The C preprocessor provides the ability to perform macro substitution, conditional compilation, and inclusion of named files. Preprocessor directives, lines beginning with # and possibly preceded by white space, are used to communicate with the preprocessor.

The following sections describe the preprocessor directives and operators available with the Compaq C compiler:

• The #define and #undef directives, and the # and ## operators ( Section 8.1)
• The #if , #ifdef , #ifndef , #else , #elif , and #endif directives, and the defined operator ( Section 8.2)
• The #include directive ( Section 8.3)
• The #line directive ( Section 8.4)
• The #pragma directive ( Section 8.5)
• The #error directive ( Section 8.6)
• The null directive ( # ) ( Section 8.7)

Preprocessor directives are independent of the usual scope rules; they remain in effect from their occurrence until the end of the compilation unit or until their effect is canceled.

See Section 8.2 for more information about conditional compilation. See your platform-specific Compaq C documentation for implementation-defined information about preprocessor directives.

The ANSI standard allows only comments as text following a preprocessing directive. The Compaq C compiler issues a warning if this syntax rule is violated in all modes but the strict ANSI mode, in which it issues an error message.

8.1 Macro Definition (#define and #undef)

The #define directive specifies a macro identifier and a replacement list, and terminates with a new-line character. The replacement list, a sequence of preprocessing tokens, is substituted for every subsequent occurrence of that macro identifier in the program text, unless the identifier occurs inside a character constant, a comment, or a literal string. The #undef directive is used to cancel a definition for a macro.

A macro definition is independent of block structure, and is in effect from the #define directive that defines it until either a corresponding #undef directive or the end of the compilation unit is encountered.

The #define directive has the following syntax:

#define identifier replacement-list newline

#define identifier ( identifier-listopt ) replacement-list newline

#define identifier ( ... ) replacement-list newline

#define identifier ( identifier-list, ... replacement-list newline

If the replacement-list is empty, subsequent occurrences of the identifier are deleted from the source file.

The first form of the #define directive is called an object-like macro. The last three forms are called function-like macros.

The #undef directive has the following syntax:

#undef identifier newline

This directive cancels a previous definition of the identifier by #define . Redefining a macro previously defined is not legal, unless the new definition is precisely the same as the old.

The replacement list in the macro definition, as well as arguments in a function-like macro reference, can contain other macro references. Compaq C does not limit the depth to which such references can be nested.

For a given macro definition, any macro names contained in the replacement list are themselves replaced by their currently specified replacement lists. If a macro name being defined is contained in its own replacement list or in a nested replacement list, it is not replaced. These nonreplaced macro names are then no longer available for further replacement, even if they are later examined in contexts in which they would otherwise be replaced.

The following example shows nested #define directives:

/* Show multiple substitutions and listing format. */ #define AUTHOR james + LAST main() { int writer,james,michener,joyce; #define LAST michener writer = AUTHOR; #undef LAST #define LAST joyce writer = AUTHOR; }

After this example is compiled with the appropriate options to show intermediate macro expansions, the following listing results:

1 /* Show multiple substitutions and listing format. */ 2 3 #define AUTHOR james + LAST 4 5 main() 6 { 7 int writer, james, michener, joyce; 8 9 #define LAST michener 10 writer = AUTHOR; 10.1 james + LAST 10.2 michener 11 #undef LAST 12 #define LAST joyce 13 writer = AUTHOR; 13.1 james + LAST 13.2 joyce 14 }

On the first pass, the compiler replaces the identifier AUTHOR with the replacement list james + LAST . On the second pass, the compiler replaces the identifier LAST with its currently defined replacement list value. At line 9, the replacement list value for LAST is the identifier michener , so michener is substituted at line 10. At line 12, the replacement list value for LAST is redefined to be the identifier joyce , so joyce is substituted at line 13.

The #define directive may be continued onto subsequent lines if necessary. To do this, end each line to be continued with a backslash (\) immediately followed by a new-line character. The backslash and new-line characters do not become part of the definition. The first character in the next line is logically adjacent to the character that immediately precedes the backslash. The backslash/newline as a continuation sequence is valid anywhere. However, comments within the definition line can be continued without the backslash/newline.

If you plan to port programs to and from other C implementations, take care in choosing which macro definitions to use within your programs, because some implementations define different macros than others.

8.1.1 Object-Like Form

A preprocessing directive of the following form defines an object-like macro that causes each subsequent occurrence of the macro name to be replaced by the replacement list:

#define identifier replacement-list newline

An object like macro may be redefined by another #define directive provided that the second definition is an object-like macro definition and the two replacement lists are identical. This means that two files, each with a definition of a certain macro, must be consistent in that definition.

The object-like form of macro definition defines a descriptive name for a frequently used token. A common use of the directive is to define the end-of-file ( EOF ) indicator as follows:

#define EOF (-1)

8.1.2 Function-Like Form

The function-like form of macro definition includes a list of parameters. References to such macros look like function calls. When a function is called, control passes from the program to the function at run time; when a macro is referenced, source code is inserted into the program at compile time. The parameters are replaced by the corresponding arguments, and the text is inserted into the program stream.

If there is a ... in the identifier-list in the macro definition, then the trailing arguments, including any separating comma preprocessing tokens, are merged to form a single item: the variable arguments. The number of arguments so combined is such that, following the merger, the number of arguments is one more than the number of parameters in the macro definition (excluding the ...).

An identifier __VA_ARGS__ that occurs in the replacement-list of a function-like macro that uses ellipsis notation in the arguments is treated as if it were a parameter, and the variable arguments form the preprocessing tokens used to replace it.

If the replacement list is omitted from the macro definition, the entire macro reference disappears from the source text.

The library macro _toupper , available on some systems in the ctype.h header file, is a good example of macro replacement. This macro is defined as follows:

#define _toupper(c) ((c) >= 'a' && (c) <= 'z' ? (c) & 0X5F : (c))

When the macro _toupper is referenced, the compiler replaces the macro and its parameter with the replacement list from the directive, substituting the argument of the macro reference for each occurrence of the parameter ( c in this case) in the replacement list.

The replacement list of C source code can be translated in the following manner: if parameter c is a lowercase letter (between 'a' and 'z' ), the expression evaluates to an uppercase letter ( c & 0X5F ); otherwise, it evaluates to the character as specified. This replacement list uses the if-then-else conditional operator ( ?: ). For more information about the conditional operator, see Section 6.6. For more information about the bitwise operators, see Section 6.5.6.

8.1.2.1 Rules for Specifying Macro Definitions

Preprocessor directives and macro references have syntax that is independent of the C language. Follow these rules when specifying macro definitions:

• The macro name and the formal parameters are identifiers and are specified according to the rules for identifiers in the C language.
• Spaces, tabs, and comments may be used freely within a #define directive anywhere that the delta symbol (<ucDelta symbol>) appears in the following example:

  <ucDelta symbol> #<ucDelta symbol>define <ucDelta symbol> name(<ucDelta symbol>parm1<ucDelta symbol>,<ucDelta symbol>parm2<ucDelta symbol>)<ucDelta symbol>\ <ucDelta symbol>token-string<ucDelta symbol>

  
  Spaces, tabs, and comments are replaced by a single space.

• White space cannot appear between the name and the left parenthesis that introduces the parameter list. White space may appear inside the replacement list. Also, at least one space, tab, or comment must separate name from define .
• The identifier __VA_ARGS__ can occur only in the replacement-list of a function-like macro that uses the ellipsis notation in the arguments.

Follow these rules when specifying macro references:

• Comments and white-space characters (spaces, horizontal and vertical tabs, new-line characters, and form feeds) may be used freely within a macro reference anywhere that the delta symbol (<ucDelta symbol>) appears in the following example:

  <ucDelta symbol>name<ucDelta symbol>(<ucDelta symbol>arg1<ucDelta symbol>,<ucDelta symbol>arg2<ucDelta symbol>)

• Arguments consist of arbitrary text. Syntactically, they are not restricted to C expressions. They may contain embedded comments and white space. Comments are replaced with a single space. White space (except for leading and trailing white space) is preserved during the substitution.
• The number of arguments in the reference must match the number of parameters in the macro definition. Null arguments result in undefined behavior.
• Commas separate arguments except where the commas occur inside string or character constants, comments, or pairs of parentheses. Parentheses must be balanced within arguments.

It is not good programming practice to specify macro arguments that use the increment (++), decrement (-- --), and assignment operators (such as +=) or other arguments that can cause side effects. For example, do not pass the following argument to the _toupper macro:

_toupper(p++)

When the argument p++ is substituted in the macro definition, the effect within the program stream is as follows:

((p++) >= 'a' && (p++) <= 'z' ? (p++) & 0X5F : (p++))

Because p is being incremented, it does not have the same value for each occurrence in this macro replacement. Even if you are aware of possible side effects, the replacement lists within macro definitions can be changed, which changes the side effects without warning.

8.1.3 Conversions to String Literals (#)

The # preprocessor operator is used to convert the argument that follows it to a string literal. The preprocessor operator # can be used only in a function-like macro definition. For example:

#include <stdio.h> #define PR(id) printf("The value of " #id " is %d\n", id) main() { int i = 10; PR(i); }

The output produced is:

The value of i is 10

The macro call expands in the following steps:

/*1*/ printf("The value of " #id " is %d\n", id) /*2*/ printf("The value of " "i" " is %d\n", 10) /*3*/ printf("The value of i is %d\n", 10)

The unary # operator produces a string from its operand. This example also uses the fact that adjacent string literals are concatenated. If the operand to # contains double quotes or escape sequences, they are also expanded. For example:

#include <stdio.h> #define M(arg) printf(#arg " is %s\n", arg) main() { M("a\nb\tc"); }

The macro call expands using the following steps:

/*1*/ printf(#arg " is %s\n", arg) /*2*/ printf("\"a\\nb\\tc\"" " is %s\n", "a\nb\tc"); /*3*/ printf("\"a\\nb\\tc\" is %s\n", "a\nb\tc");

8.1.4 Token Concatenation(##)

The ## preprocessor operator is used to concatenate two tokens into a third valid token, as in the following example:

#define glue(a,b) a ## b main() { int wholenum = 5000; printf("%d", glue(whole,num)); }

The preprocessor converts the line printf("%d", glue(whole,num)); into printf("%d", wholenum); , and when executed, the program prints 5000. If the result is not a valid token, an error occurs when the tokens are concatenated.

In Compaq C, the ## operator is evaluated before any # operators on the line. ## and # operators group left-to-right.

8.2 Conditional Compilation (#if, #ifdef, #ifndef, #else, #elif, #endif, and defined)

Six directives are available to control conditional compilation. They delimit blocks of program text that are compiled only if a specified condition is true. These directives can be nested. The program text within the blocks is arbitrary and may consist of preprocessor directives, C statements, and so on. The beginning of the block of program text is marked by one of three directives:

• #if
• #ifdef
• #ifndef

Optionally, an alternative block of text can be set aside with one of two directives:

• #else
• #elif

The end of the block or alternative block is marked by the #endif directive.

If the condition checked by #if , #ifdef , or #ifndef is true (nonzero), then all lines between the matching #else (or #elif ) and an #endif directive, if present, are ignored.

If the condition is false (0), then the lines between the #if , #ifdef , or #ifndef and an #else , #elif , or #endif directive are ignored.

8.2.1 The #if Directive

The #if directive has the following syntax:

#if constant-expression newline

This directive checks whether the constant-expression is true (nonzero). The operand must be a constant integer expression that does not contain any increment (++), decrement (-- --), sizeof , pointer (*), address (&), and cast operators.

Identifiers in the constant expression either are or are not macro names. There are no keywords, enumeration constants, and so on. The constant expression can also include the defined preprocessing operator (see Section 8.2.7).

The constant expression in an #if directive is subject to text replacement and can contain references to identifiers defined in previous #define directives. The replacement occurs before the expression is evaluated. Each preprocessing token that remains after all macro replacements have occurred is in the lexical form of a token.

If an identifier used in the expression is not currently defined, the compiler treats the identifier as though it were the constant zero.

8.2.2 The #ifdef Directive

The #ifdef directive has the following syntax:

#ifdef identifier newline

This directive checks whether the identifier is currently defined. Identifiers can be defined by a #define directive or on the command line. If such identifiers have not been subsequently undefined, they are considered currently defined.

8.2.3 The #ifndef Directive

The #ifndef directive has the following syntax:

#ifndef identifier newline

This directive checks to see if the identifier is not currently defined.

8.2.4 The #else Directive

The #else directive has the following syntax:

#else newline

This directive delimits alternative source text to be compiled if the condition tested for in the corresponding #if , #ifdef , or #ifndef directive is false. An #else directive is optional.

8.2.5 The #elif Directive

The #elif directive has the following syntax:

#elif constant-expression newline

The #elif directive performs a task similar to the combined use of the else-if statements in C. This directive delimits alternative source lines to be compiled if the constant expression in the corresponding #if , #ifdef , #ifndef , or another #elif directive is false and if the additional constant expression presented in the #elif line is true. An #elif directive is optional.