such as that used for embedded processors, doesn't need to have these types. (A VCR chip probably doesn't need complex numbers to do its job.) Also, more generally, the imaginary types are optional.

In brief, there are three complex types, called float _Complex, double _Complex, and long double _Complex. A float _Complex variable, for example, would contain two float values, one representing the real part of a complex number and one representing the imaginary part. Similarly, there are three imaginary types, called float _Imaginary, double _Imaginary, and long double _Imaginary.

Including the complex.h header file lets you substitute the word complex for _Complex and the word imaginary for _Imaginary, and it allows you to use the symbol I to represent the square root of –1.

Beyond the Basic Types

That finishes the list of fundamental data types. For some of you, the list must seem long. Others of you might be thinking that more types are needed. What about a character string type? C doesn't have one, but it can still deal quite well with strings. You will take a first look at strings in Chapter 4.

C does have other types derived from the basic types. These types include arrays, pointers, structures, and unions. Although they are subject matter for later chapters, we have already smuggled some pointers into this chapter's examples. (A pointer points to the location of a variable or other data object. The & prefix used with the scanf() function creates a pointer telling scanf() where to place information.)

Type Sizes

Tables 3.3 and 3.4 show type sizes for some common C environments. (In some environments, you have a choice.) What is your system like? Try running the program in Listing 3.8 to find out.

Table 3.3 Integer Type Sizes (Bits) for Representative Systems

Table 3.4 Floating-point Facts for Representative Systems

For each type, the top row is the number of significant digits and the second row is the exponent range (base 10).

Listing 3.8 The typesize.c Program

/* typesize.c -- prints out type sizes */
#include <stdio.h>
int main(void)
{
/* c99 provides a %zd specifier for sizes */
  printf("Type int has a size of %u bytes.\n", sizeof(int));
  printf("Type char has a size of %u bytes.\n", sizeof(char));
  printf("Type long has a size of %u bytes.\n", sizeof(long));
  printf("Type double has a size of %u bytes.\n",
      sizeof(double));
  return 0;
}

C has a built-in operator called sizeof that gives sizes in bytes. (Some compilers require %lu instead of %u for printing sizeof quantities. That's because C leaves some latitude as to the actual unsigned integer type that sizeof uses to report its findings. C99 provides a %zd specifier for this type, and you should use it if your compiler supports it.) The output from List- ing 3.8 is as follows:

Type int has a size of 4 bytes.
Type char has a size of 1 bytes.
Type long has a size of 4 bytes.
Type double has a size of 8 bytes.

This program found the size of only four types, but you can easily modify it to find the size of any other type that interests you. Note that the size of char is necessarily 1 byte because C defines the size of 1 byte in terms of char. So, on a system with a 16-bit char and a 64-bit double, sizeof will report double as having a size of 4 bytes. You can check the limits.h and float.h header files for more detailed information on type limits. (The next chapter discusses these two files further.)

Incidentally, notice in the last line how the printf() statement is spread over two lines. You can do this as long as the break does not occur in the quoted section or in the middle of a word.