3 @settitle The C Preprocessor
10 @c man begin COPYRIGHT
11 Copyright @copyright{} 1987, 1989, 1991, 1992, 1993, 1994, 1995, 1996,
12 1997, 1998, 1999, 2000, 2001, 2002
13 Free Software Foundation, Inc.
15 Permission is granted to copy, distribute and/or modify this document
16 under the terms of the GNU Free Documentation License, Version 1.1 or
17 any later version published by the Free Software Foundation. A copy of
18 the license is included in the
20 section entitled ``GNU Free Documentation License''.
22 @c man begin COPYRIGHT
29 @c man begin COPYRIGHT
30 This manual contains no Invariant Sections. The Front-Cover Texts are
31 (a) (see below), and the Back-Cover Texts are (b) (see below).
33 (a) The FSF's Front-Cover Text is:
37 (b) The FSF's Back-Cover Text is:
39 You have freedom to copy and modify this GNU Manual, like GNU
40 software. Copies published by the Free Software Foundation raise
41 funds for GNU development.
45 @macro gcctabopt{body}
49 @c Create a separate index for command line options.
53 @c Used in cppopts.texi and cppenv.texi.
57 @dircategory Programming
59 * Cpp: (cpp). The GNU C preprocessor.
64 @title The C Preprocessor
65 @subtitle Last revised April 2001
66 @subtitle for GCC version 3
67 @author Richard M. Stallman
68 @author Zachary Weinberg
70 @c There is a fill at the bottom of the page, so we need a filll to
72 @vskip 0pt plus 1filll
82 The C preprocessor implements the macro language used to transform C,
83 C++, and Objective-C programs before they are compiled. It can also be
95 * Preprocessor Output::
97 * Implementation Details::
99 * Environment Variables::
100 * GNU Free Documentation License::
101 * Index of Directives::
106 --- The Detailed Node Listing ---
110 * Initial processing::
112 * The preprocessing language::
117 * Include Operation::
119 * Once-Only Headers::
120 * Computed Includes::
126 * Object-like Macros::
127 * Function-like Macros::
132 * Predefined Macros::
133 * Undefining and Redefining Macros::
134 * Directives Within Macro Arguments::
139 * Standard Predefined Macros::
140 * Common Predefined Macros::
141 * System-specific Predefined Macros::
142 * C++ Named Operators::
147 * Operator Precedence Problems::
148 * Swallowing the Semicolon::
149 * Duplication of Side Effects::
150 * Self-Referential Macros::
152 * Newlines in Arguments::
157 * Conditional Syntax::
168 Implementation Details
170 * Implementation-defined behavior::
171 * Implementation limits::
172 * Obsolete Features::
173 * Differences from previous versions::
178 * Obsolete once-only headers::
189 @c man begin DESCRIPTION
190 The C preprocessor, often known as @dfn{cpp}, is a @dfn{macro processor}
191 that is used automatically by the C compiler to transform your program
192 before compilation. It is called a macro processor because it allows
193 you to define @dfn{macros}, which are brief abbreviations for longer
196 The C preprocessor is intended to be used only with C, C++, and
197 Objective-C source code. In the past, it has been abused as a general
198 text processor. It will choke on input which does not obey C's lexical
199 rules. For example, apostrophes will be interpreted as the beginning of
200 character constants, and cause errors. Also, you cannot rely on it
201 preserving characteristics of the input which are not significant to
202 C-family languages. If a Makefile is preprocessed, all the hard tabs
203 will be removed, and the Makefile will not work.
205 Having said that, you can often get away with using cpp on things which
206 are not C@. Other Algol-ish programming languages are often safe
207 (Pascal, Ada, etc.) So is assembly, with caution. @option{-traditional-cpp}
208 mode preserves more white space, and is otherwise more permissive. Many
209 of the problems can be avoided by writing C or C++ style comments
210 instead of native language comments, and keeping macros simple.
212 Wherever possible, you should use a preprocessor geared to the language
213 you are writing in. Modern versions of the GNU assembler have macro
214 facilities. Most high level programming languages have their own
215 conditional compilation and inclusion mechanism. If all else fails,
216 try a true general text processor, such as GNU M4.
218 C preprocessors vary in some details. This manual discusses the GNU C
219 preprocessor, which provides a small superset of the features of ISO
220 Standard C@. In its default mode, the GNU C preprocessor does not do a
221 few things required by the standard. These are features which are
222 rarely, if ever, used, and may cause surprising changes to the meaning
223 of a program which does not expect them. To get strict ISO Standard C,
224 you should use the @option{-std=c89} or @option{-std=c99} options, depending
225 on which version of the standard you want. To get all the mandatory
226 diagnostics, you must also use @option{-pedantic}. @xref{Invocation}.
230 * Initial processing::
232 * The preprocessing language::
235 @node Initial processing
236 @section Initial processing
238 The preprocessor performs a series of textual transformations on its
239 input. These happen before all other processing. Conceptually, they
240 happen in a rigid order, and the entire file is run through each
241 transformation before the next one begins. GNU CPP actually does them
242 all at once, for performance reasons. These transformations correspond
243 roughly to the first three ``phases of translation'' described in the C
248 @cindex character sets
250 The input file is read into memory and broken into lines.
252 GNU CPP expects its input to be a text file, that is, an unstructured
253 stream of ASCII characters, with some characters indicating the end of a
254 line of text. Extended ASCII character sets, such as ISO Latin-1 or
255 Unicode encoded in UTF-8, are also acceptable. Character sets that are
256 not strict supersets of seven-bit ASCII will not work. We plan to add
257 complete support for international character sets in a future release.
259 Different systems use different conventions to indicate the end of a
260 line. GCC accepts the ASCII control sequences @kbd{LF}, @kbd{@w{CR
261 LF}}, @kbd{CR}, and @kbd{@w{LF CR}} as end-of-line markers. The first
262 three are the canonical sequences used by Unix, DOS and VMS, and the
263 classic Mac OS (before OSX) respectively. You may therefore safely copy
264 source code written on any of those systems to a different one and use
265 it without conversion. (GCC may lose track of the current line number
266 if a file doesn't consistently use one convention, as sometimes happens
267 when it is edited on computers with different conventions that share a
268 network file system.) @kbd{@w{LF CR}} is included because it has been
269 reported as an end-of-line marker under exotic conditions.
271 If the last line of any input file lacks an end-of-line marker, the end
272 of the file is considered to implicitly supply one. The C standard says
273 that this condition provokes undefined behavior, so GCC will emit a
278 @anchor{trigraphs}If trigraphs are enabled, they are replaced by their
279 corresponding single characters.
281 These are nine three-character sequences, all starting with @samp{??},
282 that are defined by ISO C to stand for single characters. They permit
283 obsolete systems that lack some of C's punctuation to use C@. For
284 example, @samp{??/} stands for @samp{\}, so @t{'??/n'} is a character
285 constant for a newline. By default, GCC ignores trigraphs, but if you
286 request a strictly conforming mode with the @option{-std} option, then
289 Trigraphs are not popular and many compilers implement them incorrectly.
290 Portable code should not rely on trigraphs being either converted or
291 ignored. If you use the @option{-Wall} or @option{-Wtrigraphs} options,
292 GCC will warn you when a trigraph would change the meaning of your
293 program if it were converted.
295 In a string constant, you can prevent a sequence of question marks from
296 being confused with a trigraph by inserting a backslash between the
297 question marks. @t{"(??\?)"} is the string @samp{(???)}, not
298 @samp{(?]}. Traditional C compilers do not recognize this idiom.
300 The nine trigraphs and their replacements are
303 Trigraph: ??( ??) ??< ??> ??= ??/ ??' ??! ??-
304 Replacement: [ ] @{ @} # \ ^ | ~
308 @cindex continued lines
309 @cindex backslash-newline
310 Continued lines are merged into one long line.
312 A continued line is a line which ends with a backslash, @samp{\}. The
313 backslash is removed and the following line is joined with the current
314 one. No space is inserted, so you may split a line anywhere, even in
315 the middle of a word. (It is generally more readable to split lines
316 only at white space.)
318 The trailing backslash on a continued line is commonly referred to as a
319 @dfn{backslash-newline}.
321 If there is white space between a backslash and the end of a line, that
322 is still a continued line. However, as this is usually the result of an
323 editing mistake, and many compilers will not accept it as a continued
324 line, GCC will warn you about it.
328 @cindex line comments
329 @cindex block comments
330 All comments are replaced with single spaces.
332 There are two kinds of comments. @dfn{Block comments} begin with
333 @samp{/*} and continue until the next @samp{*/}. Block comments do not
337 /* @r{this is} /* @r{one comment} */ @r{text outside comment}
340 @dfn{Line comments} begin with @samp{//} and continue to the end of the
341 current line. Line comments do not nest either, but it does not matter,
342 because they would end in the same place anyway.
345 // @r{this is} // @r{one comment}
346 @r{text outside comment}
350 It is safe to put line comments inside block comments, or vice versa.
355 // @r{contains line comment}
357 */ @r{outside comment}
359 // @r{line comment} /* @r{contains block comment} */
363 But beware of commenting out one end of a block comment with a line
368 // @r{l.c.} /* @r{block comment begins}
369 @r{oops! this isn't a comment anymore} */
373 Comments are not recognized within string literals. @t{@w{"/* blah
374 */"}} is the string constant @samp{@w{/* blah */}}, not an empty string.
376 Line comments are not in the 1989 edition of the C standard, but they
377 are recognized by GCC as an extension. In C++ and in the 1999 edition
378 of the C standard, they are an official part of the language.
380 Since these transformations happen before all other processing, you can
381 split a line mechanically with backslash-newline anywhere. You can
382 comment out the end of a line. You can continue a line comment onto the
383 next line with backslash-newline. You can even split @samp{/*},
384 @samp{*/}, and @samp{//} onto multiple lines with backslash-newline.
400 is equivalent to @code{@w{#define FOO 1020}}. All these tricks are
401 extremely confusing and should not be used in code intended to be
404 There is no way to prevent a backslash at the end of a line from being
405 interpreted as a backslash-newline. This cannot affect any correct
409 @section Tokenization
412 @cindex preprocessing tokens
413 After the textual transformations are finished, the input file is
414 converted into a sequence of @dfn{preprocessing tokens}. These mostly
415 correspond to the syntactic tokens used by the C compiler, but there are
416 a few differences. White space separates tokens; it is not itself a
417 token of any kind. Tokens do not have to be separated by white space,
418 but it is often necessary to avoid ambiguities.
420 When faced with a sequence of characters that has more than one possible
421 tokenization, the preprocessor is greedy. It always makes each token,
422 starting from the left, as big as possible before moving on to the next
423 token. For instance, @code{a+++++b} is interpreted as
424 @code{@w{a ++ ++ + b}}, not as @code{@w{a ++ + ++ b}}, even though the
425 latter tokenization could be part of a valid C program and the former
428 Once the input file is broken into tokens, the token boundaries never
429 change, except when the @samp{##} preprocessing operator is used to paste
430 tokens together. @xref{Concatenation}. For example,
442 The compiler does not re-tokenize the preprocessor's output. Each
443 preprocessing token becomes one compiler token.
446 Preprocessing tokens fall into five broad classes: identifiers,
447 preprocessing numbers, string literals, punctuators, and other. An
448 @dfn{identifier} is the same as an identifier in C: any sequence of
449 letters, digits, or underscores, which begins with a letter or
450 underscore. Keywords of C have no significance to the preprocessor;
451 they are ordinary identifiers. You can define a macro whose name is a
452 keyword, for instance. The only identifier which can be considered a
453 preprocessing keyword is @code{defined}. @xref{Defined}.
455 This is mostly true of other languages which use the C preprocessor.
456 However, a few of the keywords of C++ are significant even in the
457 preprocessor. @xref{C++ Named Operators}.
459 In the 1999 C standard, identifiers may contain letters which are not
460 part of the ``basic source character set,'' at the implementation's
461 discretion (such as accented Latin letters, Greek letters, or Chinese
462 ideograms). This may be done with an extended character set, or the
463 @samp{\u} and @samp{\U} escape sequences. GCC does not presently
464 implement either feature in the preprocessor or the compiler.
466 As an extension, GCC treats @samp{$} as a letter. This is for
467 compatibility with some systems, such as VMS, where @samp{$} is commonly
468 used in system-defined function and object names. @samp{$} is not a
469 letter in strictly conforming mode, or if you specify the @option{-$}
470 option. @xref{Invocation}.
473 @cindex preprocessing numbers
474 A @dfn{preprocessing number} has a rather bizarre definition. The
475 category includes all the normal integer and floating point constants
476 one expects of C, but also a number of other things one might not
477 initially recognize as a number. Formally, preprocessing numbers begin
478 with an optional period, a required decimal digit, and then continue
479 with any sequence of letters, digits, underscores, periods, and
480 exponents. Exponents are the two-character sequences @samp{e+},
481 @samp{e-}, @samp{E+}, @samp{E-}, @samp{p+}, @samp{p-}, @samp{P+}, and
482 @samp{P-}. (The exponents that begin with @samp{p} or @samp{P} are new
483 to C99. They are used for hexadecimal floating-point constants.)
485 The purpose of this unusual definition is to isolate the preprocessor
486 from the full complexity of numeric constants. It does not have to
487 distinguish between lexically valid and invalid floating-point numbers,
488 which is complicated. The definition also permits you to split an
489 identifier at any position and get exactly two tokens, which can then be
490 pasted back together with the @samp{##} operator.
492 It's possible for preprocessing numbers to cause programs to be
493 misinterpreted. For example, @code{0xE+12} is a preprocessing number
494 which does not translate to any valid numeric constant, therefore a
495 syntax error. It does not mean @code{@w{0xE + 12}}, which is what you
498 @cindex string literals
499 @cindex string constants
500 @cindex character constants
501 @cindex header file names
502 @c the @: prevents makeinfo from turning '' into ".
503 @dfn{String literals} are string constants, character constants, and
504 header file names (the argument of @samp{#include}).@footnote{The C
505 standard uses the term @dfn{string literal} to refer only to what we are
506 calling @dfn{string constants}.} String constants and character
507 constants are straightforward: @t{"@dots{}"} or @t{'@dots{}'}. In
508 either case embedded quotes should be escaped with a backslash:
509 @t{'\'@:'} is the character constant for @samp{'}. There is no limit on
510 the length of a character constant, but the value of a character
511 constant that contains more than one character is
512 implementation-defined. @xref{Implementation Details}.
514 Header file names either look like string constants, @t{"@dots{}"}, or are
515 written with angle brackets instead, @t{<@dots{}>}. In either case,
516 backslash is an ordinary character. There is no way to escape the
517 closing quote or angle bracket. The preprocessor looks for the header
518 file in different places depending on which form you use. @xref{Include
521 No string literal may extend past the end of a line. Older versions
522 of GCC accepted multi-line string constants. You may use continued
523 lines instead, or string constant concatenation. @xref{Differences
524 from previous versions}.
528 @cindex alternative tokens
529 @dfn{Punctuators} are all the usual bits of punctuation which are
530 meaningful to C and C++. All but three of the punctuation characters in
531 ASCII are C punctuators. The exceptions are @samp{@@}, @samp{$}, and
532 @samp{`}. In addition, all the two- and three-character operators are
533 punctuators. There are also six @dfn{digraphs}, which the C++ standard
534 calls @dfn{alternative tokens}, which are merely alternate ways to spell
535 other punctuators. This is a second attempt to work around missing
536 punctuation in obsolete systems. It has no negative side effects,
537 unlike trigraphs, but does not cover as much ground. The digraphs and
538 their corresponding normal punctuators are:
541 Digraph: <% %> <: :> %: %:%:
542 Punctuator: @{ @} [ ] # ##
546 Any other single character is considered ``other.'' It is passed on to
547 the preprocessor's output unmolested. The C compiler will almost
548 certainly reject source code containing ``other'' tokens. In ASCII, the
549 only other characters are @samp{@@}, @samp{$}, @samp{`}, and control
550 characters other than NUL (all bits zero). (Note that @samp{$} is
551 normally considered a letter.) All characters with the high bit set
552 (numeric range 0x7F--0xFF) are also ``other'' in the present
553 implementation. This will change when proper support for international
554 character sets is added to GCC@.
556 NUL is a special case because of the high probability that its
557 appearance is accidental, and because it may be invisible to the user
558 (many terminals do not display NUL at all). Within comments, NULs are
559 silently ignored, just as any other character would be. In running
560 text, NUL is considered white space. For example, these two directives
561 have the same meaning.
569 (where @samp{^@@} is ASCII NUL)@. Within string or character constants,
570 NULs are preserved. In the latter two cases the preprocessor emits a
573 @node The preprocessing language
574 @section The preprocessing language
576 @cindex preprocessing directives
577 @cindex directive line
578 @cindex directive name
580 After tokenization, the stream of tokens may simply be passed straight
581 to the compiler's parser. However, if it contains any operations in the
582 @dfn{preprocessing language}, it will be transformed first. This stage
583 corresponds roughly to the standard's ``translation phase 4'' and is
584 what most people think of as the preprocessor's job.
586 The preprocessing language consists of @dfn{directives} to be executed
587 and @dfn{macros} to be expanded. Its primary capabilities are:
591 Inclusion of header files. These are files of declarations that can be
592 substituted into your program.
595 Macro expansion. You can define @dfn{macros}, which are abbreviations
596 for arbitrary fragments of C code. The preprocessor will replace the
597 macros with their definitions throughout the program. Some macros are
598 automatically defined for you.
601 Conditional compilation. You can include or exclude parts of the
602 program according to various conditions.
605 Line control. If you use a program to combine or rearrange source files
606 into an intermediate file which is then compiled, you can use line
607 control to inform the compiler where each source line originally came
611 Diagnostics. You can detect problems at compile time and issue errors
615 There are a few more, less useful, features.
617 Except for expansion of predefined macros, all these operations are
618 triggered with @dfn{preprocessing directives}. Preprocessing directives
619 are lines in your program that start with @samp{#}. Whitespace is
620 allowed before and after the @samp{#}. The @samp{#} is followed by an
621 identifier, the @dfn{directive name}. It specifies the operation to
622 perform. Directives are commonly referred to as @samp{#@var{name}}
623 where @var{name} is the directive name. For example, @samp{#define} is
624 the directive that defines a macro.
626 The @samp{#} which begins a directive cannot come from a macro
627 expansion. Also, the directive name is not macro expanded. Thus, if
628 @code{foo} is defined as a macro expanding to @code{define}, that does
629 not make @samp{#foo} a valid preprocessing directive.
631 The set of valid directive names is fixed. Programs cannot define new
632 preprocessing directives.
634 Some directives require arguments; these make up the rest of the
635 directive line and must be separated from the directive name by
636 whitespace. For example, @samp{#define} must be followed by a macro
637 name and the intended expansion of the macro.
639 A preprocessing directive cannot cover more than one line. The line
640 may, however, be continued with backslash-newline, or by a block comment
641 which extends past the end of the line. In either case, when the
642 directive is processed, the continuations have already been merged with
643 the first line to make one long line.
646 @chapter Header Files
649 A header file is a file containing C declarations and macro definitions
650 (@pxref{Macros}) to be shared between several source files. You request
651 the use of a header file in your program by @dfn{including} it, with the
652 C preprocessing directive @samp{#include}.
654 Header files serve two purposes.
658 @cindex system header files
659 System header files declare the interfaces to parts of the operating
660 system. You include them in your program to supply the definitions and
661 declarations you need to invoke system calls and libraries.
664 Your own header files contain declarations for interfaces between the
665 source files of your program. Each time you have a group of related
666 declarations and macro definitions all or most of which are needed in
667 several different source files, it is a good idea to create a header
671 Including a header file produces the same results as copying the header
672 file into each source file that needs it. Such copying would be
673 time-consuming and error-prone. With a header file, the related
674 declarations appear in only one place. If they need to be changed, they
675 can be changed in one place, and programs that include the header file
676 will automatically use the new version when next recompiled. The header
677 file eliminates the labor of finding and changing all the copies as well
678 as the risk that a failure to find one copy will result in
679 inconsistencies within a program.
681 In C, the usual convention is to give header files names that end with
682 @file{.h}. It is most portable to use only letters, digits, dashes, and
683 underscores in header file names, and at most one dot.
687 * Include Operation::
689 * Once-Only Headers::
690 * Computed Includes::
696 @section Include Syntax
699 Both user and system header files are included using the preprocessing
700 directive @samp{#include}. It has two variants:
703 @item #include <@var{file}>
704 This variant is used for system header files. It searches for a file
705 named @var{file} in a standard list of system directories. You can prepend
706 directories to this list with the @option{-I} option (@pxref{Invocation}).
708 @item #include "@var{file}"
709 This variant is used for header files of your own program. It searches
710 for a file named @var{file} first in the directory containing the
711 current file, then in the same directories used for @code{<@var{file}>}.
714 The argument of @samp{#include}, whether delimited with quote marks or
715 angle brackets, behaves like a string constant in that comments are not
716 recognized, and macro names are not expanded. Thus, @code{@w{#include
717 <x/*y>}} specifies inclusion of a system header file named @file{x/*y}.
719 However, if backslashes occur within @var{file}, they are considered
720 ordinary text characters, not escape characters. None of the character
721 escape sequences appropriate to string constants in C are processed.
722 Thus, @code{@w{#include "x\n\\y"}} specifies a filename containing three
723 backslashes. (Some systems interpret @samp{\} as a pathname separator.
724 All of these also interpret @samp{/} the same way. It is most portable
725 to use only @samp{/}.)
727 It is an error if there is anything (other than comments) on the line
730 @node Include Operation
731 @section Include Operation
733 The @samp{#include} directive works by directing the C preprocessor to
734 scan the specified file as input before continuing with the rest of the
735 current file. The output from the preprocessor contains the output
736 already generated, followed by the output resulting from the included
737 file, followed by the output that comes from the text after the
738 @samp{#include} directive. For example, if you have a header file
739 @file{header.h} as follows,
746 and a main program called @file{program.c} that uses the header file,
761 the compiler will see the same token stream as it would if
762 @file{program.c} read
775 Included files are not limited to declarations and macro definitions;
776 those are merely the typical uses. Any fragment of a C program can be
777 included from another file. The include file could even contain the
778 beginning of a statement that is concluded in the containing file, or
779 the end of a statement that was started in the including file. However,
780 an included file must consist of complete tokens. Comments and string
781 literals which have not been closed by the end of an included file are
782 invalid. For error recovery, they are considered to end at the end of
785 To avoid confusion, it is best if header files contain only complete
786 syntactic units---function declarations or definitions, type
789 The line following the @samp{#include} directive is always treated as a
790 separate line by the C preprocessor, even if the included file lacks a
796 GCC looks in several different places for headers. On a normal Unix
797 system, if you do not instruct it otherwise, it will look for headers
798 requested with @code{@w{#include <@var{file}>}} in:
802 /usr/lib/gcc-lib/@var{target}/@var{version}/include
803 /usr/@var{target}/include
807 For C++ programs, it will also look in @file{/usr/include/g++-v3},
808 first. In the above, @var{target} is the canonical name of the system
809 GCC was configured to compile code for; often but not always the same as
810 the canonical name of the system it runs on. @var{version} is the
811 version of GCC in use.
813 You can add to this list with the @option{-I@var{dir}} command line
814 option. All the directories named by @option{-I} are searched, in
815 left-to-right order, @emph{before} the default directories. You can
816 also prevent GCC from searching any of the default directories with the
817 @option{-nostdinc} option. This is useful when you are compiling an
818 operating system kernel or some other program that does not use the
819 standard C library facilities, or the standard C library itself.
821 GCC looks for headers requested with @code{@w{#include "@var{file}"}}
822 first in the directory containing the current file, then in the same
823 places it would have looked for a header requested with angle brackets.
824 For example, if @file{/usr/include/sys/stat.h} contains
825 @code{@w{#include "types.h"}}, GCC looks for @file{types.h} first in
826 @file{/usr/include/sys}, then in its usual search path.
828 If you name a search directory with @option{-I@var{dir}} that is also a
829 system include directory, the @option{-I} wins; the directory will be
830 searched according to the @option{-I} ordering, and it will not be
831 treated as a system include directory. GCC will warn you when a system
832 include directory is hidden in this way.
834 @samp{#line} (@pxref{Line Control}) does not change GCC's idea of the
835 directory containing the current file.
837 You may put @option{-I-} at any point in your list of @option{-I} options.
838 This has two effects. First, directories appearing before the
839 @option{-I-} in the list are searched only for headers requested with
840 quote marks. Directories after @option{-I-} are searched for all
841 headers. Second, the directory containing the current file is not
842 searched for anything, unless it happens to be one of the directories
843 named by an @option{-I} switch.
845 @option{-I. -I-} is not the same as no @option{-I} options at all, and does
846 not cause the same behavior for @samp{<>} includes that @samp{""}
847 includes get with no special options. @option{-I.} searches the
848 compiler's current working directory for header files. That may or may
849 not be the same as the directory containing the current file.
851 If you need to look for headers in a directory named @file{-}, write
854 There are several more ways to adjust the header search path. They are
855 generally less useful. @xref{Invocation}.
857 @node Once-Only Headers
858 @section Once-Only Headers
859 @cindex repeated inclusion
860 @cindex including just once
861 @cindex wrapper @code{#ifndef}
863 If a header file happens to be included twice, the compiler will process
864 its contents twice. This is very likely to cause an error, e.g.@: when the
865 compiler sees the same structure definition twice. Even if it does not,
866 it will certainly waste time.
868 The standard way to prevent this is to enclose the entire real contents
869 of the file in a conditional, like this:
874 #ifndef FILE_FOO_SEEN
875 #define FILE_FOO_SEEN
877 @var{the entire file}
879 #endif /* !FILE_FOO_SEEN */
883 This construct is commonly known as a @dfn{wrapper #ifndef}.
884 When the header is included again, the conditional will be false,
885 because @code{FILE_FOO_SEEN} is defined. The preprocessor will skip
886 over the entire contents of the file, and the compiler will not see it
889 GNU CPP optimizes even further. It remembers when a header file has a
890 wrapper @samp{#ifndef}. If a subsequent @samp{#include} specifies that
891 header, and the macro in the @samp{#ifndef} is still defined, it does
892 not bother to rescan the file at all.
894 You can put comments outside the wrapper. They will not interfere with
897 @cindex controlling macro
899 The macro @code{FILE_FOO_SEEN} is called the @dfn{controlling macro} or
900 @dfn{guard macro}. In a user header file, the macro name should not
901 begin with @samp{_}. In a system header file, it should begin with
902 @samp{__} to avoid conflicts with user programs. In any kind of header
903 file, the macro name should contain the name of the file and some
904 additional text, to avoid conflicts with other header files.
906 @node Computed Includes
907 @section Computed Includes
908 @cindex computed includes
909 @cindex macros in include
911 Sometimes it is necessary to select one of several different header
912 files to be included into your program. They might specify
913 configuration parameters to be used on different sorts of operating
914 systems, for instance. You could do this with a series of conditionals,
918 # include "system_1.h"
920 # include "system_2.h"
926 That rapidly becomes tedious. Instead, the preprocessor offers the
927 ability to use a macro for the header name. This is called a
928 @dfn{computed include}. Instead of writing a header name as the direct
929 argument of @samp{#include}, you simply put a macro name there instead:
932 #define SYSTEM_H "system_1.h"
938 @code{SYSTEM_H} will be expanded, and the preprocessor will look for
939 @file{system_1.h} as if the @samp{#include} had been written that way
940 originally. @code{SYSTEM_H} could be defined by your Makefile with a
943 You must be careful when you define the macro. @samp{#define} saves
944 tokens, not text. The preprocessor has no way of knowing that the macro
945 will be used as the argument of @samp{#include}, so it generates
946 ordinary tokens, not a header name. This is unlikely to cause problems
947 if you use double-quote includes, which are close enough to string
948 constants. If you use angle brackets, however, you may have trouble.
950 The syntax of a computed include is actually a bit more general than the
951 above. If the first non-whitespace character after @samp{#include} is
952 not @samp{"} or @samp{<}, then the entire line is macro-expanded
953 like running text would be.
955 If the line expands to a single string constant, the contents of that
956 string constant are the file to be included. CPP does not re-examine the
957 string for embedded quotes, but neither does it process backslash
958 escapes in the string. Therefore
961 #define HEADER "a\"b"
966 looks for a file named @file{a\"b}. CPP searches for the file according
967 to the rules for double-quoted includes.
969 If the line expands to a token stream beginning with a @samp{<} token
970 and including a @samp{>} token, then the tokens between the @samp{<} and
971 the first @samp{>} are combined to form the filename to be included.
972 Any whitespace between tokens is reduced to a single space; then any
973 space after the initial @samp{<} is retained, but a trailing space
974 before the closing @samp{>} is ignored. CPP searches for the file
975 according to the rules for angle-bracket includes.
977 In either case, if there are any tokens on the line after the file name,
978 an error occurs and the directive is not processed. It is also an error
979 if the result of expansion does not match either of the two expected
982 These rules are implementation-defined behavior according to the C
983 standard. To minimize the risk of different compilers interpreting your
984 computed includes differently, we recommend you use only a single
985 object-like macro which expands to a string constant. This will also
986 minimize confusion for people reading your program.
988 @node Wrapper Headers
989 @section Wrapper Headers
990 @cindex wrapper headers
991 @cindex overriding a header file
992 @findex #include_next
994 Sometimes it is necessary to adjust the contents of a system-provided
995 header file without editing it directly. GCC's @command{fixincludes}
996 operation does this, for example. One way to do that would be to create
997 a new header file with the same name and insert it in the search path
998 before the original header. That works fine as long as you're willing
999 to replace the old header entirely. But what if you want to refer to
1000 the old header from the new one?
1002 You cannot simply include the old header with @samp{#include}. That
1003 will start from the beginning, and find your new header again. If your
1004 header is not protected from multiple inclusion (@pxref{Once-Only
1005 Headers}), it will recurse infinitely and cause a fatal error.
1007 You could include the old header with an absolute pathname:
1009 #include "/usr/include/old-header.h"
1012 This works, but is not clean; should the system headers ever move, you
1013 would have to edit the new headers to match.
1015 There is no way to solve this problem within the C standard, but you can
1016 use the GNU extension @samp{#include_next}. It means, ``Include the
1017 @emph{next} file with this name.'' This directive works like
1018 @samp{#include} except in searching for the specified file: it starts
1019 searching the list of header file directories @emph{after} the directory
1020 in which the current file was found.
1022 Suppose you specify @option{-I /usr/local/include}, and the list of
1023 directories to search also includes @file{/usr/include}; and suppose
1024 both directories contain @file{signal.h}. Ordinary @code{@w{#include
1025 <signal.h>}} finds the file under @file{/usr/local/include}. If that
1026 file contains @code{@w{#include_next <signal.h>}}, it starts searching
1027 after that directory, and finds the file in @file{/usr/include}.
1029 @samp{#include_next} does not distinguish between @code{<@var{file}>}
1030 and @code{"@var{file}"} inclusion, nor does it check that the file you
1031 specify has the same name as the current file. It simply looks for the
1032 file named, starting with the directory in the search path after the one
1033 where the current file was found.
1035 The use of @samp{#include_next} can lead to great confusion. We
1036 recommend it be used only when there is no other alternative. In
1037 particular, it should not be used in the headers belonging to a specific
1038 program; it should be used only to make global corrections along the
1039 lines of @command{fixincludes}.
1041 @node System Headers
1042 @section System Headers
1043 @cindex system header files
1045 The header files declaring interfaces to the operating system and
1046 runtime libraries often cannot be written in strictly conforming C@.
1047 Therefore, GCC gives code found in @dfn{system headers} special
1048 treatment. All warnings, other than those generated by @samp{#warning}
1049 (@pxref{Diagnostics}), are suppressed while GCC is processing a system
1050 header. Macros defined in a system header are immune to a few warnings
1051 wherever they are expanded. This immunity is granted on an ad-hoc
1052 basis, when we find that a warning generates lots of false positives
1053 because of code in macros defined in system headers.
1055 Normally, only the headers found in specific directories are considered
1056 system headers. These directories are determined when GCC is compiled.
1057 There are, however, two ways to make normal headers into system headers.
1059 The @option{-isystem} command line option adds its argument to the list of
1060 directories to search for headers, just like @option{-I}. Any headers
1061 found in that directory will be considered system headers.
1063 All directories named by @option{-isystem} are searched @emph{after} all
1064 directories named by @option{-I}, no matter what their order was on the
1065 command line. If the same directory is named by both @option{-I} and
1066 @option{-isystem}, @option{-I} wins; it is as if the @option{-isystem} option
1067 had never been specified at all. GCC warns you when this happens.
1069 @findex #pragma GCC system_header
1070 There is also a directive, @code{@w{#pragma GCC system_header}}, which
1071 tells GCC to consider the rest of the current include file a system
1072 header, no matter where it was found. Code that comes before the
1073 @samp{#pragma} in the file will not be affected. @code{@w{#pragma GCC
1074 system_header}} has no effect in the primary source file.
1076 On very old systems, some of the pre-defined system header directories
1077 get even more special treatment. GNU C++ considers code in headers
1078 found in those directories to be surrounded by an @code{@w{extern "C"}}
1079 block. There is no way to request this behavior with a @samp{#pragma},
1080 or from the command line.
1085 A @dfn{macro} is a fragment of code which has been given a name.
1086 Whenever the name is used, it is replaced by the contents of the macro.
1087 There are two kinds of macros. They differ mostly in what they look
1088 like when they are used. @dfn{Object-like} macros resemble data objects
1089 when used, @dfn{function-like} macros resemble function calls.
1091 You may define any valid identifier as a macro, even if it is a C
1092 keyword. The preprocessor does not know anything about keywords. This
1093 can be useful if you wish to hide a keyword such as @code{const} from an
1094 older compiler that does not understand it. However, the preprocessor
1095 operator @code{defined} (@pxref{Defined}) can never be defined as a
1096 macro, and C++'s named operators (@pxref{C++ Named Operators}) cannot be
1097 macros when you are compiling C++.
1100 * Object-like Macros::
1101 * Function-like Macros::
1106 * Predefined Macros::
1107 * Undefining and Redefining Macros::
1108 * Directives Within Macro Arguments::
1112 @node Object-like Macros
1113 @section Object-like Macros
1114 @cindex object-like macro
1115 @cindex symbolic constants
1116 @cindex manifest constants
1118 An @dfn{object-like macro} is a simple identifier which will be replaced
1119 by a code fragment. It is called object-like because it looks like a
1120 data object in code that uses it. They are most commonly used to give
1121 symbolic names to numeric constants.
1124 You create macros with the @samp{#define} directive. @samp{#define} is
1125 followed by the name of the macro and then the token sequence it should
1126 be an abbreviation for, which is variously referred to as the macro's
1127 @dfn{body}, @dfn{expansion} or @dfn{replacement list}. For example,
1130 #define BUFFER_SIZE 1024
1134 defines a macro named @code{BUFFER_SIZE} as an abbreviation for the
1135 token @code{1024}. If somewhere after this @samp{#define} directive
1136 there comes a C statement of the form
1139 foo = (char *) malloc (BUFFER_SIZE);
1143 then the C preprocessor will recognize and @dfn{expand} the macro
1144 @code{BUFFER_SIZE}. The C compiler will see the same tokens as it would
1148 foo = (char *) malloc (1024);
1151 By convention, macro names are written in upper case. Programs are
1152 easier to read when it is possible to tell at a glance which names are
1155 The macro's body ends at the end of the @samp{#define} line. You may
1156 continue the definition onto multiple lines, if necessary, using
1157 backslash-newline. When the macro is expanded, however, it will all
1158 come out on one line. For example,
1161 #define NUMBERS 1, \
1164 int x[] = @{ NUMBERS @};
1165 @expansion{} int x[] = @{ 1, 2, 3 @};
1169 The most common visible consequence of this is surprising line numbers
1172 There is no restriction on what can go in a macro body provided it
1173 decomposes into valid preprocessing tokens. Parentheses need not
1174 balance, and the body need not resemble valid C code. (If it does not,
1175 you may get error messages from the C compiler when you use the macro.)
1177 The C preprocessor scans your program sequentially. Macro definitions
1178 take effect at the place you write them. Therefore, the following input
1179 to the C preprocessor
1195 When the preprocessor expands a macro name, the macro's expansion
1196 replaces the macro invocation, then the expansion is examined for more
1197 macros to expand. For example,
1201 #define TABLESIZE BUFSIZE
1202 #define BUFSIZE 1024
1204 @expansion{} BUFSIZE
1210 @code{TABLESIZE} is expanded first to produce @code{BUFSIZE}, then that
1211 macro is expanded to produce the final result, @code{1024}.
1213 Notice that @code{BUFSIZE} was not defined when @code{TABLESIZE} was
1214 defined. The @samp{#define} for @code{TABLESIZE} uses exactly the
1215 expansion you specify---in this case, @code{BUFSIZE}---and does not
1216 check to see whether it too contains macro names. Only when you
1217 @emph{use} @code{TABLESIZE} is the result of its expansion scanned for
1220 This makes a difference if you change the definition of @code{BUFSIZE}
1221 at some point in the source file. @code{TABLESIZE}, defined as shown,
1222 will always expand using the definition of @code{BUFSIZE} that is
1223 currently in effect:
1226 #define BUFSIZE 1020
1227 #define TABLESIZE BUFSIZE
1233 Now @code{TABLESIZE} expands (in two stages) to @code{37}.
1235 If the expansion of a macro contains its own name, either directly or
1236 via intermediate macros, it is not expanded again when the expansion is
1237 examined for more macros. This prevents infinite recursion.
1238 @xref{Self-Referential Macros}, for the precise details.
1240 @node Function-like Macros
1241 @section Function-like Macros
1242 @cindex function-like macros
1244 You can also define macros whose use looks like a function call. These
1245 are called @dfn{function-like macros}. To define a function-like macro,
1246 you use the same @samp{#define} directive, but you put a pair of
1247 parentheses immediately after the macro name. For example,
1250 #define lang_init() c_init()
1252 @expansion{} c_init()
1255 A function-like macro is only expanded if its name appears with a pair
1256 of parentheses after it. If you write just the name, it is left alone.
1257 This can be useful when you have a function and a macro of the same
1258 name, and you wish to use the function sometimes.
1261 extern void foo(void);
1262 #define foo() /* optimized inline version */
1268 Here the call to @code{foo()} will use the macro, but the function
1269 pointer will get the address of the real function. If the macro were to
1270 be expanded, it would cause a syntax error.
1272 If you put spaces between the macro name and the parentheses in the
1273 macro definition, that does not define a function-like macro, it defines
1274 an object-like macro whose expansion happens to begin with a pair of
1278 #define lang_init () c_init()
1280 @expansion{} () c_init()()
1283 The first two pairs of parentheses in this expansion come from the
1284 macro. The third is the pair that was originally after the macro
1285 invocation. Since @code{lang_init} is an object-like macro, it does not
1286 consume those parentheses.
1288 @node Macro Arguments
1289 @section Macro Arguments
1291 @cindex macros with arguments
1292 @cindex arguments in macro definitions
1294 Function-like macros can take @dfn{arguments}, just like true functions.
1295 To define a macro that uses arguments, you insert @dfn{parameters}
1296 between the pair of parentheses in the macro definition that make the
1297 macro function-like. The parameters must be valid C identifiers,
1298 separated by commas and optionally whitespace.
1300 To invoke a macro that takes arguments, you write the name of the macro
1301 followed by a list of @dfn{actual arguments} in parentheses, separated
1302 by commas. The invocation of the macro need not be restricted to a
1303 single logical line---it can cross as many lines in the source file as
1304 you wish. The number of arguments you give must match the number of
1305 parameters in the macro definition. When the macro is expanded, each
1306 use of a parameter in its body is replaced by the tokens of the
1307 corresponding argument. (You need not use all of the parameters in the
1310 As an example, here is a macro that computes the minimum of two numeric
1311 values, as it is defined in many C programs, and some uses.
1314 #define min(X, Y) ((X) < (Y) ? (X) : (Y))
1315 x = min(a, b); @expansion{} x = ((a) < (b) ? (a) : (b));
1316 y = min(1, 2); @expansion{} y = ((1) < (2) ? (1) : (2));
1317 z = min(a + 28, *p); @expansion{} z = ((a + 28) < (*p) ? (a + 28) : (*p));
1321 (In this small example you can already see several of the dangers of
1322 macro arguments. @xref{Macro Pitfalls}, for detailed explanations.)
1324 Leading and trailing whitespace in each argument is dropped, and all
1325 whitespace between the tokens of an argument is reduced to a single
1326 space. Parentheses within each argument must balance; a comma within
1327 such parentheses does not end the argument. However, there is no
1328 requirement for square brackets or braces to balance, and they do not
1329 prevent a comma from separating arguments. Thus,
1332 macro (array[x = y, x + 1])
1336 passes two arguments to @code{macro}: @code{array[x = y} and @code{x +
1337 1]}. If you want to supply @code{array[x = y, x + 1]} as an argument,
1338 you can write it as @code{array[(x = y, x + 1)]}, which is equivalent C
1341 All arguments to a macro are completely macro-expanded before they are
1342 substituted into the macro body. After substitution, the complete text
1343 is scanned again for macros to expand, including the arguments. This rule
1344 may seem strange, but it is carefully designed so you need not worry
1345 about whether any function call is actually a macro invocation. You can
1346 run into trouble if you try to be too clever, though. @xref{Argument
1347 Prescan}, for detailed discussion.
1349 For example, @code{min (min (a, b), c)} is first expanded to
1352 min (((a) < (b) ? (a) : (b)), (c))
1360 ((((a) < (b) ? (a) : (b))) < (c)
1361 ? (((a) < (b) ? (a) : (b)))
1367 (Line breaks shown here for clarity would not actually be generated.)
1369 @cindex empty macro arguments
1370 You can leave macro arguments empty; this is not an error to the
1371 preprocessor (but many macros will then expand to invalid code).
1372 You cannot leave out arguments entirely; if a macro takes two arguments,
1373 there must be exactly one comma at the top level of its argument list.
1374 Here are some silly examples using @code{min}:
1377 min(, b) @expansion{} (( ) < (b) ? ( ) : (b))
1378 min(a, ) @expansion{} ((a ) < ( ) ? (a ) : ( ))
1379 min(,) @expansion{} (( ) < ( ) ? ( ) : ( ))
1380 min((,),) @expansion{} (((,)) < ( ) ? ((,)) : ( ))
1382 min() @error{} macro "min" requires 2 arguments, but only 1 given
1383 min(,,) @error{} macro "min" passed 3 arguments, but takes just 2
1386 Whitespace is not a preprocessing token, so if a macro @code{foo} takes
1387 one argument, @code{@w{foo ()}} and @code{@w{foo ( )}} both supply it an
1388 empty argument. Previous GNU preprocessor implementations and
1389 documentation were incorrect on this point, insisting that a
1390 function-like macro that takes a single argument be passed a space if an
1391 empty argument was required.
1393 Macro parameters appearing inside string literals are not replaced by
1394 their corresponding actual arguments.
1397 #define foo(x) x, "x"
1398 foo(bar) @expansion{} bar, "x"
1401 @node Stringification
1402 @section Stringification
1403 @cindex stringification
1404 @cindex @samp{#} operator
1406 Sometimes you may want to convert a macro argument into a string
1407 constant. Parameters are not replaced inside string constants, but you
1408 can use the @samp{#} preprocessing operator instead. When a macro
1409 parameter is used with a leading @samp{#}, the preprocessor replaces it
1410 with the literal text of the actual argument, converted to a string
1411 constant. Unlike normal parameter replacement, the argument is not
1412 macro-expanded first. This is called @dfn{stringification}.
1414 There is no way to combine an argument with surrounding text and
1415 stringify it all together. Instead, you can write a series of adjacent
1416 string constants and stringified arguments. The preprocessor will
1417 replace the stringified arguments with string constants. The C
1418 compiler will then combine all the adjacent string constants into one
1421 Here is an example of a macro definition that uses stringification:
1425 #define WARN_IF(EXP) \
1427 fprintf (stderr, "Warning: " #EXP "\n"); @} \
1430 @expansion{} do @{ if (x == 0)
1431 fprintf (stderr, "Warning: " "x == 0" "\n"); @} while (0);
1436 The argument for @code{EXP} is substituted once, as-is, into the
1437 @code{if} statement, and once, stringified, into the argument to
1438 @code{fprintf}. If @code{x} were a macro, it would be expanded in the
1439 @code{if} statement, but not in the string.
1441 The @code{do} and @code{while (0)} are a kludge to make it possible to
1442 write @code{WARN_IF (@var{arg});}, which the resemblance of
1443 @code{WARN_IF} to a function would make C programmers want to do; see
1444 @ref{Swallowing the Semicolon}.
1446 Stringification in C involves more than putting double-quote characters
1447 around the fragment. The preprocessor backslash-escapes the quotes
1448 surrounding embedded string constants, and all backslashes within string and
1449 character constants, in order to get a valid C string constant with the
1450 proper contents. Thus, stringifying @code{@w{p = "foo\n";}} results in
1451 @t{@w{"p = \"foo\\n\";"}}. However, backslashes that are not inside string
1452 or character constants are not duplicated: @samp{\n} by itself
1453 stringifies to @t{"\n"}.
1455 All leading and trailing whitespace in text being stringified is
1456 ignored. Any sequence of whitespace in the middle of the text is
1457 converted to a single space in the stringified result. Comments are
1458 replaced by whitespace long before stringification happens, so they
1459 never appear in stringified text.
1461 There is no way to convert a macro argument into a character constant.
1463 If you want to stringify the result of expansion of a macro argument,
1464 you have to use two levels of macros.
1467 #define xstr(s) str(s)
1473 @expansion{} xstr (4)
1474 @expansion{} str (4)
1478 @code{s} is stringified when it is used in @code{str}, so it is not
1479 macro-expanded first. But @code{s} is an ordinary argument to
1480 @code{xstr}, so it is completely macro-expanded before @code{xstr}
1481 itself is expanded (@pxref{Argument Prescan}). Therefore, by the time
1482 @code{str} gets to its argument, it has already been macro-expanded.
1485 @section Concatenation
1486 @cindex concatenation
1487 @cindex token pasting
1488 @cindex token concatenation
1489 @cindex @samp{##} operator
1491 It is often useful to merge two tokens into one while expanding macros.
1492 This is called @dfn{token pasting} or @dfn{token concatenation}. The
1493 @samp{##} preprocessing operator performs token pasting. When a macro
1494 is expanded, the two tokens on either side of each @samp{##} operator
1495 are combined into a single token, which then replaces the @samp{##} and
1496 the two original tokens in the macro expansion. Usually both will be
1497 identifiers, or one will be an identifier and the other a preprocessing
1498 number. When pasted, they make a longer identifier. This isn't the
1499 only valid case. It is also possible to concatenate two numbers (or a
1500 number and a name, such as @code{1.5} and @code{e3}) into a number.
1501 Also, multi-character operators such as @code{+=} can be formed by
1504 However, two tokens that don't together form a valid token cannot be
1505 pasted together. For example, you cannot concatenate @code{x} with
1506 @code{+} in either order. If you try, the preprocessor issues a warning
1507 and emits the two tokens. Whether it puts white space between the
1508 tokens is undefined. It is common to find unnecessary uses of @samp{##}
1509 in complex macros. If you get this warning, it is likely that you can
1510 simply remove the @samp{##}.
1512 Both the tokens combined by @samp{##} could come from the macro body,
1513 but you could just as well write them as one token in the first place.
1514 Token pasting is most useful when one or both of the tokens comes from a
1515 macro argument. If either of the tokens next to an @samp{##} is a
1516 parameter name, it is replaced by its actual argument before @samp{##}
1517 executes. As with stringification, the actual argument is not
1518 macro-expanded first. If the argument is empty, that @samp{##} has no
1521 Keep in mind that the C preprocessor converts comments to whitespace
1522 before macros are even considered. Therefore, you cannot create a
1523 comment by concatenating @samp{/} and @samp{*}. You can put as much
1524 whitespace between @samp{##} and its operands as you like, including
1525 comments, and you can put comments in arguments that will be
1526 concatenated. However, it is an error if @samp{##} appears at either
1527 end of a macro body.
1529 Consider a C program that interprets named commands. There probably
1530 needs to be a table of commands, perhaps an array of structures declared
1538 void (*function) (void);
1543 struct command commands[] =
1545 @{ "quit", quit_command @},
1546 @{ "help", help_command @},
1552 It would be cleaner not to have to give each command name twice, once in
1553 the string constant and once in the function name. A macro which takes the
1554 name of a command as an argument can make this unnecessary. The string
1555 constant can be created with stringification, and the function name by
1556 concatenating the argument with @samp{_command}. Here is how it is done:
1559 #define COMMAND(NAME) @{ #NAME, NAME ## _command @}
1561 struct command commands[] =
1569 @node Variadic Macros
1570 @section Variadic Macros
1571 @cindex variable number of arguments
1572 @cindex macros with variable arguments
1573 @cindex variadic macros
1575 A macro can be declared to accept a variable number of arguments much as
1576 a function can. The syntax for defining the macro is similar to that of
1577 a function. Here is an example:
1580 #define eprintf(@dots{}) fprintf (stderr, __VA_ARGS__)
1583 This kind of macro is called @dfn{variadic}. When the macro is invoked,
1584 all the tokens in its argument list after the last named argument (this
1585 macro has none), including any commas, become the @dfn{variable
1586 argument}. This sequence of tokens replaces the identifier
1587 @code{@w{__VA_ARGS__}} in the macro body wherever it appears. Thus, we
1588 have this expansion:
1591 eprintf ("%s:%d: ", input_file, lineno)
1592 @expansion{} fprintf (stderr, "%s:%d: ", input_file, lineno)
1595 The variable argument is completely macro-expanded before it is inserted
1596 into the macro expansion, just like an ordinary argument. You may use
1597 the @samp{#} and @samp{##} operators to stringify the variable argument
1598 or to paste its leading or trailing token with another token. (But see
1599 below for an important special case for @samp{##}.)
1601 If your macro is complicated, you may want a more descriptive name for
1602 the variable argument than @code{@w{__VA_ARGS__}}. GNU CPP permits
1603 this, as an extension. You may write an argument name immediately
1604 before the @samp{@dots{}}; that name is used for the variable argument.
1605 The @code{eprintf} macro above could be written
1608 #define eprintf(args@dots{}) fprintf (stderr, args)
1612 using this extension. You cannot use @code{__VA_ARGS__} and this
1613 extension in the same macro.
1615 You can have named arguments as well as variable arguments in a variadic
1616 macro. We could define @code{eprintf} like this, instead:
1619 #define eprintf(format, @dots{}) fprintf (stderr, format, __VA_ARGS__)
1623 This formulation looks more descriptive, but unfortunately it is less
1624 flexible: you must now supply at least one argument after the format
1625 string. In standard C, you cannot omit the comma separating the named
1626 argument from the variable arguments. Furthermore, if you leave the
1627 variable argument empty, you will get a syntax error, because
1628 there will be an extra comma after the format string.
1631 eprintf("success!\n", );
1632 @expansion{} fprintf(stderr, "success!\n", );
1635 GNU CPP has a pair of extensions which deal with this problem. First,
1636 you are allowed to leave the variable argument out entirely:
1639 eprintf ("success!\n")
1640 @expansion{} fprintf(stderr, "success!\n", );
1644 Second, the @samp{##} token paste operator has a special meaning when
1645 placed between a comma and a variable argument. If you write
1648 #define eprintf(format, @dots{}) fprintf (stderr, format, ##__VA_ARGS__)
1652 and the variable argument is left out when the @code{eprintf} macro is
1653 used, then the comma before the @samp{##} will be deleted. This does
1654 @emph{not} happen if you pass an empty argument, nor does it happen if
1655 the token preceding @samp{##} is anything other than a comma.
1658 eprintf ("success!\n")
1659 @expansion{} fprintf(stderr, "success!\n");
1662 C99 mandates that the only place the identifier @code{@w{__VA_ARGS__}}
1663 can appear is in the replacement list of a variadic macro. It may not
1664 be used as a macro name, macro argument name, or within a different type
1665 of macro. It may also be forbidden in open text; the standard is
1666 ambiguous. We recommend you avoid using it except for its defined
1669 Variadic macros are a new feature in C99. GNU CPP has supported them
1670 for a long time, but only with a named variable argument
1671 (@samp{args@dots{}}, not @samp{@dots{}} and @code{@w{__VA_ARGS__}}). If you are
1672 concerned with portability to previous versions of GCC, you should use
1673 only named variable arguments. On the other hand, if you are concerned
1674 with portability to other conforming implementations of C99, you should
1675 use only @code{@w{__VA_ARGS__}}.
1677 Previous versions of GNU CPP implemented the comma-deletion extension
1678 much more generally. We have restricted it in this release to minimize
1679 the differences from C99. To get the same effect with both this and
1680 previous versions of GCC, the token preceding the special @samp{##} must
1681 be a comma, and there must be white space between that comma and
1682 whatever comes immediately before it:
1685 #define eprintf(format, args@dots{}) fprintf (stderr, format , ##args)
1689 @xref{Differences from previous versions}, for the gory details.
1691 @node Predefined Macros
1692 @section Predefined Macros
1694 @cindex predefined macros
1695 Several object-like macros are predefined; you use them without
1696 supplying their definitions. They fall into three classes: standard,
1697 common, and system-specific.
1699 In C++, there is a fourth category, the named operators. They act like
1700 predefined macros, but you cannot undefine them.
1703 * Standard Predefined Macros::
1704 * Common Predefined Macros::
1705 * System-specific Predefined Macros::
1706 * C++ Named Operators::
1709 @node Standard Predefined Macros
1710 @subsection Standard Predefined Macros
1711 @cindex standard predefined macros.
1713 The standard predefined macros are specified by the C and/or C++
1714 language standards, so they are available with all compilers that
1715 implement those standards. Older compilers may not provide all of
1716 them. Their names all start with double underscores.
1720 This macro expands to the name of the current input file, in the form of
1721 a C string constant. This is the path by which the preprocessor opened
1722 the file, not the short name specified in @samp{#include} or as the
1723 input file name argument. For example,
1724 @code{"/usr/local/include/myheader.h"} is a possible expansion of this
1728 This macro expands to the current input line number, in the form of a
1729 decimal integer constant. While we call it a predefined macro, it's
1730 a pretty strange macro, since its ``definition'' changes with each
1731 new line of source code.
1734 @code{__FILE__} and @code{__LINE__} are useful in generating an error
1735 message to report an inconsistency detected by the program; the message
1736 can state the source line at which the inconsistency was detected. For
1740 fprintf (stderr, "Internal error: "
1741 "negative string length "
1742 "%d at %s, line %d.",
1743 length, __FILE__, __LINE__);
1746 An @samp{#include} directive changes the expansions of @code{__FILE__}
1747 and @code{__LINE__} to correspond to the included file. At the end of
1748 that file, when processing resumes on the input file that contained
1749 the @samp{#include} directive, the expansions of @code{__FILE__} and
1750 @code{__LINE__} revert to the values they had before the
1751 @samp{#include} (but @code{__LINE__} is then incremented by one as
1752 processing moves to the line after the @samp{#include}).
1754 A @samp{#line} directive changes @code{__LINE__}, and may change
1755 @code{__FILE__} as well. @xref{Line Control}.
1757 C99 introduces @code{__func__}, and GCC has provided @code{__FUNCTION__}
1758 for a long time. Both of these are strings containing the name of the
1759 current function (there are slight semantic differences; see the GCC
1760 manual). Neither of them is a macro; the preprocessor does not know the
1761 name of the current function. They tend to be useful in conjunction
1762 with @code{__FILE__} and @code{__LINE__}, though.
1767 This macro expands to a string constant that describes the date on which
1768 the preprocessor is being run. The string constant contains eleven
1769 characters and looks like @code{@w{"Feb 12 1996"}}. If the day of the
1770 month is less than 10, it is padded with a space on the left.
1773 This macro expands to a string constant that describes the time at
1774 which the preprocessor is being run. The string constant contains
1775 eight characters and looks like @code{"23:59:01"}.
1778 In normal operation, this macro expands to the constant 1, to signify
1779 that this compiler conforms to ISO Standard C@. If GNU CPP is used with
1780 a compiler other than GCC, this is not necessarily true; however, the
1781 preprocessor always conforms to the standard, unless the
1782 @option{-traditional-cpp} option is used.
1784 This macro is not defined if the @option{-traditional-cpp} option is used.
1786 On some hosts, the system compiler uses a different convention, where
1787 @code{__STDC__} is normally 0, but is 1 if the user specifies strict
1788 conformance to the C Standard. GNU CPP follows the host convention when
1789 processing system header files, but when processing user files
1790 @code{__STDC__} is always 1. This has been reported to cause problems;
1791 for instance, some versions of Solaris provide X Windows headers that
1792 expect @code{__STDC__} to be either undefined or 1. You may be able to
1793 work around this sort of problem by using an @option{-I} option to
1794 cancel treatment of those headers as system headers. @xref{Invocation}.
1796 @item __STDC_VERSION__
1797 This macro expands to the C Standard's version number, a long integer
1798 constant of the form @code{@var{yyyy}@var{mm}L} where @var{yyyy} and
1799 @var{mm} are the year and month of the Standard version. This signifies
1800 which version of the C Standard the compiler conforms to. Like
1801 @code{__STDC__}, this is not necessarily accurate for the entire
1802 implementation, unless GNU CPP is being used with GCC@.
1804 The value @code{199409L} signifies the 1989 C standard as amended in
1805 1994, which is the current default; the value @code{199901L} signifies
1806 the 1999 revision of the C standard. Support for the 1999 revision is
1809 This macro is not defined if the @option{-traditional-cpp} option is
1810 used, nor when compiling C++ or Objective-C@.
1812 @item __STDC_HOSTED__
1813 This macro is defined, with value 1, if the compiler's target is a
1814 @dfn{hosted environment}. A hosted environment has the complete
1815 facilities of the standard C library available.
1818 This macro is defined when the C++ compiler is in use. You can use
1819 @code{__cplusplus} to test whether a header is compiled by a C compiler
1820 or a C++ compiler. This macro is similar to @code{__STDC_VERSION__}, in
1821 that it expands to a version number. A fully conforming implementation
1822 of the 1998 C++ standard will define this macro to @code{199711L}. The
1823 GNU C++ compiler is not yet fully conforming, so it uses @code{1}
1824 instead. We hope to complete our implementation in the near future.
1828 @node Common Predefined Macros
1829 @subsection Common Predefined Macros
1830 @cindex common predefined macros
1832 The common predefined macros are GNU C extensions. They are available
1833 with the same meanings regardless of the machine or operating system on
1834 which you are using GNU C@. Their names all start with double
1840 @itemx __GNUC_MINOR__
1841 @itemx __GNUC_PATCHLEVEL__
1842 These macros are defined by all GNU compilers that use the C
1843 preprocessor: C, C++, and Objective-C@. Their values are the major
1844 version, minor version, and patch level of the compiler, as integer
1845 constants. For example, GCC 3.2.1 will define @code{__GNUC__} to 3,
1846 @code{__GNUC_MINOR__} to 2, and @code{__GNUC_PATCHLEVEL__} to 1. They
1847 are defined only when the entire compiler is in use; if you invoke the
1848 preprocessor directly, they are not defined.
1850 @code{__GNUC_PATCHLEVEL__} is new to GCC 3.0; it is also present in the
1851 widely-used development snapshots leading up to 3.0 (which identify
1852 themselves as GCC 2.96 or 2.97, depending on which snapshot you have).
1854 If all you need to know is whether or not your program is being compiled
1855 by GCC, you can simply test @code{__GNUC__}. If you need to write code
1856 which depends on a specific version, you must be more careful. Each
1857 time the minor version is increased, the patch level is reset to zero;
1858 each time the major version is increased (which happens rarely), the
1859 minor version and patch level are reset. If you wish to use the
1860 predefined macros directly in the conditional, you will need to write it
1864 /* @r{Test for GCC > 3.2.0} */
1865 #if __GNUC__ > 3 || \
1866 (__GNUC__ == 3 && (__GNUC_MINOR__ > 2 || \
1867 (__GNUC_MINOR__ == 2 && \
1868 __GNUC_PATCHLEVEL__ > 0))
1872 Another approach is to use the predefined macros to
1873 calculate a single number, then compare that against a threshold:
1876 #define GCC_VERSION (__GNUC__ * 10000 \
1877 + __GNUC_MINOR__ * 100 \
1878 + __GNUC_PATCHLEVEL__)
1880 /* @r{Test for GCC > 3.2.0} */
1881 #if GCC_VERSION > 30200
1885 Many people find this form easier to understand.
1888 This macro is defined, with value 1, when the Objective-C compiler is in
1889 use. You can use @code{__OBJC__} to test whether a header is compiled
1890 by a C compiler or a Objective-C compiler.
1893 The GNU C++ compiler defines this. Testing it is equivalent to
1894 testing @code{@w{(__GNUC__ && __cplusplus)}}.
1896 @item __STRICT_ANSI__
1897 GCC defines this macro if and only if the @option{-ansi} switch, or a
1898 @option{-std} switch specifying strict conformance to some version of ISO C,
1899 was specified when GCC was invoked. It is defined to @samp{1}.
1900 This macro exists primarily to direct GNU libc's header files to
1901 restrict their definitions to the minimal set found in the 1989 C
1905 This macro expands to the name of the main input file, in the form
1906 of a C string constant. This is the source file that was specified
1907 on the command line of the preprocessor or C compiler.
1909 @item __INCLUDE_LEVEL__
1910 This macro expands to a decimal integer constant that represents the
1911 depth of nesting in include files. The value of this macro is
1912 incremented on every @samp{#include} directive and decremented at the
1913 end of every included file. It starts out at 0, it's value within the
1914 base file specified on the command line.
1917 This macro expands to a string constant which describes the version of
1918 the compiler in use. You should not rely on its contents having any
1919 particular form, but it can be counted on to contain at least the
1923 @itemx __OPTIMIZE_SIZE__
1924 @itemx __NO_INLINE__
1925 These macros describe the compilation mode. @code{__OPTIMIZE__} is
1926 defined in all optimizing compilations. @code{__OPTIMIZE_SIZE__} is
1927 defined if the compiler is optimizing for size, not speed.
1928 @code{__NO_INLINE__} is defined if no functions will be inlined into
1929 their callers (when not optimizing, or when inlining has been
1930 specifically disabled by @option{-fno-inline}).
1932 These macros cause certain GNU header files to provide optimized
1933 definitions, using macros or inline functions, of system library
1934 functions. You should not use these macros in any way unless you make
1935 sure that programs will execute with the same effect whether or not they
1936 are defined. If they are defined, their value is 1.
1938 @item __CHAR_UNSIGNED__
1939 GCC defines this macro if and only if the data type @code{char} is
1940 unsigned on the target machine. It exists to cause the standard header
1941 file @file{limits.h} to work correctly. You should not use this macro
1942 yourself; instead, refer to the standard macros defined in @file{limits.h}.
1944 @item __REGISTER_PREFIX__
1945 This macro expands to a single token (not a string constant) which is
1946 the prefix applied to CPU register names in assembly language for this
1947 target. You can use it to write assembly that is usable in multiple
1948 environments. For example, in the @code{m68k-aout} environment it
1949 expands to nothing, but in the @code{m68k-coff} environment it expands
1950 to a single @samp{%}.
1952 @item __USER_LABEL_PREFIX__
1953 This macro expands to a single token which is the prefix applied to
1954 user labels (symbols visible to C code) in assembly. For example, in
1955 the @code{m68k-aout} environment it expands to an @samp{_}, but in the
1956 @code{m68k-coff} environment it expands to nothing.
1958 This macro will have the correct definition even if
1959 @option{-f(no-)underscores} is in use, but it will not be correct if
1960 target-specific options that adjust this prefix are used (e.g.@: the
1961 OSF/rose @option{-mno-underscores} option).
1964 @itemx __PTRDIFF_TYPE__
1965 @itemx __WCHAR_TYPE__
1966 @itemx __WINT_TYPE__
1967 These macros are defined to the correct underlying types for the
1968 @code{size_t}, @code{ptrdiff_t}, @code{wchar_t}, and @code{wint_t}
1969 typedefs, respectively. They exist to make the standard header files
1970 @file{stddef.h} and @file{wchar.h} work correctly. You should not use
1971 these macros directly; instead, include the appropriate headers and use
1974 @item __USING_SJLJ_EXCEPTIONS__
1975 This macro is defined, with value 1, if the compiler uses the old
1976 mechanism based on @code{setjmp} and @code{longjmp} for exception
1980 @node System-specific Predefined Macros
1981 @subsection System-specific Predefined Macros
1983 @cindex system-specific predefined macros
1984 @cindex predefined macros, system-specific
1985 @cindex reserved namespace
1987 The C preprocessor normally predefines several macros that indicate what
1988 type of system and machine is in use. They are obviously different on
1989 each target supported by GCC@. This manual, being for all systems and
1990 machines, cannot tell you what their names are, but you can use
1991 @command{cpp -dM} to see them all. @xref{Invocation}. All system-specific
1992 predefined macros expand to the constant 1, so you can test them with
1993 either @samp{#ifdef} or @samp{#if}.
1995 The C standard requires that all system-specific macros be part of the
1996 @dfn{reserved namespace}. All names which begin with two underscores,
1997 or an underscore and a capital letter, are reserved for the compiler and
1998 library to use as they wish. However, historically system-specific
1999 macros have had names with no special prefix; for instance, it is common
2000 to find @code{unix} defined on Unix systems. For all such macros, GCC
2001 provides a parallel macro with two underscores added at the beginning
2002 and the end. If @code{unix} is defined, @code{__unix__} will be defined
2003 too. There will never be more than two underscores; the parallel of
2004 @code{_mips} is @code{__mips__}.
2006 When the @option{-ansi} option, or any @option{-std} option that
2007 requests strict conformance, is given to the compiler, all the
2008 system-specific predefined macros outside the reserved namespace are
2009 suppressed. The parallel macros, inside the reserved namespace, remain
2012 We are slowly phasing out all predefined macros which are outside the
2013 reserved namespace. You should never use them in new programs, and we
2014 encourage you to correct older code to use the parallel macros whenever
2015 you find it. We don't recommend you use the system-specific macros that
2016 are in the reserved namespace, either. It is better in the long run to
2017 check specifically for features you need, using a tool such as
2020 @node C++ Named Operators
2021 @subsection C++ Named Operators
2022 @cindex named operators
2023 @cindex C++ named operators
2026 In C++, there are eleven keywords which are simply alternate spellings
2027 of operators normally written with punctuation. These keywords are
2028 treated as such even in the preprocessor. They function as operators in
2029 @samp{#if}, and they cannot be defined as macros or poisoned. In C, you
2030 can request that those keywords take their C++ meaning by including
2031 @file{iso646.h}. That header defines each one as a normal object-like
2032 macro expanding to the appropriate punctuator.
2034 These are the named operators and their corresponding punctuators:
2036 @multitable {Named Operator} {Punctuator}
2037 @item Named Operator @tab Punctuator
2038 @item @code{and} @tab @code{&&}
2039 @item @code{and_eq} @tab @code{&=}
2040 @item @code{bitand} @tab @code{&}
2041 @item @code{bitor} @tab @code{|}
2042 @item @code{compl} @tab @code{~}
2043 @item @code{not} @tab @code{!}
2044 @item @code{not_eq} @tab @code{!=}
2045 @item @code{or} @tab @code{||}
2046 @item @code{or_eq} @tab @code{|=}
2047 @item @code{xor} @tab @code{^}
2048 @item @code{xor_eq} @tab @code{^=}
2051 @node Undefining and Redefining Macros
2052 @section Undefining and Redefining Macros
2053 @cindex undefining macros
2054 @cindex redefining macros
2057 If a macro ceases to be useful, it may be @dfn{undefined} with the
2058 @samp{#undef} directive. @samp{#undef} takes a single argument, the
2059 name of the macro to undefine. You use the bare macro name, even if the
2060 macro is function-like. It is an error if anything appears on the line
2061 after the macro name. @samp{#undef} has no effect if the name is not a
2066 x = FOO; @expansion{} x = 4;
2068 x = FOO; @expansion{} x = FOO;
2071 Once a macro has been undefined, that identifier may be @dfn{redefined}
2072 as a macro by a subsequent @samp{#define} directive. The new definition
2073 need not have any resemblance to the old definition.
2075 However, if an identifier which is currently a macro is redefined, then
2076 the new definition must be @dfn{effectively the same} as the old one.
2077 Two macro definitions are effectively the same if:
2079 @item Both are the same type of macro (object- or function-like).
2080 @item All the tokens of the replacement list are the same.
2081 @item If there are any parameters, they are the same.
2082 @item Whitespace appears in the same places in both. It need not be
2083 exactly the same amount of whitespace, though. Remember that comments
2084 count as whitespace.
2088 These definitions are effectively the same:
2090 #define FOUR (2 + 2)
2091 #define FOUR (2 + 2)
2092 #define FOUR (2 /* two */ + 2)
2097 #define FOUR (2 + 2)
2098 #define FOUR ( 2+2 )
2099 #define FOUR (2 * 2)
2100 #define FOUR(score,and,seven,years,ago) (2 + 2)
2103 If a macro is redefined with a definition that is not effectively the
2104 same as the old one, the preprocessor issues a warning and changes the
2105 macro to use the new definition. If the new definition is effectively
2106 the same, the redefinition is silently ignored. This allows, for
2107 instance, two different headers to define a common macro. The
2108 preprocessor will only complain if the definitions do not match.
2110 @node Directives Within Macro Arguments
2111 @section Directives Within Macro Arguments
2112 @cindex macro arguments and directives
2114 Occasionally it is convenient to use preprocessor directives within
2115 the arguments of a macro. The C and C++ standards declare that
2116 behavior in these cases is undefined.
2118 Versions of GNU CPP prior to 3.2 would reject such constructs with an
2119 error message. This was the only syntactic difference between normal
2120 functions and function-like macros, so it seemed attractive to remove
2121 this limitation, and people would often be surprised that they could
2122 not use macros in this way. Moreover, sometimes people would use
2123 conditional compilation in the argument list to a normal library
2124 function like @samp{printf}, only to find that after a library upgrade
2125 @samp{printf} had changed to be a function-like macro, and their code
2126 would no longer compile. So from version 3.2 we changed CPP to
2127 successfully process arbitrary directives within macro arguments in
2128 exactly the same way as it would have processed the directive were the
2129 function-like macro invocation not present.
2131 If, within a macro invocation, that macro is redefined, then the new
2132 definition takes effect in time for argument pre-expansion, but the
2133 original definition is still used for argument replacement. Here is a
2134 pathological example:
2152 with the semantics described above.
2154 @node Macro Pitfalls
2155 @section Macro Pitfalls
2156 @cindex problems with macros
2157 @cindex pitfalls of macros
2159 In this section we describe some special rules that apply to macros and
2160 macro expansion, and point out certain cases in which the rules have
2161 counter-intuitive consequences that you must watch out for.
2165 * Operator Precedence Problems::
2166 * Swallowing the Semicolon::
2167 * Duplication of Side Effects::
2168 * Self-Referential Macros::
2169 * Argument Prescan::
2170 * Newlines in Arguments::
2174 @subsection Misnesting
2176 When a macro is called with arguments, the arguments are substituted
2177 into the macro body and the result is checked, together with the rest of
2178 the input file, for more macro calls. It is possible to piece together
2179 a macro call coming partially from the macro body and partially from the
2180 arguments. For example,
2183 #define twice(x) (2*(x))
2184 #define call_with_1(x) x(1)
2186 @expansion{} twice(1)
2187 @expansion{} (2*(1))
2190 Macro definitions do not have to have balanced parentheses. By writing
2191 an unbalanced open parenthesis in a macro body, it is possible to create
2192 a macro call that begins inside the macro body but ends outside of it.
2196 #define strange(file) fprintf (file, "%s %d",
2198 strange(stderr) p, 35)
2199 @expansion{} fprintf (stderr, "%s %d", p, 35)
2202 The ability to piece together a macro call can be useful, but the use of
2203 unbalanced open parentheses in a macro body is just confusing, and
2206 @node Operator Precedence Problems
2207 @subsection Operator Precedence Problems
2208 @cindex parentheses in macro bodies
2210 You may have noticed that in most of the macro definition examples shown
2211 above, each occurrence of a macro argument name had parentheses around
2212 it. In addition, another pair of parentheses usually surround the
2213 entire macro definition. Here is why it is best to write macros that
2216 Suppose you define a macro as follows,
2219 #define ceil_div(x, y) (x + y - 1) / y
2223 whose purpose is to divide, rounding up. (One use for this operation is
2224 to compute how many @code{int} objects are needed to hold a certain
2225 number of @code{char} objects.) Then suppose it is used as follows:
2228 a = ceil_div (b & c, sizeof (int));
2229 @expansion{} a = (b & c + sizeof (int) - 1) / sizeof (int);
2233 This does not do what is intended. The operator-precedence rules of
2234 C make it equivalent to this:
2237 a = (b & (c + sizeof (int) - 1)) / sizeof (int);
2241 What we want is this:
2244 a = ((b & c) + sizeof (int) - 1)) / sizeof (int);
2248 Defining the macro as
2251 #define ceil_div(x, y) ((x) + (y) - 1) / (y)
2255 provides the desired result.
2257 Unintended grouping can result in another way. Consider @code{sizeof
2258 ceil_div(1, 2)}. That has the appearance of a C expression that would
2259 compute the size of the type of @code{ceil_div (1, 2)}, but in fact it
2260 means something very different. Here is what it expands to:
2263 sizeof ((1) + (2) - 1) / (2)
2267 This would take the size of an integer and divide it by two. The
2268 precedence rules have put the division outside the @code{sizeof} when it
2269 was intended to be inside.
2271 Parentheses around the entire macro definition prevent such problems.
2272 Here, then, is the recommended way to define @code{ceil_div}:
2275 #define ceil_div(x, y) (((x) + (y) - 1) / (y))
2278 @node Swallowing the Semicolon
2279 @subsection Swallowing the Semicolon
2280 @cindex semicolons (after macro calls)
2282 Often it is desirable to define a macro that expands into a compound
2283 statement. Consider, for example, the following macro, that advances a
2284 pointer (the argument @code{p} says where to find it) across whitespace
2288 #define SKIP_SPACES(p, limit) \
2289 @{ char *lim = (limit); \
2290 while (p < lim) @{ \
2291 if (*p++ != ' ') @{ \
2296 Here backslash-newline is used to split the macro definition, which must
2297 be a single logical line, so that it resembles the way such code would
2298 be laid out if not part of a macro definition.
2300 A call to this macro might be @code{SKIP_SPACES (p, lim)}. Strictly
2301 speaking, the call expands to a compound statement, which is a complete
2302 statement with no need for a semicolon to end it. However, since it
2303 looks like a function call, it minimizes confusion if you can use it
2304 like a function call, writing a semicolon afterward, as in
2305 @code{SKIP_SPACES (p, lim);}
2307 This can cause trouble before @code{else} statements, because the
2308 semicolon is actually a null statement. Suppose you write
2312 SKIP_SPACES (p, lim);
2317 The presence of two statements---the compound statement and a null
2318 statement---in between the @code{if} condition and the @code{else}
2319 makes invalid C code.
2321 The definition of the macro @code{SKIP_SPACES} can be altered to solve
2322 this problem, using a @code{do @dots{} while} statement. Here is how:
2325 #define SKIP_SPACES(p, limit) \
2326 do @{ char *lim = (limit); \
2327 while (p < lim) @{ \
2328 if (*p++ != ' ') @{ \
2329 p--; break; @}@}@} \
2333 Now @code{SKIP_SPACES (p, lim);} expands into
2336 do @{@dots{}@} while (0);
2340 which is one statement. The loop executes exactly once; most compilers
2341 generate no extra code for it.
2343 @node Duplication of Side Effects
2344 @subsection Duplication of Side Effects
2346 @cindex side effects (in macro arguments)
2347 @cindex unsafe macros
2348 Many C programs define a macro @code{min}, for ``minimum'', like this:
2351 #define min(X, Y) ((X) < (Y) ? (X) : (Y))
2354 When you use this macro with an argument containing a side effect,
2358 next = min (x + y, foo (z));
2362 it expands as follows:
2365 next = ((x + y) < (foo (z)) ? (x + y) : (foo (z)));
2369 where @code{x + y} has been substituted for @code{X} and @code{foo (z)}
2372 The function @code{foo} is used only once in the statement as it appears
2373 in the program, but the expression @code{foo (z)} has been substituted
2374 twice into the macro expansion. As a result, @code{foo} might be called
2375 two times when the statement is executed. If it has side effects or if
2376 it takes a long time to compute, the results might not be what you
2377 intended. We say that @code{min} is an @dfn{unsafe} macro.
2379 The best solution to this problem is to define @code{min} in a way that
2380 computes the value of @code{foo (z)} only once. The C language offers
2381 no standard way to do this, but it can be done with GNU extensions as
2386 (@{ typeof (X) x_ = (X); \
2387 typeof (Y) y_ = (Y); \
2388 (x_ < y_) ? x_ : y_; @})
2391 The @samp{(@{ @dots{} @})} notation produces a compound statement that
2392 acts as an expression. Its value is the value of its last statement.
2393 This permits us to define local variables and assign each argument to
2394 one. The local variables have underscores after their names to reduce
2395 the risk of conflict with an identifier of wider scope (it is impossible
2396 to avoid this entirely). Now each argument is evaluated exactly once.
2398 If you do not wish to use GNU C extensions, the only solution is to be
2399 careful when @emph{using} the macro @code{min}. For example, you can
2400 calculate the value of @code{foo (z)}, save it in a variable, and use
2401 that variable in @code{min}:
2405 #define min(X, Y) ((X) < (Y) ? (X) : (Y))
2409 next = min (x + y, tem);
2415 (where we assume that @code{foo} returns type @code{int}).
2417 @node Self-Referential Macros
2418 @subsection Self-Referential Macros
2419 @cindex self-reference
2421 A @dfn{self-referential} macro is one whose name appears in its
2422 definition. Recall that all macro definitions are rescanned for more
2423 macros to replace. If the self-reference were considered a use of the
2424 macro, it would produce an infinitely large expansion. To prevent this,
2425 the self-reference is not considered a macro call. It is passed into
2426 the preprocessor output unchanged. Let's consider an example:
2429 #define foo (4 + foo)
2433 where @code{foo} is also a variable in your program.
2435 Following the ordinary rules, each reference to @code{foo} will expand
2436 into @code{(4 + foo)}; then this will be rescanned and will expand into
2437 @code{(4 + (4 + foo))}; and so on until the computer runs out of memory.
2439 The self-reference rule cuts this process short after one step, at
2440 @code{(4 + foo)}. Therefore, this macro definition has the possibly
2441 useful effect of causing the program to add 4 to the value of @code{foo}
2442 wherever @code{foo} is referred to.
2444 In most cases, it is a bad idea to take advantage of this feature. A
2445 person reading the program who sees that @code{foo} is a variable will
2446 not expect that it is a macro as well. The reader will come across the
2447 identifier @code{foo} in the program and think its value should be that
2448 of the variable @code{foo}, whereas in fact the value is four greater.
2450 One common, useful use of self-reference is to create a macro which
2451 expands to itself. If you write
2458 then the macro @code{EPERM} expands to @code{EPERM}. Effectively, it is
2459 left alone by the preprocessor whenever it's used in running text. You
2460 can tell that it's a macro with @samp{#ifdef}. You might do this if you
2461 want to define numeric constants with an @code{enum}, but have
2462 @samp{#ifdef} be true for each constant.
2464 If a macro @code{x} expands to use a macro @code{y}, and the expansion of
2465 @code{y} refers to the macro @code{x}, that is an @dfn{indirect
2466 self-reference} of @code{x}. @code{x} is not expanded in this case
2467 either. Thus, if we have
2475 then @code{x} and @code{y} expand as follows:
2479 x @expansion{} (4 + y)
2480 @expansion{} (4 + (2 * x))
2482 y @expansion{} (2 * x)
2483 @expansion{} (2 * (4 + y))
2488 Each macro is expanded when it appears in the definition of the other
2489 macro, but not when it indirectly appears in its own definition.
2491 @node Argument Prescan
2492 @subsection Argument Prescan
2493 @cindex expansion of arguments
2494 @cindex macro argument expansion
2495 @cindex prescan of macro arguments
2497 Macro arguments are completely macro-expanded before they are
2498 substituted into a macro body, unless they are stringified or pasted
2499 with other tokens. After substitution, the entire macro body, including
2500 the substituted arguments, is scanned again for macros to be expanded.
2501 The result is that the arguments are scanned @emph{twice} to expand
2502 macro calls in them.
2504 Most of the time, this has no effect. If the argument contained any
2505 macro calls, they are expanded during the first scan. The result
2506 therefore contains no macro calls, so the second scan does not change
2507 it. If the argument were substituted as given, with no prescan, the
2508 single remaining scan would find the same macro calls and produce the
2511 You might expect the double scan to change the results when a
2512 self-referential macro is used in an argument of another macro
2513 (@pxref{Self-Referential Macros}): the self-referential macro would be
2514 expanded once in the first scan, and a second time in the second scan.
2515 However, this is not what happens. The self-references that do not
2516 expand in the first scan are marked so that they will not expand in the
2519 You might wonder, ``Why mention the prescan, if it makes no difference?
2520 And why not skip it and make the preprocessor faster?'' The answer is
2521 that the prescan does make a difference in three special cases:
2525 Nested calls to a macro.
2527 We say that @dfn{nested} calls to a macro occur when a macro's argument
2528 contains a call to that very macro. For example, if @code{f} is a macro
2529 that expects one argument, @code{f (f (1))} is a nested pair of calls to
2530 @code{f}. The desired expansion is made by expanding @code{f (1)} and
2531 substituting that into the definition of @code{f}. The prescan causes
2532 the expected result to happen. Without the prescan, @code{f (1)} itself
2533 would be substituted as an argument, and the inner use of @code{f} would
2534 appear during the main scan as an indirect self-reference and would not
2538 Macros that call other macros that stringify or concatenate.
2540 If an argument is stringified or concatenated, the prescan does not
2541 occur. If you @emph{want} to expand a macro, then stringify or
2542 concatenate its expansion, you can do that by causing one macro to call
2543 another macro that does the stringification or concatenation. For
2544 instance, if you have
2547 #define AFTERX(x) X_ ## x
2548 #define XAFTERX(x) AFTERX(x)
2549 #define TABLESIZE 1024
2550 #define BUFSIZE TABLESIZE
2553 then @code{AFTERX(BUFSIZE)} expands to @code{X_BUFSIZE}, and
2554 @code{XAFTERX(BUFSIZE)} expands to @code{X_1024}. (Not to
2555 @code{X_TABLESIZE}. Prescan always does a complete expansion.)
2558 Macros used in arguments, whose expansions contain unshielded commas.
2560 This can cause a macro expanded on the second scan to be called with the
2561 wrong number of arguments. Here is an example:
2565 #define bar(x) lose(x)
2566 #define lose(x) (1 + (x))
2569 We would like @code{bar(foo)} to turn into @code{(1 + (foo))}, which
2570 would then turn into @code{(1 + (a,b))}. Instead, @code{bar(foo)}
2571 expands into @code{lose(a,b)}, and you get an error because @code{lose}
2572 requires a single argument. In this case, the problem is easily solved
2573 by the same parentheses that ought to be used to prevent misnesting of
2574 arithmetic operations:
2579 #define bar(x) lose((x))
2582 The extra pair of parentheses prevents the comma in @code{foo}'s
2583 definition from being interpreted as an argument separator.
2587 @node Newlines in Arguments
2588 @subsection Newlines in Arguments
2589 @cindex newlines in macro arguments
2591 The invocation of a function-like macro can extend over many logical
2592 lines. However, in the present implementation, the entire expansion
2593 comes out on one line. Thus line numbers emitted by the compiler or
2594 debugger refer to the line the invocation started on, which might be
2595 different to the line containing the argument causing the problem.
2597 Here is an example illustrating this:
2600 #define ignore_second_arg(a,b,c) a; c
2602 ignore_second_arg (foo (),
2608 The syntax error triggered by the tokens @code{syntax error} results in
2609 an error message citing line three---the line of ignore_second_arg---
2610 even though the problematic code comes from line five.
2612 We consider this a bug, and intend to fix it in the near future.
2615 @chapter Conditionals
2616 @cindex conditionals
2618 A @dfn{conditional} is a directive that instructs the preprocessor to
2619 select whether or not to include a chunk of code in the final token
2620 stream passed to the compiler. Preprocessor conditionals can test
2621 arithmetic expressions, or whether a name is defined as a macro, or both
2622 simultaneously using the special @code{defined} operator.
2624 A conditional in the C preprocessor resembles in some ways an @code{if}
2625 statement in C, but it is important to understand the difference between
2626 them. The condition in an @code{if} statement is tested during the
2627 execution of your program. Its purpose is to allow your program to
2628 behave differently from run to run, depending on the data it is
2629 operating on. The condition in a preprocessing conditional directive is
2630 tested when your program is compiled. Its purpose is to allow different
2631 code to be included in the program depending on the situation at the
2632 time of compilation.
2634 However, the distinction is becoming less clear. Modern compilers often
2635 do test @code{if} statements when a program is compiled, if their
2636 conditions are known not to vary at run time, and eliminate code which
2637 can never be executed. If you can count on your compiler to do this,
2638 you may find that your program is more readable if you use @code{if}
2639 statements with constant conditions (perhaps determined by macros). Of
2640 course, you can only use this to exclude code, not type definitions or
2641 other preprocessing directives, and you can only do it if the code
2642 remains syntactically valid when it is not to be used.
2644 GCC version 3 eliminates this kind of never-executed code even when
2645 not optimizing. Older versions did it only when optimizing.
2648 * Conditional Uses::
2649 * Conditional Syntax::
2653 @node Conditional Uses
2654 @section Conditional Uses
2656 There are three general reasons to use a conditional.
2660 A program may need to use different code depending on the machine or
2661 operating system it is to run on. In some cases the code for one
2662 operating system may be erroneous on another operating system; for
2663 example, it might refer to data types or constants that do not exist on
2664 the other system. When this happens, it is not enough to avoid
2665 executing the invalid code. Its mere presence will cause the compiler
2666 to reject the program. With a preprocessing conditional, the offending
2667 code can be effectively excised from the program when it is not valid.
2670 You may want to be able to compile the same source file into two
2671 different programs. One version might make frequent time-consuming
2672 consistency checks on its intermediate data, or print the values of
2673 those data for debugging, and the other not.
2676 A conditional whose condition is always false is one way to exclude code
2677 from the program but keep it as a sort of comment for future reference.
2680 Simple programs that do not need system-specific logic or complex
2681 debugging hooks generally will not need to use preprocessing
2684 @node Conditional Syntax
2685 @section Conditional Syntax
2688 A conditional in the C preprocessor begins with a @dfn{conditional
2689 directive}: @samp{#if}, @samp{#ifdef} or @samp{#ifndef}.
2704 The simplest sort of conditional is
2710 @var{controlled text}
2712 #endif /* @var{MACRO} */
2716 @cindex conditional group
2717 This block is called a @dfn{conditional group}. @var{controlled text}
2718 will be included in the output of the preprocessor if and only if
2719 @var{MACRO} is defined. We say that the conditional @dfn{succeeds} if
2720 @var{MACRO} is defined, @dfn{fails} if it is not.
2722 The @var{controlled text} inside of a conditional can include
2723 preprocessing directives. They are executed only if the conditional
2724 succeeds. You can nest conditional groups inside other conditional
2725 groups, but they must be completely nested. In other words,
2726 @samp{#endif} always matches the nearest @samp{#ifdef} (or
2727 @samp{#ifndef}, or @samp{#if}). Also, you cannot start a conditional
2728 group in one file and end it in another.
2730 Even if a conditional fails, the @var{controlled text} inside it is
2731 still run through initial transformations and tokenization. Therefore,
2732 it must all be lexically valid C@. Normally the only way this matters is
2733 that all comments and string literals inside a failing conditional group
2734 must still be properly ended.
2736 The comment following the @samp{#endif} is not required, but it is a
2737 good practice if there is a lot of @var{controlled text}, because it
2738 helps people match the @samp{#endif} to the corresponding @samp{#ifdef}.
2739 Older programs sometimes put @var{MACRO} directly after the
2740 @samp{#endif} without enclosing it in a comment. This is invalid code
2741 according to the C standard. GNU CPP accepts it with a warning. It
2742 never affects which @samp{#ifndef} the @samp{#endif} matches.
2745 Sometimes you wish to use some code if a macro is @emph{not} defined.
2746 You can do this by writing @samp{#ifndef} instead of @samp{#ifdef}.
2747 One common use of @samp{#ifndef} is to include code only the first
2748 time a header file is included. @xref{Once-Only Headers}.
2750 Macro definitions can vary between compilations for several reasons.
2751 Here are some samples.
2755 Some macros are predefined on each kind of machine
2756 (@pxref{System-specific Predefined Macros}). This allows you to provide
2757 code specially tuned for a particular machine.
2760 System header files define more macros, associated with the features
2761 they implement. You can test these macros with conditionals to avoid
2762 using a system feature on a machine where it is not implemented.
2765 Macros can be defined or undefined with the @option{-D} and @option{-U}
2766 command line options when you compile the program. You can arrange to
2767 compile the same source file into two different programs by choosing a
2768 macro name to specify which program you want, writing conditionals to
2769 test whether or how this macro is defined, and then controlling the
2770 state of the macro with command line options, perhaps set in the
2771 Makefile. @xref{Invocation}.
2774 Your program might have a special header file (often called
2775 @file{config.h}) that is adjusted when the program is compiled. It can
2776 define or not define macros depending on the features of the system and
2777 the desired capabilities of the program. The adjustment can be
2778 automated by a tool such as @command{autoconf}, or done by hand.
2784 The @samp{#if} directive allows you to test the value of an arithmetic
2785 expression, rather than the mere existence of one macro. Its syntax is
2789 #if @var{expression}
2791 @var{controlled text}
2793 #endif /* @var{expression} */
2797 @var{expression} is a C expression of integer type, subject to stringent
2798 restrictions. It may contain
2805 Character constants, which are interpreted as they would be in normal
2809 Arithmetic operators for addition, subtraction, multiplication,
2810 division, bitwise operations, shifts, comparisons, and logical
2811 operations (@code{&&} and @code{||}). The latter two obey the usual
2812 short-circuiting rules of standard C@.
2815 Macros. All macros in the expression are expanded before actual
2816 computation of the expression's value begins.
2819 Uses of the @code{defined} operator, which lets you check whether macros
2820 are defined in the middle of an @samp{#if}.
2823 Identifiers that are not macros, which are all considered to be the
2824 number zero. This allows you to write @code{@w{#if MACRO}} instead of
2825 @code{@w{#ifdef MACRO}}, if you know that MACRO, when defined, will
2826 always have a nonzero value. Function-like macros used without their
2827 function call parentheses are also treated as zero.
2829 In some contexts this shortcut is undesirable. The @option{-Wundef}
2830 option causes GCC to warn whenever it encounters an identifier which is
2831 not a macro in an @samp{#if}.
2834 The preprocessor does not know anything about types in the language.
2835 Therefore, @code{sizeof} operators are not recognized in @samp{#if}, and
2836 neither are @code{enum} constants. They will be taken as identifiers
2837 which are not macros, and replaced by zero. In the case of
2838 @code{sizeof}, this is likely to cause the expression to be invalid.
2840 The preprocessor calculates the value of @var{expression}. It carries
2841 out all calculations in the widest integer type known to the compiler;
2842 on most machines supported by GCC this is 64 bits. This is not the same
2843 rule as the compiler uses to calculate the value of a constant
2844 expression, and may give different results in some cases. If the value
2845 comes out to be nonzero, the @samp{#if} succeeds and the @var{controlled
2846 text} is included; otherwise it is skipped.
2848 If @var{expression} is not correctly formed, GCC issues an error and
2849 treats the conditional as having failed.
2854 @cindex @code{defined}
2855 The special operator @code{defined} is used in @samp{#if} and
2856 @samp{#elif} expressions to test whether a certain name is defined as a
2857 macro. @code{defined @var{name}} and @code{defined (@var{name})} are
2858 both expressions whose value is 1 if @var{name} is defined as a macro at
2859 the current point in the program, and 0 otherwise. Thus, @code{@w{#if
2860 defined MACRO}} is precisely equivalent to @code{@w{#ifdef MACRO}}.
2862 @code{defined} is useful when you wish to test more than one macro for
2863 existence at once. For example,
2866 #if defined (__vax__) || defined (__ns16000__)
2870 would succeed if either of the names @code{__vax__} or
2871 @code{__ns16000__} is defined as a macro.
2873 Conditionals written like this:
2876 #if defined BUFSIZE && BUFSIZE >= 1024
2880 can generally be simplified to just @code{@w{#if BUFSIZE >= 1024}},
2881 since if @code{BUFSIZE} is not defined, it will be interpreted as having
2884 If the @code{defined} operator appears as a result of a macro expansion,
2885 the C standard says the behavior is undefined. GNU cpp treats it as a
2886 genuine @code{defined} operator and evaluates it normally. It will warn
2887 wherever your code uses this feature if you use the command-line option
2888 @option{-pedantic}, since other compilers may handle it differently.
2894 The @samp{#else} directive can be added to a conditional to provide
2895 alternative text to be used if the condition fails. This is what it
2900 #if @var{expression}
2902 #else /* Not @var{expression} */
2904 #endif /* Not @var{expression} */
2909 If @var{expression} is nonzero, the @var{text-if-true} is included and
2910 the @var{text-if-false} is skipped. If @var{expression} is zero, the
2913 You can use @samp{#else} with @samp{#ifdef} and @samp{#ifndef}, too.
2919 One common case of nested conditionals is used to check for more than two
2920 possible alternatives. For example, you might have
2934 Another conditional directive, @samp{#elif}, allows this to be
2935 abbreviated as follows:
2942 #else /* X != 2 and X != 1*/
2944 #endif /* X != 2 and X != 1*/
2947 @samp{#elif} stands for ``else if''. Like @samp{#else}, it goes in the
2948 middle of a conditional group and subdivides it; it does not require a
2949 matching @samp{#endif} of its own. Like @samp{#if}, the @samp{#elif}
2950 directive includes an expression to be tested. The text following the
2951 @samp{#elif} is processed only if the original @samp{#if}-condition
2952 failed and the @samp{#elif} condition succeeds.
2954 More than one @samp{#elif} can go in the same conditional group. Then
2955 the text after each @samp{#elif} is processed only if the @samp{#elif}
2956 condition succeeds after the original @samp{#if} and all previous
2957 @samp{#elif} directives within it have failed.
2959 @samp{#else} is allowed after any number of @samp{#elif} directives, but
2960 @samp{#elif} may not follow @samp{#else}.
2963 @section Deleted Code
2964 @cindex commenting out code
2966 If you replace or delete a part of the program but want to keep the old
2967 code around for future reference, you often cannot simply comment it
2968 out. Block comments do not nest, so the first comment inside the old
2969 code will end the commenting-out. The probable result is a flood of
2972 One way to avoid this problem is to use an always-false conditional
2973 instead. For instance, put @code{#if 0} before the deleted code and
2974 @code{#endif} after it. This works even if the code being turned
2975 off contains conditionals, but they must be entire conditionals
2976 (balanced @samp{#if} and @samp{#endif}).
2978 Some people use @code{#ifdef notdef} instead. This is risky, because
2979 @code{notdef} might be accidentally defined as a macro, and then the
2980 conditional would succeed. @code{#if 0} can be counted on to fail.
2982 Do not use @code{#if 0} for comments which are not C code. Use a real
2983 comment, instead. The interior of @code{#if 0} must consist of complete
2984 tokens; in particular, single-quote characters must balance. Comments
2985 often contain unbalanced single-quote characters (known in English as
2986 apostrophes). These confuse @code{#if 0}. They don't confuse
2990 @chapter Diagnostics
2992 @cindex reporting errors
2993 @cindex reporting warnings
2996 The directive @samp{#error} causes the preprocessor to report a fatal
2997 error. The tokens forming the rest of the line following @samp{#error}
2998 are used as the error message.
3000 You would use @samp{#error} inside of a conditional that detects a
3001 combination of parameters which you know the program does not properly
3002 support. For example, if you know that the program will not run
3003 properly on a VAX, you might write
3008 #error "Won't work on VAXen. See comments at get_last_object."
3013 If you have several configuration parameters that must be set up by
3014 the installation in a consistent way, you can use conditionals to detect
3015 an inconsistency and report it with @samp{#error}. For example,
3018 #if !defined(UNALIGNED_INT_ASM_OP) && defined(DWARF2_DEBUGGING_INFO)
3019 #error "DWARF2_DEBUGGING_INFO requires UNALIGNED_INT_ASM_OP."
3024 The directive @samp{#warning} is like @samp{#error}, but causes the
3025 preprocessor to issue a warning and continue preprocessing. The tokens
3026 following @samp{#warning} are used as the warning message.
3028 You might use @samp{#warning} in obsolete header files, with a message
3029 directing the user to the header file which should be used instead.
3031 Neither @samp{#error} nor @samp{#warning} macro-expands its argument.
3032 Internal whitespace sequences are each replaced with a single space.
3033 The line must consist of complete tokens. It is wisest to make the
3034 argument of these directives be a single string constant; this avoids
3035 problems with apostrophes and the like.
3038 @chapter Line Control
3039 @cindex line control
3041 The C preprocessor informs the C compiler of the location in your source
3042 code where each token came from. Presently, this is just the file name
3043 and line number. All the tokens resulting from macro expansion are
3044 reported as having appeared on the line of the source file where the
3045 outermost macro was used. We intend to be more accurate in the future.
3047 If you write a program which generates source code, such as the
3048 @command{bison} parser generator, you may want to adjust the preprocessor's
3049 notion of the current file name and line number by hand. Parts of the
3050 output from @command{bison} are generated from scratch, other parts come
3051 from a standard parser file. The rest are copied verbatim from
3052 @command{bison}'s input. You would like compiler error messages and
3053 symbolic debuggers to be able to refer to @code{bison}'s input file.
3056 @command{bison} or any such program can arrange this by writing
3057 @samp{#line} directives into the output file. @samp{#line} is a
3058 directive that specifies the original line number and source file name
3059 for subsequent input in the current preprocessor input file.
3060 @samp{#line} has three variants:
3063 @item #line @var{linenum}
3064 @var{linenum} is a non-negative decimal integer constant. It specifies
3065 the line number which should be reported for the following line of
3066 input. Subsequent lines are counted from @var{linenum}.
3068 @item #line @var{linenum} @var{filename}
3069 @var{linenum} is the same as for the first form, and has the same
3070 effect. In addition, @var{filename} is a string constant. The
3071 following line and all subsequent lines are reported to come from the
3072 file it specifies, until something else happens to change that.
3073 @var{filename} is interpreted according to the normal rules for a string
3074 constant: backslash escapes are interpreted. This is different from
3077 Previous versions of GNU CPP did not interpret escapes in @samp{#line};
3078 we have changed it because the standard requires they be interpreted,
3079 and most other compilers do.
3081 @item #line @var{anything else}
3082 @var{anything else} is checked for macro calls, which are expanded.
3083 The result should match one of the above two forms.
3086 @samp{#line} directives alter the results of the @code{__FILE__} and
3087 @code{__LINE__} predefined macros from that point on. @xref{Standard
3088 Predefined Macros}. They do not have any effect on @samp{#include}'s
3089 idea of the directory containing the current file. This is a change
3090 from GCC 2.95. Previously, a file reading
3093 #line 1 "../src/gram.y"
3097 would search for @file{gram.h} in @file{../src}, then the @option{-I}
3098 chain; the directory containing the physical source file would not be
3099 searched. In GCC 3.0 and later, the @samp{#include} is not affected by
3100 the presence of a @samp{#line} referring to a different directory.
3102 We made this change because the old behavior caused problems when
3103 generated source files were transported between machines. For instance,
3104 it is common practice to ship generated parsers with a source release,
3105 so that people building the distribution do not need to have yacc or
3106 Bison installed. These files frequently have @samp{#line} directives
3107 referring to the directory tree of the system where the distribution was
3108 created. If GCC tries to search for headers in those directories, the
3109 build is likely to fail.
3111 The new behavior can cause failures too, if the generated file is not
3112 in the same directory as its source and it attempts to include a header
3113 which would be visible searching from the directory containing the
3114 source file. However, this problem is easily solved with an additional
3115 @option{-I} switch on the command line. The failures caused by the old
3116 semantics could sometimes be corrected only by editing the generated
3117 files, which is difficult and error-prone.
3122 The @samp{#pragma} directive is the method specified by the C standard
3123 for providing additional information to the compiler, beyond what is
3124 conveyed in the language itself. Three forms of this directive
3125 (commonly known as @dfn{pragmas}) are specified by the 1999 C standard.
3126 A C compiler is free to attach any meaning it likes to other pragmas.
3128 GCC has historically preferred to use extensions to the syntax of the
3129 language, such as @code{__attribute__}, for this purpose. However, GCC
3130 does define a few pragmas of its own. These mostly have effects on the
3131 entire translation unit or source file.
3133 In GCC version 3, all GNU-defined, supported pragmas have been given a
3134 @code{GCC} prefix. This is in line with the @code{STDC} prefix on all
3135 pragmas defined by C99. For backward compatibility, pragmas which were
3136 recognized by previous versions are still recognized without the
3137 @code{GCC} prefix, but that usage is deprecated. Some older pragmas are
3138 deprecated in their entirety. They are not recognized with the
3139 @code{GCC} prefix. @xref{Obsolete Features}.
3141 @cindex @code{_Pragma}
3142 C99 introduces the @code{@w{_Pragma}} operator. This feature addresses a
3143 major problem with @samp{#pragma}: being a directive, it cannot be
3144 produced as the result of macro expansion. @code{@w{_Pragma}} is an
3145 operator, much like @code{sizeof} or @code{defined}, and can be embedded
3148 Its syntax is @code{@w{_Pragma (@var{string-literal})}}, where
3149 @var{string-literal} can be either a normal or wide-character string
3150 literal. It is destringized, by replacing all @samp{\\} with a single
3151 @samp{\} and all @samp{\"} with a @samp{"}. The result is then
3152 processed as if it had appeared as the right hand side of a
3153 @samp{#pragma} directive. For example,
3156 _Pragma ("GCC dependency \"parse.y\"")
3160 has the same effect as @code{#pragma GCC dependency "parse.y"}. The
3161 same effect could be achieved using macros, for example
3164 #define DO_PRAGMA(x) _Pragma (#x)
3165 DO_PRAGMA (GCC dependency "parse.y")
3168 The standard is unclear on where a @code{_Pragma} operator can appear.
3169 The preprocessor does not accept it within a preprocessing conditional
3170 directive like @samp{#if}. To be safe, you are probably best keeping it
3171 out of directives other than @samp{#define}, and putting it on a line of
3174 This manual documents the pragmas which are meaningful to the
3175 preprocessor itself. Other pragmas are meaningful to the C or C++
3176 compilers. They are documented in the GCC manual.
3179 @item #pragma GCC dependency
3180 @code{#pragma GCC dependency} allows you to check the relative dates of
3181 the current file and another file. If the other file is more recent than
3182 the current file, a warning is issued. This is useful if the current
3183 file is derived from the other file, and should be regenerated. The
3184 other file is searched for using the normal include search path.
3185 Optional trailing text can be used to give more information in the
3189 #pragma GCC dependency "parse.y"
3190 #pragma GCC dependency "/usr/include/time.h" rerun fixincludes
3193 @item #pragma GCC poison
3194 Sometimes, there is an identifier that you want to remove completely
3195 from your program, and make sure that it never creeps back in. To
3196 enforce this, you can @dfn{poison} the identifier with this pragma.
3197 @code{#pragma GCC poison} is followed by a list of identifiers to
3198 poison. If any of those identifiers appears anywhere in the source
3199 after the directive, it is a hard error. For example,
3202 #pragma GCC poison printf sprintf fprintf
3203 sprintf(some_string, "hello");
3207 will produce an error.
3209 If a poisoned identifier appears as part of the expansion of a macro
3210 which was defined before the identifier was poisoned, it will @emph{not}
3211 cause an error. This lets you poison an identifier without worrying
3212 about system headers defining macros that use it.
3217 #define strrchr rindex
3218 #pragma GCC poison rindex
3219 strrchr(some_string, 'h');
3223 will not produce an error.
3225 @item #pragma GCC system_header
3226 This pragma takes no arguments. It causes the rest of the code in the
3227 current file to be treated as if it came from a system header.
3228 @xref{System Headers}.
3232 @node Other Directives
3233 @chapter Other Directives
3236 The @samp{#ident} directive takes one argument, a string constant. On
3237 some systems, that string constant is copied into a special segment of
3238 the object file. On other systems, the directive is ignored.
3240 This directive is not part of the C standard, but it is not an official
3241 GNU extension either. We believe it came from System V@.
3244 The @samp{#sccs} directive is recognized on some systems, because it
3245 appears in their header files. It is a very old, obscure, extension
3246 which we did not invent, and we have been unable to find any
3247 documentation of what it should do, so GCC simply ignores it.
3249 @cindex null directive
3250 The @dfn{null directive} consists of a @samp{#} followed by a newline,
3251 with only whitespace (including comments) in between. A null directive
3252 is understood as a preprocessing directive but has no effect on the
3253 preprocessor output. The primary significance of the existence of the
3254 null directive is that an input line consisting of just a @samp{#} will
3255 produce no output, rather than a line of output containing just a
3256 @samp{#}. Supposedly some old C programs contain such lines.
3258 @node Preprocessor Output
3259 @chapter Preprocessor Output
3261 When the C preprocessor is used with the C, C++, or Objective-C
3262 compilers, it is integrated into the compiler and communicates a stream
3263 of binary tokens directly to the compiler's parser. However, it can
3264 also be used in the more conventional standalone mode, where it produces
3266 @c FIXME: Document the library interface.
3268 @cindex output format
3269 The output from the C preprocessor looks much like the input, except
3270 that all preprocessing directive lines have been replaced with blank
3271 lines and all comments with spaces. Long runs of blank lines are
3274 The ISO standard specifies that it is implementation defined whether a
3275 preprocessor preserves whitespace between tokens, or replaces it with
3276 e.g.@: a single space. In GNU CPP, whitespace between tokens is collapsed
3277 to become a single space, with the exception that the first token on a
3278 non-directive line is preceded with sufficient spaces that it appears in
3279 the same column in the preprocessed output that it appeared in the
3280 original source file. This is so the output is easy to read.
3281 @xref{Differences from previous versions}. CPP does not insert any
3282 whitespace where there was none in the original source, except where
3283 necessary to prevent an accidental token paste.
3286 Source file name and line number information is conveyed by lines
3290 # @var{linenum} @var{filename} @var{flags}
3294 These are called @dfn{linemarkers}. They are inserted as needed into
3295 the output (but never within a string or character constant). They mean
3296 that the following line originated in file @var{filename} at line
3297 @var{linenum}. @var{filename} will never contain any non-printing
3298 characters; they are replaced with octal escape sequences.
3300 After the file name comes zero or more flags, which are @samp{1},
3301 @samp{2}, @samp{3}, or @samp{4}. If there are multiple flags, spaces
3302 separate them. Here is what the flags mean:
3306 This indicates the start of a new file.
3308 This indicates returning to a file (after having included another file).
3310 This indicates that the following text comes from a system header file,
3311 so certain warnings should be suppressed.
3313 This indicates that the following text should be treated as being
3314 wrapped in an implicit @code{extern "C"} block.
3315 @c maybe cross reference NO_IMPLICIT_EXTERN_C
3318 As an extension, the preprocessor accepts linemarkers in non-assembler
3319 input files. They are treated like the corresponding @samp{#line}
3320 directive, (@pxref{Line Control}), except that trailing flags are
3321 permitted, and are interpreted with the meanings described above. If
3322 multiple flags are given, they must be in ascending order.
3324 Some directives may be duplicated in the output of the preprocessor.
3325 These are @samp{#ident} (always), @samp{#pragma} (only if the
3326 preprocessor does not handle the pragma itself), and @samp{#define} and
3327 @samp{#undef} (with certain debugging options). If this happens, the
3328 @samp{#} of the directive will always be in the first column, and there
3329 will be no space between the @samp{#} and the directive name. If macro
3330 expansion happens to generate tokens which might be mistaken for a
3331 duplicated directive, a space will be inserted between the @samp{#} and
3334 @node Traditional Mode
3335 @chapter Traditional Mode
3337 Traditional (pre-standard) C preprocessing is rather different from
3338 the preprocessing specified by the standard. When GCC is given the
3339 @option{-traditional-cpp} option, it attempts to emulate a traditional
3342 GCC versions 3.2 and later only support traditional mode semantics in
3343 the preprocessor, and not in the compiler. This chapter outlines the
3344 semantics we implemented in the traditional preprocessor that is
3345 integrated into the compiler front end.
3347 The implementation does not correspond precisely to the behavior of
3348 earlier versions of GCC, nor to any true traditional preprocessor.
3349 After all, inconsistencies among traditional implementations were a
3350 major motivation for C standardization. However, we intend that it
3351 should be compatible with true traditional preprocessors in all ways
3352 that actually matter.
3355 * Traditional lexical analysis::
3356 * Traditional macros::
3357 * Traditional miscellany::
3358 * Traditional warnings::
3361 @node Traditional lexical analysis
3362 @section Traditional lexical analysis
3364 The traditional preprocessor does not decompose its input into tokens
3365 the same way a standards-conforming preprocessor does. The input is
3366 simply treated as a stream of text with minimal form imposed on it.
3368 This implementation does not treat trigraphs (@pxref{trigraphs})
3369 specially since they were created later during standardization. It
3370 handles arbitrarily-positioned escaped newlines properly and splices
3371 the lines as you would expect; many traditional preprocessors did not
3374 The form of horizontal whitespace in the input file is preserved in
3375 the output. In particular, hard tabs remain hard tabs. This can be
3376 useful if, for example, you are preprocessing a Makefile.
3378 Traditional CPP only recognizes C-style block comments, and treats the
3379 @samp{/*} sequence as introducing a comment only if it lies outside
3380 quoted text. Quoted text is introduced by the usual single and double
3381 quotes, and also by @samp{<} in a @code{#include} directive.
3383 Traditionally, comments are completely removed and are not replaced
3384 with a space. Since a traditional compiler does its own tokenization
3385 of the output of the preprocessor, comments can effectively be used as
3386 token paste operators. However, comments behave like separators for
3387 text handled by the preprocessor itself. For example, in
3394 @samp{foo} and @samp{bar} are distinct identifiers and expanded
3395 separately if they happen to be macros. In other words, this
3396 directive is equivalent to
3409 Generally speaking, in traditional mode an opening quote need not have
3410 a matching closing quote. In particular, a macro may be defined with
3411 replacement text that contains an unmatched quote. Of course, if you
3412 attempt to compile preprocessed output containing an unmatched quote
3413 you will get a syntax error.
3415 However, all preprocessing directives other than @code{#define}
3416 require matching quotes. For example:
3419 #define m This macro's fine and has an unmatched quote
3420 "/* This is not a comment. */
3421 /* This is a comment. The following #include directive
3426 Just as for the ISO preprocessor, what would be a closing quote can be
3427 escaped with a backslash to prevent the quoted text from closing.
3429 @node Traditional macros
3430 @section Traditional macros
3432 The major difference between traditional and ISO macros is that the
3433 former expand to text rather than to a token sequence. CPP removes
3434 all leading and trailing horizontal whitespace from a macro's
3435 replacement text before storing it, but preserves the form of internal
3438 One consequence is that it is legitimate for the replacement text to
3439 contain an unmatched quote (@pxref{Traditional lexical analysis}). An
3440 unclosed string or character constant continues into the text
3441 following the macro call. Similarly, the text at the end of a macro's
3442 expansion can run together with the text after the macro invocation to
3443 produce a single token.
3445 Normally comments are removed from the replacement text after the
3446 macro is expanded, but if the @option{-CC} option is passed on the
3447 command line comments are preserved. (In fact, the current
3448 implementation removes comments even before saving the macro
3449 replacement text, but it careful to do it in such a way that the
3450 observed effect is identical even in the function-like macro case.)
3452 The ISO stringification operator @samp{#} and token paste operator
3453 @samp{##} have no special meaning. As explained later, an effect
3454 similar to these operators can be obtained in a different way. Macro
3455 names that are embedded in quotes, either from the main file or after
3456 macro replacement, do not expand.
3458 CPP replaces an unquoted object-like macro name with its replacement
3459 text, and then rescans it for further macros to replace. Unlike
3460 standard macro expansion, traditional macro expansion has no provision
3461 to prevent recursion. If an object-like macro appears unquoted in its
3462 replacement text, it will be replaced again during the rescan pass,
3463 and so on @emph{ad infinitum}. GCC detects when it is expanding
3464 recursive macros, emits an error message, and continues after the
3465 offending macro invocation.
3469 #define INC(x) PLUS+x
3474 Function-like macros are similar in form but quite different in
3475 behavior to their ISO counterparts. Their arguments are contained
3476 within parentheses, are comma-separated, and can cross physical lines.
3477 Commas within nested parentheses are not treated as argument
3478 separators. Similarly, a quote in an argument cannot be left
3479 unclosed; in other words a comma or parenthesis in quotes is treated
3480 like any other character. There is no facility for handling variadic
3483 This implementation removes all comments from macro arguments, unless
3484 the @option{-C} option is given. The form of all other horizontal
3485 whitespace in arguments is preserved, including leading and trailing
3486 whitespace. In particular
3493 is treated as an invocation of the macro @samp{f} with a single
3494 argument consisting of a single space. If you want to invoke a
3495 function-like macro that takes no arguments, you must not leave any
3496 whitespace between the parentheses.
3498 If a macro argument crosses a new line, the new line is replaced with
3499 a space when forming the argument. If the previous line contained an
3500 unterminated quote, the following line inherits the quoted state.
3502 Traditional preprocessors replace parameters in the replacement text
3503 with their arguments regardless of whether the parameters are within
3504 quotes or not. This provides a way to stringize arguments. For
3509 str(/* A comment */ some text)
3510 @expansion{} " some text"
3514 Note that the comment is removed, but that the leading space is
3515 preserved. Here is an example of using a comment to effect token
3519 #define suffix(x) foo_/**/x
3521 @expansion{} foo_bar
3524 @node Traditional miscellany
3525 @section Traditional miscellany
3527 Here are some things to be aware of when using the traditional
3532 Preprocessing directives are recognized only when their leading
3533 @samp{#} appears in the first column. There can be no whitespace
3534 between the beginning of the line and the @samp{#}, but whitespace can
3535 follow the @samp{#}.
3538 A true traditional C preprocessor does not recognize @samp{#error} or
3539 @samp{#pragma}, and may not recognize @samp{#elif}. CPP supports all
3540 the directives in traditional mode that it supports in ISO mode,
3541 including extensions, with the exception that the effects of
3542 @samp{#pragma GCC poison} are undefined.
3545 __STDC__ is not defined.
3548 If you use digraphs the behaviour is undefined.
3552 @node Traditional warnings
3553 @section Traditional warnings
3554 You can request warnings about features that did not exist, or worked
3555 differently, in traditional C with the @option{-Wtraditional} option.
3556 GCC does not warn about features of ISO C which you must use when you
3557 are using a conforming compiler, such as the @samp{#} and @samp{##}
3560 Presently @option{-Wtraditional} warns about:
3564 Macro parameters that appear within string literals in the macro body.
3565 In traditional C macro replacement takes place within string literals,
3566 but does not in ISO C@.
3569 In traditional C, some preprocessor directives did not exist.
3570 Traditional preprocessors would only consider a line to be a directive
3571 if the @samp{#} appeared in column 1 on the line. Therefore
3572 @option{-Wtraditional} warns about directives that traditional C
3573 understands but would ignore because the @samp{#} does not appear as the
3574 first character on the line. It also suggests you hide directives like
3575 @samp{#pragma} not understood by traditional C by indenting them. Some
3576 traditional implementations would not recognize @samp{#elif}, so it
3577 suggests avoiding it altogether.
3580 A function-like macro that appears without an argument list. In some
3581 traditional preprocessors this was an error. In ISO C it merely means
3582 that the macro is not expanded.
3585 The unary plus operator. This did not exist in traditional C@.
3588 The @samp{U} and @samp{LL} integer constant suffixes, which were not
3589 available in traditional C@. (Traditional C does support the @samp{L}
3590 suffix for simple long integer constants.) You are not warned about
3591 uses of these suffixes in macros defined in system headers. For
3592 instance, @code{UINT_MAX} may well be defined as @code{4294967295U}, but
3593 you will not be warned if you use @code{UINT_MAX}.
3595 You can usually avoid the warning, and the related warning about
3596 constants which are so large that they are unsigned, by writing the
3597 integer constant in question in hexadecimal, with no U suffix. Take
3598 care, though, because this gives the wrong result in exotic cases.
3601 @node Implementation Details
3602 @chapter Implementation Details
3604 Here we document details of how the preprocessor's implementation
3605 affects its user-visible behavior. You should try to avoid undue
3606 reliance on behavior described here, as it is possible that it will
3607 change subtly in future implementations.
3609 Also documented here are obsolete features and changes from previous
3610 versions of GNU CPP@.
3613 * Implementation-defined behavior::
3614 * Implementation limits::
3615 * Obsolete Features::
3616 * Differences from previous versions::
3619 @node Implementation-defined behavior
3620 @section Implementation-defined behavior
3621 @cindex implementation-defined behavior
3623 This is how GNU CPP behaves in all the cases which the C standard
3624 describes as @dfn{implementation-defined}. This term means that the
3625 implementation is free to do what it likes, but must document its choice
3627 @c FIXME: Check the C++ standard for more implementation-defined stuff.
3631 @item The mapping of physical source file multi-byte characters to the
3632 execution character set.
3634 Currently, GNU cpp only supports character sets that are strict supersets
3635 of ASCII, and performs no translation of characters.
3637 @item Non-empty sequences of whitespace characters.
3639 In textual output, each whitespace sequence is collapsed to a single
3640 space. For aesthetic reasons, the first token on each non-directive
3641 line of output is preceded with sufficient spaces that it appears in the
3642 same column as it did in the original source file.
3644 @item The numeric value of character constants in preprocessor expressions.
3646 The preprocessor and compiler interpret character constants in the
3647 same way; i.e.@: escape sequences such as @samp{\a} are given the
3648 values they would have on the target machine.
3650 The compiler values a multi-character character constant a character
3651 at a time, shifting the previous value left by the number of bits per
3652 target character, and then or-ing in the bit-pattern of the new
3653 character truncated to the width of a target character. The final
3654 bit-pattern is given type @code{int}, and is therefore signed,
3655 regardless of whether single characters are signed or not (a slight
3656 change from versions 3.1 and earlier of GCC). If there are more
3657 characters in the constant than would fit in the target @code{int} the
3658 compiler issues a warning, and the excess leading characters are
3661 For example, 'ab' for a target with an 8-bit @code{char} would be
3662 interpreted as @w{(int) ((unsigned char) 'a' * 256 + (unsigned char)
3663 'b')}, and '\234a' as @w{(int) ((unsigned char) '\234' * 256 + (unsigned
3666 @item Source file inclusion.
3668 For a discussion on how the preprocessor locates header files,
3669 @ref{Include Operation}.
3671 @item Interpretation of the filename resulting from a macro-expanded
3672 @samp{#include} directive.
3674 @xref{Computed Includes}.
3676 @item Treatment of a @samp{#pragma} directive that after macro-expansion
3677 results in a standard pragma.
3679 No macro expansion occurs on any @samp{#pragma} directive line, so the
3680 question does not arise.
3682 Note that GCC does not yet implement any of the standard
3687 @node Implementation limits
3688 @section Implementation limits
3689 @cindex implementation limits
3691 GNU CPP has a small number of internal limits. This section lists the
3692 limits which the C standard requires to be no lower than some minimum,
3693 and all the others we are aware of. We intend there to be as few limits
3694 as possible. If you encounter an undocumented or inconvenient limit,
3695 please report that to us as a bug. (See the section on reporting bugs in
3698 Where we say something is limited @dfn{only by available memory}, that
3699 means that internal data structures impose no intrinsic limit, and space
3700 is allocated with @code{malloc} or equivalent. The actual limit will
3701 therefore depend on many things, such as the size of other things
3702 allocated by the compiler at the same time, the amount of memory
3703 consumed by other processes on the same computer, etc.
3707 @item Nesting levels of @samp{#include} files.
3709 We impose an arbitrary limit of 200 levels, to avoid runaway recursion.
3710 The standard requires at least 15 levels.
3712 @item Nesting levels of conditional inclusion.
3714 The C standard mandates this be at least 63. GNU CPP is limited only by
3717 @item Levels of parenthesised expressions within a full expression.
3719 The C standard requires this to be at least 63. In preprocessor
3720 conditional expressions, it is limited only by available memory.
3722 @item Significant initial characters in an identifier or macro name.
3724 The preprocessor treats all characters as significant. The C standard
3725 requires only that the first 63 be significant.
3727 @item Number of macros simultaneously defined in a single translation unit.
3729 The standard requires at least 4095 be possible. GNU CPP is limited only
3730 by available memory.
3732 @item Number of parameters in a macro definition and arguments in a macro call.
3734 We allow @code{USHRT_MAX}, which is no smaller than 65,535. The minimum
3735 required by the standard is 127.
3737 @item Number of characters on a logical source line.
3739 The C standard requires a minimum of 4096 be permitted. GNU CPP places
3740 no limits on this, but you may get incorrect column numbers reported in
3741 diagnostics for lines longer than 65,535 characters.
3743 @item Maximum size of a source file.
3745 The standard does not specify any lower limit on the maximum size of a
3746 source file. GNU cpp maps files into memory, so it is limited by the
3747 available address space. This is generally at least two gigabytes.
3748 Depending on the operating system, the size of physical memory may or
3749 may not be a limitation.
3753 @node Obsolete Features
3754 @section Obsolete Features
3756 GNU CPP has a number of features which are present mainly for
3757 compatibility with older programs. We discourage their use in new code.
3758 In some cases, we plan to remove the feature in a future version of GCC@.
3762 * Obsolete once-only headers::
3766 @subsection Assertions
3769 @dfn{Assertions} are a deprecated alternative to macros in writing
3770 conditionals to test what sort of computer or system the compiled
3771 program will run on. Assertions are usually predefined, but you can
3772 define them with preprocessing directives or command-line options.
3774 Assertions were intended to provide a more systematic way to describe
3775 the compiler's target system. However, in practice they are just as
3776 unpredictable as the system-specific predefined macros. In addition, they
3777 are not part of any standard, and only a few compilers support them.
3778 Therefore, the use of assertions is @strong{less} portable than the use
3779 of system-specific predefined macros. We recommend you do not use them at
3783 An assertion looks like this:
3786 #@var{predicate} (@var{answer})
3790 @var{predicate} must be a single identifier. @var{answer} can be any
3791 sequence of tokens; all characters are significant except for leading
3792 and trailing whitespace, and differences in internal whitespace
3793 sequences are ignored. (This is similar to the rules governing macro
3794 redefinition.) Thus, @code{(x + y)} is different from @code{(x+y)} but
3795 equivalent to @code{@w{( x + y )}}. Parentheses do not nest inside an
3798 @cindex testing predicates
3799 To test an assertion, you write it in an @samp{#if}. For example, this
3800 conditional succeeds if either @code{vax} or @code{ns16000} has been
3801 asserted as an answer for @code{machine}.
3804 #if #machine (vax) || #machine (ns16000)
3808 You can test whether @emph{any} answer is asserted for a predicate by
3809 omitting the answer in the conditional:
3816 Assertions are made with the @samp{#assert} directive. Its sole
3817 argument is the assertion to make, without the leading @samp{#} that
3818 identifies assertions in conditionals.
3821 #assert @var{predicate} (@var{answer})
3825 You may make several assertions with the same predicate and different
3826 answers. Subsequent assertions do not override previous ones for the
3827 same predicate. All the answers for any given predicate are
3828 simultaneously true.
3830 @cindex assertions, cancelling
3832 Assertions can be cancelled with the @samp{#unassert} directive. It
3833 has the same syntax as @samp{#assert}. In that form it cancels only the
3834 answer which was specified on the @samp{#unassert} line; other answers
3835 for that predicate remain true. You can cancel an entire predicate by
3836 leaving out the answer:
3839 #unassert @var{predicate}
3843 In either form, if no such assertion has been made, @samp{#unassert} has
3846 You can also make or cancel assertions using command line options.
3849 @node Obsolete once-only headers
3850 @subsection Obsolete once-only headers
3852 GNU CPP supports two more ways of indicating that a header file should be
3853 read only once. Neither one is as portable as a wrapper @samp{#ifndef},
3854 and we recommend you do not use them in new programs.
3857 In the Objective-C language, there is a variant of @samp{#include}
3858 called @samp{#import} which includes a file, but does so at most once.
3859 If you use @samp{#import} instead of @samp{#include}, then you don't
3860 need the conditionals inside the header file to prevent multiple
3861 inclusion of the contents. GCC permits the use of @samp{#import} in C
3862 and C++ as well as Objective-C@. However, it is not in standard C or C++
3863 and should therefore not be used by portable programs.
3865 @samp{#import} is not a well designed feature. It requires the users of
3866 a header file to know that it should only be included once. It is much
3867 better for the header file's implementor to write the file so that users
3868 don't need to know this. Using a wrapper @samp{#ifndef} accomplishes
3871 In the present implementation, a single use of @samp{#import} will
3872 prevent the file from ever being read again, by either @samp{#import} or
3873 @samp{#include}. You should not rely on this; do not use both
3874 @samp{#import} and @samp{#include} to refer to the same header file.
3876 Another way to prevent a header file from being included more than once
3877 is with the @samp{#pragma once} directive. If @samp{#pragma once} is
3878 seen when scanning a header file, that file will never be read again, no
3881 @samp{#pragma once} does not have the problems that @samp{#import} does,
3882 but it is not recognized by all preprocessors, so you cannot rely on it
3883 in a portable program.
3885 @node Differences from previous versions
3886 @section Differences from previous versions
3887 @cindex differences from previous versions
3889 This section details behavior which has changed from previous versions
3890 of GNU CPP@. We do not plan to change it again in the near future, but
3891 we do not promise not to, either.
3893 The ``previous versions'' discussed here are 2.95 and before. The
3894 behavior of GCC 3.0 is mostly the same as the behavior of the widely
3895 used 2.96 and 2.97 development snapshots. Where there are differences,
3896 they generally represent bugs in the snapshots.
3900 @item Order of evaluation of @samp{#} and @samp{##} operators
3902 The standard does not specify the order of evaluation of a chain of
3903 @samp{##} operators, nor whether @samp{#} is evaluated before, after, or
3904 at the same time as @samp{##}. You should therefore not write any code
3905 which depends on any specific ordering. It is possible to guarantee an
3906 ordering, if you need one, by suitable use of nested macros.
3908 An example of where this might matter is pasting the arguments @samp{1},
3909 @samp{e} and @samp{-2}. This would be fine for left-to-right pasting,
3910 but right-to-left pasting would produce an invalid token @samp{e-2}.
3912 GCC 3.0 evaluates @samp{#} and @samp{##} at the same time and strictly
3913 left to right. Older versions evaluated all @samp{#} operators first,
3914 then all @samp{##} operators, in an unreliable order.
3916 @item The form of whitespace betwen tokens in preprocessor output
3918 @xref{Preprocessor Output}, for the current textual format. This is
3919 also the format used by stringification. Normally, the preprocessor
3920 communicates tokens directly to the compiler's parser, and whitespace
3921 does not come up at all.
3923 Older versions of GCC preserved all whitespace provided by the user and
3924 inserted lots more whitespace of their own, because they could not
3925 accurately predict when extra spaces were needed to prevent accidental
3928 @item Optional argument when invoking rest argument macros
3930 As an extension, GCC permits you to omit the variable arguments entirely
3931 when you use a variable argument macro. This is forbidden by the 1999 C
3932 standard, and will provoke a pedantic warning with GCC 3.0. Previous
3933 versions accepted it silently.
3935 @item @samp{##} swallowing preceding text in rest argument macros
3937 Formerly, in a macro expansion, if @samp{##} appeared before a variable
3938 arguments parameter, and the set of tokens specified for that argument
3939 in the macro invocation was empty, previous versions of GNU CPP would
3940 back up and remove the preceding sequence of non-whitespace characters
3941 (@strong{not} the preceding token). This extension is in direct
3942 conflict with the 1999 C standard and has been drastically pared back.
3944 In the current version of the preprocessor, if @samp{##} appears between
3945 a comma and a variable arguments parameter, and the variable argument is
3946 omitted entirely, the comma will be removed from the expansion. If the
3947 variable argument is empty, or the token before @samp{##} is not a
3948 comma, then @samp{##} behaves as a normal token paste.
3950 @item @samp{#line} and @samp{#include}
3952 The @samp{#line} directive used to change GCC's notion of the
3953 ``directory containing the current file,'' used by @samp{#include} with
3954 a double-quoted header file name. In 3.0 and later, it does not.
3955 @xref{Line Control}, for further explanation.
3957 @item Syntax of @samp{#line}
3959 In GCC 2.95 and previous, the string constant argument to @samp{#line}
3960 was treated the same way as the argument to @samp{#include}: backslash
3961 escapes were not honored, and the string ended at the second @samp{"}.
3962 This is not compliant with the C standard. In GCC 3.0, an attempt was
3963 made to correct the behavior, so that the string was treated as a real
3964 string constant, but it turned out to be buggy. In 3.1, the bugs have
3965 been fixed. (We are not fixing the bugs in 3.0 because they affect
3966 relatively few people and the fix is quite invasive.)
3973 @cindex command line
3975 Most often when you use the C preprocessor you will not have to invoke it
3976 explicitly: the C compiler will do so automatically. However, the
3977 preprocessor is sometimes useful on its own. All the options listed
3978 here are also acceptable to the C compiler and have the same meaning,
3979 except that the C compiler has different rules for specifying the output
3982 @strong{Note:} Whether you use the preprocessor by way of @command{gcc}
3983 or @command{cpp}, the @dfn{compiler driver} is run first. This
3984 program's purpose is to translate your command into invocations of the
3985 programs that do the actual work. Their command line interfaces are
3986 similar but not identical to the documented interface, and may change
3990 @c man begin SYNOPSIS
3991 cpp [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
3992 [@option{-I}@var{dir}@dots{}] [@option{-W}@var{warn}@dots{}]
3993 [@option{-M}|@option{-MM}] [@option{-MG}] [@option{-MF} @var{filename}]
3994 [@option{-MP}] [@option{-MQ} @var{target}@dots{}] [@option{-MT} @var{target}@dots{}]
3995 [@option{-x} @var{language}] [@option{-std=}@var{standard}]
3996 @var{infile} @var{outfile}
3998 Only the most useful options are listed here; see below for the remainder.
4000 @c man begin SEEALSO
4001 gpl(7), gfdl(7), fsf-funding(7),
4002 gcc(1), as(1), ld(1), and the Info entries for @file{cpp}, @file{gcc}, and
4007 @c man begin OPTIONS
4008 The C preprocessor expects two file names as arguments, @var{infile} and
4009 @var{outfile}. The preprocessor reads @var{infile} together with any
4010 other files it specifies with @samp{#include}. All the output generated
4011 by the combined input files is written in @var{outfile}.
4013 Either @var{infile} or @var{outfile} may be @option{-}, which as
4014 @var{infile} means to read from standard input and as @var{outfile}
4015 means to write to standard output. Also, if either file is omitted, it
4016 means the same as if @option{-} had been specified for that file.
4018 Unless otherwise noted, or the option ends in @samp{=}, all options
4019 which take an argument may have that argument appear either immediately
4020 after the option, or with a space between option and argument:
4021 @option{-Ifoo} and @option{-I foo} have the same effect.
4023 @cindex grouping options
4024 @cindex options, grouping
4025 Many options have multi-letter names; therefore multiple single-letter
4026 options may @emph{not} be grouped: @option{-dM} is very different from
4030 @include cppopts.texi
4033 @node Environment Variables
4034 @chapter Environment Variables
4035 @cindex environment variables
4036 @c man begin ENVIRONMENT
4038 This section describes the environment variables that affect how CPP
4039 operates. You can use them to specify directories or prefixes to use
4040 when searching for include files, or to control dependency output.
4042 Note that you can also specify places to search using options such as
4043 @option{-I}, and control dependency output with options like
4044 @option{-M} (@pxref{Invocation}). These take precedence over
4045 environment variables, which in turn take precedence over the
4046 configuration of GCC@.
4048 @include cppenv.texi
4055 @node Index of Directives
4056 @unnumbered Index of Directives
4060 @unnumbered Option Index
4062 CPP's command line options and environment variables are indexed here
4063 without any initial @samp{-} or @samp{--}.
4068 @unnumbered Concept Index