3 @settitle The C Preprocessor
9 @include gcc-common.texi
12 @c man begin COPYRIGHT
13 Copyright @copyright{} 1987-2016 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 1.3 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
27 @c man begin COPYRIGHT
28 This manual contains no Invariant Sections. The Front-Cover Texts are
29 (a) (see below), and the Back-Cover Texts are (b) (see below).
31 (a) The FSF's Front-Cover Text is:
35 (b) The FSF's Back-Cover Text is:
37 You have freedom to copy and modify this GNU Manual, like GNU
38 software. Copies published by the Free Software Foundation raise
39 funds for GNU development.
43 @c Create a separate index for command line options.
47 @c Used in cppopts.texi and cppenv.texi.
51 @dircategory Software development
53 * Cpp: (cpp). The GNU C preprocessor.
58 @title The C Preprocessor
60 @author Richard M. Stallman, Zachary Weinberg
62 @c There is a fill at the bottom of the page, so we need a filll to
64 @vskip 0pt plus 1filll
73 The C preprocessor implements the macro language used to transform C,
74 C++, and Objective-C programs before they are compiled. It can also be
86 * Preprocessor Output::
88 * Implementation Details::
90 * Environment Variables::
91 * GNU Free Documentation License::
92 * Index of Directives::
97 --- The Detailed Node Listing ---
102 * Initial processing::
104 * The preprocessing language::
109 * Include Operation::
111 * Once-Only Headers::
112 * Alternatives to Wrapper #ifndef::
113 * Computed Includes::
119 * Object-like Macros::
120 * Function-like Macros::
125 * Predefined Macros::
126 * Undefining and Redefining Macros::
127 * Directives Within Macro Arguments::
132 * Standard Predefined Macros::
133 * Common Predefined Macros::
134 * System-specific Predefined Macros::
135 * C++ Named Operators::
140 * Operator Precedence Problems::
141 * Swallowing the Semicolon::
142 * Duplication of Side Effects::
143 * Self-Referential Macros::
145 * Newlines in Arguments::
150 * Conditional Syntax::
161 Implementation Details
163 * Implementation-defined behavior::
164 * Implementation limits::
165 * Obsolete Features::
166 * Differences from previous versions::
170 * Obsolete Features::
180 @c man begin DESCRIPTION
181 The C preprocessor, often known as @dfn{cpp}, is a @dfn{macro processor}
182 that is used automatically by the C compiler to transform your program
183 before compilation. It is called a macro processor because it allows
184 you to define @dfn{macros}, which are brief abbreviations for longer
187 The C preprocessor is intended to be used only with C, C++, and
188 Objective-C source code. In the past, it has been abused as a general
189 text processor. It will choke on input which does not obey C's lexical
190 rules. For example, apostrophes will be interpreted as the beginning of
191 character constants, and cause errors. Also, you cannot rely on it
192 preserving characteristics of the input which are not significant to
193 C-family languages. If a Makefile is preprocessed, all the hard tabs
194 will be removed, and the Makefile will not work.
196 Having said that, you can often get away with using cpp on things which
197 are not C@. Other Algol-ish programming languages are often safe
198 (Pascal, Ada, etc.) So is assembly, with caution. @option{-traditional-cpp}
199 mode preserves more white space, and is otherwise more permissive. Many
200 of the problems can be avoided by writing C or C++ style comments
201 instead of native language comments, and keeping macros simple.
203 Wherever possible, you should use a preprocessor geared to the language
204 you are writing in. Modern versions of the GNU assembler have macro
205 facilities. Most high level programming languages have their own
206 conditional compilation and inclusion mechanism. If all else fails,
207 try a true general text processor, such as GNU M4.
209 C preprocessors vary in some details. This manual discusses the GNU C
210 preprocessor, which provides a small superset of the features of ISO
211 Standard C@. In its default mode, the GNU C preprocessor does not do a
212 few things required by the standard. These are features which are
213 rarely, if ever, used, and may cause surprising changes to the meaning
214 of a program which does not expect them. To get strict ISO Standard C,
215 you should use the @option{-std=c90}, @option{-std=c99} or
216 @option{-std=c11} options, depending
217 on which version of the standard you want. To get all the mandatory
218 diagnostics, you must also use @option{-pedantic}. @xref{Invocation}.
220 This manual describes the behavior of the ISO preprocessor. To
221 minimize gratuitous differences, where the ISO preprocessor's
222 behavior does not conflict with traditional semantics, the
223 traditional preprocessor should behave the same way. The various
224 differences that do exist are detailed in the section @ref{Traditional
227 For clarity, unless noted otherwise, references to @samp{CPP} in this
228 manual refer to GNU CPP@.
233 * Initial processing::
235 * The preprocessing language::
239 @section Character sets
241 Source code character set processing in C and related languages is
242 rather complicated. The C standard discusses two character sets, but
243 there are really at least four.
245 The files input to CPP might be in any character set at all. CPP's
246 very first action, before it even looks for line boundaries, is to
247 convert the file into the character set it uses for internal
248 processing. That set is what the C standard calls the @dfn{source}
249 character set. It must be isomorphic with ISO 10646, also known as
250 Unicode. CPP uses the UTF-8 encoding of Unicode.
252 The character sets of the input files are specified using the
253 @option{-finput-charset=} option.
255 All preprocessing work (the subject of the rest of this manual) is
256 carried out in the source character set. If you request textual
257 output from the preprocessor with the @option{-E} option, it will be
260 After preprocessing is complete, string and character constants are
261 converted again, into the @dfn{execution} character set. This
262 character set is under control of the user; the default is UTF-8,
263 matching the source character set. Wide string and character
264 constants have their own character set, which is not called out
265 specifically in the standard. Again, it is under control of the user.
266 The default is UTF-16 or UTF-32, whichever fits in the target's
267 @code{wchar_t} type, in the target machine's byte
268 order.@footnote{UTF-16 does not meet the requirements of the C
269 standard for a wide character set, but the choice of 16-bit
270 @code{wchar_t} is enshrined in some system ABIs so we cannot fix
271 this.} Octal and hexadecimal escape sequences do not undergo
272 conversion; @t{'\x12'} has the value 0x12 regardless of the currently
273 selected execution character set. All other escapes are replaced by
274 the character in the source character set that they represent, then
275 converted to the execution character set, just like unescaped
278 In identifiers, characters outside the ASCII range can only be
279 specified with the @samp{\u} and @samp{\U} escapes, not used
280 directly. If strict ISO C90 conformance is specified with an option
281 such as @option{-std=c90}, or @option{-fno-extended-identifiers} is
282 used, then those escapes are not permitted in identifiers.
284 @node Initial processing
285 @section Initial processing
287 The preprocessor performs a series of textual transformations on its
288 input. These happen before all other processing. Conceptually, they
289 happen in a rigid order, and the entire file is run through each
290 transformation before the next one begins. CPP actually does them
291 all at once, for performance reasons. These transformations correspond
292 roughly to the first three ``phases of translation'' described in the C
298 The input file is read into memory and broken into lines.
300 Different systems use different conventions to indicate the end of a
301 line. GCC accepts the ASCII control sequences @kbd{LF}, @kbd{@w{CR
302 LF}} and @kbd{CR} as end-of-line markers. These are the canonical
303 sequences used by Unix, DOS and VMS, and the classic Mac OS (before
304 OSX) respectively. You may therefore safely copy source code written
305 on any of those systems to a different one and use it without
306 conversion. (GCC may lose track of the current line number if a file
307 doesn't consistently use one convention, as sometimes happens when it
308 is edited on computers with different conventions that share a network
311 If the last line of any input file lacks an end-of-line marker, the end
312 of the file is considered to implicitly supply one. The C standard says
313 that this condition provokes undefined behavior, so GCC will emit a
318 @anchor{trigraphs}If trigraphs are enabled, they are replaced by their
319 corresponding single characters. By default GCC ignores trigraphs,
320 but if you request a strictly conforming mode with the @option{-std}
321 option, or you specify the @option{-trigraphs} option, then it
324 These are nine three-character sequences, all starting with @samp{??},
325 that are defined by ISO C to stand for single characters. They permit
326 obsolete systems that lack some of C's punctuation to use C@. For
327 example, @samp{??/} stands for @samp{\}, so @t{'??/n'} is a character
328 constant for a newline.
330 Trigraphs are not popular and many compilers implement them
331 incorrectly. Portable code should not rely on trigraphs being either
332 converted or ignored. With @option{-Wtrigraphs} GCC will warn you
333 when a trigraph may change the meaning of your program if it were
334 converted. @xref{Wtrigraphs}.
336 In a string constant, you can prevent a sequence of question marks
337 from being confused with a trigraph by inserting a backslash between
338 the question marks, or by separating the string literal at the
339 trigraph and making use of string literal concatenation. @t{"(??\?)"}
340 is the string @samp{(???)}, not @samp{(?]}. Traditional C compilers
341 do not recognize these idioms.
343 The nine trigraphs and their replacements are
346 Trigraph: ??( ??) ??< ??> ??= ??/ ??' ??! ??-
347 Replacement: [ ] @{ @} # \ ^ | ~
351 @cindex continued lines
352 @cindex backslash-newline
353 Continued lines are merged into one long line.
355 A continued line is a line which ends with a backslash, @samp{\}. The
356 backslash is removed and the following line is joined with the current
357 one. No space is inserted, so you may split a line anywhere, even in
358 the middle of a word. (It is generally more readable to split lines
359 only at white space.)
361 The trailing backslash on a continued line is commonly referred to as a
362 @dfn{backslash-newline}.
364 If there is white space between a backslash and the end of a line, that
365 is still a continued line. However, as this is usually the result of an
366 editing mistake, and many compilers will not accept it as a continued
367 line, GCC will warn you about it.
371 @cindex line comments
372 @cindex block comments
373 All comments are replaced with single spaces.
375 There are two kinds of comments. @dfn{Block comments} begin with
376 @samp{/*} and continue until the next @samp{*/}. Block comments do not
380 /* @r{this is} /* @r{one comment} */ @r{text outside comment}
383 @dfn{Line comments} begin with @samp{//} and continue to the end of the
384 current line. Line comments do not nest either, but it does not matter,
385 because they would end in the same place anyway.
388 // @r{this is} // @r{one comment}
389 @r{text outside comment}
393 It is safe to put line comments inside block comments, or vice versa.
398 // @r{contains line comment}
400 */ @r{outside comment}
402 // @r{line comment} /* @r{contains block comment} */
406 But beware of commenting out one end of a block comment with a line
411 // @r{l.c.} /* @r{block comment begins}
412 @r{oops! this isn't a comment anymore} */
416 Comments are not recognized within string literals.
417 @t{@w{"/* blah */"}} is the string constant @samp{@w{/* blah */}}, not
420 Line comments are not in the 1989 edition of the C standard, but they
421 are recognized by GCC as an extension. In C++ and in the 1999 edition
422 of the C standard, they are an official part of the language.
424 Since these transformations happen before all other processing, you can
425 split a line mechanically with backslash-newline anywhere. You can
426 comment out the end of a line. You can continue a line comment onto the
427 next line with backslash-newline. You can even split @samp{/*},
428 @samp{*/}, and @samp{//} onto multiple lines with backslash-newline.
444 is equivalent to @code{@w{#define FOO 1020}}. All these tricks are
445 extremely confusing and should not be used in code intended to be
448 There is no way to prevent a backslash at the end of a line from being
449 interpreted as a backslash-newline. This cannot affect any correct
453 @section Tokenization
456 @cindex preprocessing tokens
457 After the textual transformations are finished, the input file is
458 converted into a sequence of @dfn{preprocessing tokens}. These mostly
459 correspond to the syntactic tokens used by the C compiler, but there are
460 a few differences. White space separates tokens; it is not itself a
461 token of any kind. Tokens do not have to be separated by white space,
462 but it is often necessary to avoid ambiguities.
464 When faced with a sequence of characters that has more than one possible
465 tokenization, the preprocessor is greedy. It always makes each token,
466 starting from the left, as big as possible before moving on to the next
467 token. For instance, @code{a+++++b} is interpreted as
468 @code{@w{a ++ ++ + b}}, not as @code{@w{a ++ + ++ b}}, even though the
469 latter tokenization could be part of a valid C program and the former
472 Once the input file is broken into tokens, the token boundaries never
473 change, except when the @samp{##} preprocessing operator is used to paste
474 tokens together. @xref{Concatenation}. For example,
486 The compiler does not re-tokenize the preprocessor's output. Each
487 preprocessing token becomes one compiler token.
490 Preprocessing tokens fall into five broad classes: identifiers,
491 preprocessing numbers, string literals, punctuators, and other. An
492 @dfn{identifier} is the same as an identifier in C: any sequence of
493 letters, digits, or underscores, which begins with a letter or
494 underscore. Keywords of C have no significance to the preprocessor;
495 they are ordinary identifiers. You can define a macro whose name is a
496 keyword, for instance. The only identifier which can be considered a
497 preprocessing keyword is @code{defined}. @xref{Defined}.
499 This is mostly true of other languages which use the C preprocessor.
500 However, a few of the keywords of C++ are significant even in the
501 preprocessor. @xref{C++ Named Operators}.
503 In the 1999 C standard, identifiers may contain letters which are not
504 part of the ``basic source character set'', at the implementation's
505 discretion (such as accented Latin letters, Greek letters, or Chinese
506 ideograms). This may be done with an extended character set, or the
507 @samp{\u} and @samp{\U} escape sequences. GCC only accepts such
508 characters in the @samp{\u} and @samp{\U} forms.
510 As an extension, GCC treats @samp{$} as a letter. This is for
511 compatibility with some systems, such as VMS, where @samp{$} is commonly
512 used in system-defined function and object names. @samp{$} is not a
513 letter in strictly conforming mode, or if you specify the @option{-$}
514 option. @xref{Invocation}.
517 @cindex preprocessing numbers
518 A @dfn{preprocessing number} has a rather bizarre definition. The
519 category includes all the normal integer and floating point constants
520 one expects of C, but also a number of other things one might not
521 initially recognize as a number. Formally, preprocessing numbers begin
522 with an optional period, a required decimal digit, and then continue
523 with any sequence of letters, digits, underscores, periods, and
524 exponents. Exponents are the two-character sequences @samp{e+},
525 @samp{e-}, @samp{E+}, @samp{E-}, @samp{p+}, @samp{p-}, @samp{P+}, and
526 @samp{P-}. (The exponents that begin with @samp{p} or @samp{P} are new
527 to C99. They are used for hexadecimal floating-point constants.)
529 The purpose of this unusual definition is to isolate the preprocessor
530 from the full complexity of numeric constants. It does not have to
531 distinguish between lexically valid and invalid floating-point numbers,
532 which is complicated. The definition also permits you to split an
533 identifier at any position and get exactly two tokens, which can then be
534 pasted back together with the @samp{##} operator.
536 It's possible for preprocessing numbers to cause programs to be
537 misinterpreted. For example, @code{0xE+12} is a preprocessing number
538 which does not translate to any valid numeric constant, therefore a
539 syntax error. It does not mean @code{@w{0xE + 12}}, which is what you
542 @cindex string literals
543 @cindex string constants
544 @cindex character constants
545 @cindex header file names
546 @c the @: prevents makeinfo from turning '' into ".
547 @dfn{String literals} are string constants, character constants, and
548 header file names (the argument of @samp{#include}).@footnote{The C
549 standard uses the term @dfn{string literal} to refer only to what we are
550 calling @dfn{string constants}.} String constants and character
551 constants are straightforward: @t{"@dots{}"} or @t{'@dots{}'}. In
552 either case embedded quotes should be escaped with a backslash:
553 @t{'\'@:'} is the character constant for @samp{'}. There is no limit on
554 the length of a character constant, but the value of a character
555 constant that contains more than one character is
556 implementation-defined. @xref{Implementation Details}.
558 Header file names either look like string constants, @t{"@dots{}"}, or are
559 written with angle brackets instead, @t{<@dots{}>}. In either case,
560 backslash is an ordinary character. There is no way to escape the
561 closing quote or angle bracket. The preprocessor looks for the header
562 file in different places depending on which form you use. @xref{Include
565 No string literal may extend past the end of a line. Older versions
566 of GCC accepted multi-line string constants. You may use continued
567 lines instead, or string constant concatenation. @xref{Differences
568 from previous versions}.
572 @cindex alternative tokens
573 @dfn{Punctuators} are all the usual bits of punctuation which are
574 meaningful to C and C++. All but three of the punctuation characters in
575 ASCII are C punctuators. The exceptions are @samp{@@}, @samp{$}, and
576 @samp{`}. In addition, all the two- and three-character operators are
577 punctuators. There are also six @dfn{digraphs}, which the C++ standard
578 calls @dfn{alternative tokens}, which are merely alternate ways to spell
579 other punctuators. This is a second attempt to work around missing
580 punctuation in obsolete systems. It has no negative side effects,
581 unlike trigraphs, but does not cover as much ground. The digraphs and
582 their corresponding normal punctuators are:
585 Digraph: <% %> <: :> %: %:%:
586 Punctuator: @{ @} [ ] # ##
590 Any other single character is considered ``other''. It is passed on to
591 the preprocessor's output unmolested. The C compiler will almost
592 certainly reject source code containing ``other'' tokens. In ASCII, the
593 only other characters are @samp{@@}, @samp{$}, @samp{`}, and control
594 characters other than NUL (all bits zero). (Note that @samp{$} is
595 normally considered a letter.) All characters with the high bit set
596 (numeric range 0x7F--0xFF) are also ``other'' in the present
597 implementation. This will change when proper support for international
598 character sets is added to GCC@.
600 NUL is a special case because of the high probability that its
601 appearance is accidental, and because it may be invisible to the user
602 (many terminals do not display NUL at all). Within comments, NULs are
603 silently ignored, just as any other character would be. In running
604 text, NUL is considered white space. For example, these two directives
605 have the same meaning.
613 (where @samp{^@@} is ASCII NUL)@. Within string or character constants,
614 NULs are preserved. In the latter two cases the preprocessor emits a
617 @node The preprocessing language
618 @section The preprocessing language
620 @cindex preprocessing directives
621 @cindex directive line
622 @cindex directive name
624 After tokenization, the stream of tokens may simply be passed straight
625 to the compiler's parser. However, if it contains any operations in the
626 @dfn{preprocessing language}, it will be transformed first. This stage
627 corresponds roughly to the standard's ``translation phase 4'' and is
628 what most people think of as the preprocessor's job.
630 The preprocessing language consists of @dfn{directives} to be executed
631 and @dfn{macros} to be expanded. Its primary capabilities are:
635 Inclusion of header files. These are files of declarations that can be
636 substituted into your program.
639 Macro expansion. You can define @dfn{macros}, which are abbreviations
640 for arbitrary fragments of C code. The preprocessor will replace the
641 macros with their definitions throughout the program. Some macros are
642 automatically defined for you.
645 Conditional compilation. You can include or exclude parts of the
646 program according to various conditions.
649 Line control. If you use a program to combine or rearrange source files
650 into an intermediate file which is then compiled, you can use line
651 control to inform the compiler where each source line originally came
655 Diagnostics. You can detect problems at compile time and issue errors
659 There are a few more, less useful, features.
661 Except for expansion of predefined macros, all these operations are
662 triggered with @dfn{preprocessing directives}. Preprocessing directives
663 are lines in your program that start with @samp{#}. Whitespace is
664 allowed before and after the @samp{#}. The @samp{#} is followed by an
665 identifier, the @dfn{directive name}. It specifies the operation to
666 perform. Directives are commonly referred to as @samp{#@var{name}}
667 where @var{name} is the directive name. For example, @samp{#define} is
668 the directive that defines a macro.
670 The @samp{#} which begins a directive cannot come from a macro
671 expansion. Also, the directive name is not macro expanded. Thus, if
672 @code{foo} is defined as a macro expanding to @code{define}, that does
673 not make @samp{#foo} a valid preprocessing directive.
675 The set of valid directive names is fixed. Programs cannot define new
676 preprocessing directives.
678 Some directives require arguments; these make up the rest of the
679 directive line and must be separated from the directive name by
680 whitespace. For example, @samp{#define} must be followed by a macro
681 name and the intended expansion of the macro.
683 A preprocessing directive cannot cover more than one line. The line
684 may, however, be continued with backslash-newline, or by a block comment
685 which extends past the end of the line. In either case, when the
686 directive is processed, the continuations have already been merged with
687 the first line to make one long line.
690 @chapter Header Files
693 A header file is a file containing C declarations and macro definitions
694 (@pxref{Macros}) to be shared between several source files. You request
695 the use of a header file in your program by @dfn{including} it, with the
696 C preprocessing directive @samp{#include}.
698 Header files serve two purposes.
702 @cindex system header files
703 System header files declare the interfaces to parts of the operating
704 system. You include them in your program to supply the definitions and
705 declarations you need to invoke system calls and libraries.
708 Your own header files contain declarations for interfaces between the
709 source files of your program. Each time you have a group of related
710 declarations and macro definitions all or most of which are needed in
711 several different source files, it is a good idea to create a header
715 Including a header file produces the same results as copying the header
716 file into each source file that needs it. Such copying would be
717 time-consuming and error-prone. With a header file, the related
718 declarations appear in only one place. If they need to be changed, they
719 can be changed in one place, and programs that include the header file
720 will automatically use the new version when next recompiled. The header
721 file eliminates the labor of finding and changing all the copies as well
722 as the risk that a failure to find one copy will result in
723 inconsistencies within a program.
725 In C, the usual convention is to give header files names that end with
726 @file{.h}. It is most portable to use only letters, digits, dashes, and
727 underscores in header file names, and at most one dot.
731 * Include Operation::
733 * Once-Only Headers::
734 * Alternatives to Wrapper #ifndef::
735 * Computed Includes::
741 @section Include Syntax
744 Both user and system header files are included using the preprocessing
745 directive @samp{#include}. It has two variants:
748 @item #include <@var{file}>
749 This variant is used for system header files. It searches for a file
750 named @var{file} in a standard list of system directories. You can prepend
751 directories to this list with the @option{-I} option (@pxref{Invocation}).
753 @item #include "@var{file}"
754 This variant is used for header files of your own program. It
755 searches for a file named @var{file} first in the directory containing
756 the current file, then in the quote directories and then the same
757 directories used for @code{<@var{file}>}. You can prepend directories
758 to the list of quote directories with the @option{-iquote} option.
761 The argument of @samp{#include}, whether delimited with quote marks or
762 angle brackets, behaves like a string constant in that comments are not
763 recognized, and macro names are not expanded. Thus, @code{@w{#include
764 <x/*y>}} specifies inclusion of a system header file named @file{x/*y}.
766 However, if backslashes occur within @var{file}, they are considered
767 ordinary text characters, not escape characters. None of the character
768 escape sequences appropriate to string constants in C are processed.
769 Thus, @code{@w{#include "x\n\\y"}} specifies a filename containing three
770 backslashes. (Some systems interpret @samp{\} as a pathname separator.
771 All of these also interpret @samp{/} the same way. It is most portable
772 to use only @samp{/}.)
774 It is an error if there is anything (other than comments) on the line
777 @node Include Operation
778 @section Include Operation
780 The @samp{#include} directive works by directing the C preprocessor to
781 scan the specified file as input before continuing with the rest of the
782 current file. The output from the preprocessor contains the output
783 already generated, followed by the output resulting from the included
784 file, followed by the output that comes from the text after the
785 @samp{#include} directive. For example, if you have a header file
786 @file{header.h} as follows,
793 and a main program called @file{program.c} that uses the header file,
808 the compiler will see the same token stream as it would if
809 @file{program.c} read
822 Included files are not limited to declarations and macro definitions;
823 those are merely the typical uses. Any fragment of a C program can be
824 included from another file. The include file could even contain the
825 beginning of a statement that is concluded in the containing file, or
826 the end of a statement that was started in the including file. However,
827 an included file must consist of complete tokens. Comments and string
828 literals which have not been closed by the end of an included file are
829 invalid. For error recovery, they are considered to end at the end of
832 To avoid confusion, it is best if header files contain only complete
833 syntactic units---function declarations or definitions, type
836 The line following the @samp{#include} directive is always treated as a
837 separate line by the C preprocessor, even if the included file lacks a
843 GCC looks in several different places for headers. On a normal Unix
844 system, if you do not instruct it otherwise, it will look for headers
845 requested with @code{@w{#include <@var{file}>}} in:
849 @var{libdir}/gcc/@var{target}/@var{version}/include
850 /usr/@var{target}/include
854 For C++ programs, it will also look in
855 @file{@var{libdir}/../include/c++/@var{version}},
856 first. In the above, @var{target} is the canonical name of the system
857 GCC was configured to compile code for; often but not always the same as
858 the canonical name of the system it runs on. @var{version} is the
859 version of GCC in use.
861 You can add to this list with the @option{-I@var{dir}} command-line
862 option. All the directories named by @option{-I} are searched, in
863 left-to-right order, @emph{before} the default directories. The only
864 exception is when @file{dir} is already searched by default. In
865 this case, the option is ignored and the search order for system
866 directories remains unchanged.
868 Duplicate directories are removed from the quote and bracket search
869 chains before the two chains are merged to make the final search chain.
870 Thus, it is possible for a directory to occur twice in the final search
871 chain if it was specified in both the quote and bracket chains.
873 You can prevent GCC from searching any of the default directories with
874 the @option{-nostdinc} option. This is useful when you are compiling an
875 operating system kernel or some other program that does not use the
876 standard C library facilities, or the standard C library itself.
877 @option{-I} options are not ignored as described above when
878 @option{-nostdinc} is in effect.
880 GCC looks for headers requested with @code{@w{#include "@var{file}"}}
881 first in the directory containing the current file, then in the
882 directories as specified by @option{-iquote} options, then in the same
883 places it would have looked for a header requested with angle
884 brackets. For example, if @file{/usr/include/sys/stat.h} contains
885 @code{@w{#include "types.h"}}, GCC looks for @file{types.h} first in
886 @file{/usr/include/sys}, then in its usual search path.
888 @samp{#line} (@pxref{Line Control}) does not change GCC's idea of the
889 directory containing the current file.
891 You may put @option{-I-} at any point in your list of @option{-I} options.
892 This has two effects. First, directories appearing before the
893 @option{-I-} in the list are searched only for headers requested with
894 quote marks. Directories after @option{-I-} are searched for all
895 headers. Second, the directory containing the current file is not
896 searched for anything, unless it happens to be one of the directories
897 named by an @option{-I} switch. @option{-I-} is deprecated, @option{-iquote}
898 should be used instead.
900 @option{-I. -I-} is not the same as no @option{-I} options at all, and does
901 not cause the same behavior for @samp{<>} includes that @samp{""}
902 includes get with no special options. @option{-I.} searches the
903 compiler's current working directory for header files. That may or may
904 not be the same as the directory containing the current file.
906 If you need to look for headers in a directory named @file{-}, write
909 There are several more ways to adjust the header search path. They are
910 generally less useful. @xref{Invocation}.
912 @node Once-Only Headers
913 @section Once-Only Headers
914 @cindex repeated inclusion
915 @cindex including just once
916 @cindex wrapper @code{#ifndef}
918 If a header file happens to be included twice, the compiler will process
919 its contents twice. This is very likely to cause an error, e.g.@: when the
920 compiler sees the same structure definition twice. Even if it does not,
921 it will certainly waste time.
923 The standard way to prevent this is to enclose the entire real contents
924 of the file in a conditional, like this:
929 #ifndef FILE_FOO_SEEN
930 #define FILE_FOO_SEEN
932 @var{the entire file}
934 #endif /* !FILE_FOO_SEEN */
938 This construct is commonly known as a @dfn{wrapper #ifndef}.
939 When the header is included again, the conditional will be false,
940 because @code{FILE_FOO_SEEN} is defined. The preprocessor will skip
941 over the entire contents of the file, and the compiler will not see it
944 CPP optimizes even further. It remembers when a header file has a
945 wrapper @samp{#ifndef}. If a subsequent @samp{#include} specifies that
946 header, and the macro in the @samp{#ifndef} is still defined, it does
947 not bother to rescan the file at all.
949 You can put comments outside the wrapper. They will not interfere with
952 @cindex controlling macro
954 The macro @code{FILE_FOO_SEEN} is called the @dfn{controlling macro} or
955 @dfn{guard macro}. In a user header file, the macro name should not
956 begin with @samp{_}. In a system header file, it should begin with
957 @samp{__} to avoid conflicts with user programs. In any kind of header
958 file, the macro name should contain the name of the file and some
959 additional text, to avoid conflicts with other header files.
961 @node Alternatives to Wrapper #ifndef
962 @section Alternatives to Wrapper #ifndef
964 CPP supports two more ways of indicating that a header file should be
965 read only once. Neither one is as portable as a wrapper @samp{#ifndef}
966 and we recommend you do not use them in new programs, with the caveat
967 that @samp{#import} is standard practice in Objective-C.
970 CPP supports a variant of @samp{#include} called @samp{#import} which
971 includes a file, but does so at most once. If you use @samp{#import}
972 instead of @samp{#include}, then you don't need the conditionals
973 inside the header file to prevent multiple inclusion of the contents.
974 @samp{#import} is standard in Objective-C, but is considered a
975 deprecated extension in C and C++.
977 @samp{#import} is not a well designed feature. It requires the users of
978 a header file to know that it should only be included once. It is much
979 better for the header file's implementor to write the file so that users
980 don't need to know this. Using a wrapper @samp{#ifndef} accomplishes
983 In the present implementation, a single use of @samp{#import} will
984 prevent the file from ever being read again, by either @samp{#import} or
985 @samp{#include}. You should not rely on this; do not use both
986 @samp{#import} and @samp{#include} to refer to the same header file.
988 Another way to prevent a header file from being included more than once
989 is with the @samp{#pragma once} directive. If @samp{#pragma once} is
990 seen when scanning a header file, that file will never be read again, no
993 @samp{#pragma once} does not have the problems that @samp{#import} does,
994 but it is not recognized by all preprocessors, so you cannot rely on it
995 in a portable program.
997 @node Computed Includes
998 @section Computed Includes
999 @cindex computed includes
1000 @cindex macros in include
1002 Sometimes it is necessary to select one of several different header
1003 files to be included into your program. They might specify
1004 configuration parameters to be used on different sorts of operating
1005 systems, for instance. You could do this with a series of conditionals,
1009 # include "system_1.h"
1011 # include "system_2.h"
1017 That rapidly becomes tedious. Instead, the preprocessor offers the
1018 ability to use a macro for the header name. This is called a
1019 @dfn{computed include}. Instead of writing a header name as the direct
1020 argument of @samp{#include}, you simply put a macro name there instead:
1023 #define SYSTEM_H "system_1.h"
1029 @code{SYSTEM_H} will be expanded, and the preprocessor will look for
1030 @file{system_1.h} as if the @samp{#include} had been written that way
1031 originally. @code{SYSTEM_H} could be defined by your Makefile with a
1034 You must be careful when you define the macro. @samp{#define} saves
1035 tokens, not text. The preprocessor has no way of knowing that the macro
1036 will be used as the argument of @samp{#include}, so it generates
1037 ordinary tokens, not a header name. This is unlikely to cause problems
1038 if you use double-quote includes, which are close enough to string
1039 constants. If you use angle brackets, however, you may have trouble.
1041 The syntax of a computed include is actually a bit more general than the
1042 above. If the first non-whitespace character after @samp{#include} is
1043 not @samp{"} or @samp{<}, then the entire line is macro-expanded
1044 like running text would be.
1046 If the line expands to a single string constant, the contents of that
1047 string constant are the file to be included. CPP does not re-examine the
1048 string for embedded quotes, but neither does it process backslash
1049 escapes in the string. Therefore
1052 #define HEADER "a\"b"
1057 looks for a file named @file{a\"b}. CPP searches for the file according
1058 to the rules for double-quoted includes.
1060 If the line expands to a token stream beginning with a @samp{<} token
1061 and including a @samp{>} token, then the tokens between the @samp{<} and
1062 the first @samp{>} are combined to form the filename to be included.
1063 Any whitespace between tokens is reduced to a single space; then any
1064 space after the initial @samp{<} is retained, but a trailing space
1065 before the closing @samp{>} is ignored. CPP searches for the file
1066 according to the rules for angle-bracket includes.
1068 In either case, if there are any tokens on the line after the file name,
1069 an error occurs and the directive is not processed. It is also an error
1070 if the result of expansion does not match either of the two expected
1073 These rules are implementation-defined behavior according to the C
1074 standard. To minimize the risk of different compilers interpreting your
1075 computed includes differently, we recommend you use only a single
1076 object-like macro which expands to a string constant. This will also
1077 minimize confusion for people reading your program.
1079 @node Wrapper Headers
1080 @section Wrapper Headers
1081 @cindex wrapper headers
1082 @cindex overriding a header file
1083 @findex #include_next
1085 Sometimes it is necessary to adjust the contents of a system-provided
1086 header file without editing it directly. GCC's @command{fixincludes}
1087 operation does this, for example. One way to do that would be to create
1088 a new header file with the same name and insert it in the search path
1089 before the original header. That works fine as long as you're willing
1090 to replace the old header entirely. But what if you want to refer to
1091 the old header from the new one?
1093 You cannot simply include the old header with @samp{#include}. That
1094 will start from the beginning, and find your new header again. If your
1095 header is not protected from multiple inclusion (@pxref{Once-Only
1096 Headers}), it will recurse infinitely and cause a fatal error.
1098 You could include the old header with an absolute pathname:
1100 #include "/usr/include/old-header.h"
1103 This works, but is not clean; should the system headers ever move, you
1104 would have to edit the new headers to match.
1106 There is no way to solve this problem within the C standard, but you can
1107 use the GNU extension @samp{#include_next}. It means, ``Include the
1108 @emph{next} file with this name''. This directive works like
1109 @samp{#include} except in searching for the specified file: it starts
1110 searching the list of header file directories @emph{after} the directory
1111 in which the current file was found.
1113 Suppose you specify @option{-I /usr/local/include}, and the list of
1114 directories to search also includes @file{/usr/include}; and suppose
1115 both directories contain @file{signal.h}. Ordinary @code{@w{#include
1116 <signal.h>}} finds the file under @file{/usr/local/include}. If that
1117 file contains @code{@w{#include_next <signal.h>}}, it starts searching
1118 after that directory, and finds the file in @file{/usr/include}.
1120 @samp{#include_next} does not distinguish between @code{<@var{file}>}
1121 and @code{"@var{file}"} inclusion, nor does it check that the file you
1122 specify has the same name as the current file. It simply looks for the
1123 file named, starting with the directory in the search path after the one
1124 where the current file was found.
1126 The use of @samp{#include_next} can lead to great confusion. We
1127 recommend it be used only when there is no other alternative. In
1128 particular, it should not be used in the headers belonging to a specific
1129 program; it should be used only to make global corrections along the
1130 lines of @command{fixincludes}.
1132 @node System Headers
1133 @section System Headers
1134 @cindex system header files
1136 The header files declaring interfaces to the operating system and
1137 runtime libraries often cannot be written in strictly conforming C@.
1138 Therefore, GCC gives code found in @dfn{system headers} special
1139 treatment. All warnings, other than those generated by @samp{#warning}
1140 (@pxref{Diagnostics}), are suppressed while GCC is processing a system
1141 header. Macros defined in a system header are immune to a few warnings
1142 wherever they are expanded. This immunity is granted on an ad-hoc
1143 basis, when we find that a warning generates lots of false positives
1144 because of code in macros defined in system headers.
1146 Normally, only the headers found in specific directories are considered
1147 system headers. These directories are determined when GCC is compiled.
1148 There are, however, two ways to make normal headers into system headers.
1150 The @option{-isystem} command-line option adds its argument to the list of
1151 directories to search for headers, just like @option{-I}. Any headers
1152 found in that directory will be considered system headers.
1154 All directories named by @option{-isystem} are searched @emph{after} all
1155 directories named by @option{-I}, no matter what their order was on the
1156 command line. If the same directory is named by both @option{-I} and
1157 @option{-isystem}, the @option{-I} option is ignored. GCC provides an
1158 informative message when this occurs if @option{-v} is used.
1160 @findex #pragma GCC system_header
1161 There is also a directive, @code{@w{#pragma GCC system_header}}, which
1162 tells GCC to consider the rest of the current include file a system
1163 header, no matter where it was found. Code that comes before the
1164 @samp{#pragma} in the file will not be affected. @code{@w{#pragma GCC
1165 system_header}} has no effect in the primary source file.
1167 On very old systems, some of the pre-defined system header directories
1168 get even more special treatment. GNU C++ considers code in headers
1169 found in those directories to be surrounded by an @code{@w{extern "C"}}
1170 block. There is no way to request this behavior with a @samp{#pragma},
1171 or from the command line.
1176 A @dfn{macro} is a fragment of code which has been given a name.
1177 Whenever the name is used, it is replaced by the contents of the macro.
1178 There are two kinds of macros. They differ mostly in what they look
1179 like when they are used. @dfn{Object-like} macros resemble data objects
1180 when used, @dfn{function-like} macros resemble function calls.
1182 You may define any valid identifier as a macro, even if it is a C
1183 keyword. The preprocessor does not know anything about keywords. This
1184 can be useful if you wish to hide a keyword such as @code{const} from an
1185 older compiler that does not understand it. However, the preprocessor
1186 operator @code{defined} (@pxref{Defined}) can never be defined as a
1187 macro, and C++'s named operators (@pxref{C++ Named Operators}) cannot be
1188 macros when you are compiling C++.
1191 * Object-like Macros::
1192 * Function-like Macros::
1197 * Predefined Macros::
1198 * Undefining and Redefining Macros::
1199 * Directives Within Macro Arguments::
1203 @node Object-like Macros
1204 @section Object-like Macros
1205 @cindex object-like macro
1206 @cindex symbolic constants
1207 @cindex manifest constants
1209 An @dfn{object-like macro} is a simple identifier which will be replaced
1210 by a code fragment. It is called object-like because it looks like a
1211 data object in code that uses it. They are most commonly used to give
1212 symbolic names to numeric constants.
1215 You create macros with the @samp{#define} directive. @samp{#define} is
1216 followed by the name of the macro and then the token sequence it should
1217 be an abbreviation for, which is variously referred to as the macro's
1218 @dfn{body}, @dfn{expansion} or @dfn{replacement list}. For example,
1221 #define BUFFER_SIZE 1024
1225 defines a macro named @code{BUFFER_SIZE} as an abbreviation for the
1226 token @code{1024}. If somewhere after this @samp{#define} directive
1227 there comes a C statement of the form
1230 foo = (char *) malloc (BUFFER_SIZE);
1234 then the C preprocessor will recognize and @dfn{expand} the macro
1235 @code{BUFFER_SIZE}. The C compiler will see the same tokens as it would
1239 foo = (char *) malloc (1024);
1242 By convention, macro names are written in uppercase. Programs are
1243 easier to read when it is possible to tell at a glance which names are
1246 The macro's body ends at the end of the @samp{#define} line. You may
1247 continue the definition onto multiple lines, if necessary, using
1248 backslash-newline. When the macro is expanded, however, it will all
1249 come out on one line. For example,
1252 #define NUMBERS 1, \
1255 int x[] = @{ NUMBERS @};
1256 @expansion{} int x[] = @{ 1, 2, 3 @};
1260 The most common visible consequence of this is surprising line numbers
1263 There is no restriction on what can go in a macro body provided it
1264 decomposes into valid preprocessing tokens. Parentheses need not
1265 balance, and the body need not resemble valid C code. (If it does not,
1266 you may get error messages from the C compiler when you use the macro.)
1268 The C preprocessor scans your program sequentially. Macro definitions
1269 take effect at the place you write them. Therefore, the following input
1270 to the C preprocessor
1286 When the preprocessor expands a macro name, the macro's expansion
1287 replaces the macro invocation, then the expansion is examined for more
1288 macros to expand. For example,
1292 #define TABLESIZE BUFSIZE
1293 #define BUFSIZE 1024
1295 @expansion{} BUFSIZE
1301 @code{TABLESIZE} is expanded first to produce @code{BUFSIZE}, then that
1302 macro is expanded to produce the final result, @code{1024}.
1304 Notice that @code{BUFSIZE} was not defined when @code{TABLESIZE} was
1305 defined. The @samp{#define} for @code{TABLESIZE} uses exactly the
1306 expansion you specify---in this case, @code{BUFSIZE}---and does not
1307 check to see whether it too contains macro names. Only when you
1308 @emph{use} @code{TABLESIZE} is the result of its expansion scanned for
1311 This makes a difference if you change the definition of @code{BUFSIZE}
1312 at some point in the source file. @code{TABLESIZE}, defined as shown,
1313 will always expand using the definition of @code{BUFSIZE} that is
1314 currently in effect:
1317 #define BUFSIZE 1020
1318 #define TABLESIZE BUFSIZE
1324 Now @code{TABLESIZE} expands (in two stages) to @code{37}.
1326 If the expansion of a macro contains its own name, either directly or
1327 via intermediate macros, it is not expanded again when the expansion is
1328 examined for more macros. This prevents infinite recursion.
1329 @xref{Self-Referential Macros}, for the precise details.
1331 @node Function-like Macros
1332 @section Function-like Macros
1333 @cindex function-like macros
1335 You can also define macros whose use looks like a function call. These
1336 are called @dfn{function-like macros}. To define a function-like macro,
1337 you use the same @samp{#define} directive, but you put a pair of
1338 parentheses immediately after the macro name. For example,
1341 #define lang_init() c_init()
1343 @expansion{} c_init()
1346 A function-like macro is only expanded if its name appears with a pair
1347 of parentheses after it. If you write just the name, it is left alone.
1348 This can be useful when you have a function and a macro of the same
1349 name, and you wish to use the function sometimes.
1352 extern void foo(void);
1353 #define foo() /* @r{optimized inline version} */
1359 Here the call to @code{foo()} will use the macro, but the function
1360 pointer will get the address of the real function. If the macro were to
1361 be expanded, it would cause a syntax error.
1363 If you put spaces between the macro name and the parentheses in the
1364 macro definition, that does not define a function-like macro, it defines
1365 an object-like macro whose expansion happens to begin with a pair of
1369 #define lang_init () c_init()
1371 @expansion{} () c_init()()
1374 The first two pairs of parentheses in this expansion come from the
1375 macro. The third is the pair that was originally after the macro
1376 invocation. Since @code{lang_init} is an object-like macro, it does not
1377 consume those parentheses.
1379 @node Macro Arguments
1380 @section Macro Arguments
1382 @cindex macros with arguments
1383 @cindex arguments in macro definitions
1385 Function-like macros can take @dfn{arguments}, just like true functions.
1386 To define a macro that uses arguments, you insert @dfn{parameters}
1387 between the pair of parentheses in the macro definition that make the
1388 macro function-like. The parameters must be valid C identifiers,
1389 separated by commas and optionally whitespace.
1391 To invoke a macro that takes arguments, you write the name of the macro
1392 followed by a list of @dfn{actual arguments} in parentheses, separated
1393 by commas. The invocation of the macro need not be restricted to a
1394 single logical line---it can cross as many lines in the source file as
1395 you wish. The number of arguments you give must match the number of
1396 parameters in the macro definition. When the macro is expanded, each
1397 use of a parameter in its body is replaced by the tokens of the
1398 corresponding argument. (You need not use all of the parameters in the
1401 As an example, here is a macro that computes the minimum of two numeric
1402 values, as it is defined in many C programs, and some uses.
1405 #define min(X, Y) ((X) < (Y) ? (X) : (Y))
1406 x = min(a, b); @expansion{} x = ((a) < (b) ? (a) : (b));
1407 y = min(1, 2); @expansion{} y = ((1) < (2) ? (1) : (2));
1408 z = min(a + 28, *p); @expansion{} z = ((a + 28) < (*p) ? (a + 28) : (*p));
1412 (In this small example you can already see several of the dangers of
1413 macro arguments. @xref{Macro Pitfalls}, for detailed explanations.)
1415 Leading and trailing whitespace in each argument is dropped, and all
1416 whitespace between the tokens of an argument is reduced to a single
1417 space. Parentheses within each argument must balance; a comma within
1418 such parentheses does not end the argument. However, there is no
1419 requirement for square brackets or braces to balance, and they do not
1420 prevent a comma from separating arguments. Thus,
1423 macro (array[x = y, x + 1])
1427 passes two arguments to @code{macro}: @code{array[x = y} and @code{x +
1428 1]}. If you want to supply @code{array[x = y, x + 1]} as an argument,
1429 you can write it as @code{array[(x = y, x + 1)]}, which is equivalent C
1432 All arguments to a macro are completely macro-expanded before they are
1433 substituted into the macro body. After substitution, the complete text
1434 is scanned again for macros to expand, including the arguments. This rule
1435 may seem strange, but it is carefully designed so you need not worry
1436 about whether any function call is actually a macro invocation. You can
1437 run into trouble if you try to be too clever, though. @xref{Argument
1438 Prescan}, for detailed discussion.
1440 For example, @code{min (min (a, b), c)} is first expanded to
1443 min (((a) < (b) ? (a) : (b)), (c))
1451 ((((a) < (b) ? (a) : (b))) < (c)
1452 ? (((a) < (b) ? (a) : (b)))
1458 (Line breaks shown here for clarity would not actually be generated.)
1460 @cindex empty macro arguments
1461 You can leave macro arguments empty; this is not an error to the
1462 preprocessor (but many macros will then expand to invalid code).
1463 You cannot leave out arguments entirely; if a macro takes two arguments,
1464 there must be exactly one comma at the top level of its argument list.
1465 Here are some silly examples using @code{min}:
1468 min(, b) @expansion{} (( ) < (b) ? ( ) : (b))
1469 min(a, ) @expansion{} ((a ) < ( ) ? (a ) : ( ))
1470 min(,) @expansion{} (( ) < ( ) ? ( ) : ( ))
1471 min((,),) @expansion{} (((,)) < ( ) ? ((,)) : ( ))
1473 min() @error{} macro "min" requires 2 arguments, but only 1 given
1474 min(,,) @error{} macro "min" passed 3 arguments, but takes just 2
1477 Whitespace is not a preprocessing token, so if a macro @code{foo} takes
1478 one argument, @code{@w{foo ()}} and @code{@w{foo ( )}} both supply it an
1479 empty argument. Previous GNU preprocessor implementations and
1480 documentation were incorrect on this point, insisting that a
1481 function-like macro that takes a single argument be passed a space if an
1482 empty argument was required.
1484 Macro parameters appearing inside string literals are not replaced by
1485 their corresponding actual arguments.
1488 #define foo(x) x, "x"
1489 foo(bar) @expansion{} bar, "x"
1492 @node Stringification
1493 @section Stringification
1494 @cindex stringification
1495 @cindex @samp{#} operator
1497 Sometimes you may want to convert a macro argument into a string
1498 constant. Parameters are not replaced inside string constants, but you
1499 can use the @samp{#} preprocessing operator instead. When a macro
1500 parameter is used with a leading @samp{#}, the preprocessor replaces it
1501 with the literal text of the actual argument, converted to a string
1502 constant. Unlike normal parameter replacement, the argument is not
1503 macro-expanded first. This is called @dfn{stringification}.
1505 There is no way to combine an argument with surrounding text and
1506 stringify it all together. Instead, you can write a series of adjacent
1507 string constants and stringified arguments. The preprocessor will
1508 replace the stringified arguments with string constants. The C
1509 compiler will then combine all the adjacent string constants into one
1512 Here is an example of a macro definition that uses stringification:
1516 #define WARN_IF(EXP) \
1518 fprintf (stderr, "Warning: " #EXP "\n"); @} \
1521 @expansion{} do @{ if (x == 0)
1522 fprintf (stderr, "Warning: " "x == 0" "\n"); @} while (0);
1527 The argument for @code{EXP} is substituted once, as-is, into the
1528 @code{if} statement, and once, stringified, into the argument to
1529 @code{fprintf}. If @code{x} were a macro, it would be expanded in the
1530 @code{if} statement, but not in the string.
1532 The @code{do} and @code{while (0)} are a kludge to make it possible to
1533 write @code{WARN_IF (@var{arg});}, which the resemblance of
1534 @code{WARN_IF} to a function would make C programmers want to do; see
1535 @ref{Swallowing the Semicolon}.
1537 Stringification in C involves more than putting double-quote characters
1538 around the fragment. The preprocessor backslash-escapes the quotes
1539 surrounding embedded string constants, and all backslashes within string and
1540 character constants, in order to get a valid C string constant with the
1541 proper contents. Thus, stringifying @code{@w{p = "foo\n";}} results in
1542 @t{@w{"p = \"foo\\n\";"}}. However, backslashes that are not inside string
1543 or character constants are not duplicated: @samp{\n} by itself
1544 stringifies to @t{"\n"}.
1546 All leading and trailing whitespace in text being stringified is
1547 ignored. Any sequence of whitespace in the middle of the text is
1548 converted to a single space in the stringified result. Comments are
1549 replaced by whitespace long before stringification happens, so they
1550 never appear in stringified text.
1552 There is no way to convert a macro argument into a character constant.
1554 If you want to stringify the result of expansion of a macro argument,
1555 you have to use two levels of macros.
1558 #define xstr(s) str(s)
1564 @expansion{} xstr (4)
1565 @expansion{} str (4)
1569 @code{s} is stringified when it is used in @code{str}, so it is not
1570 macro-expanded first. But @code{s} is an ordinary argument to
1571 @code{xstr}, so it is completely macro-expanded before @code{xstr}
1572 itself is expanded (@pxref{Argument Prescan}). Therefore, by the time
1573 @code{str} gets to its argument, it has already been macro-expanded.
1576 @section Concatenation
1577 @cindex concatenation
1578 @cindex token pasting
1579 @cindex token concatenation
1580 @cindex @samp{##} operator
1582 It is often useful to merge two tokens into one while expanding macros.
1583 This is called @dfn{token pasting} or @dfn{token concatenation}. The
1584 @samp{##} preprocessing operator performs token pasting. When a macro
1585 is expanded, the two tokens on either side of each @samp{##} operator
1586 are combined into a single token, which then replaces the @samp{##} and
1587 the two original tokens in the macro expansion. Usually both will be
1588 identifiers, or one will be an identifier and the other a preprocessing
1589 number. When pasted, they make a longer identifier. This isn't the
1590 only valid case. It is also possible to concatenate two numbers (or a
1591 number and a name, such as @code{1.5} and @code{e3}) into a number.
1592 Also, multi-character operators such as @code{+=} can be formed by
1595 However, two tokens that don't together form a valid token cannot be
1596 pasted together. For example, you cannot concatenate @code{x} with
1597 @code{+} in either order. If you try, the preprocessor issues a warning
1598 and emits the two tokens. Whether it puts white space between the
1599 tokens is undefined. It is common to find unnecessary uses of @samp{##}
1600 in complex macros. If you get this warning, it is likely that you can
1601 simply remove the @samp{##}.
1603 Both the tokens combined by @samp{##} could come from the macro body,
1604 but you could just as well write them as one token in the first place.
1605 Token pasting is most useful when one or both of the tokens comes from a
1606 macro argument. If either of the tokens next to an @samp{##} is a
1607 parameter name, it is replaced by its actual argument before @samp{##}
1608 executes. As with stringification, the actual argument is not
1609 macro-expanded first. If the argument is empty, that @samp{##} has no
1612 Keep in mind that the C preprocessor converts comments to whitespace
1613 before macros are even considered. Therefore, you cannot create a
1614 comment by concatenating @samp{/} and @samp{*}. You can put as much
1615 whitespace between @samp{##} and its operands as you like, including
1616 comments, and you can put comments in arguments that will be
1617 concatenated. However, it is an error if @samp{##} appears at either
1618 end of a macro body.
1620 Consider a C program that interprets named commands. There probably
1621 needs to be a table of commands, perhaps an array of structures declared
1629 void (*function) (void);
1634 struct command commands[] =
1636 @{ "quit", quit_command @},
1637 @{ "help", help_command @},
1643 It would be cleaner not to have to give each command name twice, once in
1644 the string constant and once in the function name. A macro which takes the
1645 name of a command as an argument can make this unnecessary. The string
1646 constant can be created with stringification, and the function name by
1647 concatenating the argument with @samp{_command}. Here is how it is done:
1650 #define COMMAND(NAME) @{ #NAME, NAME ## _command @}
1652 struct command commands[] =
1660 @node Variadic Macros
1661 @section Variadic Macros
1662 @cindex variable number of arguments
1663 @cindex macros with variable arguments
1664 @cindex variadic macros
1666 A macro can be declared to accept a variable number of arguments much as
1667 a function can. The syntax for defining the macro is similar to that of
1668 a function. Here is an example:
1671 #define eprintf(@dots{}) fprintf (stderr, __VA_ARGS__)
1674 This kind of macro is called @dfn{variadic}. When the macro is invoked,
1675 all the tokens in its argument list after the last named argument (this
1676 macro has none), including any commas, become the @dfn{variable
1677 argument}. This sequence of tokens replaces the identifier
1678 @code{@w{__VA_ARGS__}} in the macro body wherever it appears. Thus, we
1679 have this expansion:
1682 eprintf ("%s:%d: ", input_file, lineno)
1683 @expansion{} fprintf (stderr, "%s:%d: ", input_file, lineno)
1686 The variable argument is completely macro-expanded before it is inserted
1687 into the macro expansion, just like an ordinary argument. You may use
1688 the @samp{#} and @samp{##} operators to stringify the variable argument
1689 or to paste its leading or trailing token with another token. (But see
1690 below for an important special case for @samp{##}.)
1692 If your macro is complicated, you may want a more descriptive name for
1693 the variable argument than @code{@w{__VA_ARGS__}}. CPP permits
1694 this, as an extension. You may write an argument name immediately
1695 before the @samp{@dots{}}; that name is used for the variable argument.
1696 The @code{eprintf} macro above could be written
1699 #define eprintf(args@dots{}) fprintf (stderr, args)
1703 using this extension. You cannot use @code{@w{__VA_ARGS__}} and this
1704 extension in the same macro.
1706 You can have named arguments as well as variable arguments in a variadic
1707 macro. We could define @code{eprintf} like this, instead:
1710 #define eprintf(format, @dots{}) fprintf (stderr, format, __VA_ARGS__)
1714 This formulation looks more descriptive, but unfortunately it is less
1715 flexible: you must now supply at least one argument after the format
1716 string. In standard C, you cannot omit the comma separating the named
1717 argument from the variable arguments. Furthermore, if you leave the
1718 variable argument empty, you will get a syntax error, because
1719 there will be an extra comma after the format string.
1722 eprintf("success!\n", );
1723 @expansion{} fprintf(stderr, "success!\n", );
1726 GNU CPP has a pair of extensions which deal with this problem. First,
1727 you are allowed to leave the variable argument out entirely:
1730 eprintf ("success!\n")
1731 @expansion{} fprintf(stderr, "success!\n", );
1735 Second, the @samp{##} token paste operator has a special meaning when
1736 placed between a comma and a variable argument. If you write
1739 #define eprintf(format, @dots{}) fprintf (stderr, format, ##__VA_ARGS__)
1743 and the variable argument is left out when the @code{eprintf} macro is
1744 used, then the comma before the @samp{##} will be deleted. This does
1745 @emph{not} happen if you pass an empty argument, nor does it happen if
1746 the token preceding @samp{##} is anything other than a comma.
1749 eprintf ("success!\n")
1750 @expansion{} fprintf(stderr, "success!\n");
1754 The above explanation is ambiguous about the case where the only macro
1755 parameter is a variable arguments parameter, as it is meaningless to
1756 try to distinguish whether no argument at all is an empty argument or
1757 a missing argument. In this case the C99 standard is clear that the
1758 comma must remain, however the existing GCC extension used to swallow
1759 the comma. So CPP retains the comma when conforming to a specific C
1760 standard, and drops it otherwise.
1762 C99 mandates that the only place the identifier @code{@w{__VA_ARGS__}}
1763 can appear is in the replacement list of a variadic macro. It may not
1764 be used as a macro name, macro argument name, or within a different type
1765 of macro. It may also be forbidden in open text; the standard is
1766 ambiguous. We recommend you avoid using it except for its defined
1769 Variadic macros are a new feature in C99. GNU CPP has supported them
1770 for a long time, but only with a named variable argument
1771 (@samp{args@dots{}}, not @samp{@dots{}} and @code{@w{__VA_ARGS__}}). If you are
1772 concerned with portability to previous versions of GCC, you should use
1773 only named variable arguments. On the other hand, if you are concerned
1774 with portability to other conforming implementations of C99, you should
1775 use only @code{@w{__VA_ARGS__}}.
1777 Previous versions of CPP implemented the comma-deletion extension
1778 much more generally. We have restricted it in this release to minimize
1779 the differences from C99. To get the same effect with both this and
1780 previous versions of GCC, the token preceding the special @samp{##} must
1781 be a comma, and there must be white space between that comma and
1782 whatever comes immediately before it:
1785 #define eprintf(format, args@dots{}) fprintf (stderr, format , ##args)
1789 @xref{Differences from previous versions}, for the gory details.
1791 @node Predefined Macros
1792 @section Predefined Macros
1794 @cindex predefined macros
1795 Several object-like macros are predefined; you use them without
1796 supplying their definitions. They fall into three classes: standard,
1797 common, and system-specific.
1799 In C++, there is a fourth category, the named operators. They act like
1800 predefined macros, but you cannot undefine them.
1803 * Standard Predefined Macros::
1804 * Common Predefined Macros::
1805 * System-specific Predefined Macros::
1806 * C++ Named Operators::
1809 @node Standard Predefined Macros
1810 @subsection Standard Predefined Macros
1811 @cindex standard predefined macros.
1813 The standard predefined macros are specified by the relevant
1814 language standards, so they are available with all compilers that
1815 implement those standards. Older compilers may not provide all of
1816 them. Their names all start with double underscores.
1820 This macro expands to the name of the current input file, in the form of
1821 a C string constant. This is the path by which the preprocessor opened
1822 the file, not the short name specified in @samp{#include} or as the
1823 input file name argument. For example,
1824 @code{"/usr/local/include/myheader.h"} is a possible expansion of this
1828 This macro expands to the current input line number, in the form of a
1829 decimal integer constant. While we call it a predefined macro, it's
1830 a pretty strange macro, since its ``definition'' changes with each
1831 new line of source code.
1834 @code{__FILE__} and @code{__LINE__} are useful in generating an error
1835 message to report an inconsistency detected by the program; the message
1836 can state the source line at which the inconsistency was detected. For
1840 fprintf (stderr, "Internal error: "
1841 "negative string length "
1842 "%d at %s, line %d.",
1843 length, __FILE__, __LINE__);
1846 An @samp{#include} directive changes the expansions of @code{__FILE__}
1847 and @code{__LINE__} to correspond to the included file. At the end of
1848 that file, when processing resumes on the input file that contained
1849 the @samp{#include} directive, the expansions of @code{__FILE__} and
1850 @code{__LINE__} revert to the values they had before the
1851 @samp{#include} (but @code{__LINE__} is then incremented by one as
1852 processing moves to the line after the @samp{#include}).
1854 A @samp{#line} directive changes @code{__LINE__}, and may change
1855 @code{__FILE__} as well. @xref{Line Control}.
1857 C99 introduces @code{__func__}, and GCC has provided @code{__FUNCTION__}
1858 for a long time. Both of these are strings containing the name of the
1859 current function (there are slight semantic differences; see the GCC
1860 manual). Neither of them is a macro; the preprocessor does not know the
1861 name of the current function. They tend to be useful in conjunction
1862 with @code{__FILE__} and @code{__LINE__}, though.
1867 This macro expands to a string constant that describes the date on which
1868 the preprocessor is being run. The string constant contains eleven
1869 characters and looks like @code{@w{"Feb 12 1996"}}. If the day of the
1870 month is less than 10, it is padded with a space on the left.
1872 If GCC cannot determine the current date, it will emit a warning message
1873 (once per compilation) and @code{__DATE__} will expand to
1874 @code{@w{"??? ?? ????"}}.
1877 This macro expands to a string constant that describes the time at
1878 which the preprocessor is being run. The string constant contains
1879 eight characters and looks like @code{"23:59:01"}.
1881 If GCC cannot determine the current time, it will emit a warning message
1882 (once per compilation) and @code{__TIME__} will expand to
1886 In normal operation, this macro expands to the constant 1, to signify
1887 that this compiler conforms to ISO Standard C@. If GNU CPP is used with
1888 a compiler other than GCC, this is not necessarily true; however, the
1889 preprocessor always conforms to the standard unless the
1890 @option{-traditional-cpp} option is used.
1892 This macro is not defined if the @option{-traditional-cpp} option is used.
1894 On some hosts, the system compiler uses a different convention, where
1895 @code{__STDC__} is normally 0, but is 1 if the user specifies strict
1896 conformance to the C Standard. CPP follows the host convention when
1897 processing system header files, but when processing user files
1898 @code{__STDC__} is always 1. This has been reported to cause problems;
1899 for instance, some versions of Solaris provide X Windows headers that
1900 expect @code{__STDC__} to be either undefined or 1. @xref{Invocation}.
1902 @item __STDC_VERSION__
1903 This macro expands to the C Standard's version number, a long integer
1904 constant of the form @code{@var{yyyy}@var{mm}L} where @var{yyyy} and
1905 @var{mm} are the year and month of the Standard version. This signifies
1906 which version of the C Standard the compiler conforms to. Like
1907 @code{__STDC__}, this is not necessarily accurate for the entire
1908 implementation, unless GNU CPP is being used with GCC@.
1910 The value @code{199409L} signifies the 1989 C standard as amended in
1911 1994, which is the current default; the value @code{199901L} signifies
1912 the 1999 revision of the C standard. Support for the 1999 revision is
1915 This macro is not defined if the @option{-traditional-cpp} option is
1916 used, nor when compiling C++ or Objective-C@.
1918 @item __STDC_HOSTED__
1919 This macro is defined, with value 1, if the compiler's target is a
1920 @dfn{hosted environment}. A hosted environment has the complete
1921 facilities of the standard C library available.
1924 This macro is defined when the C++ compiler is in use. You can use
1925 @code{__cplusplus} to test whether a header is compiled by a C compiler
1926 or a C++ compiler. This macro is similar to @code{__STDC_VERSION__}, in
1927 that it expands to a version number. Depending on the language standard
1928 selected, the value of the macro is
1929 @code{199711L} for the 1998 C++ standard,
1930 @code{201103L} for the 2011 C++ standard,
1931 @code{201402L} for the 2014 C++ standard,
1932 or an unspecified value strictly larger than @code{201402L} for the
1933 experimental languages enabled by @option{-std=c++1z} and
1934 @option{-std=gnu++1z}.
1937 This macro is defined, with value 1, when the Objective-C compiler is in
1938 use. You can use @code{__OBJC__} to test whether a header is compiled
1939 by a C compiler or an Objective-C compiler.
1942 This macro is defined with value 1 when preprocessing assembly
1947 @node Common Predefined Macros
1948 @subsection Common Predefined Macros
1949 @cindex common predefined macros
1951 The common predefined macros are GNU C extensions. They are available
1952 with the same meanings regardless of the machine or operating system on
1953 which you are using GNU C or GNU Fortran. Their names all start with
1959 This macro expands to sequential integral values starting from 0. In
1960 conjunction with the @code{##} operator, this provides a convenient means to
1961 generate unique identifiers. Care must be taken to ensure that
1962 @code{__COUNTER__} is not expanded prior to inclusion of precompiled headers
1963 which use it. Otherwise, the precompiled headers will not be used.
1966 The GNU Fortran compiler defines this.
1969 @itemx __GNUC_MINOR__
1970 @itemx __GNUC_PATCHLEVEL__
1971 These macros are defined by all GNU compilers that use the C
1972 preprocessor: C, C++, Objective-C and Fortran. Their values are the major
1973 version, minor version, and patch level of the compiler, as integer
1974 constants. For example, GCC 3.2.1 will define @code{__GNUC__} to 3,
1975 @code{__GNUC_MINOR__} to 2, and @code{__GNUC_PATCHLEVEL__} to 1. These
1976 macros are also defined if you invoke the preprocessor directly.
1978 @code{__GNUC_PATCHLEVEL__} is new to GCC 3.0; it is also present in the
1979 widely-used development snapshots leading up to 3.0 (which identify
1980 themselves as GCC 2.96 or 2.97, depending on which snapshot you have).
1982 If all you need to know is whether or not your program is being compiled
1983 by GCC, or a non-GCC compiler that claims to accept the GNU C dialects,
1984 you can simply test @code{__GNUC__}. If you need to write code
1985 which depends on a specific version, you must be more careful. Each
1986 time the minor version is increased, the patch level is reset to zero;
1987 each time the major version is increased, the
1988 minor version and patch level are reset. If you wish to use the
1989 predefined macros directly in the conditional, you will need to write it
1993 /* @r{Test for GCC > 3.2.0} */
1994 #if __GNUC__ > 3 || \
1995 (__GNUC__ == 3 && (__GNUC_MINOR__ > 2 || \
1996 (__GNUC_MINOR__ == 2 && \
1997 __GNUC_PATCHLEVEL__ > 0))
2001 Another approach is to use the predefined macros to
2002 calculate a single number, then compare that against a threshold:
2005 #define GCC_VERSION (__GNUC__ * 10000 \
2006 + __GNUC_MINOR__ * 100 \
2007 + __GNUC_PATCHLEVEL__)
2009 /* @r{Test for GCC > 3.2.0} */
2010 #if GCC_VERSION > 30200
2014 Many people find this form easier to understand.
2017 The GNU C++ compiler defines this. Testing it is equivalent to
2018 testing @code{@w{(__GNUC__ && __cplusplus)}}.
2020 @item __STRICT_ANSI__
2021 GCC defines this macro if and only if the @option{-ansi} switch, or a
2022 @option{-std} switch specifying strict conformance to some version of ISO C
2023 or ISO C++, was specified when GCC was invoked. It is defined to @samp{1}.
2024 This macro exists primarily to direct GNU libc's header files to
2025 restrict their definitions to the minimal set found in the 1989 C
2029 This macro expands to the name of the main input file, in the form
2030 of a C string constant. This is the source file that was specified
2031 on the command line of the preprocessor or C compiler.
2033 @item __INCLUDE_LEVEL__
2034 This macro expands to a decimal integer constant that represents the
2035 depth of nesting in include files. The value of this macro is
2036 incremented on every @samp{#include} directive and decremented at the
2037 end of every included file. It starts out at 0, its value within the
2038 base file specified on the command line.
2041 This macro is defined if the target uses the ELF object format.
2044 This macro expands to a string constant which describes the version of
2045 the compiler in use. You should not rely on its contents having any
2046 particular form, but it can be counted on to contain at least the
2050 @itemx __OPTIMIZE_SIZE__
2051 @itemx __NO_INLINE__
2052 These macros describe the compilation mode. @code{__OPTIMIZE__} is
2053 defined in all optimizing compilations. @code{__OPTIMIZE_SIZE__} is
2054 defined if the compiler is optimizing for size, not speed.
2055 @code{__NO_INLINE__} is defined if no functions will be inlined into
2056 their callers (when not optimizing, or when inlining has been
2057 specifically disabled by @option{-fno-inline}).
2059 These macros cause certain GNU header files to provide optimized
2060 definitions, using macros or inline functions, of system library
2061 functions. You should not use these macros in any way unless you make
2062 sure that programs will execute with the same effect whether or not they
2063 are defined. If they are defined, their value is 1.
2065 @item __GNUC_GNU_INLINE__
2066 GCC defines this macro if functions declared @code{inline} will be
2067 handled in GCC's traditional gnu90 mode. Object files will contain
2068 externally visible definitions of all functions declared @code{inline}
2069 without @code{extern} or @code{static}. They will not contain any
2070 definitions of any functions declared @code{extern inline}.
2072 @item __GNUC_STDC_INLINE__
2073 GCC defines this macro if functions declared @code{inline} will be
2074 handled according to the ISO C99 standard. Object files will contain
2075 externally visible definitions of all functions declared @code{extern
2076 inline}. They will not contain definitions of any functions declared
2077 @code{inline} without @code{extern}.
2079 If this macro is defined, GCC supports the @code{gnu_inline} function
2080 attribute as a way to always get the gnu90 behavior. Support for
2081 this and @code{__GNUC_GNU_INLINE__} was added in GCC 4.1.3. If
2082 neither macro is defined, an older version of GCC is being used:
2083 @code{inline} functions will be compiled in gnu90 mode, and the
2084 @code{gnu_inline} function attribute will not be recognized.
2086 @item __CHAR_UNSIGNED__
2087 GCC defines this macro if and only if the data type @code{char} is
2088 unsigned on the target machine. It exists to cause the standard header
2089 file @file{limits.h} to work correctly. You should not use this macro
2090 yourself; instead, refer to the standard macros defined in @file{limits.h}.
2092 @item __WCHAR_UNSIGNED__
2093 Like @code{__CHAR_UNSIGNED__}, this macro is defined if and only if the
2094 data type @code{wchar_t} is unsigned and the front-end is in C++ mode.
2096 @item __REGISTER_PREFIX__
2097 This macro expands to a single token (not a string constant) which is
2098 the prefix applied to CPU register names in assembly language for this
2099 target. You can use it to write assembly that is usable in multiple
2100 environments. For example, in the @code{m68k-aout} environment it
2101 expands to nothing, but in the @code{m68k-coff} environment it expands
2102 to a single @samp{%}.
2104 @item __USER_LABEL_PREFIX__
2105 This macro expands to a single token which is the prefix applied to
2106 user labels (symbols visible to C code) in assembly. For example, in
2107 the @code{m68k-aout} environment it expands to an @samp{_}, but in the
2108 @code{m68k-coff} environment it expands to nothing.
2110 This macro will have the correct definition even if
2111 @option{-f(no-)underscores} is in use, but it will not be correct if
2112 target-specific options that adjust this prefix are used (e.g.@: the
2113 OSF/rose @option{-mno-underscores} option).
2116 @itemx __PTRDIFF_TYPE__
2117 @itemx __WCHAR_TYPE__
2118 @itemx __WINT_TYPE__
2119 @itemx __INTMAX_TYPE__
2120 @itemx __UINTMAX_TYPE__
2121 @itemx __SIG_ATOMIC_TYPE__
2122 @itemx __INT8_TYPE__
2123 @itemx __INT16_TYPE__
2124 @itemx __INT32_TYPE__
2125 @itemx __INT64_TYPE__
2126 @itemx __UINT8_TYPE__
2127 @itemx __UINT16_TYPE__
2128 @itemx __UINT32_TYPE__
2129 @itemx __UINT64_TYPE__
2130 @itemx __INT_LEAST8_TYPE__
2131 @itemx __INT_LEAST16_TYPE__
2132 @itemx __INT_LEAST32_TYPE__
2133 @itemx __INT_LEAST64_TYPE__
2134 @itemx __UINT_LEAST8_TYPE__
2135 @itemx __UINT_LEAST16_TYPE__
2136 @itemx __UINT_LEAST32_TYPE__
2137 @itemx __UINT_LEAST64_TYPE__
2138 @itemx __INT_FAST8_TYPE__
2139 @itemx __INT_FAST16_TYPE__
2140 @itemx __INT_FAST32_TYPE__
2141 @itemx __INT_FAST64_TYPE__
2142 @itemx __UINT_FAST8_TYPE__
2143 @itemx __UINT_FAST16_TYPE__
2144 @itemx __UINT_FAST32_TYPE__
2145 @itemx __UINT_FAST64_TYPE__
2146 @itemx __INTPTR_TYPE__
2147 @itemx __UINTPTR_TYPE__
2148 These macros are defined to the correct underlying types for the
2149 @code{size_t}, @code{ptrdiff_t}, @code{wchar_t}, @code{wint_t},
2150 @code{intmax_t}, @code{uintmax_t}, @code{sig_atomic_t}, @code{int8_t},
2151 @code{int16_t}, @code{int32_t}, @code{int64_t}, @code{uint8_t},
2152 @code{uint16_t}, @code{uint32_t}, @code{uint64_t},
2153 @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
2154 @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
2155 @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
2156 @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
2157 @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
2158 @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t} typedefs,
2159 respectively. They exist to make the standard header files
2160 @file{stddef.h}, @file{stdint.h}, and @file{wchar.h} work correctly.
2161 You should not use these macros directly; instead, include the
2162 appropriate headers and use the typedefs. Some of these macros may
2163 not be defined on particular systems if GCC does not provide a
2164 @file{stdint.h} header on those systems.
2167 Defined to the number of bits used in the representation of the
2168 @code{char} data type. It exists to make the standard header given
2169 numerical limits work correctly. You should not use
2170 this macro directly; instead, include the appropriate headers.
2173 @itemx __WCHAR_MAX__
2177 @itemx __LONG_LONG_MAX__
2180 @itemx __PTRDIFF_MAX__
2181 @itemx __INTMAX_MAX__
2182 @itemx __UINTMAX_MAX__
2183 @itemx __SIG_ATOMIC_MAX__
2185 @itemx __INT16_MAX__
2186 @itemx __INT32_MAX__
2187 @itemx __INT64_MAX__
2188 @itemx __UINT8_MAX__
2189 @itemx __UINT16_MAX__
2190 @itemx __UINT32_MAX__
2191 @itemx __UINT64_MAX__
2192 @itemx __INT_LEAST8_MAX__
2193 @itemx __INT_LEAST16_MAX__
2194 @itemx __INT_LEAST32_MAX__
2195 @itemx __INT_LEAST64_MAX__
2196 @itemx __UINT_LEAST8_MAX__
2197 @itemx __UINT_LEAST16_MAX__
2198 @itemx __UINT_LEAST32_MAX__
2199 @itemx __UINT_LEAST64_MAX__
2200 @itemx __INT_FAST8_MAX__
2201 @itemx __INT_FAST16_MAX__
2202 @itemx __INT_FAST32_MAX__
2203 @itemx __INT_FAST64_MAX__
2204 @itemx __UINT_FAST8_MAX__
2205 @itemx __UINT_FAST16_MAX__
2206 @itemx __UINT_FAST32_MAX__
2207 @itemx __UINT_FAST64_MAX__
2208 @itemx __INTPTR_MAX__
2209 @itemx __UINTPTR_MAX__
2210 @itemx __WCHAR_MIN__
2212 @itemx __SIG_ATOMIC_MIN__
2213 Defined to the maximum value of the @code{signed char}, @code{wchar_t},
2214 @code{signed short},
2215 @code{signed int}, @code{signed long}, @code{signed long long},
2216 @code{wint_t}, @code{size_t}, @code{ptrdiff_t},
2217 @code{intmax_t}, @code{uintmax_t}, @code{sig_atomic_t}, @code{int8_t},
2218 @code{int16_t}, @code{int32_t}, @code{int64_t}, @code{uint8_t},
2219 @code{uint16_t}, @code{uint32_t}, @code{uint64_t},
2220 @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
2221 @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
2222 @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
2223 @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
2224 @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
2225 @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t} types and
2226 to the minimum value of the @code{wchar_t}, @code{wint_t}, and
2227 @code{sig_atomic_t} types respectively. They exist to make the
2228 standard header given numerical limits work correctly. You should not
2229 use these macros directly; instead, include the appropriate headers.
2230 Some of these macros may not be defined on particular systems if GCC
2231 does not provide a @file{stdint.h} header on those systems.
2243 Defined to implementations of the standard @file{stdint.h} macros with
2244 the same names without the leading @code{__}. They exist the make the
2245 implementation of that header work correctly. You should not use
2246 these macros directly; instead, include the appropriate headers. Some
2247 of these macros may not be defined on particular systems if GCC does
2248 not provide a @file{stdint.h} header on those systems.
2250 @item __SCHAR_WIDTH__
2251 @itemx __SHRT_WIDTH__
2252 @itemx __INT_WIDTH__
2253 @itemx __LONG_WIDTH__
2254 @itemx __LONG_LONG_WIDTH__
2255 @itemx __PTRDIFF_WIDTH__
2256 @itemx __SIG_ATOMIC_WIDTH__
2257 @itemx __SIZE_WIDTH__
2258 @itemx __WCHAR_WIDTH__
2259 @itemx __WINT_WIDTH__
2260 @itemx __INT_LEAST8_WIDTH__
2261 @itemx __INT_LEAST16_WIDTH__
2262 @itemx __INT_LEAST32_WIDTH__
2263 @itemx __INT_LEAST64_WIDTH__
2264 @itemx __INT_FAST8_WIDTH__
2265 @itemx __INT_FAST16_WIDTH__
2266 @itemx __INT_FAST32_WIDTH__
2267 @itemx __INT_FAST64_WIDTH__
2268 @itemx __INTPTR_WIDTH__
2269 @itemx __INTMAX_WIDTH__
2270 Defined to the bit widths of the corresponding types. They exist to
2271 make the implementations of @file{limits.h} and @file{stdint.h} behave
2272 correctly. You should not use these macros directly; instead, include
2273 the appropriate headers. Some of these macros may not be defined on
2274 particular systems if GCC does not provide a @file{stdint.h} header on
2277 @item __SIZEOF_INT__
2278 @itemx __SIZEOF_LONG__
2279 @itemx __SIZEOF_LONG_LONG__
2280 @itemx __SIZEOF_SHORT__
2281 @itemx __SIZEOF_POINTER__
2282 @itemx __SIZEOF_FLOAT__
2283 @itemx __SIZEOF_DOUBLE__
2284 @itemx __SIZEOF_LONG_DOUBLE__
2285 @itemx __SIZEOF_SIZE_T__
2286 @itemx __SIZEOF_WCHAR_T__
2287 @itemx __SIZEOF_WINT_T__
2288 @itemx __SIZEOF_PTRDIFF_T__
2289 Defined to the number of bytes of the C standard data types: @code{int},
2290 @code{long}, @code{long long}, @code{short}, @code{void *}, @code{float},
2291 @code{double}, @code{long double}, @code{size_t}, @code{wchar_t}, @code{wint_t}
2292 and @code{ptrdiff_t}.
2294 @item __BYTE_ORDER__
2295 @itemx __ORDER_LITTLE_ENDIAN__
2296 @itemx __ORDER_BIG_ENDIAN__
2297 @itemx __ORDER_PDP_ENDIAN__
2298 @code{__BYTE_ORDER__} is defined to one of the values
2299 @code{__ORDER_LITTLE_ENDIAN__}, @code{__ORDER_BIG_ENDIAN__}, or
2300 @code{__ORDER_PDP_ENDIAN__} to reflect the layout of multi-byte and
2301 multi-word quantities in memory. If @code{__BYTE_ORDER__} is equal to
2302 @code{__ORDER_LITTLE_ENDIAN__} or @code{__ORDER_BIG_ENDIAN__}, then
2303 multi-byte and multi-word quantities are laid out identically: the
2304 byte (word) at the lowest address is the least significant or most
2305 significant byte (word) of the quantity, respectively. If
2306 @code{__BYTE_ORDER__} is equal to @code{__ORDER_PDP_ENDIAN__}, then
2307 bytes in 16-bit words are laid out in a little-endian fashion, whereas
2308 the 16-bit subwords of a 32-bit quantity are laid out in big-endian
2311 You should use these macros for testing like this:
2314 /* @r{Test for a little-endian machine} */
2315 #if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
2318 @item __FLOAT_WORD_ORDER__
2319 @code{__FLOAT_WORD_ORDER__} is defined to one of the values
2320 @code{__ORDER_LITTLE_ENDIAN__} or @code{__ORDER_BIG_ENDIAN__} to reflect
2321 the layout of the words of multi-word floating-point quantities.
2324 This macro is defined, with value 1, when compiling a C++ source file
2325 with warnings about deprecated constructs enabled. These warnings are
2326 enabled by default, but can be disabled with @option{-Wno-deprecated}.
2329 This macro is defined, with value 1, when compiling a C++ source file
2330 with exceptions enabled. If @option{-fno-exceptions} is used when
2331 compiling the file, then this macro is not defined.
2334 This macro is defined, with value 1, when compiling a C++ source file
2335 with runtime type identification enabled. If @option{-fno-rtti} is
2336 used when compiling the file, then this macro is not defined.
2338 @item __USING_SJLJ_EXCEPTIONS__
2339 This macro is defined, with value 1, if the compiler uses the old
2340 mechanism based on @code{setjmp} and @code{longjmp} for exception
2343 @item __GXX_EXPERIMENTAL_CXX0X__
2344 This macro is defined when compiling a C++ source file with the option
2345 @option{-std=c++0x} or @option{-std=gnu++0x}. It indicates that some
2346 features likely to be included in C++0x are available. Note that these
2347 features are experimental, and may change or be removed in future
2351 This macro is defined when compiling a C++ source file. It has the
2352 value 1 if the compiler will use weak symbols, COMDAT sections, or
2353 other similar techniques to collapse symbols with ``vague linkage''
2354 that are defined in multiple translation units. If the compiler will
2355 not collapse such symbols, this macro is defined with value 0. In
2356 general, user code should not need to make use of this macro; the
2357 purpose of this macro is to ease implementation of the C++ runtime
2358 library provided with G++.
2360 @item __NEXT_RUNTIME__
2361 This macro is defined, with value 1, if (and only if) the NeXT runtime
2362 (as in @option{-fnext-runtime}) is in use for Objective-C@. If the GNU
2363 runtime is used, this macro is not defined, so that you can use this
2364 macro to determine which runtime (NeXT or GNU) is being used.
2368 These macros are defined, with value 1, if (and only if) the compilation
2369 is for a target where @code{long int} and pointer both use 64-bits and
2370 @code{int} uses 32-bit.
2373 This macro is defined, with value 1, when @option{-fstack-protector} is in
2377 This macro is defined, with value 2, when @option{-fstack-protector-all} is
2380 @item __SSP_STRONG__
2381 This macro is defined, with value 3, when @option{-fstack-protector-strong} is
2384 @item __SSP_EXPLICIT__
2385 This macro is defined, with value 4, when @option{-fstack-protector-explicit} is
2388 @item __SANITIZE_ADDRESS__
2389 This macro is defined, with value 1, when @option{-fsanitize=address}
2390 or @option{-fsanitize=kernel-address} are in use.
2392 @item __SANITIZE_THREAD__
2393 This macro is defined, with value 1, when @option{-fsanitize=thread} is in use.
2396 This macro expands to a string constant that describes the date and time
2397 of the last modification of the current source file. The string constant
2398 contains abbreviated day of the week, month, day of the month, time in
2399 hh:mm:ss form, year and looks like @code{@w{"Sun Sep 16 01:03:52 1973"}}.
2400 If the day of the month is less than 10, it is padded with a space on the left.
2402 If GCC cannot determine the current date, it will emit a warning message
2403 (once per compilation) and @code{__TIMESTAMP__} will expand to
2404 @code{@w{"??? ??? ?? ??:??:?? ????"}}.
2406 @item __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1
2407 @itemx __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2
2408 @itemx __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4
2409 @itemx __GCC_HAVE_SYNC_COMPARE_AND_SWAP_8
2410 @itemx __GCC_HAVE_SYNC_COMPARE_AND_SWAP_16
2411 These macros are defined when the target processor supports atomic compare
2412 and swap operations on operands 1, 2, 4, 8 or 16 bytes in length, respectively.
2414 @item __GCC_HAVE_DWARF2_CFI_ASM
2415 This macro is defined when the compiler is emitting DWARF CFI directives
2416 to the assembler. When this is defined, it is possible to emit those same
2417 directives in inline assembly.
2420 @itemx __FP_FAST_FMAF
2421 @itemx __FP_FAST_FMAL
2422 These macros are defined with value 1 if the backend supports the
2423 @code{fma}, @code{fmaf}, and @code{fmal} builtin functions, so that
2424 the include file @file{math.h} can define the macros
2425 @code{FP_FAST_FMA}, @code{FP_FAST_FMAF}, and @code{FP_FAST_FMAL}
2426 for compatibility with the 1999 C standard.
2429 This macro is defined to indicate the intended level of support for
2430 IEEE 754 (IEC 60559) floating-point arithmetic. It expands to a
2431 nonnegative integer value. If 0, it indicates that the combination of
2432 the compiler configuration and the command-line options is not
2433 intended to support IEEE 754 arithmetic for @code{float} and
2434 @code{double} as defined in C99 and C11 Annex F (for example, that the
2435 standard rounding modes and exceptions are not supported, or that
2436 optimizations are enabled that conflict with IEEE 754 semantics). If
2437 1, it indicates that IEEE 754 arithmetic is intended to be supported;
2438 this does not mean that all relevant language features are supported
2439 by GCC. If 2 or more, it additionally indicates support for IEEE
2440 754-2008 (in particular, that the binary encodings for quiet and
2441 signaling NaNs are as specified in IEEE 754-2008).
2443 This macro does not indicate the default state of command-line options
2444 that control optimizations that C99 and C11 permit to be controlled by
2445 standard pragmas, where those standards do not require a particular
2446 default state. It does not indicate whether optimizations respect
2447 signaling NaN semantics (the macro for that is
2448 @code{__SUPPORT_SNAN__}). It does not indicate support for decimal
2449 floating point or the IEEE 754 binary16 and binary128 types.
2451 @item __GCC_IEC_559_COMPLEX
2452 This macro is defined to indicate the intended level of support for
2453 IEEE 754 (IEC 60559) floating-point arithmetic for complex numbers, as
2454 defined in C99 and C11 Annex G. It expands to a nonnegative integer
2455 value. If 0, it indicates that the combination of the compiler
2456 configuration and the command-line options is not intended to support
2457 Annex G requirements (for example, because @option{-fcx-limited-range}
2458 was used). If 1 or more, it indicates that it is intended to support
2459 those requirements; this does not mean that all relevant language
2460 features are supported by GCC.
2462 @item __NO_MATH_ERRNO__
2463 This macro is defined if @option{-fno-math-errno} is used, or enabled
2464 by another option such as @option{-ffast-math} or by default.
2467 @node System-specific Predefined Macros
2468 @subsection System-specific Predefined Macros
2470 @cindex system-specific predefined macros
2471 @cindex predefined macros, system-specific
2472 @cindex reserved namespace
2474 The C preprocessor normally predefines several macros that indicate what
2475 type of system and machine is in use. They are obviously different on
2476 each target supported by GCC@. This manual, being for all systems and
2477 machines, cannot tell you what their names are, but you can use
2478 @command{cpp -dM} to see them all. @xref{Invocation}. All system-specific
2479 predefined macros expand to a constant value, so you can test them with
2480 either @samp{#ifdef} or @samp{#if}.
2482 The C standard requires that all system-specific macros be part of the
2483 @dfn{reserved namespace}. All names which begin with two underscores,
2484 or an underscore and a capital letter, are reserved for the compiler and
2485 library to use as they wish. However, historically system-specific
2486 macros have had names with no special prefix; for instance, it is common
2487 to find @code{unix} defined on Unix systems. For all such macros, GCC
2488 provides a parallel macro with two underscores added at the beginning
2489 and the end. If @code{unix} is defined, @code{__unix__} will be defined
2490 too. There will never be more than two underscores; the parallel of
2491 @code{_mips} is @code{__mips__}.
2493 When the @option{-ansi} option, or any @option{-std} option that
2494 requests strict conformance, is given to the compiler, all the
2495 system-specific predefined macros outside the reserved namespace are
2496 suppressed. The parallel macros, inside the reserved namespace, remain
2499 We are slowly phasing out all predefined macros which are outside the
2500 reserved namespace. You should never use them in new programs, and we
2501 encourage you to correct older code to use the parallel macros whenever
2502 you find it. We don't recommend you use the system-specific macros that
2503 are in the reserved namespace, either. It is better in the long run to
2504 check specifically for features you need, using a tool such as
2507 @node C++ Named Operators
2508 @subsection C++ Named Operators
2509 @cindex named operators
2510 @cindex C++ named operators
2511 @cindex @file{iso646.h}
2513 In C++, there are eleven keywords which are simply alternate spellings
2514 of operators normally written with punctuation. These keywords are
2515 treated as such even in the preprocessor. They function as operators in
2516 @samp{#if}, and they cannot be defined as macros or poisoned. In C, you
2517 can request that those keywords take their C++ meaning by including
2518 @file{iso646.h}. That header defines each one as a normal object-like
2519 macro expanding to the appropriate punctuator.
2521 These are the named operators and their corresponding punctuators:
2523 @multitable {Named Operator} {Punctuator}
2524 @item Named Operator @tab Punctuator
2525 @item @code{and} @tab @code{&&}
2526 @item @code{and_eq} @tab @code{&=}
2527 @item @code{bitand} @tab @code{&}
2528 @item @code{bitor} @tab @code{|}
2529 @item @code{compl} @tab @code{~}
2530 @item @code{not} @tab @code{!}
2531 @item @code{not_eq} @tab @code{!=}
2532 @item @code{or} @tab @code{||}
2533 @item @code{or_eq} @tab @code{|=}
2534 @item @code{xor} @tab @code{^}
2535 @item @code{xor_eq} @tab @code{^=}
2538 @node Undefining and Redefining Macros
2539 @section Undefining and Redefining Macros
2540 @cindex undefining macros
2541 @cindex redefining macros
2544 If a macro ceases to be useful, it may be @dfn{undefined} with the
2545 @samp{#undef} directive. @samp{#undef} takes a single argument, the
2546 name of the macro to undefine. You use the bare macro name, even if the
2547 macro is function-like. It is an error if anything appears on the line
2548 after the macro name. @samp{#undef} has no effect if the name is not a
2553 x = FOO; @expansion{} x = 4;
2555 x = FOO; @expansion{} x = FOO;
2558 Once a macro has been undefined, that identifier may be @dfn{redefined}
2559 as a macro by a subsequent @samp{#define} directive. The new definition
2560 need not have any resemblance to the old definition.
2562 However, if an identifier which is currently a macro is redefined, then
2563 the new definition must be @dfn{effectively the same} as the old one.
2564 Two macro definitions are effectively the same if:
2566 @item Both are the same type of macro (object- or function-like).
2567 @item All the tokens of the replacement list are the same.
2568 @item If there are any parameters, they are the same.
2569 @item Whitespace appears in the same places in both. It need not be
2570 exactly the same amount of whitespace, though. Remember that comments
2571 count as whitespace.
2575 These definitions are effectively the same:
2577 #define FOUR (2 + 2)
2578 #define FOUR (2 + 2)
2579 #define FOUR (2 /* @r{two} */ + 2)
2584 #define FOUR (2 + 2)
2585 #define FOUR ( 2+2 )
2586 #define FOUR (2 * 2)
2587 #define FOUR(score,and,seven,years,ago) (2 + 2)
2590 If a macro is redefined with a definition that is not effectively the
2591 same as the old one, the preprocessor issues a warning and changes the
2592 macro to use the new definition. If the new definition is effectively
2593 the same, the redefinition is silently ignored. This allows, for
2594 instance, two different headers to define a common macro. The
2595 preprocessor will only complain if the definitions do not match.
2597 @node Directives Within Macro Arguments
2598 @section Directives Within Macro Arguments
2599 @cindex macro arguments and directives
2601 Occasionally it is convenient to use preprocessor directives within
2602 the arguments of a macro. The C and C++ standards declare that
2603 behavior in these cases is undefined.
2605 Versions of CPP prior to 3.2 would reject such constructs with an
2606 error message. This was the only syntactic difference between normal
2607 functions and function-like macros, so it seemed attractive to remove
2608 this limitation, and people would often be surprised that they could
2609 not use macros in this way. Moreover, sometimes people would use
2610 conditional compilation in the argument list to a normal library
2611 function like @samp{printf}, only to find that after a library upgrade
2612 @samp{printf} had changed to be a function-like macro, and their code
2613 would no longer compile. So from version 3.2 we changed CPP to
2614 successfully process arbitrary directives within macro arguments in
2615 exactly the same way as it would have processed the directive were the
2616 function-like macro invocation not present.
2618 If, within a macro invocation, that macro is redefined, then the new
2619 definition takes effect in time for argument pre-expansion, but the
2620 original definition is still used for argument replacement. Here is a
2621 pathological example:
2639 with the semantics described above.
2641 @node Macro Pitfalls
2642 @section Macro Pitfalls
2643 @cindex problems with macros
2644 @cindex pitfalls of macros
2646 In this section we describe some special rules that apply to macros and
2647 macro expansion, and point out certain cases in which the rules have
2648 counter-intuitive consequences that you must watch out for.
2652 * Operator Precedence Problems::
2653 * Swallowing the Semicolon::
2654 * Duplication of Side Effects::
2655 * Self-Referential Macros::
2656 * Argument Prescan::
2657 * Newlines in Arguments::
2661 @subsection Misnesting
2663 When a macro is called with arguments, the arguments are substituted
2664 into the macro body and the result is checked, together with the rest of
2665 the input file, for more macro calls. It is possible to piece together
2666 a macro call coming partially from the macro body and partially from the
2667 arguments. For example,
2670 #define twice(x) (2*(x))
2671 #define call_with_1(x) x(1)
2673 @expansion{} twice(1)
2674 @expansion{} (2*(1))
2677 Macro definitions do not have to have balanced parentheses. By writing
2678 an unbalanced open parenthesis in a macro body, it is possible to create
2679 a macro call that begins inside the macro body but ends outside of it.
2683 #define strange(file) fprintf (file, "%s %d",
2685 strange(stderr) p, 35)
2686 @expansion{} fprintf (stderr, "%s %d", p, 35)
2689 The ability to piece together a macro call can be useful, but the use of
2690 unbalanced open parentheses in a macro body is just confusing, and
2693 @node Operator Precedence Problems
2694 @subsection Operator Precedence Problems
2695 @cindex parentheses in macro bodies
2697 You may have noticed that in most of the macro definition examples shown
2698 above, each occurrence of a macro argument name had parentheses around
2699 it. In addition, another pair of parentheses usually surround the
2700 entire macro definition. Here is why it is best to write macros that
2703 Suppose you define a macro as follows,
2706 #define ceil_div(x, y) (x + y - 1) / y
2710 whose purpose is to divide, rounding up. (One use for this operation is
2711 to compute how many @code{int} objects are needed to hold a certain
2712 number of @code{char} objects.) Then suppose it is used as follows:
2715 a = ceil_div (b & c, sizeof (int));
2716 @expansion{} a = (b & c + sizeof (int) - 1) / sizeof (int);
2720 This does not do what is intended. The operator-precedence rules of
2721 C make it equivalent to this:
2724 a = (b & (c + sizeof (int) - 1)) / sizeof (int);
2728 What we want is this:
2731 a = ((b & c) + sizeof (int) - 1)) / sizeof (int);
2735 Defining the macro as
2738 #define ceil_div(x, y) ((x) + (y) - 1) / (y)
2742 provides the desired result.
2744 Unintended grouping can result in another way. Consider @code{sizeof
2745 ceil_div(1, 2)}. That has the appearance of a C expression that would
2746 compute the size of the type of @code{ceil_div (1, 2)}, but in fact it
2747 means something very different. Here is what it expands to:
2750 sizeof ((1) + (2) - 1) / (2)
2754 This would take the size of an integer and divide it by two. The
2755 precedence rules have put the division outside the @code{sizeof} when it
2756 was intended to be inside.
2758 Parentheses around the entire macro definition prevent such problems.
2759 Here, then, is the recommended way to define @code{ceil_div}:
2762 #define ceil_div(x, y) (((x) + (y) - 1) / (y))
2765 @node Swallowing the Semicolon
2766 @subsection Swallowing the Semicolon
2767 @cindex semicolons (after macro calls)
2769 Often it is desirable to define a macro that expands into a compound
2770 statement. Consider, for example, the following macro, that advances a
2771 pointer (the argument @code{p} says where to find it) across whitespace
2775 #define SKIP_SPACES(p, limit) \
2776 @{ char *lim = (limit); \
2777 while (p < lim) @{ \
2778 if (*p++ != ' ') @{ \
2783 Here backslash-newline is used to split the macro definition, which must
2784 be a single logical line, so that it resembles the way such code would
2785 be laid out if not part of a macro definition.
2787 A call to this macro might be @code{SKIP_SPACES (p, lim)}. Strictly
2788 speaking, the call expands to a compound statement, which is a complete
2789 statement with no need for a semicolon to end it. However, since it
2790 looks like a function call, it minimizes confusion if you can use it
2791 like a function call, writing a semicolon afterward, as in
2792 @code{SKIP_SPACES (p, lim);}
2794 This can cause trouble before @code{else} statements, because the
2795 semicolon is actually a null statement. Suppose you write
2799 SKIP_SPACES (p, lim);
2804 The presence of two statements---the compound statement and a null
2805 statement---in between the @code{if} condition and the @code{else}
2806 makes invalid C code.
2808 The definition of the macro @code{SKIP_SPACES} can be altered to solve
2809 this problem, using a @code{do @dots{} while} statement. Here is how:
2812 #define SKIP_SPACES(p, limit) \
2813 do @{ char *lim = (limit); \
2814 while (p < lim) @{ \
2815 if (*p++ != ' ') @{ \
2816 p--; break; @}@}@} \
2820 Now @code{SKIP_SPACES (p, lim);} expands into
2823 do @{@dots{}@} while (0);
2827 which is one statement. The loop executes exactly once; most compilers
2828 generate no extra code for it.
2830 @node Duplication of Side Effects
2831 @subsection Duplication of Side Effects
2833 @cindex side effects (in macro arguments)
2834 @cindex unsafe macros
2835 Many C programs define a macro @code{min}, for ``minimum'', like this:
2838 #define min(X, Y) ((X) < (Y) ? (X) : (Y))
2841 When you use this macro with an argument containing a side effect,
2845 next = min (x + y, foo (z));
2849 it expands as follows:
2852 next = ((x + y) < (foo (z)) ? (x + y) : (foo (z)));
2856 where @code{x + y} has been substituted for @code{X} and @code{foo (z)}
2859 The function @code{foo} is used only once in the statement as it appears
2860 in the program, but the expression @code{foo (z)} has been substituted
2861 twice into the macro expansion. As a result, @code{foo} might be called
2862 two times when the statement is executed. If it has side effects or if
2863 it takes a long time to compute, the results might not be what you
2864 intended. We say that @code{min} is an @dfn{unsafe} macro.
2866 The best solution to this problem is to define @code{min} in a way that
2867 computes the value of @code{foo (z)} only once. The C language offers
2868 no standard way to do this, but it can be done with GNU extensions as
2873 (@{ typeof (X) x_ = (X); \
2874 typeof (Y) y_ = (Y); \
2875 (x_ < y_) ? x_ : y_; @})
2878 The @samp{(@{ @dots{} @})} notation produces a compound statement that
2879 acts as an expression. Its value is the value of its last statement.
2880 This permits us to define local variables and assign each argument to
2881 one. The local variables have underscores after their names to reduce
2882 the risk of conflict with an identifier of wider scope (it is impossible
2883 to avoid this entirely). Now each argument is evaluated exactly once.
2885 If you do not wish to use GNU C extensions, the only solution is to be
2886 careful when @emph{using} the macro @code{min}. For example, you can
2887 calculate the value of @code{foo (z)}, save it in a variable, and use
2888 that variable in @code{min}:
2892 #define min(X, Y) ((X) < (Y) ? (X) : (Y))
2896 next = min (x + y, tem);
2902 (where we assume that @code{foo} returns type @code{int}).
2904 @node Self-Referential Macros
2905 @subsection Self-Referential Macros
2906 @cindex self-reference
2908 A @dfn{self-referential} macro is one whose name appears in its
2909 definition. Recall that all macro definitions are rescanned for more
2910 macros to replace. If the self-reference were considered a use of the
2911 macro, it would produce an infinitely large expansion. To prevent this,
2912 the self-reference is not considered a macro call. It is passed into
2913 the preprocessor output unchanged. Consider an example:
2916 #define foo (4 + foo)
2920 where @code{foo} is also a variable in your program.
2922 Following the ordinary rules, each reference to @code{foo} will expand
2923 into @code{(4 + foo)}; then this will be rescanned and will expand into
2924 @code{(4 + (4 + foo))}; and so on until the computer runs out of memory.
2926 The self-reference rule cuts this process short after one step, at
2927 @code{(4 + foo)}. Therefore, this macro definition has the possibly
2928 useful effect of causing the program to add 4 to the value of @code{foo}
2929 wherever @code{foo} is referred to.
2931 In most cases, it is a bad idea to take advantage of this feature. A
2932 person reading the program who sees that @code{foo} is a variable will
2933 not expect that it is a macro as well. The reader will come across the
2934 identifier @code{foo} in the program and think its value should be that
2935 of the variable @code{foo}, whereas in fact the value is four greater.
2937 One common, useful use of self-reference is to create a macro which
2938 expands to itself. If you write
2945 then the macro @code{EPERM} expands to @code{EPERM}. Effectively, it is
2946 left alone by the preprocessor whenever it's used in running text. You
2947 can tell that it's a macro with @samp{#ifdef}. You might do this if you
2948 want to define numeric constants with an @code{enum}, but have
2949 @samp{#ifdef} be true for each constant.
2951 If a macro @code{x} expands to use a macro @code{y}, and the expansion of
2952 @code{y} refers to the macro @code{x}, that is an @dfn{indirect
2953 self-reference} of @code{x}. @code{x} is not expanded in this case
2954 either. Thus, if we have
2962 then @code{x} and @code{y} expand as follows:
2966 x @expansion{} (4 + y)
2967 @expansion{} (4 + (2 * x))
2969 y @expansion{} (2 * x)
2970 @expansion{} (2 * (4 + y))
2975 Each macro is expanded when it appears in the definition of the other
2976 macro, but not when it indirectly appears in its own definition.
2978 @node Argument Prescan
2979 @subsection Argument Prescan
2980 @cindex expansion of arguments
2981 @cindex macro argument expansion
2982 @cindex prescan of macro arguments
2984 Macro arguments are completely macro-expanded before they are
2985 substituted into a macro body, unless they are stringified or pasted
2986 with other tokens. After substitution, the entire macro body, including
2987 the substituted arguments, is scanned again for macros to be expanded.
2988 The result is that the arguments are scanned @emph{twice} to expand
2989 macro calls in them.
2991 Most of the time, this has no effect. If the argument contained any
2992 macro calls, they are expanded during the first scan. The result
2993 therefore contains no macro calls, so the second scan does not change
2994 it. If the argument were substituted as given, with no prescan, the
2995 single remaining scan would find the same macro calls and produce the
2998 You might expect the double scan to change the results when a
2999 self-referential macro is used in an argument of another macro
3000 (@pxref{Self-Referential Macros}): the self-referential macro would be
3001 expanded once in the first scan, and a second time in the second scan.
3002 However, this is not what happens. The self-references that do not
3003 expand in the first scan are marked so that they will not expand in the
3006 You might wonder, ``Why mention the prescan, if it makes no difference?
3007 And why not skip it and make the preprocessor faster?'' The answer is
3008 that the prescan does make a difference in three special cases:
3012 Nested calls to a macro.
3014 We say that @dfn{nested} calls to a macro occur when a macro's argument
3015 contains a call to that very macro. For example, if @code{f} is a macro
3016 that expects one argument, @code{f (f (1))} is a nested pair of calls to
3017 @code{f}. The desired expansion is made by expanding @code{f (1)} and
3018 substituting that into the definition of @code{f}. The prescan causes
3019 the expected result to happen. Without the prescan, @code{f (1)} itself
3020 would be substituted as an argument, and the inner use of @code{f} would
3021 appear during the main scan as an indirect self-reference and would not
3025 Macros that call other macros that stringify or concatenate.
3027 If an argument is stringified or concatenated, the prescan does not
3028 occur. If you @emph{want} to expand a macro, then stringify or
3029 concatenate its expansion, you can do that by causing one macro to call
3030 another macro that does the stringification or concatenation. For
3031 instance, if you have
3034 #define AFTERX(x) X_ ## x
3035 #define XAFTERX(x) AFTERX(x)
3036 #define TABLESIZE 1024
3037 #define BUFSIZE TABLESIZE
3040 then @code{AFTERX(BUFSIZE)} expands to @code{X_BUFSIZE}, and
3041 @code{XAFTERX(BUFSIZE)} expands to @code{X_1024}. (Not to
3042 @code{X_TABLESIZE}. Prescan always does a complete expansion.)
3045 Macros used in arguments, whose expansions contain unshielded commas.
3047 This can cause a macro expanded on the second scan to be called with the
3048 wrong number of arguments. Here is an example:
3052 #define bar(x) lose(x)
3053 #define lose(x) (1 + (x))
3056 We would like @code{bar(foo)} to turn into @code{(1 + (foo))}, which
3057 would then turn into @code{(1 + (a,b))}. Instead, @code{bar(foo)}
3058 expands into @code{lose(a,b)}, and you get an error because @code{lose}
3059 requires a single argument. In this case, the problem is easily solved
3060 by the same parentheses that ought to be used to prevent misnesting of
3061 arithmetic operations:
3066 #define bar(x) lose((x))
3069 The extra pair of parentheses prevents the comma in @code{foo}'s
3070 definition from being interpreted as an argument separator.
3074 @node Newlines in Arguments
3075 @subsection Newlines in Arguments
3076 @cindex newlines in macro arguments
3078 The invocation of a function-like macro can extend over many logical
3079 lines. However, in the present implementation, the entire expansion
3080 comes out on one line. Thus line numbers emitted by the compiler or
3081 debugger refer to the line the invocation started on, which might be
3082 different to the line containing the argument causing the problem.
3084 Here is an example illustrating this:
3087 #define ignore_second_arg(a,b,c) a; c
3089 ignore_second_arg (foo (),
3095 The syntax error triggered by the tokens @code{syntax error} results in
3096 an error message citing line three---the line of ignore_second_arg---
3097 even though the problematic code comes from line five.
3099 We consider this a bug, and intend to fix it in the near future.
3102 @chapter Conditionals
3103 @cindex conditionals
3105 A @dfn{conditional} is a directive that instructs the preprocessor to
3106 select whether or not to include a chunk of code in the final token
3107 stream passed to the compiler. Preprocessor conditionals can test
3108 arithmetic expressions, or whether a name is defined as a macro, or both
3109 simultaneously using the special @code{defined} operator.
3111 A conditional in the C preprocessor resembles in some ways an @code{if}
3112 statement in C, but it is important to understand the difference between
3113 them. The condition in an @code{if} statement is tested during the
3114 execution of your program. Its purpose is to allow your program to
3115 behave differently from run to run, depending on the data it is
3116 operating on. The condition in a preprocessing conditional directive is
3117 tested when your program is compiled. Its purpose is to allow different
3118 code to be included in the program depending on the situation at the
3119 time of compilation.
3121 However, the distinction is becoming less clear. Modern compilers often
3122 do test @code{if} statements when a program is compiled, if their
3123 conditions are known not to vary at run time, and eliminate code which
3124 can never be executed. If you can count on your compiler to do this,
3125 you may find that your program is more readable if you use @code{if}
3126 statements with constant conditions (perhaps determined by macros). Of
3127 course, you can only use this to exclude code, not type definitions or
3128 other preprocessing directives, and you can only do it if the code
3129 remains syntactically valid when it is not to be used.
3131 GCC version 3 eliminates this kind of never-executed code even when
3132 not optimizing. Older versions did it only when optimizing.
3135 * Conditional Uses::
3136 * Conditional Syntax::
3140 @node Conditional Uses
3141 @section Conditional Uses
3143 There are three general reasons to use a conditional.
3147 A program may need to use different code depending on the machine or
3148 operating system it is to run on. In some cases the code for one
3149 operating system may be erroneous on another operating system; for
3150 example, it might refer to data types or constants that do not exist on
3151 the other system. When this happens, it is not enough to avoid
3152 executing the invalid code. Its mere presence will cause the compiler
3153 to reject the program. With a preprocessing conditional, the offending
3154 code can be effectively excised from the program when it is not valid.
3157 You may want to be able to compile the same source file into two
3158 different programs. One version might make frequent time-consuming
3159 consistency checks on its intermediate data, or print the values of
3160 those data for debugging, and the other not.
3163 A conditional whose condition is always false is one way to exclude code
3164 from the program but keep it as a sort of comment for future reference.
3167 Simple programs that do not need system-specific logic or complex
3168 debugging hooks generally will not need to use preprocessing
3171 @node Conditional Syntax
3172 @section Conditional Syntax
3175 A conditional in the C preprocessor begins with a @dfn{conditional
3176 directive}: @samp{#if}, @samp{#ifdef} or @samp{#ifndef}.
3191 The simplest sort of conditional is
3197 @var{controlled text}
3199 #endif /* @var{MACRO} */
3203 @cindex conditional group
3204 This block is called a @dfn{conditional group}. @var{controlled text}
3205 will be included in the output of the preprocessor if and only if
3206 @var{MACRO} is defined. We say that the conditional @dfn{succeeds} if
3207 @var{MACRO} is defined, @dfn{fails} if it is not.
3209 The @var{controlled text} inside of a conditional can include
3210 preprocessing directives. They are executed only if the conditional
3211 succeeds. You can nest conditional groups inside other conditional
3212 groups, but they must be completely nested. In other words,
3213 @samp{#endif} always matches the nearest @samp{#ifdef} (or
3214 @samp{#ifndef}, or @samp{#if}). Also, you cannot start a conditional
3215 group in one file and end it in another.
3217 Even if a conditional fails, the @var{controlled text} inside it is
3218 still run through initial transformations and tokenization. Therefore,
3219 it must all be lexically valid C@. Normally the only way this matters is
3220 that all comments and string literals inside a failing conditional group
3221 must still be properly ended.
3223 The comment following the @samp{#endif} is not required, but it is a
3224 good practice if there is a lot of @var{controlled text}, because it
3225 helps people match the @samp{#endif} to the corresponding @samp{#ifdef}.
3226 Older programs sometimes put @var{MACRO} directly after the
3227 @samp{#endif} without enclosing it in a comment. This is invalid code
3228 according to the C standard. CPP accepts it with a warning. It
3229 never affects which @samp{#ifndef} the @samp{#endif} matches.
3232 Sometimes you wish to use some code if a macro is @emph{not} defined.
3233 You can do this by writing @samp{#ifndef} instead of @samp{#ifdef}.
3234 One common use of @samp{#ifndef} is to include code only the first
3235 time a header file is included. @xref{Once-Only Headers}.
3237 Macro definitions can vary between compilations for several reasons.
3238 Here are some samples.
3242 Some macros are predefined on each kind of machine
3243 (@pxref{System-specific Predefined Macros}). This allows you to provide
3244 code specially tuned for a particular machine.
3247 System header files define more macros, associated with the features
3248 they implement. You can test these macros with conditionals to avoid
3249 using a system feature on a machine where it is not implemented.
3252 Macros can be defined or undefined with the @option{-D} and @option{-U}
3253 command-line options when you compile the program. You can arrange to
3254 compile the same source file into two different programs by choosing a
3255 macro name to specify which program you want, writing conditionals to
3256 test whether or how this macro is defined, and then controlling the
3257 state of the macro with command-line options, perhaps set in the
3258 Makefile. @xref{Invocation}.
3261 Your program might have a special header file (often called
3262 @file{config.h}) that is adjusted when the program is compiled. It can
3263 define or not define macros depending on the features of the system and
3264 the desired capabilities of the program. The adjustment can be
3265 automated by a tool such as @command{autoconf}, or done by hand.
3271 The @samp{#if} directive allows you to test the value of an arithmetic
3272 expression, rather than the mere existence of one macro. Its syntax is
3276 #if @var{expression}
3278 @var{controlled text}
3280 #endif /* @var{expression} */
3284 @var{expression} is a C expression of integer type, subject to stringent
3285 restrictions. It may contain
3292 Character constants, which are interpreted as they would be in normal
3296 Arithmetic operators for addition, subtraction, multiplication,
3297 division, bitwise operations, shifts, comparisons, and logical
3298 operations (@code{&&} and @code{||}). The latter two obey the usual
3299 short-circuiting rules of standard C@.
3302 Macros. All macros in the expression are expanded before actual
3303 computation of the expression's value begins.
3306 Uses of the @code{defined} operator, which lets you check whether macros
3307 are defined in the middle of an @samp{#if}.
3310 Identifiers that are not macros, which are all considered to be the
3311 number zero. This allows you to write @code{@w{#if MACRO}} instead of
3312 @code{@w{#ifdef MACRO}}, if you know that MACRO, when defined, will
3313 always have a nonzero value. Function-like macros used without their
3314 function call parentheses are also treated as zero.
3316 In some contexts this shortcut is undesirable. The @option{-Wundef}
3317 option causes GCC to warn whenever it encounters an identifier which is
3318 not a macro in an @samp{#if}.
3321 The preprocessor does not know anything about types in the language.
3322 Therefore, @code{sizeof} operators are not recognized in @samp{#if}, and
3323 neither are @code{enum} constants. They will be taken as identifiers
3324 which are not macros, and replaced by zero. In the case of
3325 @code{sizeof}, this is likely to cause the expression to be invalid.
3327 The preprocessor calculates the value of @var{expression}. It carries
3328 out all calculations in the widest integer type known to the compiler;
3329 on most machines supported by GCC this is 64 bits. This is not the same
3330 rule as the compiler uses to calculate the value of a constant
3331 expression, and may give different results in some cases. If the value
3332 comes out to be nonzero, the @samp{#if} succeeds and the @var{controlled
3333 text} is included; otherwise it is skipped.
3338 @cindex @code{defined}
3339 The special operator @code{defined} is used in @samp{#if} and
3340 @samp{#elif} expressions to test whether a certain name is defined as a
3341 macro. @code{defined @var{name}} and @code{defined (@var{name})} are
3342 both expressions whose value is 1 if @var{name} is defined as a macro at
3343 the current point in the program, and 0 otherwise. Thus, @code{@w{#if
3344 defined MACRO}} is precisely equivalent to @code{@w{#ifdef MACRO}}.
3346 @code{defined} is useful when you wish to test more than one macro for
3347 existence at once. For example,
3350 #if defined (__vax__) || defined (__ns16000__)
3354 would succeed if either of the names @code{__vax__} or
3355 @code{__ns16000__} is defined as a macro.
3357 Conditionals written like this:
3360 #if defined BUFSIZE && BUFSIZE >= 1024
3364 can generally be simplified to just @code{@w{#if BUFSIZE >= 1024}},
3365 since if @code{BUFSIZE} is not defined, it will be interpreted as having
3368 If the @code{defined} operator appears as a result of a macro expansion,
3369 the C standard says the behavior is undefined. GNU cpp treats it as a
3370 genuine @code{defined} operator and evaluates it normally. It will warn
3371 wherever your code uses this feature if you use the command-line option
3372 @option{-pedantic}, since other compilers may handle it differently.
3378 The @samp{#else} directive can be added to a conditional to provide
3379 alternative text to be used if the condition fails. This is what it
3384 #if @var{expression}
3386 #else /* Not @var{expression} */
3388 #endif /* Not @var{expression} */
3393 If @var{expression} is nonzero, the @var{text-if-true} is included and
3394 the @var{text-if-false} is skipped. If @var{expression} is zero, the
3397 You can use @samp{#else} with @samp{#ifdef} and @samp{#ifndef}, too.
3403 One common case of nested conditionals is used to check for more than two
3404 possible alternatives. For example, you might have
3418 Another conditional directive, @samp{#elif}, allows this to be
3419 abbreviated as follows:
3426 #else /* X != 2 and X != 1*/
3428 #endif /* X != 2 and X != 1*/
3431 @samp{#elif} stands for ``else if''. Like @samp{#else}, it goes in the
3432 middle of a conditional group and subdivides it; it does not require a
3433 matching @samp{#endif} of its own. Like @samp{#if}, the @samp{#elif}
3434 directive includes an expression to be tested. The text following the
3435 @samp{#elif} is processed only if the original @samp{#if}-condition
3436 failed and the @samp{#elif} condition succeeds.
3438 More than one @samp{#elif} can go in the same conditional group. Then
3439 the text after each @samp{#elif} is processed only if the @samp{#elif}
3440 condition succeeds after the original @samp{#if} and all previous
3441 @samp{#elif} directives within it have failed.
3443 @samp{#else} is allowed after any number of @samp{#elif} directives, but
3444 @samp{#elif} may not follow @samp{#else}.
3447 @section Deleted Code
3448 @cindex commenting out code
3450 If you replace or delete a part of the program but want to keep the old
3451 code around for future reference, you often cannot simply comment it
3452 out. Block comments do not nest, so the first comment inside the old
3453 code will end the commenting-out. The probable result is a flood of
3456 One way to avoid this problem is to use an always-false conditional
3457 instead. For instance, put @code{#if 0} before the deleted code and
3458 @code{#endif} after it. This works even if the code being turned
3459 off contains conditionals, but they must be entire conditionals
3460 (balanced @samp{#if} and @samp{#endif}).
3462 Some people use @code{#ifdef notdef} instead. This is risky, because
3463 @code{notdef} might be accidentally defined as a macro, and then the
3464 conditional would succeed. @code{#if 0} can be counted on to fail.
3466 Do not use @code{#if 0} for comments which are not C code. Use a real
3467 comment, instead. The interior of @code{#if 0} must consist of complete
3468 tokens; in particular, single-quote characters must balance. Comments
3469 often contain unbalanced single-quote characters (known in English as
3470 apostrophes). These confuse @code{#if 0}. They don't confuse
3474 @chapter Diagnostics
3476 @cindex reporting errors
3477 @cindex reporting warnings
3480 The directive @samp{#error} causes the preprocessor to report a fatal
3481 error. The tokens forming the rest of the line following @samp{#error}
3482 are used as the error message.
3484 You would use @samp{#error} inside of a conditional that detects a
3485 combination of parameters which you know the program does not properly
3486 support. For example, if you know that the program will not run
3487 properly on a VAX, you might write
3492 #error "Won't work on VAXen. See comments at get_last_object."
3497 If you have several configuration parameters that must be set up by
3498 the installation in a consistent way, you can use conditionals to detect
3499 an inconsistency and report it with @samp{#error}. For example,
3502 #if !defined(FOO) && defined(BAR)
3503 #error "BAR requires FOO."
3508 The directive @samp{#warning} is like @samp{#error}, but causes the
3509 preprocessor to issue a warning and continue preprocessing. The tokens
3510 following @samp{#warning} are used as the warning message.
3512 You might use @samp{#warning} in obsolete header files, with a message
3513 directing the user to the header file which should be used instead.
3515 Neither @samp{#error} nor @samp{#warning} macro-expands its argument.
3516 Internal whitespace sequences are each replaced with a single space.
3517 The line must consist of complete tokens. It is wisest to make the
3518 argument of these directives be a single string constant; this avoids
3519 problems with apostrophes and the like.
3522 @chapter Line Control
3523 @cindex line control
3525 The C preprocessor informs the C compiler of the location in your source
3526 code where each token came from. Presently, this is just the file name
3527 and line number. All the tokens resulting from macro expansion are
3528 reported as having appeared on the line of the source file where the
3529 outermost macro was used. We intend to be more accurate in the future.
3531 If you write a program which generates source code, such as the
3532 @command{bison} parser generator, you may want to adjust the preprocessor's
3533 notion of the current file name and line number by hand. Parts of the
3534 output from @command{bison} are generated from scratch, other parts come
3535 from a standard parser file. The rest are copied verbatim from
3536 @command{bison}'s input. You would like compiler error messages and
3537 symbolic debuggers to be able to refer to @code{bison}'s input file.
3540 @command{bison} or any such program can arrange this by writing
3541 @samp{#line} directives into the output file. @samp{#line} is a
3542 directive that specifies the original line number and source file name
3543 for subsequent input in the current preprocessor input file.
3544 @samp{#line} has three variants:
3547 @item #line @var{linenum}
3548 @var{linenum} is a non-negative decimal integer constant. It specifies
3549 the line number which should be reported for the following line of
3550 input. Subsequent lines are counted from @var{linenum}.
3552 @item #line @var{linenum} @var{filename}
3553 @var{linenum} is the same as for the first form, and has the same
3554 effect. In addition, @var{filename} is a string constant. The
3555 following line and all subsequent lines are reported to come from the
3556 file it specifies, until something else happens to change that.
3557 @var{filename} is interpreted according to the normal rules for a string
3558 constant: backslash escapes are interpreted. This is different from
3561 Previous versions of CPP did not interpret escapes in @samp{#line};
3562 we have changed it because the standard requires they be interpreted,
3563 and most other compilers do.
3565 @item #line @var{anything else}
3566 @var{anything else} is checked for macro calls, which are expanded.
3567 The result should match one of the above two forms.
3570 @samp{#line} directives alter the results of the @code{__FILE__} and
3571 @code{__LINE__} predefined macros from that point on. @xref{Standard
3572 Predefined Macros}. They do not have any effect on @samp{#include}'s
3573 idea of the directory containing the current file. This is a change
3574 from GCC 2.95. Previously, a file reading
3577 #line 1 "../src/gram.y"
3581 would search for @file{gram.h} in @file{../src}, then the @option{-I}
3582 chain; the directory containing the physical source file would not be
3583 searched. In GCC 3.0 and later, the @samp{#include} is not affected by
3584 the presence of a @samp{#line} referring to a different directory.
3586 We made this change because the old behavior caused problems when
3587 generated source files were transported between machines. For instance,
3588 it is common practice to ship generated parsers with a source release,
3589 so that people building the distribution do not need to have yacc or
3590 Bison installed. These files frequently have @samp{#line} directives
3591 referring to the directory tree of the system where the distribution was
3592 created. If GCC tries to search for headers in those directories, the
3593 build is likely to fail.
3595 The new behavior can cause failures too, if the generated file is not
3596 in the same directory as its source and it attempts to include a header
3597 which would be visible searching from the directory containing the
3598 source file. However, this problem is easily solved with an additional
3599 @option{-I} switch on the command line. The failures caused by the old
3600 semantics could sometimes be corrected only by editing the generated
3601 files, which is difficult and error-prone.
3606 The @samp{#pragma} directive is the method specified by the C standard
3607 for providing additional information to the compiler, beyond what is
3608 conveyed in the language itself. Three forms of this directive
3609 (commonly known as @dfn{pragmas}) are specified by the 1999 C standard.
3610 A C compiler is free to attach any meaning it likes to other pragmas.
3612 GCC has historically preferred to use extensions to the syntax of the
3613 language, such as @code{__attribute__}, for this purpose. However, GCC
3614 does define a few pragmas of its own. These mostly have effects on the
3615 entire translation unit or source file.
3617 In GCC version 3, all GNU-defined, supported pragmas have been given a
3618 @code{GCC} prefix. This is in line with the @code{STDC} prefix on all
3619 pragmas defined by C99. For backward compatibility, pragmas which were
3620 recognized by previous versions are still recognized without the
3621 @code{GCC} prefix, but that usage is deprecated. Some older pragmas are
3622 deprecated in their entirety. They are not recognized with the
3623 @code{GCC} prefix. @xref{Obsolete Features}.
3625 @cindex @code{_Pragma}
3626 C99 introduces the @code{@w{_Pragma}} operator. This feature addresses a
3627 major problem with @samp{#pragma}: being a directive, it cannot be
3628 produced as the result of macro expansion. @code{@w{_Pragma}} is an
3629 operator, much like @code{sizeof} or @code{defined}, and can be embedded
3632 Its syntax is @code{@w{_Pragma (@var{string-literal})}}, where
3633 @var{string-literal} can be either a normal or wide-character string
3634 literal. It is destringized, by replacing all @samp{\\} with a single
3635 @samp{\} and all @samp{\"} with a @samp{"}. The result is then
3636 processed as if it had appeared as the right hand side of a
3637 @samp{#pragma} directive. For example,
3640 _Pragma ("GCC dependency \"parse.y\"")
3644 has the same effect as @code{#pragma GCC dependency "parse.y"}. The
3645 same effect could be achieved using macros, for example
3648 #define DO_PRAGMA(x) _Pragma (#x)
3649 DO_PRAGMA (GCC dependency "parse.y")
3652 The standard is unclear on where a @code{_Pragma} operator can appear.
3653 The preprocessor does not accept it within a preprocessing conditional
3654 directive like @samp{#if}. To be safe, you are probably best keeping it
3655 out of directives other than @samp{#define}, and putting it on a line of
3658 This manual documents the pragmas which are meaningful to the
3659 preprocessor itself. Other pragmas are meaningful to the C or C++
3660 compilers. They are documented in the GCC manual.
3662 GCC plugins may provide their own pragmas.
3665 @item #pragma GCC dependency
3666 @code{#pragma GCC dependency} allows you to check the relative dates of
3667 the current file and another file. If the other file is more recent than
3668 the current file, a warning is issued. This is useful if the current
3669 file is derived from the other file, and should be regenerated. The
3670 other file is searched for using the normal include search path.
3671 Optional trailing text can be used to give more information in the
3675 #pragma GCC dependency "parse.y"
3676 #pragma GCC dependency "/usr/include/time.h" rerun fixincludes
3679 @item #pragma GCC poison
3680 Sometimes, there is an identifier that you want to remove completely
3681 from your program, and make sure that it never creeps back in. To
3682 enforce this, you can @dfn{poison} the identifier with this pragma.
3683 @code{#pragma GCC poison} is followed by a list of identifiers to
3684 poison. If any of those identifiers appears anywhere in the source
3685 after the directive, it is a hard error. For example,
3688 #pragma GCC poison printf sprintf fprintf
3689 sprintf(some_string, "hello");
3693 will produce an error.
3695 If a poisoned identifier appears as part of the expansion of a macro
3696 which was defined before the identifier was poisoned, it will @emph{not}
3697 cause an error. This lets you poison an identifier without worrying
3698 about system headers defining macros that use it.
3703 #define strrchr rindex
3704 #pragma GCC poison rindex
3705 strrchr(some_string, 'h');
3709 will not produce an error.
3711 @item #pragma GCC system_header
3712 This pragma takes no arguments. It causes the rest of the code in the
3713 current file to be treated as if it came from a system header.
3714 @xref{System Headers}.
3716 @item #pragma GCC warning
3717 @itemx #pragma GCC error
3718 @code{#pragma GCC warning "message"} causes the preprocessor to issue
3719 a warning diagnostic with the text @samp{message}. The message
3720 contained in the pragma must be a single string literal. Similarly,
3721 @code{#pragma GCC error "message"} issues an error message. Unlike
3722 the @samp{#warning} and @samp{#error} directives, these pragmas can be
3723 embedded in preprocessor macros using @samp{_Pragma}.
3727 @node Other Directives
3728 @chapter Other Directives
3732 The @samp{#ident} directive takes one argument, a string constant. On
3733 some systems, that string constant is copied into a special segment of
3734 the object file. On other systems, the directive is ignored. The
3735 @samp{#sccs} directive is a synonym for @samp{#ident}.
3737 These directives are not part of the C standard, but they are not
3738 official GNU extensions either. What historical information we have
3739 been able to find, suggests they originated with System V@.
3741 @cindex null directive
3742 The @dfn{null directive} consists of a @samp{#} followed by a newline,
3743 with only whitespace (including comments) in between. A null directive
3744 is understood as a preprocessing directive but has no effect on the
3745 preprocessor output. The primary significance of the existence of the
3746 null directive is that an input line consisting of just a @samp{#} will
3747 produce no output, rather than a line of output containing just a
3748 @samp{#}. Supposedly some old C programs contain such lines.
3750 @node Preprocessor Output
3751 @chapter Preprocessor Output
3753 When the C preprocessor is used with the C, C++, or Objective-C
3754 compilers, it is integrated into the compiler and communicates a stream
3755 of binary tokens directly to the compiler's parser. However, it can
3756 also be used in the more conventional standalone mode, where it produces
3758 @c FIXME: Document the library interface.
3760 @cindex output format
3761 The output from the C preprocessor looks much like the input, except
3762 that all preprocessing directive lines have been replaced with blank
3763 lines and all comments with spaces. Long runs of blank lines are
3766 The ISO standard specifies that it is implementation defined whether a
3767 preprocessor preserves whitespace between tokens, or replaces it with
3768 e.g.@: a single space. In GNU CPP, whitespace between tokens is collapsed
3769 to become a single space, with the exception that the first token on a
3770 non-directive line is preceded with sufficient spaces that it appears in
3771 the same column in the preprocessed output that it appeared in the
3772 original source file. This is so the output is easy to read.
3773 @xref{Differences from previous versions}. CPP does not insert any
3774 whitespace where there was none in the original source, except where
3775 necessary to prevent an accidental token paste.
3778 Source file name and line number information is conveyed by lines
3782 # @var{linenum} @var{filename} @var{flags}
3786 These are called @dfn{linemarkers}. They are inserted as needed into
3787 the output (but never within a string or character constant). They mean
3788 that the following line originated in file @var{filename} at line
3789 @var{linenum}. @var{filename} will never contain any non-printing
3790 characters; they are replaced with octal escape sequences.
3792 After the file name comes zero or more flags, which are @samp{1},
3793 @samp{2}, @samp{3}, or @samp{4}. If there are multiple flags, spaces
3794 separate them. Here is what the flags mean:
3798 This indicates the start of a new file.
3800 This indicates returning to a file (after having included another file).
3802 This indicates that the following text comes from a system header file,
3803 so certain warnings should be suppressed.
3805 This indicates that the following text should be treated as being
3806 wrapped in an implicit @code{extern "C"} block.
3807 @c maybe cross reference NO_IMPLICIT_EXTERN_C
3810 As an extension, the preprocessor accepts linemarkers in non-assembler
3811 input files. They are treated like the corresponding @samp{#line}
3812 directive, (@pxref{Line Control}), except that trailing flags are
3813 permitted, and are interpreted with the meanings described above. If
3814 multiple flags are given, they must be in ascending order.
3816 Some directives may be duplicated in the output of the preprocessor.
3817 These are @samp{#ident} (always), @samp{#pragma} (only if the
3818 preprocessor does not handle the pragma itself), and @samp{#define} and
3819 @samp{#undef} (with certain debugging options). If this happens, the
3820 @samp{#} of the directive will always be in the first column, and there
3821 will be no space between the @samp{#} and the directive name. If macro
3822 expansion happens to generate tokens which might be mistaken for a
3823 duplicated directive, a space will be inserted between the @samp{#} and
3826 @node Traditional Mode
3827 @chapter Traditional Mode
3829 Traditional (pre-standard) C preprocessing is rather different from
3830 the preprocessing specified by the standard. When GCC is given the
3831 @option{-traditional-cpp} option, it attempts to emulate a traditional
3834 GCC versions 3.2 and later only support traditional mode semantics in
3835 the preprocessor, and not in the compiler front ends. This chapter
3836 outlines the traditional preprocessor semantics we implemented.
3838 The implementation does not correspond precisely to the behavior of
3839 earlier versions of GCC, nor to any true traditional preprocessor.
3840 After all, inconsistencies among traditional implementations were a
3841 major motivation for C standardization. However, we intend that it
3842 should be compatible with true traditional preprocessors in all ways
3843 that actually matter.
3846 * Traditional lexical analysis::
3847 * Traditional macros::
3848 * Traditional miscellany::
3849 * Traditional warnings::
3852 @node Traditional lexical analysis
3853 @section Traditional lexical analysis
3855 The traditional preprocessor does not decompose its input into tokens
3856 the same way a standards-conforming preprocessor does. The input is
3857 simply treated as a stream of text with minimal internal form.
3859 This implementation does not treat trigraphs (@pxref{trigraphs})
3860 specially since they were an invention of the standards committee. It
3861 handles arbitrarily-positioned escaped newlines properly and splices
3862 the lines as you would expect; many traditional preprocessors did not
3865 The form of horizontal whitespace in the input file is preserved in
3866 the output. In particular, hard tabs remain hard tabs. This can be
3867 useful if, for example, you are preprocessing a Makefile.
3869 Traditional CPP only recognizes C-style block comments, and treats the
3870 @samp{/*} sequence as introducing a comment only if it lies outside
3871 quoted text. Quoted text is introduced by the usual single and double
3872 quotes, and also by an initial @samp{<} in a @code{#include}
3875 Traditionally, comments are completely removed and are not replaced
3876 with a space. Since a traditional compiler does its own tokenization
3877 of the output of the preprocessor, this means that comments can
3878 effectively be used as token paste operators. However, comments
3879 behave like separators for text handled by the preprocessor itself,
3880 since it doesn't re-lex its input. For example, in
3887 @samp{foo} and @samp{bar} are distinct identifiers and expanded
3888 separately if they happen to be macros. In other words, this
3889 directive is equivalent to
3902 Generally speaking, in traditional mode an opening quote need not have
3903 a matching closing quote. In particular, a macro may be defined with
3904 replacement text that contains an unmatched quote. Of course, if you
3905 attempt to compile preprocessed output containing an unmatched quote
3906 you will get a syntax error.
3908 However, all preprocessing directives other than @code{#define}
3909 require matching quotes. For example:
3912 #define m This macro's fine and has an unmatched quote
3913 "/* This is not a comment. */
3914 /* @r{This is a comment. The following #include directive
3919 Just as for the ISO preprocessor, what would be a closing quote can be
3920 escaped with a backslash to prevent the quoted text from closing.
3922 @node Traditional macros
3923 @section Traditional macros
3925 The major difference between traditional and ISO macros is that the
3926 former expand to text rather than to a token sequence. CPP removes
3927 all leading and trailing horizontal whitespace from a macro's
3928 replacement text before storing it, but preserves the form of internal
3931 One consequence is that it is legitimate for the replacement text to
3932 contain an unmatched quote (@pxref{Traditional lexical analysis}). An
3933 unclosed string or character constant continues into the text
3934 following the macro call. Similarly, the text at the end of a macro's
3935 expansion can run together with the text after the macro invocation to
3936 produce a single token.
3938 Normally comments are removed from the replacement text after the
3939 macro is expanded, but if the @option{-CC} option is passed on the
3940 command-line comments are preserved. (In fact, the current
3941 implementation removes comments even before saving the macro
3942 replacement text, but it careful to do it in such a way that the
3943 observed effect is identical even in the function-like macro case.)
3945 The ISO stringification operator @samp{#} and token paste operator
3946 @samp{##} have no special meaning. As explained later, an effect
3947 similar to these operators can be obtained in a different way. Macro
3948 names that are embedded in quotes, either from the main file or after
3949 macro replacement, do not expand.
3951 CPP replaces an unquoted object-like macro name with its replacement
3952 text, and then rescans it for further macros to replace. Unlike
3953 standard macro expansion, traditional macro expansion has no provision
3954 to prevent recursion. If an object-like macro appears unquoted in its
3955 replacement text, it will be replaced again during the rescan pass,
3956 and so on @emph{ad infinitum}. GCC detects when it is expanding
3957 recursive macros, emits an error message, and continues after the
3958 offending macro invocation.
3962 #define INC(x) PLUS+x
3967 Function-like macros are similar in form but quite different in
3968 behavior to their ISO counterparts. Their arguments are contained
3969 within parentheses, are comma-separated, and can cross physical lines.
3970 Commas within nested parentheses are not treated as argument
3971 separators. Similarly, a quote in an argument cannot be left
3972 unclosed; a following comma or parenthesis that comes before the
3973 closing quote is treated like any other character. There is no
3974 facility for handling variadic macros.
3976 This implementation removes all comments from macro arguments, unless
3977 the @option{-C} option is given. The form of all other horizontal
3978 whitespace in arguments is preserved, including leading and trailing
3979 whitespace. In particular
3986 is treated as an invocation of the macro @samp{f} with a single
3987 argument consisting of a single space. If you want to invoke a
3988 function-like macro that takes no arguments, you must not leave any
3989 whitespace between the parentheses.
3991 If a macro argument crosses a new line, the new line is replaced with
3992 a space when forming the argument. If the previous line contained an
3993 unterminated quote, the following line inherits the quoted state.
3995 Traditional preprocessors replace parameters in the replacement text
3996 with their arguments regardless of whether the parameters are within
3997 quotes or not. This provides a way to stringize arguments. For
4002 str(/* @r{A comment} */some text )
4003 @expansion{} "some text "
4007 Note that the comment is removed, but that the trailing space is
4008 preserved. Here is an example of using a comment to effect token
4012 #define suffix(x) foo_/**/x
4014 @expansion{} foo_bar
4017 @node Traditional miscellany
4018 @section Traditional miscellany
4020 Here are some things to be aware of when using the traditional
4025 Preprocessing directives are recognized only when their leading
4026 @samp{#} appears in the first column. There can be no whitespace
4027 between the beginning of the line and the @samp{#}, but whitespace can
4028 follow the @samp{#}.
4031 A true traditional C preprocessor does not recognize @samp{#error} or
4032 @samp{#pragma}, and may not recognize @samp{#elif}. CPP supports all
4033 the directives in traditional mode that it supports in ISO mode,
4034 including extensions, with the exception that the effects of
4035 @samp{#pragma GCC poison} are undefined.
4038 __STDC__ is not defined.
4041 If you use digraphs the behavior is undefined.
4044 If a line that looks like a directive appears within macro arguments,
4045 the behavior is undefined.
4049 @node Traditional warnings
4050 @section Traditional warnings
4051 You can request warnings about features that did not exist, or worked
4052 differently, in traditional C with the @option{-Wtraditional} option.
4053 GCC does not warn about features of ISO C which you must use when you
4054 are using a conforming compiler, such as the @samp{#} and @samp{##}
4057 Presently @option{-Wtraditional} warns about:
4061 Macro parameters that appear within string literals in the macro body.
4062 In traditional C macro replacement takes place within string literals,
4063 but does not in ISO C@.
4066 In traditional C, some preprocessor directives did not exist.
4067 Traditional preprocessors would only consider a line to be a directive
4068 if the @samp{#} appeared in column 1 on the line. Therefore
4069 @option{-Wtraditional} warns about directives that traditional C
4070 understands but would ignore because the @samp{#} does not appear as the
4071 first character on the line. It also suggests you hide directives like
4072 @samp{#pragma} not understood by traditional C by indenting them. Some
4073 traditional implementations would not recognize @samp{#elif}, so it
4074 suggests avoiding it altogether.
4077 A function-like macro that appears without an argument list. In some
4078 traditional preprocessors this was an error. In ISO C it merely means
4079 that the macro is not expanded.
4082 The unary plus operator. This did not exist in traditional C@.
4085 The @samp{U} and @samp{LL} integer constant suffixes, which were not
4086 available in traditional C@. (Traditional C does support the @samp{L}
4087 suffix for simple long integer constants.) You are not warned about
4088 uses of these suffixes in macros defined in system headers. For
4089 instance, @code{UINT_MAX} may well be defined as @code{4294967295U}, but
4090 you will not be warned if you use @code{UINT_MAX}.
4092 You can usually avoid the warning, and the related warning about
4093 constants which are so large that they are unsigned, by writing the
4094 integer constant in question in hexadecimal, with no U suffix. Take
4095 care, though, because this gives the wrong result in exotic cases.
4098 @node Implementation Details
4099 @chapter Implementation Details
4101 Here we document details of how the preprocessor's implementation
4102 affects its user-visible behavior. You should try to avoid undue
4103 reliance on behavior described here, as it is possible that it will
4104 change subtly in future implementations.
4106 Also documented here are obsolete features and changes from previous
4110 * Implementation-defined behavior::
4111 * Implementation limits::
4112 * Obsolete Features::
4113 * Differences from previous versions::
4116 @node Implementation-defined behavior
4117 @section Implementation-defined behavior
4118 @cindex implementation-defined behavior
4120 This is how CPP behaves in all the cases which the C standard
4121 describes as @dfn{implementation-defined}. This term means that the
4122 implementation is free to do what it likes, but must document its choice
4124 @c FIXME: Check the C++ standard for more implementation-defined stuff.
4128 @item The mapping of physical source file multi-byte characters to the
4129 execution character set.
4131 The input character set can be specified using the
4132 @option{-finput-charset} option, while the execution character set may
4133 be controlled using the @option{-fexec-charset} and
4134 @option{-fwide-exec-charset} options.
4136 @item Identifier characters.
4137 @anchor{Identifier characters}
4139 The C and C++ standards allow identifiers to be composed of @samp{_}
4140 and the alphanumeric characters. C++ and C99 also allow universal
4141 character names, and C99 further permits implementation-defined
4144 GCC allows the @samp{$} character in identifiers as an extension for
4145 most targets. This is true regardless of the @option{std=} switch,
4146 since this extension cannot conflict with standards-conforming
4147 programs. When preprocessing assembler, however, dollars are not
4148 identifier characters by default.
4150 Currently the targets that by default do not permit @samp{$} are AVR,
4151 IP2K, MMIX, MIPS Irix 3, ARM aout, and PowerPC targets for the AIX
4154 You can override the default with @option{-fdollars-in-identifiers} or
4155 @option{fno-dollars-in-identifiers}. @xref{fdollars-in-identifiers}.
4157 @item Non-empty sequences of whitespace characters.
4159 In textual output, each whitespace sequence is collapsed to a single
4160 space. For aesthetic reasons, the first token on each non-directive
4161 line of output is preceded with sufficient spaces that it appears in the
4162 same column as it did in the original source file.
4164 @item The numeric value of character constants in preprocessor expressions.
4166 The preprocessor and compiler interpret character constants in the
4167 same way; i.e.@: escape sequences such as @samp{\a} are given the
4168 values they would have on the target machine.
4170 The compiler evaluates a multi-character character constant a character
4171 at a time, shifting the previous value left by the number of bits per
4172 target character, and then or-ing in the bit-pattern of the new
4173 character truncated to the width of a target character. The final
4174 bit-pattern is given type @code{int}, and is therefore signed,
4175 regardless of whether single characters are signed or not (a slight
4176 change from versions 3.1 and earlier of GCC)@. If there are more
4177 characters in the constant than would fit in the target @code{int} the
4178 compiler issues a warning, and the excess leading characters are
4181 For example, @code{'ab'} for a target with an 8-bit @code{char} would be
4182 interpreted as @w{@samp{(int) ((unsigned char) 'a' * 256 + (unsigned char)
4183 'b')}}, and @code{'\234a'} as @w{@samp{(int) ((unsigned char) '\234' *
4184 256 + (unsigned char) 'a')}}.
4186 @item Source file inclusion.
4188 For a discussion on how the preprocessor locates header files,
4189 @ref{Include Operation}.
4191 @item Interpretation of the filename resulting from a macro-expanded
4192 @samp{#include} directive.
4194 @xref{Computed Includes}.
4196 @item Treatment of a @samp{#pragma} directive that after macro-expansion
4197 results in a standard pragma.
4199 No macro expansion occurs on any @samp{#pragma} directive line, so the
4200 question does not arise.
4202 Note that GCC does not yet implement any of the standard
4207 @node Implementation limits
4208 @section Implementation limits
4209 @cindex implementation limits
4211 CPP has a small number of internal limits. This section lists the
4212 limits which the C standard requires to be no lower than some minimum,
4213 and all the others known. It is intended that there should be as few limits
4214 as possible. If you encounter an undocumented or inconvenient limit,
4215 please report that as a bug. @xref{Bugs, , Reporting Bugs, gcc, Using
4216 the GNU Compiler Collection (GCC)}.
4218 Where we say something is limited @dfn{only by available memory}, that
4219 means that internal data structures impose no intrinsic limit, and space
4220 is allocated with @code{malloc} or equivalent. The actual limit will
4221 therefore depend on many things, such as the size of other things
4222 allocated by the compiler at the same time, the amount of memory
4223 consumed by other processes on the same computer, etc.
4227 @item Nesting levels of @samp{#include} files.
4229 We impose an arbitrary limit of 200 levels, to avoid runaway recursion.
4230 The standard requires at least 15 levels.
4232 @item Nesting levels of conditional inclusion.
4234 The C standard mandates this be at least 63. CPP is limited only by
4237 @item Levels of parenthesized expressions within a full expression.
4239 The C standard requires this to be at least 63. In preprocessor
4240 conditional expressions, it is limited only by available memory.
4242 @item Significant initial characters in an identifier or macro name.
4244 The preprocessor treats all characters as significant. The C standard
4245 requires only that the first 63 be significant.
4247 @item Number of macros simultaneously defined in a single translation unit.
4249 The standard requires at least 4095 be possible. CPP is limited only
4250 by available memory.
4252 @item Number of parameters in a macro definition and arguments in a macro call.
4254 We allow @code{USHRT_MAX}, which is no smaller than 65,535. The minimum
4255 required by the standard is 127.
4257 @item Number of characters on a logical source line.
4259 The C standard requires a minimum of 4096 be permitted. CPP places
4260 no limits on this, but you may get incorrect column numbers reported in
4261 diagnostics for lines longer than 65,535 characters.
4263 @item Maximum size of a source file.
4265 The standard does not specify any lower limit on the maximum size of a
4266 source file. GNU cpp maps files into memory, so it is limited by the
4267 available address space. This is generally at least two gigabytes.
4268 Depending on the operating system, the size of physical memory may or
4269 may not be a limitation.
4273 @node Obsolete Features
4274 @section Obsolete Features
4276 CPP has some features which are present mainly for compatibility with
4277 older programs. We discourage their use in new code. In some cases,
4278 we plan to remove the feature in a future version of GCC@.
4280 @subsection Assertions
4283 @dfn{Assertions} are a deprecated alternative to macros in writing
4284 conditionals to test what sort of computer or system the compiled
4285 program will run on. Assertions are usually predefined, but you can
4286 define them with preprocessing directives or command-line options.
4288 Assertions were intended to provide a more systematic way to describe
4289 the compiler's target system and we added them for compatibility with
4290 existing compilers. In practice they are just as unpredictable as the
4291 system-specific predefined macros. In addition, they are not part of
4292 any standard, and only a few compilers support them.
4293 Therefore, the use of assertions is @strong{less} portable than the use
4294 of system-specific predefined macros. We recommend you do not use them at
4298 An assertion looks like this:
4301 #@var{predicate} (@var{answer})
4305 @var{predicate} must be a single identifier. @var{answer} can be any
4306 sequence of tokens; all characters are significant except for leading
4307 and trailing whitespace, and differences in internal whitespace
4308 sequences are ignored. (This is similar to the rules governing macro
4309 redefinition.) Thus, @code{(x + y)} is different from @code{(x+y)} but
4310 equivalent to @code{@w{( x + y )}}. Parentheses do not nest inside an
4313 @cindex testing predicates
4314 To test an assertion, you write it in an @samp{#if}. For example, this
4315 conditional succeeds if either @code{vax} or @code{ns16000} has been
4316 asserted as an answer for @code{machine}.
4319 #if #machine (vax) || #machine (ns16000)
4323 You can test whether @emph{any} answer is asserted for a predicate by
4324 omitting the answer in the conditional:
4331 Assertions are made with the @samp{#assert} directive. Its sole
4332 argument is the assertion to make, without the leading @samp{#} that
4333 identifies assertions in conditionals.
4336 #assert @var{predicate} (@var{answer})
4340 You may make several assertions with the same predicate and different
4341 answers. Subsequent assertions do not override previous ones for the
4342 same predicate. All the answers for any given predicate are
4343 simultaneously true.
4345 @cindex assertions, canceling
4347 Assertions can be canceled with the @samp{#unassert} directive. It
4348 has the same syntax as @samp{#assert}. In that form it cancels only the
4349 answer which was specified on the @samp{#unassert} line; other answers
4350 for that predicate remain true. You can cancel an entire predicate by
4351 leaving out the answer:
4354 #unassert @var{predicate}
4358 In either form, if no such assertion has been made, @samp{#unassert} has
4361 You can also make or cancel assertions using command-line options.
4364 @node Differences from previous versions
4365 @section Differences from previous versions
4366 @cindex differences from previous versions
4368 This section details behavior which has changed from previous versions
4369 of CPP@. We do not plan to change it again in the near future, but
4370 we do not promise not to, either.
4372 The ``previous versions'' discussed here are 2.95 and before. The
4373 behavior of GCC 3.0 is mostly the same as the behavior of the widely
4374 used 2.96 and 2.97 development snapshots. Where there are differences,
4375 they generally represent bugs in the snapshots.
4379 @item -I- deprecated
4381 This option has been deprecated in 4.0. @option{-iquote} is meant to
4382 replace the need for this option.
4384 @item Order of evaluation of @samp{#} and @samp{##} operators
4386 The standard does not specify the order of evaluation of a chain of
4387 @samp{##} operators, nor whether @samp{#} is evaluated before, after, or
4388 at the same time as @samp{##}. You should therefore not write any code
4389 which depends on any specific ordering. It is possible to guarantee an
4390 ordering, if you need one, by suitable use of nested macros.
4392 An example of where this might matter is pasting the arguments @samp{1},
4393 @samp{e} and @samp{-2}. This would be fine for left-to-right pasting,
4394 but right-to-left pasting would produce an invalid token @samp{e-2}.
4396 GCC 3.0 evaluates @samp{#} and @samp{##} at the same time and strictly
4397 left to right. Older versions evaluated all @samp{#} operators first,
4398 then all @samp{##} operators, in an unreliable order.
4400 @item The form of whitespace between tokens in preprocessor output
4402 @xref{Preprocessor Output}, for the current textual format. This is
4403 also the format used by stringification. Normally, the preprocessor
4404 communicates tokens directly to the compiler's parser, and whitespace
4405 does not come up at all.
4407 Older versions of GCC preserved all whitespace provided by the user and
4408 inserted lots more whitespace of their own, because they could not
4409 accurately predict when extra spaces were needed to prevent accidental
4412 @item Optional argument when invoking rest argument macros
4414 As an extension, GCC permits you to omit the variable arguments entirely
4415 when you use a variable argument macro. This is forbidden by the 1999 C
4416 standard, and will provoke a pedantic warning with GCC 3.0. Previous
4417 versions accepted it silently.
4419 @item @samp{##} swallowing preceding text in rest argument macros
4421 Formerly, in a macro expansion, if @samp{##} appeared before a variable
4422 arguments parameter, and the set of tokens specified for that argument
4423 in the macro invocation was empty, previous versions of CPP would
4424 back up and remove the preceding sequence of non-whitespace characters
4425 (@strong{not} the preceding token). This extension is in direct
4426 conflict with the 1999 C standard and has been drastically pared back.
4428 In the current version of the preprocessor, if @samp{##} appears between
4429 a comma and a variable arguments parameter, and the variable argument is
4430 omitted entirely, the comma will be removed from the expansion. If the
4431 variable argument is empty, or the token before @samp{##} is not a
4432 comma, then @samp{##} behaves as a normal token paste.
4434 @item @samp{#line} and @samp{#include}
4436 The @samp{#line} directive used to change GCC's notion of the
4437 ``directory containing the current file'', used by @samp{#include} with
4438 a double-quoted header file name. In 3.0 and later, it does not.
4439 @xref{Line Control}, for further explanation.
4441 @item Syntax of @samp{#line}
4443 In GCC 2.95 and previous, the string constant argument to @samp{#line}
4444 was treated the same way as the argument to @samp{#include}: backslash
4445 escapes were not honored, and the string ended at the second @samp{"}.
4446 This is not compliant with the C standard. In GCC 3.0, an attempt was
4447 made to correct the behavior, so that the string was treated as a real
4448 string constant, but it turned out to be buggy. In 3.1, the bugs have
4449 been fixed. (We are not fixing the bugs in 3.0 because they affect
4450 relatively few people and the fix is quite invasive.)
4457 @cindex command line
4459 Most often when you use the C preprocessor you will not have to invoke it
4460 explicitly: the C compiler will do so automatically. However, the
4461 preprocessor is sometimes useful on its own. All the options listed
4462 here are also acceptable to the C compiler and have the same meaning,
4463 except that the C compiler has different rules for specifying the output
4466 @emph{Note:} Whether you use the preprocessor by way of @command{gcc}
4467 or @command{cpp}, the @dfn{compiler driver} is run first. This
4468 program's purpose is to translate your command into invocations of the
4469 programs that do the actual work. Their command-line interfaces are
4470 similar but not identical to the documented interface, and may change
4474 @c man begin SYNOPSIS
4475 cpp [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
4476 [@option{-I}@var{dir}@dots{}] [@option{-iquote}@var{dir}@dots{}]
4477 [@option{-W}@var{warn}@dots{}]
4478 [@option{-M}|@option{-MM}] [@option{-MG}] [@option{-MF} @var{filename}]
4479 [@option{-MP}] [@option{-MQ} @var{target}@dots{}]
4480 [@option{-MT} @var{target}@dots{}]
4481 [@option{-P}] [@option{-fno-working-directory}]
4482 [@option{-x} @var{language}] [@option{-std=}@var{standard}]
4483 @var{infile} @var{outfile}
4485 Only the most useful options are listed here; see below for the remainder.
4487 @c man begin SEEALSO
4488 gpl(7), gfdl(7), fsf-funding(7),
4489 gcc(1), as(1), ld(1), and the Info entries for @file{cpp}, @file{gcc}, and
4494 @c man begin OPTIONS
4495 The C preprocessor expects two file names as arguments, @var{infile} and
4496 @var{outfile}. The preprocessor reads @var{infile} together with any
4497 other files it specifies with @samp{#include}. All the output generated
4498 by the combined input files is written in @var{outfile}.
4500 Either @var{infile} or @var{outfile} may be @option{-}, which as
4501 @var{infile} means to read from standard input and as @var{outfile}
4502 means to write to standard output. Also, if either file is omitted, it
4503 means the same as if @option{-} had been specified for that file.
4505 Unless otherwise noted, or the option ends in @samp{=}, all options
4506 which take an argument may have that argument appear either immediately
4507 after the option, or with a space between option and argument:
4508 @option{-Ifoo} and @option{-I foo} have the same effect.
4510 @cindex grouping options
4511 @cindex options, grouping
4512 Many options have multi-letter names; therefore multiple single-letter
4513 options may @emph{not} be grouped: @option{-dM} is very different from
4517 @include cppopts.texi
4520 @node Environment Variables
4521 @chapter Environment Variables
4522 @cindex environment variables
4523 @c man begin ENVIRONMENT
4525 This section describes the environment variables that affect how CPP
4526 operates. You can use them to specify directories or prefixes to use
4527 when searching for include files, or to control dependency output.
4529 Note that you can also specify places to search using options such as
4530 @option{-I}, and control dependency output with options like
4531 @option{-M} (@pxref{Invocation}). These take precedence over
4532 environment variables, which in turn take precedence over the
4533 configuration of GCC@.
4535 @include cppenv.texi
4542 @node Index of Directives
4543 @unnumbered Index of Directives
4547 @unnumbered Option Index
4549 CPP's command-line options and environment variables are indexed here
4550 without any initial @samp{-} or @samp{--}.
4555 @unnumbered Concept Index