1 This is doc/cpp.info, produced by makeinfo version 4.8 from
2 ../../../toolchain/android-toolchain/gcc-4.3.1/gcc/doc/cpp.texi.
4 Copyright (C) 1987, 1989, 1991, 1992, 1993, 1994, 1995, 1996, 1997,
5 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007 Free
6 Software Foundation, Inc.
8 Permission is granted to copy, distribute and/or modify this document
9 under the terms of the GNU Free Documentation License, Version 1.1 or
10 any later version published by the Free Software Foundation. A copy of
11 the license is included in the section entitled "GNU Free Documentation
14 This manual contains no Invariant Sections. The Front-Cover Texts
15 are (a) (see below), and the Back-Cover Texts are (b) (see below).
17 (a) The FSF's Front-Cover Text is:
21 (b) The FSF's Back-Cover Text is:
23 You have freedom to copy and modify this GNU Manual, like GNU
24 software. Copies published by the Free Software Foundation raise
25 funds for GNU development.
27 INFO-DIR-SECTION Software development
29 * Cpp: (cpp). The GNU C preprocessor.
33 File: cpp.info, Node: Top, Next: Overview, Up: (dir)
38 The C preprocessor implements the macro language used to transform C,
39 C++, and Objective-C programs before they are compiled. It can also be
52 * Preprocessor Output::
54 * Implementation Details::
56 * Environment Variables::
57 * GNU Free Documentation License::
58 * Index of Directives::
62 --- The Detailed Node Listing ---
67 * Initial processing::
69 * The preprocessing language::
83 * Object-like Macros::
84 * Function-like Macros::
90 * Undefining and Redefining Macros::
91 * Directives Within Macro Arguments::
96 * Standard Predefined Macros::
97 * Common Predefined Macros::
98 * System-specific Predefined Macros::
99 * C++ Named Operators::
104 * Operator Precedence Problems::
105 * Swallowing the Semicolon::
106 * Duplication of Side Effects::
107 * Self-Referential Macros::
109 * Newlines in Arguments::
114 * Conditional Syntax::
125 Implementation Details
127 * Implementation-defined behavior::
128 * Implementation limits::
129 * Obsolete Features::
130 * Differences from previous versions::
135 * Obsolete once-only headers::
137 Copyright (C) 1987, 1989, 1991, 1992, 1993, 1994, 1995, 1996, 1997,
138 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007 Free
139 Software Foundation, Inc.
141 Permission is granted to copy, distribute and/or modify this document
142 under the terms of the GNU Free Documentation License, Version 1.1 or
143 any later version published by the Free Software Foundation. A copy of
144 the license is included in the section entitled "GNU Free Documentation
147 This manual contains no Invariant Sections. The Front-Cover Texts
148 are (a) (see below), and the Back-Cover Texts are (b) (see below).
150 (a) The FSF's Front-Cover Text is:
154 (b) The FSF's Back-Cover Text is:
156 You have freedom to copy and modify this GNU Manual, like GNU
157 software. Copies published by the Free Software Foundation raise
158 funds for GNU development.
161 File: cpp.info, Node: Overview, Next: Header Files, Prev: Top, Up: Top
166 The C preprocessor, often known as "cpp", is a "macro processor" that
167 is used automatically by the C compiler to transform your program
168 before compilation. It is called a macro processor because it allows
169 you to define "macros", which are brief abbreviations for longer
172 The C preprocessor is intended to be used only with C, C++, and
173 Objective-C source code. In the past, it has been abused as a general
174 text processor. It will choke on input which does not obey C's lexical
175 rules. For example, apostrophes will be interpreted as the beginning of
176 character constants, and cause errors. Also, you cannot rely on it
177 preserving characteristics of the input which are not significant to
178 C-family languages. If a Makefile is preprocessed, all the hard tabs
179 will be removed, and the Makefile will not work.
181 Having said that, you can often get away with using cpp on things
182 which are not C. Other Algol-ish programming languages are often safe
183 (Pascal, Ada, etc.) So is assembly, with caution. `-traditional-cpp'
184 mode preserves more white space, and is otherwise more permissive. Many
185 of the problems can be avoided by writing C or C++ style comments
186 instead of native language comments, and keeping macros simple.
188 Wherever possible, you should use a preprocessor geared to the
189 language you are writing in. Modern versions of the GNU assembler have
190 macro facilities. Most high level programming languages have their own
191 conditional compilation and inclusion mechanism. If all else fails,
192 try a true general text processor, such as GNU M4.
194 C preprocessors vary in some details. This manual discusses the GNU
195 C preprocessor, which provides a small superset of the features of ISO
196 Standard C. In its default mode, the GNU C preprocessor does not do a
197 few things required by the standard. These are features which are
198 rarely, if ever, used, and may cause surprising changes to the meaning
199 of a program which does not expect them. To get strict ISO Standard C,
200 you should use the `-std=c89' or `-std=c99' options, depending on which
201 version of the standard you want. To get all the mandatory
202 diagnostics, you must also use `-pedantic'. *Note Invocation::.
204 This manual describes the behavior of the ISO preprocessor. To
205 minimize gratuitous differences, where the ISO preprocessor's behavior
206 does not conflict with traditional semantics, the traditional
207 preprocessor should behave the same way. The various differences that
208 do exist are detailed in the section *Note Traditional Mode::.
210 For clarity, unless noted otherwise, references to `CPP' in this
211 manual refer to GNU CPP.
216 * Initial processing::
218 * The preprocessing language::
221 File: cpp.info, Node: Character sets, Next: Initial processing, Up: Overview
226 Source code character set processing in C and related languages is
227 rather complicated. The C standard discusses two character sets, but
228 there are really at least four.
230 The files input to CPP might be in any character set at all. CPP's
231 very first action, before it even looks for line boundaries, is to
232 convert the file into the character set it uses for internal
233 processing. That set is what the C standard calls the "source"
234 character set. It must be isomorphic with ISO 10646, also known as
235 Unicode. CPP uses the UTF-8 encoding of Unicode.
237 The character sets of the input files are specified using the
238 `-finput-charset=' option.
240 All preprocessing work (the subject of the rest of this manual) is
241 carried out in the source character set. If you request textual output
242 from the preprocessor with the `-E' option, it will be in UTF-8.
244 After preprocessing is complete, string and character constants are
245 converted again, into the "execution" character set. This character
246 set is under control of the user; the default is UTF-8, matching the
247 source character set. Wide string and character constants have their
248 own character set, which is not called out specifically in the
249 standard. Again, it is under control of the user. The default is
250 UTF-16 or UTF-32, whichever fits in the target's `wchar_t' type, in the
251 target machine's byte order.(1) Octal and hexadecimal escape sequences
252 do not undergo conversion; '\x12' has the value 0x12 regardless of the
253 currently selected execution character set. All other escapes are
254 replaced by the character in the source character set that they
255 represent, then converted to the execution character set, just like
256 unescaped characters.
258 Unless the experimental `-fextended-identifiers' option is used, GCC
259 does not permit the use of characters outside the ASCII range, nor `\u'
260 and `\U' escapes, in identifiers. Even with that option, characters
261 outside the ASCII range can only be specified with the `\u' and `\U'
262 escapes, not used directly in identifiers.
264 ---------- Footnotes ----------
266 (1) UTF-16 does not meet the requirements of the C standard for a
267 wide character set, but the choice of 16-bit `wchar_t' is enshrined in
268 some system ABIs so we cannot fix this.
271 File: cpp.info, Node: Initial processing, Next: Tokenization, Prev: Character sets, Up: Overview
273 1.2 Initial processing
274 ======================
276 The preprocessor performs a series of textual transformations on its
277 input. These happen before all other processing. Conceptually, they
278 happen in a rigid order, and the entire file is run through each
279 transformation before the next one begins. CPP actually does them all
280 at once, for performance reasons. These transformations correspond
281 roughly to the first three "phases of translation" described in the C
284 1. The input file is read into memory and broken into lines.
286 Different systems use different conventions to indicate the end of
287 a line. GCC accepts the ASCII control sequences `LF', `CR LF' and
288 `CR' as end-of-line markers. These are the canonical sequences
289 used by Unix, DOS and VMS, and the classic Mac OS (before OSX)
290 respectively. You may therefore safely copy source code written
291 on any of those systems to a different one and use it without
292 conversion. (GCC may lose track of the current line number if a
293 file doesn't consistently use one convention, as sometimes happens
294 when it is edited on computers with different conventions that
295 share a network file system.)
297 If the last line of any input file lacks an end-of-line marker,
298 the end of the file is considered to implicitly supply one. The C
299 standard says that this condition provokes undefined behavior, so
300 GCC will emit a warning message.
302 2. If trigraphs are enabled, they are replaced by their corresponding
303 single characters. By default GCC ignores trigraphs, but if you
304 request a strictly conforming mode with the `-std' option, or you
305 specify the `-trigraphs' option, then it converts them.
307 These are nine three-character sequences, all starting with `??',
308 that are defined by ISO C to stand for single characters. They
309 permit obsolete systems that lack some of C's punctuation to use
310 C. For example, `??/' stands for `\', so '??/n' is a character
311 constant for a newline.
313 Trigraphs are not popular and many compilers implement them
314 incorrectly. Portable code should not rely on trigraphs being
315 either converted or ignored. With `-Wtrigraphs' GCC will warn you
316 when a trigraph may change the meaning of your program if it were
317 converted. *Note Wtrigraphs::.
319 In a string constant, you can prevent a sequence of question marks
320 from being confused with a trigraph by inserting a backslash
321 between the question marks, or by separating the string literal at
322 the trigraph and making use of string literal concatenation.
323 "(??\?)" is the string `(???)', not `(?]'. Traditional C
324 compilers do not recognize these idioms.
326 The nine trigraphs and their replacements are
328 Trigraph: ??( ??) ??< ??> ??= ??/ ??' ??! ??-
329 Replacement: [ ] { } # \ ^ | ~
331 3. Continued lines are merged into one long line.
333 A continued line is a line which ends with a backslash, `\'. The
334 backslash is removed and the following line is joined with the
335 current one. No space is inserted, so you may split a line
336 anywhere, even in the middle of a word. (It is generally more
337 readable to split lines only at white space.)
339 The trailing backslash on a continued line is commonly referred to
340 as a "backslash-newline".
342 If there is white space between a backslash and the end of a line,
343 that is still a continued line. However, as this is usually the
344 result of an editing mistake, and many compilers will not accept
345 it as a continued line, GCC will warn you about it.
347 4. All comments are replaced with single spaces.
349 There are two kinds of comments. "Block comments" begin with `/*'
350 and continue until the next `*/'. Block comments do not nest:
352 /* this is /* one comment */ text outside comment
354 "Line comments" begin with `//' and continue to the end of the
355 current line. Line comments do not nest either, but it does not
356 matter, because they would end in the same place anyway.
358 // this is // one comment
361 It is safe to put line comments inside block comments, or vice versa.
364 // contains line comment
368 // line comment /* contains block comment */
370 But beware of commenting out one end of a block comment with a line
373 // l.c. /* block comment begins
374 oops! this isn't a comment anymore */
376 Comments are not recognized within string literals. "/* blah */" is
377 the string constant `/* blah */', not an empty string.
379 Line comments are not in the 1989 edition of the C standard, but they
380 are recognized by GCC as an extension. In C++ and in the 1999 edition
381 of the C standard, they are an official part of the language.
383 Since these transformations happen before all other processing, you
384 can split a line mechanically with backslash-newline anywhere. You can
385 comment out the end of a line. You can continue a line comment onto the
386 next line with backslash-newline. You can even split `/*', `*/', and
387 `//' onto multiple lines with backslash-newline. For example:
397 is equivalent to `#define FOO 1020'. All these tricks are extremely
398 confusing and should not be used in code intended to be readable.
400 There is no way to prevent a backslash at the end of a line from
401 being interpreted as a backslash-newline. This cannot affect any
402 correct program, however.
405 File: cpp.info, Node: Tokenization, Next: The preprocessing language, Prev: Initial processing, Up: Overview
410 After the textual transformations are finished, the input file is
411 converted into a sequence of "preprocessing tokens". These mostly
412 correspond to the syntactic tokens used by the C compiler, but there are
413 a few differences. White space separates tokens; it is not itself a
414 token of any kind. Tokens do not have to be separated by white space,
415 but it is often necessary to avoid ambiguities.
417 When faced with a sequence of characters that has more than one
418 possible tokenization, the preprocessor is greedy. It always makes
419 each token, starting from the left, as big as possible before moving on
420 to the next token. For instance, `a+++++b' is interpreted as
421 `a ++ ++ + b', not as `a ++ + ++ b', even though the latter
422 tokenization could be part of a valid C program and the former could
425 Once the input file is broken into tokens, the token boundaries never
426 change, except when the `##' preprocessing operator is used to paste
427 tokens together. *Note Concatenation::. For example,
435 The compiler does not re-tokenize the preprocessor's output. Each
436 preprocessing token becomes one compiler token.
438 Preprocessing tokens fall into five broad classes: identifiers,
439 preprocessing numbers, string literals, punctuators, and other. An
440 "identifier" is the same as an identifier in C: any sequence of
441 letters, digits, or underscores, which begins with a letter or
442 underscore. Keywords of C have no significance to the preprocessor;
443 they are ordinary identifiers. You can define a macro whose name is a
444 keyword, for instance. The only identifier which can be considered a
445 preprocessing keyword is `defined'. *Note Defined::.
447 This is mostly true of other languages which use the C preprocessor.
448 However, a few of the keywords of C++ are significant even in the
449 preprocessor. *Note C++ Named Operators::.
451 In the 1999 C standard, identifiers may contain letters which are not
452 part of the "basic source character set", at the implementation's
453 discretion (such as accented Latin letters, Greek letters, or Chinese
454 ideograms). This may be done with an extended character set, or the
455 `\u' and `\U' escape sequences. The implementation of this feature in
456 GCC is experimental; such characters are only accepted in the `\u' and
457 `\U' forms and only if `-fextended-identifiers' is used.
459 As an extension, GCC treats `$' as a letter. This is for
460 compatibility with some systems, such as VMS, where `$' is commonly
461 used in system-defined function and object names. `$' is not a letter
462 in strictly conforming mode, or if you specify the `-$' option. *Note
465 A "preprocessing number" has a rather bizarre definition. The
466 category includes all the normal integer and floating point constants
467 one expects of C, but also a number of other things one might not
468 initially recognize as a number. Formally, preprocessing numbers begin
469 with an optional period, a required decimal digit, and then continue
470 with any sequence of letters, digits, underscores, periods, and
471 exponents. Exponents are the two-character sequences `e+', `e-', `E+',
472 `E-', `p+', `p-', `P+', and `P-'. (The exponents that begin with `p'
473 or `P' are new to C99. They are used for hexadecimal floating-point
476 The purpose of this unusual definition is to isolate the preprocessor
477 from the full complexity of numeric constants. It does not have to
478 distinguish between lexically valid and invalid floating-point numbers,
479 which is complicated. The definition also permits you to split an
480 identifier at any position and get exactly two tokens, which can then be
481 pasted back together with the `##' operator.
483 It's possible for preprocessing numbers to cause programs to be
484 misinterpreted. For example, `0xE+12' is a preprocessing number which
485 does not translate to any valid numeric constant, therefore a syntax
486 error. It does not mean `0xE + 12', which is what you might have
489 "String literals" are string constants, character constants, and
490 header file names (the argument of `#include').(1) String constants
491 and character constants are straightforward: "..." or '...'. In either
492 case embedded quotes should be escaped with a backslash: '\'' is the
493 character constant for `''. There is no limit on the length of a
494 character constant, but the value of a character constant that contains
495 more than one character is implementation-defined. *Note
496 Implementation Details::.
498 Header file names either look like string constants, "...", or are
499 written with angle brackets instead, <...>. In either case, backslash
500 is an ordinary character. There is no way to escape the closing quote
501 or angle bracket. The preprocessor looks for the header file in
502 different places depending on which form you use. *Note Include
505 No string literal may extend past the end of a line. Older versions
506 of GCC accepted multi-line string constants. You may use continued
507 lines instead, or string constant concatenation. *Note Differences
508 from previous versions::.
510 "Punctuators" are all the usual bits of punctuation which are
511 meaningful to C and C++. All but three of the punctuation characters in
512 ASCII are C punctuators. The exceptions are `@', `$', and ``'. In
513 addition, all the two- and three-character operators are punctuators.
514 There are also six "digraphs", which the C++ standard calls
515 "alternative tokens", which are merely alternate ways to spell other
516 punctuators. This is a second attempt to work around missing
517 punctuation in obsolete systems. It has no negative side effects,
518 unlike trigraphs, but does not cover as much ground. The digraphs and
519 their corresponding normal punctuators are:
521 Digraph: <% %> <: :> %: %:%:
522 Punctuator: { } [ ] # ##
524 Any other single character is considered "other". It is passed on to
525 the preprocessor's output unmolested. The C compiler will almost
526 certainly reject source code containing "other" tokens. In ASCII, the
527 only other characters are `@', `$', ``', and control characters other
528 than NUL (all bits zero). (Note that `$' is normally considered a
529 letter.) All characters with the high bit set (numeric range
530 0x7F-0xFF) are also "other" in the present implementation. This will
531 change when proper support for international character sets is added to
534 NUL is a special case because of the high probability that its
535 appearance is accidental, and because it may be invisible to the user
536 (many terminals do not display NUL at all). Within comments, NULs are
537 silently ignored, just as any other character would be. In running
538 text, NUL is considered white space. For example, these two directives
539 have the same meaning.
544 (where `^@' is ASCII NUL). Within string or character constants, NULs
545 are preserved. In the latter two cases the preprocessor emits a
548 ---------- Footnotes ----------
550 (1) The C standard uses the term "string literal" to refer only to
551 what we are calling "string constants".
554 File: cpp.info, Node: The preprocessing language, Prev: Tokenization, Up: Overview
556 1.4 The preprocessing language
557 ==============================
559 After tokenization, the stream of tokens may simply be passed straight
560 to the compiler's parser. However, if it contains any operations in the
561 "preprocessing language", it will be transformed first. This stage
562 corresponds roughly to the standard's "translation phase 4" and is what
563 most people think of as the preprocessor's job.
565 The preprocessing language consists of "directives" to be executed
566 and "macros" to be expanded. Its primary capabilities are:
568 * Inclusion of header files. These are files of declarations that
569 can be substituted into your program.
571 * Macro expansion. You can define "macros", which are abbreviations
572 for arbitrary fragments of C code. The preprocessor will replace
573 the macros with their definitions throughout the program. Some
574 macros are automatically defined for you.
576 * Conditional compilation. You can include or exclude parts of the
577 program according to various conditions.
579 * Line control. If you use a program to combine or rearrange source
580 files into an intermediate file which is then compiled, you can
581 use line control to inform the compiler where each source line
582 originally came from.
584 * Diagnostics. You can detect problems at compile time and issue
587 There are a few more, less useful, features.
589 Except for expansion of predefined macros, all these operations are
590 triggered with "preprocessing directives". Preprocessing directives
591 are lines in your program that start with `#'. Whitespace is allowed
592 before and after the `#'. The `#' is followed by an identifier, the
593 "directive name". It specifies the operation to perform. Directives
594 are commonly referred to as `#NAME' where NAME is the directive name.
595 For example, `#define' is the directive that defines a macro.
597 The `#' which begins a directive cannot come from a macro expansion.
598 Also, the directive name is not macro expanded. Thus, if `foo' is
599 defined as a macro expanding to `define', that does not make `#foo' a
600 valid preprocessing directive.
602 The set of valid directive names is fixed. Programs cannot define
603 new preprocessing directives.
605 Some directives require arguments; these make up the rest of the
606 directive line and must be separated from the directive name by
607 whitespace. For example, `#define' must be followed by a macro name
608 and the intended expansion of the macro.
610 A preprocessing directive cannot cover more than one line. The line
611 may, however, be continued with backslash-newline, or by a block comment
612 which extends past the end of the line. In either case, when the
613 directive is processed, the continuations have already been merged with
614 the first line to make one long line.
617 File: cpp.info, Node: Header Files, Next: Macros, Prev: Overview, Up: Top
622 A header file is a file containing C declarations and macro definitions
623 (*note Macros::) to be shared between several source files. You request
624 the use of a header file in your program by "including" it, with the C
625 preprocessing directive `#include'.
627 Header files serve two purposes.
629 * System header files declare the interfaces to parts of the
630 operating system. You include them in your program to supply the
631 definitions and declarations you need to invoke system calls and
634 * Your own header files contain declarations for interfaces between
635 the source files of your program. Each time you have a group of
636 related declarations and macro definitions all or most of which
637 are needed in several different source files, it is a good idea to
638 create a header file for them.
640 Including a header file produces the same results as copying the
641 header file into each source file that needs it. Such copying would be
642 time-consuming and error-prone. With a header file, the related
643 declarations appear in only one place. If they need to be changed, they
644 can be changed in one place, and programs that include the header file
645 will automatically use the new version when next recompiled. The header
646 file eliminates the labor of finding and changing all the copies as well
647 as the risk that a failure to find one copy will result in
648 inconsistencies within a program.
650 In C, the usual convention is to give header files names that end
651 with `.h'. It is most portable to use only letters, digits, dashes, and
652 underscores in header file names, and at most one dot.
657 * Include Operation::
659 * Once-Only Headers::
660 * Computed Includes::
665 File: cpp.info, Node: Include Syntax, Next: Include Operation, Up: Header Files
670 Both user and system header files are included using the preprocessing
671 directive `#include'. It has two variants:
674 This variant is used for system header files. It searches for a
675 file named FILE in a standard list of system directories. You can
676 prepend directories to this list with the `-I' option (*note
680 This variant is used for header files of your own program. It
681 searches for a file named FILE first in the directory containing
682 the current file, then in the quote directories and then the same
683 directories used for `<FILE>'. You can prepend directories to the
684 list of quote directories with the `-iquote' option.
686 The argument of `#include', whether delimited with quote marks or
687 angle brackets, behaves like a string constant in that comments are not
688 recognized, and macro names are not expanded. Thus, `#include <x/*y>'
689 specifies inclusion of a system header file named `x/*y'.
691 However, if backslashes occur within FILE, they are considered
692 ordinary text characters, not escape characters. None of the character
693 escape sequences appropriate to string constants in C are processed.
694 Thus, `#include "x\n\\y"' specifies a filename containing three
695 backslashes. (Some systems interpret `\' as a pathname separator. All
696 of these also interpret `/' the same way. It is most portable to use
699 It is an error if there is anything (other than comments) on the line
703 File: cpp.info, Node: Include Operation, Next: Search Path, Prev: Include Syntax, Up: Header Files
705 2.2 Include Operation
706 =====================
708 The `#include' directive works by directing the C preprocessor to scan
709 the specified file as input before continuing with the rest of the
710 current file. The output from the preprocessor contains the output
711 already generated, followed by the output resulting from the included
712 file, followed by the output that comes from the text after the
713 `#include' directive. For example, if you have a header file
714 `header.h' as follows,
718 and a main program called `program.c' that uses the header file, like
730 the compiler will see the same token stream as it would if `program.c'
742 Included files are not limited to declarations and macro definitions;
743 those are merely the typical uses. Any fragment of a C program can be
744 included from another file. The include file could even contain the
745 beginning of a statement that is concluded in the containing file, or
746 the end of a statement that was started in the including file. However,
747 an included file must consist of complete tokens. Comments and string
748 literals which have not been closed by the end of an included file are
749 invalid. For error recovery, they are considered to end at the end of
752 To avoid confusion, it is best if header files contain only complete
753 syntactic units--function declarations or definitions, type
756 The line following the `#include' directive is always treated as a
757 separate line by the C preprocessor, even if the included file lacks a
761 File: cpp.info, Node: Search Path, Next: Once-Only Headers, Prev: Include Operation, Up: Header Files
766 GCC looks in several different places for headers. On a normal Unix
767 system, if you do not instruct it otherwise, it will look for headers
768 requested with `#include <FILE>' in:
771 LIBDIR/gcc/TARGET/VERSION/include
775 For C++ programs, it will also look in `/usr/include/g++-v3', first.
776 In the above, TARGET is the canonical name of the system GCC was
777 configured to compile code for; often but not always the same as the
778 canonical name of the system it runs on. VERSION is the version of GCC
781 You can add to this list with the `-IDIR' command line option. All
782 the directories named by `-I' are searched, in left-to-right order,
783 _before_ the default directories. The only exception is when `dir' is
784 already searched by default. In this case, the option is ignored and
785 the search order for system directories remains unchanged.
787 Duplicate directories are removed from the quote and bracket search
788 chains before the two chains are merged to make the final search chain.
789 Thus, it is possible for a directory to occur twice in the final search
790 chain if it was specified in both the quote and bracket chains.
792 You can prevent GCC from searching any of the default directories
793 with the `-nostdinc' option. This is useful when you are compiling an
794 operating system kernel or some other program that does not use the
795 standard C library facilities, or the standard C library itself. `-I'
796 options are not ignored as described above when `-nostdinc' is in
799 GCC looks for headers requested with `#include "FILE"' first in the
800 directory containing the current file, then in the directories as
801 specified by `-iquote' options, then in the same places it would have
802 looked for a header requested with angle brackets. For example, if
803 `/usr/include/sys/stat.h' contains `#include "types.h"', GCC looks for
804 `types.h' first in `/usr/include/sys', then in its usual search path.
806 `#line' (*note Line Control::) does not change GCC's idea of the
807 directory containing the current file.
809 You may put `-I-' at any point in your list of `-I' options. This
810 has two effects. First, directories appearing before the `-I-' in the
811 list are searched only for headers requested with quote marks.
812 Directories after `-I-' are searched for all headers. Second, the
813 directory containing the current file is not searched for anything,
814 unless it happens to be one of the directories named by an `-I' switch.
815 `-I-' is deprecated, `-iquote' should be used instead.
817 `-I. -I-' is not the same as no `-I' options at all, and does not
818 cause the same behavior for `<>' includes that `""' includes get with
819 no special options. `-I.' searches the compiler's current working
820 directory for header files. That may or may not be the same as the
821 directory containing the current file.
823 If you need to look for headers in a directory named `-', write
826 There are several more ways to adjust the header search path. They
827 are generally less useful. *Note Invocation::.
830 File: cpp.info, Node: Once-Only Headers, Next: Computed Includes, Prev: Search Path, Up: Header Files
832 2.4 Once-Only Headers
833 =====================
835 If a header file happens to be included twice, the compiler will process
836 its contents twice. This is very likely to cause an error, e.g. when
837 the compiler sees the same structure definition twice. Even if it does
838 not, it will certainly waste time.
840 The standard way to prevent this is to enclose the entire real
841 contents of the file in a conditional, like this:
844 #ifndef FILE_FOO_SEEN
845 #define FILE_FOO_SEEN
849 #endif /* !FILE_FOO_SEEN */
851 This construct is commonly known as a "wrapper #ifndef". When the
852 header is included again, the conditional will be false, because
853 `FILE_FOO_SEEN' is defined. The preprocessor will skip over the entire
854 contents of the file, and the compiler will not see it twice.
856 CPP optimizes even further. It remembers when a header file has a
857 wrapper `#ifndef'. If a subsequent `#include' specifies that header,
858 and the macro in the `#ifndef' is still defined, it does not bother to
859 rescan the file at all.
861 You can put comments outside the wrapper. They will not interfere
862 with this optimization.
864 The macro `FILE_FOO_SEEN' is called the "controlling macro" or
865 "guard macro". In a user header file, the macro name should not begin
866 with `_'. In a system header file, it should begin with `__' to avoid
867 conflicts with user programs. In any kind of header file, the macro
868 name should contain the name of the file and some additional text, to
869 avoid conflicts with other header files.
872 File: cpp.info, Node: Computed Includes, Next: Wrapper Headers, Prev: Once-Only Headers, Up: Header Files
874 2.5 Computed Includes
875 =====================
877 Sometimes it is necessary to select one of several different header
878 files to be included into your program. They might specify
879 configuration parameters to be used on different sorts of operating
880 systems, for instance. You could do this with a series of conditionals,
883 # include "system_1.h"
885 # include "system_2.h"
890 That rapidly becomes tedious. Instead, the preprocessor offers the
891 ability to use a macro for the header name. This is called a "computed
892 include". Instead of writing a header name as the direct argument of
893 `#include', you simply put a macro name there instead:
895 #define SYSTEM_H "system_1.h"
899 `SYSTEM_H' will be expanded, and the preprocessor will look for
900 `system_1.h' as if the `#include' had been written that way originally.
901 `SYSTEM_H' could be defined by your Makefile with a `-D' option.
903 You must be careful when you define the macro. `#define' saves
904 tokens, not text. The preprocessor has no way of knowing that the macro
905 will be used as the argument of `#include', so it generates ordinary
906 tokens, not a header name. This is unlikely to cause problems if you
907 use double-quote includes, which are close enough to string constants.
908 If you use angle brackets, however, you may have trouble.
910 The syntax of a computed include is actually a bit more general than
911 the above. If the first non-whitespace character after `#include' is
912 not `"' or `<', then the entire line is macro-expanded like running
915 If the line expands to a single string constant, the contents of that
916 string constant are the file to be included. CPP does not re-examine
917 the string for embedded quotes, but neither does it process backslash
918 escapes in the string. Therefore
920 #define HEADER "a\"b"
923 looks for a file named `a\"b'. CPP searches for the file according to
924 the rules for double-quoted includes.
926 If the line expands to a token stream beginning with a `<' token and
927 including a `>' token, then the tokens between the `<' and the first
928 `>' are combined to form the filename to be included. Any whitespace
929 between tokens is reduced to a single space; then any space after the
930 initial `<' is retained, but a trailing space before the closing `>' is
931 ignored. CPP searches for the file according to the rules for
932 angle-bracket includes.
934 In either case, if there are any tokens on the line after the file
935 name, an error occurs and the directive is not processed. It is also
936 an error if the result of expansion does not match either of the two
939 These rules are implementation-defined behavior according to the C
940 standard. To minimize the risk of different compilers interpreting your
941 computed includes differently, we recommend you use only a single
942 object-like macro which expands to a string constant. This will also
943 minimize confusion for people reading your program.
946 File: cpp.info, Node: Wrapper Headers, Next: System Headers, Prev: Computed Includes, Up: Header Files
951 Sometimes it is necessary to adjust the contents of a system-provided
952 header file without editing it directly. GCC's `fixincludes' operation
953 does this, for example. One way to do that would be to create a new
954 header file with the same name and insert it in the search path before
955 the original header. That works fine as long as you're willing to
956 replace the old header entirely. But what if you want to refer to the
957 old header from the new one?
959 You cannot simply include the old header with `#include'. That will
960 start from the beginning, and find your new header again. If your
961 header is not protected from multiple inclusion (*note Once-Only
962 Headers::), it will recurse infinitely and cause a fatal error.
964 You could include the old header with an absolute pathname:
965 #include "/usr/include/old-header.h"
966 This works, but is not clean; should the system headers ever move,
967 you would have to edit the new headers to match.
969 There is no way to solve this problem within the C standard, but you
970 can use the GNU extension `#include_next'. It means, "Include the
971 _next_ file with this name". This directive works like `#include'
972 except in searching for the specified file: it starts searching the
973 list of header file directories _after_ the directory in which the
974 current file was found.
976 Suppose you specify `-I /usr/local/include', and the list of
977 directories to search also includes `/usr/include'; and suppose both
978 directories contain `signal.h'. Ordinary `#include <signal.h>' finds
979 the file under `/usr/local/include'. If that file contains
980 `#include_next <signal.h>', it starts searching after that directory,
981 and finds the file in `/usr/include'.
983 `#include_next' does not distinguish between `<FILE>' and `"FILE"'
984 inclusion, nor does it check that the file you specify has the same
985 name as the current file. It simply looks for the file named, starting
986 with the directory in the search path after the one where the current
989 The use of `#include_next' can lead to great confusion. We
990 recommend it be used only when there is no other alternative. In
991 particular, it should not be used in the headers belonging to a specific
992 program; it should be used only to make global corrections along the
993 lines of `fixincludes'.
996 File: cpp.info, Node: System Headers, Prev: Wrapper Headers, Up: Header Files
1001 The header files declaring interfaces to the operating system and
1002 runtime libraries often cannot be written in strictly conforming C.
1003 Therefore, GCC gives code found in "system headers" special treatment.
1004 All warnings, other than those generated by `#warning' (*note
1005 Diagnostics::), are suppressed while GCC is processing a system header.
1006 Macros defined in a system header are immune to a few warnings
1007 wherever they are expanded. This immunity is granted on an ad-hoc
1008 basis, when we find that a warning generates lots of false positives
1009 because of code in macros defined in system headers.
1011 Normally, only the headers found in specific directories are
1012 considered system headers. These directories are determined when GCC
1013 is compiled. There are, however, two ways to make normal headers into
1016 The `-isystem' command line option adds its argument to the list of
1017 directories to search for headers, just like `-I'. Any headers found
1018 in that directory will be considered system headers.
1020 All directories named by `-isystem' are searched _after_ all
1021 directories named by `-I', no matter what their order was on the
1022 command line. If the same directory is named by both `-I' and
1023 `-isystem', the `-I' option is ignored. GCC provides an informative
1024 message when this occurs if `-v' is used.
1026 There is also a directive, `#pragma GCC system_header', which tells
1027 GCC to consider the rest of the current include file a system header,
1028 no matter where it was found. Code that comes before the `#pragma' in
1029 the file will not be affected. `#pragma GCC system_header' has no
1030 effect in the primary source file.
1032 On very old systems, some of the pre-defined system header
1033 directories get even more special treatment. GNU C++ considers code in
1034 headers found in those directories to be surrounded by an `extern "C"'
1035 block. There is no way to request this behavior with a `#pragma', or
1036 from the command line.
1039 File: cpp.info, Node: Macros, Next: Conditionals, Prev: Header Files, Up: Top
1044 A "macro" is a fragment of code which has been given a name. Whenever
1045 the name is used, it is replaced by the contents of the macro. There
1046 are two kinds of macros. They differ mostly in what they look like
1047 when they are used. "Object-like" macros resemble data objects when
1048 used, "function-like" macros resemble function calls.
1050 You may define any valid identifier as a macro, even if it is a C
1051 keyword. The preprocessor does not know anything about keywords. This
1052 can be useful if you wish to hide a keyword such as `const' from an
1053 older compiler that does not understand it. However, the preprocessor
1054 operator `defined' (*note Defined::) can never be defined as a macro,
1055 and C++'s named operators (*note C++ Named Operators::) cannot be
1056 macros when you are compiling C++.
1060 * Object-like Macros::
1061 * Function-like Macros::
1066 * Predefined Macros::
1067 * Undefining and Redefining Macros::
1068 * Directives Within Macro Arguments::
1072 File: cpp.info, Node: Object-like Macros, Next: Function-like Macros, Up: Macros
1074 3.1 Object-like Macros
1075 ======================
1077 An "object-like macro" is a simple identifier which will be replaced by
1078 a code fragment. It is called object-like because it looks like a data
1079 object in code that uses it. They are most commonly used to give
1080 symbolic names to numeric constants.
1082 You create macros with the `#define' directive. `#define' is
1083 followed by the name of the macro and then the token sequence it should
1084 be an abbreviation for, which is variously referred to as the macro's
1085 "body", "expansion" or "replacement list". For example,
1087 #define BUFFER_SIZE 1024
1089 defines a macro named `BUFFER_SIZE' as an abbreviation for the token
1090 `1024'. If somewhere after this `#define' directive there comes a C
1091 statement of the form
1093 foo = (char *) malloc (BUFFER_SIZE);
1095 then the C preprocessor will recognize and "expand" the macro
1096 `BUFFER_SIZE'. The C compiler will see the same tokens as it would if
1099 foo = (char *) malloc (1024);
1101 By convention, macro names are written in uppercase. Programs are
1102 easier to read when it is possible to tell at a glance which names are
1105 The macro's body ends at the end of the `#define' line. You may
1106 continue the definition onto multiple lines, if necessary, using
1107 backslash-newline. When the macro is expanded, however, it will all
1108 come out on one line. For example,
1110 #define NUMBERS 1, \
1113 int x[] = { NUMBERS };
1114 ==> int x[] = { 1, 2, 3 };
1116 The most common visible consequence of this is surprising line numbers
1119 There is no restriction on what can go in a macro body provided it
1120 decomposes into valid preprocessing tokens. Parentheses need not
1121 balance, and the body need not resemble valid C code. (If it does not,
1122 you may get error messages from the C compiler when you use the macro.)
1124 The C preprocessor scans your program sequentially. Macro
1125 definitions take effect at the place you write them. Therefore, the
1126 following input to the C preprocessor
1137 When the preprocessor expands a macro name, the macro's expansion
1138 replaces the macro invocation, then the expansion is examined for more
1139 macros to expand. For example,
1141 #define TABLESIZE BUFSIZE
1142 #define BUFSIZE 1024
1147 `TABLESIZE' is expanded first to produce `BUFSIZE', then that macro is
1148 expanded to produce the final result, `1024'.
1150 Notice that `BUFSIZE' was not defined when `TABLESIZE' was defined.
1151 The `#define' for `TABLESIZE' uses exactly the expansion you
1152 specify--in this case, `BUFSIZE'--and does not check to see whether it
1153 too contains macro names. Only when you _use_ `TABLESIZE' is the
1154 result of its expansion scanned for more macro names.
1156 This makes a difference if you change the definition of `BUFSIZE' at
1157 some point in the source file. `TABLESIZE', defined as shown, will
1158 always expand using the definition of `BUFSIZE' that is currently in
1161 #define BUFSIZE 1020
1162 #define TABLESIZE BUFSIZE
1166 Now `TABLESIZE' expands (in two stages) to `37'.
1168 If the expansion of a macro contains its own name, either directly or
1169 via intermediate macros, it is not expanded again when the expansion is
1170 examined for more macros. This prevents infinite recursion. *Note
1171 Self-Referential Macros::, for the precise details.
1174 File: cpp.info, Node: Function-like Macros, Next: Macro Arguments, Prev: Object-like Macros, Up: Macros
1176 3.2 Function-like Macros
1177 ========================
1179 You can also define macros whose use looks like a function call. These
1180 are called "function-like macros". To define a function-like macro,
1181 you use the same `#define' directive, but you put a pair of parentheses
1182 immediately after the macro name. For example,
1184 #define lang_init() c_init()
1188 A function-like macro is only expanded if its name appears with a
1189 pair of parentheses after it. If you write just the name, it is left
1190 alone. This can be useful when you have a function and a macro of the
1191 same name, and you wish to use the function sometimes.
1193 extern void foo(void);
1194 #define foo() /* optimized inline version */
1199 Here the call to `foo()' will use the macro, but the function
1200 pointer will get the address of the real function. If the macro were to
1201 be expanded, it would cause a syntax error.
1203 If you put spaces between the macro name and the parentheses in the
1204 macro definition, that does not define a function-like macro, it defines
1205 an object-like macro whose expansion happens to begin with a pair of
1208 #define lang_init () c_init()
1212 The first two pairs of parentheses in this expansion come from the
1213 macro. The third is the pair that was originally after the macro
1214 invocation. Since `lang_init' is an object-like macro, it does not
1215 consume those parentheses.
1218 File: cpp.info, Node: Macro Arguments, Next: Stringification, Prev: Function-like Macros, Up: Macros
1223 Function-like macros can take "arguments", just like true functions.
1224 To define a macro that uses arguments, you insert "parameters" between
1225 the pair of parentheses in the macro definition that make the macro
1226 function-like. The parameters must be valid C identifiers, separated
1227 by commas and optionally whitespace.
1229 To invoke a macro that takes arguments, you write the name of the
1230 macro followed by a list of "actual arguments" in parentheses, separated
1231 by commas. The invocation of the macro need not be restricted to a
1232 single logical line--it can cross as many lines in the source file as
1233 you wish. The number of arguments you give must match the number of
1234 parameters in the macro definition. When the macro is expanded, each
1235 use of a parameter in its body is replaced by the tokens of the
1236 corresponding argument. (You need not use all of the parameters in the
1239 As an example, here is a macro that computes the minimum of two
1240 numeric values, as it is defined in many C programs, and some uses.
1242 #define min(X, Y) ((X) < (Y) ? (X) : (Y))
1243 x = min(a, b); ==> x = ((a) < (b) ? (a) : (b));
1244 y = min(1, 2); ==> y = ((1) < (2) ? (1) : (2));
1245 z = min(a + 28, *p); ==> z = ((a + 28) < (*p) ? (a + 28) : (*p));
1247 (In this small example you can already see several of the dangers of
1248 macro arguments. *Note Macro Pitfalls::, for detailed explanations.)
1250 Leading and trailing whitespace in each argument is dropped, and all
1251 whitespace between the tokens of an argument is reduced to a single
1252 space. Parentheses within each argument must balance; a comma within
1253 such parentheses does not end the argument. However, there is no
1254 requirement for square brackets or braces to balance, and they do not
1255 prevent a comma from separating arguments. Thus,
1257 macro (array[x = y, x + 1])
1259 passes two arguments to `macro': `array[x = y' and `x + 1]'. If you
1260 want to supply `array[x = y, x + 1]' as an argument, you can write it
1261 as `array[(x = y, x + 1)]', which is equivalent C code.
1263 All arguments to a macro are completely macro-expanded before they
1264 are substituted into the macro body. After substitution, the complete
1265 text is scanned again for macros to expand, including the arguments.
1266 This rule may seem strange, but it is carefully designed so you need
1267 not worry about whether any function call is actually a macro
1268 invocation. You can run into trouble if you try to be too clever,
1269 though. *Note Argument Prescan::, for detailed discussion.
1271 For example, `min (min (a, b), c)' is first expanded to
1273 min (((a) < (b) ? (a) : (b)), (c))
1277 ((((a) < (b) ? (a) : (b))) < (c)
1278 ? (((a) < (b) ? (a) : (b)))
1281 (Line breaks shown here for clarity would not actually be generated.)
1283 You can leave macro arguments empty; this is not an error to the
1284 preprocessor (but many macros will then expand to invalid code). You
1285 cannot leave out arguments entirely; if a macro takes two arguments,
1286 there must be exactly one comma at the top level of its argument list.
1287 Here are some silly examples using `min':
1289 min(, b) ==> (( ) < (b) ? ( ) : (b))
1290 min(a, ) ==> ((a ) < ( ) ? (a ) : ( ))
1291 min(,) ==> (( ) < ( ) ? ( ) : ( ))
1292 min((,),) ==> (((,)) < ( ) ? ((,)) : ( ))
1294 min() error--> macro "min" requires 2 arguments, but only 1 given
1295 min(,,) error--> macro "min" passed 3 arguments, but takes just 2
1297 Whitespace is not a preprocessing token, so if a macro `foo' takes
1298 one argument, `foo ()' and `foo ( )' both supply it an empty argument.
1299 Previous GNU preprocessor implementations and documentation were
1300 incorrect on this point, insisting that a function-like macro that
1301 takes a single argument be passed a space if an empty argument was
1304 Macro parameters appearing inside string literals are not replaced by
1305 their corresponding actual arguments.
1307 #define foo(x) x, "x"
1308 foo(bar) ==> bar, "x"
1311 File: cpp.info, Node: Stringification, Next: Concatenation, Prev: Macro Arguments, Up: Macros
1316 Sometimes you may want to convert a macro argument into a string
1317 constant. Parameters are not replaced inside string constants, but you
1318 can use the `#' preprocessing operator instead. When a macro parameter
1319 is used with a leading `#', the preprocessor replaces it with the
1320 literal text of the actual argument, converted to a string constant.
1321 Unlike normal parameter replacement, the argument is not macro-expanded
1322 first. This is called "stringification".
1324 There is no way to combine an argument with surrounding text and
1325 stringify it all together. Instead, you can write a series of adjacent
1326 string constants and stringified arguments. The preprocessor will
1327 replace the stringified arguments with string constants. The C
1328 compiler will then combine all the adjacent string constants into one
1331 Here is an example of a macro definition that uses stringification:
1333 #define WARN_IF(EXP) \
1335 fprintf (stderr, "Warning: " #EXP "\n"); } \
1338 ==> do { if (x == 0)
1339 fprintf (stderr, "Warning: " "x == 0" "\n"); } while (0);
1341 The argument for `EXP' is substituted once, as-is, into the `if'
1342 statement, and once, stringified, into the argument to `fprintf'. If
1343 `x' were a macro, it would be expanded in the `if' statement, but not
1346 The `do' and `while (0)' are a kludge to make it possible to write
1347 `WARN_IF (ARG);', which the resemblance of `WARN_IF' to a function
1348 would make C programmers want to do; see *Note Swallowing the
1351 Stringification in C involves more than putting double-quote
1352 characters around the fragment. The preprocessor backslash-escapes the
1353 quotes surrounding embedded string constants, and all backslashes
1354 within string and character constants, in order to get a valid C string
1355 constant with the proper contents. Thus, stringifying `p = "foo\n";'
1356 results in "p = \"foo\\n\";". However, backslashes that are not inside
1357 string or character constants are not duplicated: `\n' by itself
1358 stringifies to "\n".
1360 All leading and trailing whitespace in text being stringified is
1361 ignored. Any sequence of whitespace in the middle of the text is
1362 converted to a single space in the stringified result. Comments are
1363 replaced by whitespace long before stringification happens, so they
1364 never appear in stringified text.
1366 There is no way to convert a macro argument into a character
1369 If you want to stringify the result of expansion of a macro argument,
1370 you have to use two levels of macros.
1372 #define xstr(s) str(s)
1382 `s' is stringified when it is used in `str', so it is not
1383 macro-expanded first. But `s' is an ordinary argument to `xstr', so it
1384 is completely macro-expanded before `xstr' itself is expanded (*note
1385 Argument Prescan::). Therefore, by the time `str' gets to its
1386 argument, it has already been macro-expanded.
1389 File: cpp.info, Node: Concatenation, Next: Variadic Macros, Prev: Stringification, Up: Macros
1394 It is often useful to merge two tokens into one while expanding macros.
1395 This is called "token pasting" or "token concatenation". The `##'
1396 preprocessing operator performs token pasting. When a macro is
1397 expanded, the two tokens on either side of each `##' operator are
1398 combined into a single token, which then replaces the `##' and the two
1399 original tokens in the macro expansion. Usually both will be
1400 identifiers, or one will be an identifier and the other a preprocessing
1401 number. When pasted, they make a longer identifier. This isn't the
1402 only valid case. It is also possible to concatenate two numbers (or a
1403 number and a name, such as `1.5' and `e3') into a number. Also,
1404 multi-character operators such as `+=' can be formed by token pasting.
1406 However, two tokens that don't together form a valid token cannot be
1407 pasted together. For example, you cannot concatenate `x' with `+' in
1408 either order. If you try, the preprocessor issues a warning and emits
1409 the two tokens. Whether it puts white space between the tokens is
1410 undefined. It is common to find unnecessary uses of `##' in complex
1411 macros. If you get this warning, it is likely that you can simply
1414 Both the tokens combined by `##' could come from the macro body, but
1415 you could just as well write them as one token in the first place.
1416 Token pasting is most useful when one or both of the tokens comes from a
1417 macro argument. If either of the tokens next to an `##' is a parameter
1418 name, it is replaced by its actual argument before `##' executes. As
1419 with stringification, the actual argument is not macro-expanded first.
1420 If the argument is empty, that `##' has no effect.
1422 Keep in mind that the C preprocessor converts comments to whitespace
1423 before macros are even considered. Therefore, you cannot create a
1424 comment by concatenating `/' and `*'. You can put as much whitespace
1425 between `##' and its operands as you like, including comments, and you
1426 can put comments in arguments that will be concatenated. However, it
1427 is an error if `##' appears at either end of a macro body.
1429 Consider a C program that interprets named commands. There probably
1430 needs to be a table of commands, perhaps an array of structures declared
1436 void (*function) (void);
1439 struct command commands[] =
1441 { "quit", quit_command },
1442 { "help", help_command },
1446 It would be cleaner not to have to give each command name twice,
1447 once in the string constant and once in the function name. A macro
1448 which takes the name of a command as an argument can make this
1449 unnecessary. The string constant can be created with stringification,
1450 and the function name by concatenating the argument with `_command'.
1451 Here is how it is done:
1453 #define COMMAND(NAME) { #NAME, NAME ## _command }
1455 struct command commands[] =
1463 File: cpp.info, Node: Variadic Macros, Next: Predefined Macros, Prev: Concatenation, Up: Macros
1468 A macro can be declared to accept a variable number of arguments much as
1469 a function can. The syntax for defining the macro is similar to that of
1470 a function. Here is an example:
1472 #define eprintf(...) fprintf (stderr, __VA_ARGS__)
1474 This kind of macro is called "variadic". When the macro is invoked,
1475 all the tokens in its argument list after the last named argument (this
1476 macro has none), including any commas, become the "variable argument".
1477 This sequence of tokens replaces the identifier `__VA_ARGS__' in the
1478 macro body wherever it appears. Thus, we have this expansion:
1480 eprintf ("%s:%d: ", input_file, lineno)
1481 ==> fprintf (stderr, "%s:%d: ", input_file, lineno)
1483 The variable argument is completely macro-expanded before it is
1484 inserted into the macro expansion, just like an ordinary argument. You
1485 may use the `#' and `##' operators to stringify the variable argument
1486 or to paste its leading or trailing token with another token. (But see
1487 below for an important special case for `##'.)
1489 If your macro is complicated, you may want a more descriptive name
1490 for the variable argument than `__VA_ARGS__'. CPP permits this, as an
1491 extension. You may write an argument name immediately before the
1492 `...'; that name is used for the variable argument. The `eprintf'
1493 macro above could be written
1495 #define eprintf(args...) fprintf (stderr, args)
1497 using this extension. You cannot use `__VA_ARGS__' and this extension
1500 You can have named arguments as well as variable arguments in a
1501 variadic macro. We could define `eprintf' like this, instead:
1503 #define eprintf(format, ...) fprintf (stderr, format, __VA_ARGS__)
1505 This formulation looks more descriptive, but unfortunately it is less
1506 flexible: you must now supply at least one argument after the format
1507 string. In standard C, you cannot omit the comma separating the named
1508 argument from the variable arguments. Furthermore, if you leave the
1509 variable argument empty, you will get a syntax error, because there
1510 will be an extra comma after the format string.
1512 eprintf("success!\n", );
1513 ==> fprintf(stderr, "success!\n", );
1515 GNU CPP has a pair of extensions which deal with this problem.
1516 First, you are allowed to leave the variable argument out entirely:
1518 eprintf ("success!\n")
1519 ==> fprintf(stderr, "success!\n", );
1521 Second, the `##' token paste operator has a special meaning when placed
1522 between a comma and a variable argument. If you write
1524 #define eprintf(format, ...) fprintf (stderr, format, ##__VA_ARGS__)
1526 and the variable argument is left out when the `eprintf' macro is used,
1527 then the comma before the `##' will be deleted. This does _not_ happen
1528 if you pass an empty argument, nor does it happen if the token
1529 preceding `##' is anything other than a comma.
1531 eprintf ("success!\n")
1532 ==> fprintf(stderr, "success!\n");
1534 The above explanation is ambiguous about the case where the only macro
1535 parameter is a variable arguments parameter, as it is meaningless to
1536 try to distinguish whether no argument at all is an empty argument or a
1537 missing argument. In this case the C99 standard is clear that the
1538 comma must remain, however the existing GCC extension used to swallow
1539 the comma. So CPP retains the comma when conforming to a specific C
1540 standard, and drops it otherwise.
1542 C99 mandates that the only place the identifier `__VA_ARGS__' can
1543 appear is in the replacement list of a variadic macro. It may not be
1544 used as a macro name, macro argument name, or within a different type
1545 of macro. It may also be forbidden in open text; the standard is
1546 ambiguous. We recommend you avoid using it except for its defined
1549 Variadic macros are a new feature in C99. GNU CPP has supported them
1550 for a long time, but only with a named variable argument (`args...',
1551 not `...' and `__VA_ARGS__'). If you are concerned with portability to
1552 previous versions of GCC, you should use only named variable arguments.
1553 On the other hand, if you are concerned with portability to other
1554 conforming implementations of C99, you should use only `__VA_ARGS__'.
1556 Previous versions of CPP implemented the comma-deletion extension
1557 much more generally. We have restricted it in this release to minimize
1558 the differences from C99. To get the same effect with both this and
1559 previous versions of GCC, the token preceding the special `##' must be
1560 a comma, and there must be white space between that comma and whatever
1561 comes immediately before it:
1563 #define eprintf(format, args...) fprintf (stderr, format , ##args)
1565 *Note Differences from previous versions::, for the gory details.
1568 File: cpp.info, Node: Predefined Macros, Next: Undefining and Redefining Macros, Prev: Variadic Macros, Up: Macros
1570 3.7 Predefined Macros
1571 =====================
1573 Several object-like macros are predefined; you use them without
1574 supplying their definitions. They fall into three classes: standard,
1575 common, and system-specific.
1577 In C++, there is a fourth category, the named operators. They act
1578 like predefined macros, but you cannot undefine them.
1582 * Standard Predefined Macros::
1583 * Common Predefined Macros::
1584 * System-specific Predefined Macros::
1585 * C++ Named Operators::
1588 File: cpp.info, Node: Standard Predefined Macros, Next: Common Predefined Macros, Up: Predefined Macros
1590 3.7.1 Standard Predefined Macros
1591 --------------------------------
1593 The standard predefined macros are specified by the relevant language
1594 standards, so they are available with all compilers that implement
1595 those standards. Older compilers may not provide all of them. Their
1596 names all start with double underscores.
1599 This macro expands to the name of the current input file, in the
1600 form of a C string constant. This is the path by which the
1601 preprocessor opened the file, not the short name specified in
1602 `#include' or as the input file name argument. For example,
1603 `"/usr/local/include/myheader.h"' is a possible expansion of this
1607 This macro expands to the current input line number, in the form
1608 of a decimal integer constant. While we call it a predefined
1609 macro, it's a pretty strange macro, since its "definition" changes
1610 with each new line of source code.
1612 `__FILE__' and `__LINE__' are useful in generating an error message
1613 to report an inconsistency detected by the program; the message can
1614 state the source line at which the inconsistency was detected. For
1617 fprintf (stderr, "Internal error: "
1618 "negative string length "
1619 "%d at %s, line %d.",
1620 length, __FILE__, __LINE__);
1622 An `#include' directive changes the expansions of `__FILE__' and
1623 `__LINE__' to correspond to the included file. At the end of that
1624 file, when processing resumes on the input file that contained the
1625 `#include' directive, the expansions of `__FILE__' and `__LINE__'
1626 revert to the values they had before the `#include' (but `__LINE__' is
1627 then incremented by one as processing moves to the line after the
1630 A `#line' directive changes `__LINE__', and may change `__FILE__' as
1631 well. *Note Line Control::.
1633 C99 introduces `__func__', and GCC has provided `__FUNCTION__' for a
1634 long time. Both of these are strings containing the name of the
1635 current function (there are slight semantic differences; see the GCC
1636 manual). Neither of them is a macro; the preprocessor does not know the
1637 name of the current function. They tend to be useful in conjunction
1638 with `__FILE__' and `__LINE__', though.
1641 This macro expands to a string constant that describes the date on
1642 which the preprocessor is being run. The string constant contains
1643 eleven characters and looks like `"Feb 12 1996"'. If the day of
1644 the month is less than 10, it is padded with a space on the left.
1646 If GCC cannot determine the current date, it will emit a warning
1647 message (once per compilation) and `__DATE__' will expand to
1651 This macro expands to a string constant that describes the time at
1652 which the preprocessor is being run. The string constant contains
1653 eight characters and looks like `"23:59:01"'.
1655 If GCC cannot determine the current time, it will emit a warning
1656 message (once per compilation) and `__TIME__' will expand to
1660 In normal operation, this macro expands to the constant 1, to
1661 signify that this compiler conforms to ISO Standard C. If GNU CPP
1662 is used with a compiler other than GCC, this is not necessarily
1663 true; however, the preprocessor always conforms to the standard
1664 unless the `-traditional-cpp' option is used.
1666 This macro is not defined if the `-traditional-cpp' option is used.
1668 On some hosts, the system compiler uses a different convention,
1669 where `__STDC__' is normally 0, but is 1 if the user specifies
1670 strict conformance to the C Standard. CPP follows the host
1671 convention when processing system header files, but when
1672 processing user files `__STDC__' is always 1. This has been
1673 reported to cause problems; for instance, some versions of Solaris
1674 provide X Windows headers that expect `__STDC__' to be either
1675 undefined or 1. *Note Invocation::.
1678 This macro expands to the C Standard's version number, a long
1679 integer constant of the form `YYYYMML' where YYYY and MM are the
1680 year and month of the Standard version. This signifies which
1681 version of the C Standard the compiler conforms to. Like
1682 `__STDC__', this is not necessarily accurate for the entire
1683 implementation, unless GNU CPP is being used with GCC.
1685 The value `199409L' signifies the 1989 C standard as amended in
1686 1994, which is the current default; the value `199901L' signifies
1687 the 1999 revision of the C standard. Support for the 1999
1688 revision is not yet complete.
1690 This macro is not defined if the `-traditional-cpp' option is
1691 used, nor when compiling C++ or Objective-C.
1694 This macro is defined, with value 1, if the compiler's target is a
1695 "hosted environment". A hosted environment has the complete
1696 facilities of the standard C library available.
1699 This macro is defined when the C++ compiler is in use. You can use
1700 `__cplusplus' to test whether a header is compiled by a C compiler
1701 or a C++ compiler. This macro is similar to `__STDC_VERSION__', in
1702 that it expands to a version number. A fully conforming
1703 implementation of the 1998 C++ standard will define this macro to
1704 `199711L'. The GNU C++ compiler is not yet fully conforming, so
1705 it uses `1' instead. It is hoped to complete the implementation
1706 of standard C++ in the near future.
1709 This macro is defined, with value 1, when the Objective-C compiler
1710 is in use. You can use `__OBJC__' to test whether a header is
1711 compiled by a C compiler or a Objective-C compiler.
1714 This macro is defined with value 1 when preprocessing assembly
1719 File: cpp.info, Node: Common Predefined Macros, Next: System-specific Predefined Macros, Prev: Standard Predefined Macros, Up: Predefined Macros
1721 3.7.2 Common Predefined Macros
1722 ------------------------------
1724 The common predefined macros are GNU C extensions. They are available
1725 with the same meanings regardless of the machine or operating system on
1726 which you are using GNU C or GNU Fortran. Their names all start with
1730 This macro expands to sequential integral values starting from 0.
1731 In conjunction with the `##' operator, this provides a convenient
1732 means to generate unique identifiers. Care must be taken to
1733 ensure that `__COUNTER__' is not expanded prior to inclusion of
1734 precompiled headers which use it. Otherwise, the precompiled
1735 headers will not be used.
1738 The GNU Fortran compiler defines this.
1742 `__GNUC_PATCHLEVEL__'
1743 These macros are defined by all GNU compilers that use the C
1744 preprocessor: C, C++, Objective-C and Fortran. Their values are
1745 the major version, minor version, and patch level of the compiler,
1746 as integer constants. For example, GCC 3.2.1 will define
1747 `__GNUC__' to 3, `__GNUC_MINOR__' to 2, and `__GNUC_PATCHLEVEL__'
1748 to 1. These macros are also defined if you invoke the
1749 preprocessor directly.
1751 `__GNUC_PATCHLEVEL__' is new to GCC 3.0; it is also present in the
1752 widely-used development snapshots leading up to 3.0 (which identify
1753 themselves as GCC 2.96 or 2.97, depending on which snapshot you
1756 If all you need to know is whether or not your program is being
1757 compiled by GCC, or a non-GCC compiler that claims to accept the
1758 GNU C dialects, you can simply test `__GNUC__'. If you need to
1759 write code which depends on a specific version, you must be more
1760 careful. Each time the minor version is increased, the patch
1761 level is reset to zero; each time the major version is increased
1762 (which happens rarely), the minor version and patch level are
1763 reset. If you wish to use the predefined macros directly in the
1764 conditional, you will need to write it like this:
1766 /* Test for GCC > 3.2.0 */
1767 #if __GNUC__ > 3 || \
1768 (__GNUC__ == 3 && (__GNUC_MINOR__ > 2 || \
1769 (__GNUC_MINOR__ == 2 && \
1770 __GNUC_PATCHLEVEL__ > 0))
1772 Another approach is to use the predefined macros to calculate a
1773 single number, then compare that against a threshold:
1775 #define GCC_VERSION (__GNUC__ * 10000 \
1776 + __GNUC_MINOR__ * 100 \
1777 + __GNUC_PATCHLEVEL__)
1779 /* Test for GCC > 3.2.0 */
1780 #if GCC_VERSION > 30200
1782 Many people find this form easier to understand.
1785 The GNU C++ compiler defines this. Testing it is equivalent to
1786 testing `(__GNUC__ && __cplusplus)'.
1789 GCC defines this macro if and only if the `-ansi' switch, or a
1790 `-std' switch specifying strict conformance to some version of ISO
1791 C, was specified when GCC was invoked. It is defined to `1'.
1792 This macro exists primarily to direct GNU libc's header files to
1793 restrict their definitions to the minimal set found in the 1989 C
1797 This macro expands to the name of the main input file, in the form
1798 of a C string constant. This is the source file that was specified
1799 on the command line of the preprocessor or C compiler.
1802 This macro expands to a decimal integer constant that represents
1803 the depth of nesting in include files. The value of this macro is
1804 incremented on every `#include' directive and decremented at the
1805 end of every included file. It starts out at 0, its value within
1806 the base file specified on the command line.
1809 This macro is defined if the target uses the ELF object format.
1812 This macro expands to a string constant which describes the
1813 version of the compiler in use. You should not rely on its
1814 contents having any particular form, but it can be counted on to
1815 contain at least the release number.
1820 These macros describe the compilation mode. `__OPTIMIZE__' is
1821 defined in all optimizing compilations. `__OPTIMIZE_SIZE__' is
1822 defined if the compiler is optimizing for size, not speed.
1823 `__NO_INLINE__' is defined if no functions will be inlined into
1824 their callers (when not optimizing, or when inlining has been
1825 specifically disabled by `-fno-inline').
1827 These macros cause certain GNU header files to provide optimized
1828 definitions, using macros or inline functions, of system library
1829 functions. You should not use these macros in any way unless you
1830 make sure that programs will execute with the same effect whether
1831 or not they are defined. If they are defined, their value is 1.
1833 `__GNUC_GNU_INLINE__'
1834 GCC defines this macro if functions declared `inline' will be
1835 handled in GCC's traditional gnu89 mode. Object files will contain
1836 externally visible definitions of all functions declared `inline'
1837 without `extern' or `static'. They will not contain any
1838 definitions of any functions declared `extern inline'.
1840 `__GNUC_STDC_INLINE__'
1841 GCC defines this macro if functions declared `inline' will be
1842 handled according to the ISO C99 standard. Object files will
1843 contain externally visible definitions of all functions declared
1844 `extern inline'. They will not contain definitions of any
1845 functions declared `inline' without `extern'.
1847 If this macro is defined, GCC supports the `gnu_inline' function
1848 attribute as a way to always get the gnu89 behavior. Support for
1849 this and `__GNUC_GNU_INLINE__' was added in GCC 4.1.3. If neither
1850 macro is defined, an older version of GCC is being used: `inline'
1851 functions will be compiled in gnu89 mode, and the `gnu_inline'
1852 function attribute will not be recognized.
1855 GCC defines this macro if and only if the data type `char' is
1856 unsigned on the target machine. It exists to cause the standard
1857 header file `limits.h' to work correctly. You should not use this
1858 macro yourself; instead, refer to the standard macros defined in
1861 `__WCHAR_UNSIGNED__'
1862 Like `__CHAR_UNSIGNED__', this macro is defined if and only if the
1863 data type `wchar_t' is unsigned and the front-end is in C++ mode.
1865 `__REGISTER_PREFIX__'
1866 This macro expands to a single token (not a string constant) which
1867 is the prefix applied to CPU register names in assembly language
1868 for this target. You can use it to write assembly that is usable
1869 in multiple environments. For example, in the `m68k-aout'
1870 environment it expands to nothing, but in the `m68k-coff'
1871 environment it expands to a single `%'.
1873 `__USER_LABEL_PREFIX__'
1874 This macro expands to a single token which is the prefix applied to
1875 user labels (symbols visible to C code) in assembly. For example,
1876 in the `m68k-aout' environment it expands to an `_', but in the
1877 `m68k-coff' environment it expands to nothing.
1879 This macro will have the correct definition even if
1880 `-f(no-)underscores' is in use, but it will not be correct if
1881 target-specific options that adjust this prefix are used (e.g. the
1882 OSF/rose `-mno-underscores' option).
1890 These macros are defined to the correct underlying types for the
1891 `size_t', `ptrdiff_t', `wchar_t', `wint_t', `intmax_t', and
1892 `uintmax_t' typedefs, respectively. They exist to make the
1893 standard header files `stddef.h' and `wchar.h' work correctly.
1894 You should not use these macros directly; instead, include the
1895 appropriate headers and use the typedefs.
1898 Defined to the number of bits used in the representation of the
1899 `char' data type. It exists to make the standard header given
1900 numerical limits work correctly. You should not use this macro
1901 directly; instead, include the appropriate headers.
1910 Defined to the maximum value of the `signed char', `wchar_t',
1911 `signed short', `signed int', `signed long', `signed long long',
1912 and `intmax_t' types respectively. They exist to make the
1913 standard header given numerical limits work correctly. You should
1914 not use these macros directly; instead, include the appropriate
1919 `__SIZEOF_LONG_LONG__'
1921 `__SIZEOF_POINTER__'
1924 `__SIZEOF_LONG_DOUBLE__'
1926 `__SIZEOF_WCHAR_T__'
1928 `__SIZEOF_PTRDIFF_T__'
1929 Defined to the number of bytes of the C standard data types: `int',
1930 `long', `long long', `short', `void *', `float', `double', `long
1931 double', `size_t', `wchar_t', `wint_t' and `ptrdiff_t'.
1934 This macro is defined, with value 1, when compiling a C++ source
1935 file with warnings about deprecated constructs enabled. These
1936 warnings are enabled by default, but can be disabled with
1940 This macro is defined, with value 1, when compiling a C++ source
1941 file with exceptions enabled. If `-fno-exceptions' is used when
1942 compiling the file, then this macro is not defined.
1945 This macro is defined, with value 1, when compiling a C++ source
1946 file with runtime type identification enabled. If `-fno-rtti' is
1947 used when compiling the file, then this macro is not defined.
1949 `__USING_SJLJ_EXCEPTIONS__'
1950 This macro is defined, with value 1, if the compiler uses the old
1951 mechanism based on `setjmp' and `longjmp' for exception handling.
1953 `__GXX_EXPERIMENTAL_CXX0X__'
1954 This macro is defined when compiling a C++ source file with the
1955 option `-std=c++0x' or `-std=gnu++0x'. It indicates that some
1956 features likely to be included in C++0x are available. Note that
1957 these features are experimental, and may change or be removed in
1958 future versions of GCC.
1961 This macro is defined when compiling a C++ source file. It has the
1962 value 1 if the compiler will use weak symbols, COMDAT sections, or
1963 other similar techniques to collapse symbols with "vague linkage"
1964 that are defined in multiple translation units. If the compiler
1965 will not collapse such symbols, this macro is defined with value
1966 0. In general, user code should not need to make use of this
1967 macro; the purpose of this macro is to ease implementation of the
1968 C++ runtime library provided with G++.
1971 This macro is defined, with value 1, if (and only if) the NeXT
1972 runtime (as in `-fnext-runtime') is in use for Objective-C. If
1973 the GNU runtime is used, this macro is not defined, so that you
1974 can use this macro to determine which runtime (NeXT or GNU) is
1979 These macros are defined, with value 1, if (and only if) the
1980 compilation is for a target where `long int' and pointer both use
1981 64-bits and `int' uses 32-bit.
1984 This macro is defined, with value 1, when `-fstack-protector' is in
1988 This macro is defined, with value 2, when `-fstack-protector-all'
1992 This macro expands to a string constant that describes the date
1993 and time of the last modification of the current source file. The
1994 string constant contains abbreviated day of the week, month, day
1995 of the month, time in hh:mm:ss form, year and looks like
1996 `"Sun Sep 16 01:03:52 1973"'. If the day of the month is less
1997 than 10, it is padded with a space on the left.
1999 If GCC cannot determine the current date, it will emit a warning
2000 message (once per compilation) and `__TIMESTAMP__' will expand to
2001 `"??? ??? ?? ??:??:?? ????"'.
2003 `__GCC_HAVE_SYNC_COMPARE_AND_SWAP_1'
2004 `__GCC_HAVE_SYNC_COMPARE_AND_SWAP_2'
2005 `__GCC_HAVE_SYNC_COMPARE_AND_SWAP_4'
2006 `__GCC_HAVE_SYNC_COMPARE_AND_SWAP_8'
2007 `__GCC_HAVE_SYNC_COMPARE_AND_SWAP_16'
2008 These macros are defined when the target processor supports atomic
2009 compare and swap operations on operands 1, 2, 4, 8 or 16 bytes in
2010 length, respectively.
2014 File: cpp.info, Node: System-specific Predefined Macros, Next: C++ Named Operators, Prev: Common Predefined Macros, Up: Predefined Macros
2016 3.7.3 System-specific Predefined Macros
2017 ---------------------------------------
2019 The C preprocessor normally predefines several macros that indicate what
2020 type of system and machine is in use. They are obviously different on
2021 each target supported by GCC. This manual, being for all systems and
2022 machines, cannot tell you what their names are, but you can use `cpp
2023 -dM' to see them all. *Note Invocation::. All system-specific
2024 predefined macros expand to the constant 1, so you can test them with
2025 either `#ifdef' or `#if'.
2027 The C standard requires that all system-specific macros be part of
2028 the "reserved namespace". All names which begin with two underscores,
2029 or an underscore and a capital letter, are reserved for the compiler and
2030 library to use as they wish. However, historically system-specific
2031 macros have had names with no special prefix; for instance, it is common
2032 to find `unix' defined on Unix systems. For all such macros, GCC
2033 provides a parallel macro with two underscores added at the beginning
2034 and the end. If `unix' is defined, `__unix__' will be defined too.
2035 There will never be more than two underscores; the parallel of `_mips'
2038 When the `-ansi' option, or any `-std' option that requests strict
2039 conformance, is given to the compiler, all the system-specific
2040 predefined macros outside the reserved namespace are suppressed. The
2041 parallel macros, inside the reserved namespace, remain defined.
2043 We are slowly phasing out all predefined macros which are outside the
2044 reserved namespace. You should never use them in new programs, and we
2045 encourage you to correct older code to use the parallel macros whenever
2046 you find it. We don't recommend you use the system-specific macros that
2047 are in the reserved namespace, either. It is better in the long run to
2048 check specifically for features you need, using a tool such as
2052 File: cpp.info, Node: C++ Named Operators, Prev: System-specific Predefined Macros, Up: Predefined Macros
2054 3.7.4 C++ Named Operators
2055 -------------------------
2057 In C++, there are eleven keywords which are simply alternate spellings
2058 of operators normally written with punctuation. These keywords are
2059 treated as such even in the preprocessor. They function as operators in
2060 `#if', and they cannot be defined as macros or poisoned. In C, you can
2061 request that those keywords take their C++ meaning by including
2062 `iso646.h'. That header defines each one as a normal object-like macro
2063 expanding to the appropriate punctuator.
2065 These are the named operators and their corresponding punctuators:
2067 Named Operator Punctuator
2081 File: cpp.info, Node: Undefining and Redefining Macros, Next: Directives Within Macro Arguments, Prev: Predefined Macros, Up: Macros
2083 3.8 Undefining and Redefining Macros
2084 ====================================
2086 If a macro ceases to be useful, it may be "undefined" with the `#undef'
2087 directive. `#undef' takes a single argument, the name of the macro to
2088 undefine. You use the bare macro name, even if the macro is
2089 function-like. It is an error if anything appears on the line after
2090 the macro name. `#undef' has no effect if the name is not a macro.
2095 x = FOO; ==> x = FOO;
2097 Once a macro has been undefined, that identifier may be "redefined"
2098 as a macro by a subsequent `#define' directive. The new definition
2099 need not have any resemblance to the old definition.
2101 However, if an identifier which is currently a macro is redefined,
2102 then the new definition must be "effectively the same" as the old one.
2103 Two macro definitions are effectively the same if:
2104 * Both are the same type of macro (object- or function-like).
2106 * All the tokens of the replacement list are the same.
2108 * If there are any parameters, they are the same.
2110 * Whitespace appears in the same places in both. It need not be
2111 exactly the same amount of whitespace, though. Remember that
2112 comments count as whitespace.
2114 These definitions are effectively the same:
2115 #define FOUR (2 + 2)
2116 #define FOUR (2 + 2)
2117 #define FOUR (2 /* two */ + 2)
2119 #define FOUR (2 + 2)
2120 #define FOUR ( 2+2 )
2121 #define FOUR (2 * 2)
2122 #define FOUR(score,and,seven,years,ago) (2 + 2)
2124 If a macro is redefined with a definition that is not effectively the
2125 same as the old one, the preprocessor issues a warning and changes the
2126 macro to use the new definition. If the new definition is effectively
2127 the same, the redefinition is silently ignored. This allows, for
2128 instance, two different headers to define a common macro. The
2129 preprocessor will only complain if the definitions do not match.
2132 File: cpp.info, Node: Directives Within Macro Arguments, Next: Macro Pitfalls, Prev: Undefining and Redefining Macros, Up: Macros
2134 3.9 Directives Within Macro Arguments
2135 =====================================
2137 Occasionally it is convenient to use preprocessor directives within the
2138 arguments of a macro. The C and C++ standards declare that behavior in
2139 these cases is undefined.
2141 Versions of CPP prior to 3.2 would reject such constructs with an
2142 error message. This was the only syntactic difference between normal
2143 functions and function-like macros, so it seemed attractive to remove
2144 this limitation, and people would often be surprised that they could
2145 not use macros in this way. Moreover, sometimes people would use
2146 conditional compilation in the argument list to a normal library
2147 function like `printf', only to find that after a library upgrade
2148 `printf' had changed to be a function-like macro, and their code would
2149 no longer compile. So from version 3.2 we changed CPP to successfully
2150 process arbitrary directives within macro arguments in exactly the same
2151 way as it would have processed the directive were the function-like
2152 macro invocation not present.
2154 If, within a macro invocation, that macro is redefined, then the new
2155 definition takes effect in time for argument pre-expansion, but the
2156 original definition is still used for argument replacement. Here is a
2157 pathological example:
2169 with the semantics described above.
2172 File: cpp.info, Node: Macro Pitfalls, Prev: Directives Within Macro Arguments, Up: Macros
2177 In this section we describe some special rules that apply to macros and
2178 macro expansion, and point out certain cases in which the rules have
2179 counter-intuitive consequences that you must watch out for.
2184 * Operator Precedence Problems::
2185 * Swallowing the Semicolon::
2186 * Duplication of Side Effects::
2187 * Self-Referential Macros::
2188 * Argument Prescan::
2189 * Newlines in Arguments::
2192 File: cpp.info, Node: Misnesting, Next: Operator Precedence Problems, Up: Macro Pitfalls
2197 When a macro is called with arguments, the arguments are substituted
2198 into the macro body and the result is checked, together with the rest of
2199 the input file, for more macro calls. It is possible to piece together
2200 a macro call coming partially from the macro body and partially from the
2201 arguments. For example,
2203 #define twice(x) (2*(x))
2204 #define call_with_1(x) x(1)
2209 Macro definitions do not have to have balanced parentheses. By
2210 writing an unbalanced open parenthesis in a macro body, it is possible
2211 to create a macro call that begins inside the macro body but ends
2212 outside of it. For example,
2214 #define strange(file) fprintf (file, "%s %d",
2216 strange(stderr) p, 35)
2217 ==> fprintf (stderr, "%s %d", p, 35)
2219 The ability to piece together a macro call can be useful, but the
2220 use of unbalanced open parentheses in a macro body is just confusing,
2221 and should be avoided.
2224 File: cpp.info, Node: Operator Precedence Problems, Next: Swallowing the Semicolon, Prev: Misnesting, Up: Macro Pitfalls
2226 3.10.2 Operator Precedence Problems
2227 -----------------------------------
2229 You may have noticed that in most of the macro definition examples shown
2230 above, each occurrence of a macro argument name had parentheses around
2231 it. In addition, another pair of parentheses usually surround the
2232 entire macro definition. Here is why it is best to write macros that
2235 Suppose you define a macro as follows,
2237 #define ceil_div(x, y) (x + y - 1) / y
2239 whose purpose is to divide, rounding up. (One use for this operation is
2240 to compute how many `int' objects are needed to hold a certain number
2241 of `char' objects.) Then suppose it is used as follows:
2243 a = ceil_div (b & c, sizeof (int));
2244 ==> a = (b & c + sizeof (int) - 1) / sizeof (int);
2246 This does not do what is intended. The operator-precedence rules of C
2247 make it equivalent to this:
2249 a = (b & (c + sizeof (int) - 1)) / sizeof (int);
2251 What we want is this:
2253 a = ((b & c) + sizeof (int) - 1)) / sizeof (int);
2255 Defining the macro as
2257 #define ceil_div(x, y) ((x) + (y) - 1) / (y)
2259 provides the desired result.
2261 Unintended grouping can result in another way. Consider `sizeof
2262 ceil_div(1, 2)'. That has the appearance of a C expression that would
2263 compute the size of the type of `ceil_div (1, 2)', but in fact it means
2264 something very different. Here is what it expands to:
2266 sizeof ((1) + (2) - 1) / (2)
2268 This would take the size of an integer and divide it by two. The
2269 precedence rules have put the division outside the `sizeof' when it was
2270 intended to be inside.
2272 Parentheses around the entire macro definition prevent such problems.
2273 Here, then, is the recommended way to define `ceil_div':
2275 #define ceil_div(x, y) (((x) + (y) - 1) / (y))
2278 File: cpp.info, Node: Swallowing the Semicolon, Next: Duplication of Side Effects, Prev: Operator Precedence Problems, Up: Macro Pitfalls
2280 3.10.3 Swallowing the Semicolon
2281 -------------------------------
2283 Often it is desirable to define a macro that expands into a compound
2284 statement. Consider, for example, the following macro, that advances a
2285 pointer (the argument `p' says where to find it) across whitespace
2288 #define SKIP_SPACES(p, limit) \
2289 { char *lim = (limit); \
2291 if (*p++ != ' ') { \
2294 Here backslash-newline is used to split the macro definition, which must
2295 be a single logical line, so that it resembles the way such code would
2296 be laid out if not part of a macro definition.
2298 A call to this macro might be `SKIP_SPACES (p, lim)'. Strictly
2299 speaking, the call expands to a compound statement, which is a complete
2300 statement with no need for a semicolon to end it. However, since it
2301 looks like a function call, it minimizes confusion if you can use it
2302 like a function call, writing a semicolon afterward, as in `SKIP_SPACES
2305 This can cause trouble before `else' statements, because the
2306 semicolon is actually a null statement. Suppose you write
2309 SKIP_SPACES (p, lim);
2312 The presence of two statements--the compound statement and a null
2313 statement--in between the `if' condition and the `else' makes invalid C
2316 The definition of the macro `SKIP_SPACES' can be altered to solve
2317 this problem, using a `do ... while' statement. Here is how:
2319 #define SKIP_SPACES(p, limit) \
2320 do { char *lim = (limit); \
2322 if (*p++ != ' ') { \
2326 Now `SKIP_SPACES (p, lim);' expands into
2330 which is one statement. The loop executes exactly once; most compilers
2331 generate no extra code for it.
2334 File: cpp.info, Node: Duplication of Side Effects, Next: Self-Referential Macros, Prev: Swallowing the Semicolon, Up: Macro Pitfalls
2336 3.10.4 Duplication of Side Effects
2337 ----------------------------------
2339 Many C programs define a macro `min', for "minimum", like this:
2341 #define min(X, Y) ((X) < (Y) ? (X) : (Y))
2343 When you use this macro with an argument containing a side effect,
2346 next = min (x + y, foo (z));
2348 it expands as follows:
2350 next = ((x + y) < (foo (z)) ? (x + y) : (foo (z)));
2352 where `x + y' has been substituted for `X' and `foo (z)' for `Y'.
2354 The function `foo' is used only once in the statement as it appears
2355 in the program, but the expression `foo (z)' has been substituted twice
2356 into the macro expansion. As a result, `foo' might be called two times
2357 when the statement is executed. If it has side effects or if it takes
2358 a long time to compute, the results might not be what you intended. We
2359 say that `min' is an "unsafe" macro.
2361 The best solution to this problem is to define `min' in a way that
2362 computes the value of `foo (z)' only once. The C language offers no
2363 standard way to do this, but it can be done with GNU extensions as
2367 ({ typeof (X) x_ = (X); \
2368 typeof (Y) y_ = (Y); \
2369 (x_ < y_) ? x_ : y_; })
2371 The `({ ... })' notation produces a compound statement that acts as
2372 an expression. Its value is the value of its last statement. This
2373 permits us to define local variables and assign each argument to one.
2374 The local variables have underscores after their names to reduce the
2375 risk of conflict with an identifier of wider scope (it is impossible to
2376 avoid this entirely). Now each argument is evaluated exactly once.
2378 If you do not wish to use GNU C extensions, the only solution is to
2379 be careful when _using_ the macro `min'. For example, you can
2380 calculate the value of `foo (z)', save it in a variable, and use that
2383 #define min(X, Y) ((X) < (Y) ? (X) : (Y))
2387 next = min (x + y, tem);
2390 (where we assume that `foo' returns type `int').
2393 File: cpp.info, Node: Self-Referential Macros, Next: Argument Prescan, Prev: Duplication of Side Effects, Up: Macro Pitfalls
2395 3.10.5 Self-Referential Macros
2396 ------------------------------
2398 A "self-referential" macro is one whose name appears in its definition.
2399 Recall that all macro definitions are rescanned for more macros to
2400 replace. If the self-reference were considered a use of the macro, it
2401 would produce an infinitely large expansion. To prevent this, the
2402 self-reference is not considered a macro call. It is passed into the
2403 preprocessor output unchanged. Consider an example:
2405 #define foo (4 + foo)
2407 where `foo' is also a variable in your program.
2409 Following the ordinary rules, each reference to `foo' will expand
2410 into `(4 + foo)'; then this will be rescanned and will expand into `(4
2411 + (4 + foo))'; and so on until the computer runs out of memory.
2413 The self-reference rule cuts this process short after one step, at
2414 `(4 + foo)'. Therefore, this macro definition has the possibly useful
2415 effect of causing the program to add 4 to the value of `foo' wherever
2416 `foo' is referred to.
2418 In most cases, it is a bad idea to take advantage of this feature. A
2419 person reading the program who sees that `foo' is a variable will not
2420 expect that it is a macro as well. The reader will come across the
2421 identifier `foo' in the program and think its value should be that of
2422 the variable `foo', whereas in fact the value is four greater.
2424 One common, useful use of self-reference is to create a macro which
2425 expands to itself. If you write
2429 then the macro `EPERM' expands to `EPERM'. Effectively, it is left
2430 alone by the preprocessor whenever it's used in running text. You can
2431 tell that it's a macro with `#ifdef'. You might do this if you want to
2432 define numeric constants with an `enum', but have `#ifdef' be true for
2435 If a macro `x' expands to use a macro `y', and the expansion of `y'
2436 refers to the macro `x', that is an "indirect self-reference" of `x'.
2437 `x' is not expanded in this case either. Thus, if we have
2442 then `x' and `y' expand as follows:
2450 Each macro is expanded when it appears in the definition of the other
2451 macro, but not when it indirectly appears in its own definition.
2454 File: cpp.info, Node: Argument Prescan, Next: Newlines in Arguments, Prev: Self-Referential Macros, Up: Macro Pitfalls
2456 3.10.6 Argument Prescan
2457 -----------------------
2459 Macro arguments are completely macro-expanded before they are
2460 substituted into a macro body, unless they are stringified or pasted
2461 with other tokens. After substitution, the entire macro body, including
2462 the substituted arguments, is scanned again for macros to be expanded.
2463 The result is that the arguments are scanned _twice_ to expand macro
2466 Most of the time, this has no effect. If the argument contained any
2467 macro calls, they are expanded during the first scan. The result
2468 therefore contains no macro calls, so the second scan does not change
2469 it. If the argument were substituted as given, with no prescan, the
2470 single remaining scan would find the same macro calls and produce the
2473 You might expect the double scan to change the results when a
2474 self-referential macro is used in an argument of another macro (*note
2475 Self-Referential Macros::): the self-referential macro would be
2476 expanded once in the first scan, and a second time in the second scan.
2477 However, this is not what happens. The self-references that do not
2478 expand in the first scan are marked so that they will not expand in the
2481 You might wonder, "Why mention the prescan, if it makes no
2482 difference? And why not skip it and make the preprocessor faster?"
2483 The answer is that the prescan does make a difference in three special
2486 * Nested calls to a macro.
2488 We say that "nested" calls to a macro occur when a macro's argument
2489 contains a call to that very macro. For example, if `f' is a macro
2490 that expects one argument, `f (f (1))' is a nested pair of calls to
2491 `f'. The desired expansion is made by expanding `f (1)' and
2492 substituting that into the definition of `f'. The prescan causes
2493 the expected result to happen. Without the prescan, `f (1)' itself
2494 would be substituted as an argument, and the inner use of `f' would
2495 appear during the main scan as an indirect self-reference and
2496 would not be expanded.
2498 * Macros that call other macros that stringify or concatenate.
2500 If an argument is stringified or concatenated, the prescan does not
2501 occur. If you _want_ to expand a macro, then stringify or
2502 concatenate its expansion, you can do that by causing one macro to
2503 call another macro that does the stringification or concatenation.
2504 For instance, if you have
2506 #define AFTERX(x) X_ ## x
2507 #define XAFTERX(x) AFTERX(x)
2508 #define TABLESIZE 1024
2509 #define BUFSIZE TABLESIZE
2511 then `AFTERX(BUFSIZE)' expands to `X_BUFSIZE', and
2512 `XAFTERX(BUFSIZE)' expands to `X_1024'. (Not to `X_TABLESIZE'.
2513 Prescan always does a complete expansion.)
2515 * Macros used in arguments, whose expansions contain unshielded
2518 This can cause a macro expanded on the second scan to be called
2519 with the wrong number of arguments. Here is an example:
2522 #define bar(x) lose(x)
2523 #define lose(x) (1 + (x))
2525 We would like `bar(foo)' to turn into `(1 + (foo))', which would
2526 then turn into `(1 + (a,b))'. Instead, `bar(foo)' expands into
2527 `lose(a,b)', and you get an error because `lose' requires a single
2528 argument. In this case, the problem is easily solved by the same
2529 parentheses that ought to be used to prevent misnesting of
2530 arithmetic operations:
2534 #define bar(x) lose((x))
2536 The extra pair of parentheses prevents the comma in `foo''s
2537 definition from being interpreted as an argument separator.
2541 File: cpp.info, Node: Newlines in Arguments, Prev: Argument Prescan, Up: Macro Pitfalls
2543 3.10.7 Newlines in Arguments
2544 ----------------------------
2546 The invocation of a function-like macro can extend over many logical
2547 lines. However, in the present implementation, the entire expansion
2548 comes out on one line. Thus line numbers emitted by the compiler or
2549 debugger refer to the line the invocation started on, which might be
2550 different to the line containing the argument causing the problem.
2552 Here is an example illustrating this:
2554 #define ignore_second_arg(a,b,c) a; c
2556 ignore_second_arg (foo (),
2560 The syntax error triggered by the tokens `syntax error' results in an
2561 error message citing line three--the line of ignore_second_arg-- even
2562 though the problematic code comes from line five.
2564 We consider this a bug, and intend to fix it in the near future.
2567 File: cpp.info, Node: Conditionals, Next: Diagnostics, Prev: Macros, Up: Top
2572 A "conditional" is a directive that instructs the preprocessor to
2573 select whether or not to include a chunk of code in the final token
2574 stream passed to the compiler. Preprocessor conditionals can test
2575 arithmetic expressions, or whether a name is defined as a macro, or both
2576 simultaneously using the special `defined' operator.
2578 A conditional in the C preprocessor resembles in some ways an `if'
2579 statement in C, but it is important to understand the difference between
2580 them. The condition in an `if' statement is tested during the
2581 execution of your program. Its purpose is to allow your program to
2582 behave differently from run to run, depending on the data it is
2583 operating on. The condition in a preprocessing conditional directive is
2584 tested when your program is compiled. Its purpose is to allow different
2585 code to be included in the program depending on the situation at the
2586 time of compilation.
2588 However, the distinction is becoming less clear. Modern compilers
2589 often do test `if' statements when a program is compiled, if their
2590 conditions are known not to vary at run time, and eliminate code which
2591 can never be executed. If you can count on your compiler to do this,
2592 you may find that your program is more readable if you use `if'
2593 statements with constant conditions (perhaps determined by macros). Of
2594 course, you can only use this to exclude code, not type definitions or
2595 other preprocessing directives, and you can only do it if the code
2596 remains syntactically valid when it is not to be used.
2598 GCC version 3 eliminates this kind of never-executed code even when
2599 not optimizing. Older versions did it only when optimizing.
2603 * Conditional Uses::
2604 * Conditional Syntax::
2608 File: cpp.info, Node: Conditional Uses, Next: Conditional Syntax, Up: Conditionals
2610 4.1 Conditional Uses
2611 ====================
2613 There are three general reasons to use a conditional.
2615 * A program may need to use different code depending on the machine
2616 or operating system it is to run on. In some cases the code for
2617 one operating system may be erroneous on another operating system;
2618 for example, it might refer to data types or constants that do not
2619 exist on the other system. When this happens, it is not enough to
2620 avoid executing the invalid code. Its mere presence will cause
2621 the compiler to reject the program. With a preprocessing
2622 conditional, the offending code can be effectively excised from
2623 the program when it is not valid.
2625 * You may want to be able to compile the same source file into two
2626 different programs. One version might make frequent time-consuming
2627 consistency checks on its intermediate data, or print the values of
2628 those data for debugging, and the other not.
2630 * A conditional whose condition is always false is one way to
2631 exclude code from the program but keep it as a sort of comment for
2634 Simple programs that do not need system-specific logic or complex
2635 debugging hooks generally will not need to use preprocessing
2639 File: cpp.info, Node: Conditional Syntax, Next: Deleted Code, Prev: Conditional Uses, Up: Conditionals
2641 4.2 Conditional Syntax
2642 ======================
2644 A conditional in the C preprocessor begins with a "conditional
2645 directive": `#if', `#ifdef' or `#ifndef'.
2656 File: cpp.info, Node: Ifdef, Next: If, Up: Conditional Syntax
2661 The simplest sort of conditional is
2669 This block is called a "conditional group". CONTROLLED TEXT will be
2670 included in the output of the preprocessor if and only if MACRO is
2671 defined. We say that the conditional "succeeds" if MACRO is defined,
2672 "fails" if it is not.
2674 The CONTROLLED TEXT inside of a conditional can include
2675 preprocessing directives. They are executed only if the conditional
2676 succeeds. You can nest conditional groups inside other conditional
2677 groups, but they must be completely nested. In other words, `#endif'
2678 always matches the nearest `#ifdef' (or `#ifndef', or `#if'). Also,
2679 you cannot start a conditional group in one file and end it in another.
2681 Even if a conditional fails, the CONTROLLED TEXT inside it is still
2682 run through initial transformations and tokenization. Therefore, it
2683 must all be lexically valid C. Normally the only way this matters is
2684 that all comments and string literals inside a failing conditional group
2685 must still be properly ended.
2687 The comment following the `#endif' is not required, but it is a good
2688 practice if there is a lot of CONTROLLED TEXT, because it helps people
2689 match the `#endif' to the corresponding `#ifdef'. Older programs
2690 sometimes put MACRO directly after the `#endif' without enclosing it in
2691 a comment. This is invalid code according to the C standard. CPP
2692 accepts it with a warning. It never affects which `#ifndef' the
2695 Sometimes you wish to use some code if a macro is _not_ defined.
2696 You can do this by writing `#ifndef' instead of `#ifdef'. One common
2697 use of `#ifndef' is to include code only the first time a header file
2698 is included. *Note Once-Only Headers::.
2700 Macro definitions can vary between compilations for several reasons.
2701 Here are some samples.
2703 * Some macros are predefined on each kind of machine (*note
2704 System-specific Predefined Macros::). This allows you to provide
2705 code specially tuned for a particular machine.
2707 * System header files define more macros, associated with the
2708 features they implement. You can test these macros with
2709 conditionals to avoid using a system feature on a machine where it
2712 * Macros can be defined or undefined with the `-D' and `-U' command
2713 line options when you compile the program. You can arrange to
2714 compile the same source file into two different programs by
2715 choosing a macro name to specify which program you want, writing
2716 conditionals to test whether or how this macro is defined, and
2717 then controlling the state of the macro with command line options,
2718 perhaps set in the Makefile. *Note Invocation::.
2720 * Your program might have a special header file (often called
2721 `config.h') that is adjusted when the program is compiled. It can
2722 define or not define macros depending on the features of the
2723 system and the desired capabilities of the program. The
2724 adjustment can be automated by a tool such as `autoconf', or done
2728 File: cpp.info, Node: If, Next: Defined, Prev: Ifdef, Up: Conditional Syntax
2733 The `#if' directive allows you to test the value of an arithmetic
2734 expression, rather than the mere existence of one macro. Its syntax is
2740 #endif /* EXPRESSION */
2742 EXPRESSION is a C expression of integer type, subject to stringent
2743 restrictions. It may contain
2745 * Integer constants.
2747 * Character constants, which are interpreted as they would be in
2750 * Arithmetic operators for addition, subtraction, multiplication,
2751 division, bitwise operations, shifts, comparisons, and logical
2752 operations (`&&' and `||'). The latter two obey the usual
2753 short-circuiting rules of standard C.
2755 * Macros. All macros in the expression are expanded before actual
2756 computation of the expression's value begins.
2758 * Uses of the `defined' operator, which lets you check whether macros
2759 are defined in the middle of an `#if'.
2761 * Identifiers that are not macros, which are all considered to be the
2762 number zero. This allows you to write `#if MACRO' instead of
2763 `#ifdef MACRO', if you know that MACRO, when defined, will always
2764 have a nonzero value. Function-like macros used without their
2765 function call parentheses are also treated as zero.
2767 In some contexts this shortcut is undesirable. The `-Wundef'
2768 option causes GCC to warn whenever it encounters an identifier
2769 which is not a macro in an `#if'.
2771 The preprocessor does not know anything about types in the language.
2772 Therefore, `sizeof' operators are not recognized in `#if', and neither
2773 are `enum' constants. They will be taken as identifiers which are not
2774 macros, and replaced by zero. In the case of `sizeof', this is likely
2775 to cause the expression to be invalid.
2777 The preprocessor calculates the value of EXPRESSION. It carries out
2778 all calculations in the widest integer type known to the compiler; on
2779 most machines supported by GCC this is 64 bits. This is not the same
2780 rule as the compiler uses to calculate the value of a constant
2781 expression, and may give different results in some cases. If the value
2782 comes out to be nonzero, the `#if' succeeds and the CONTROLLED TEXT is
2783 included; otherwise it is skipped.
2786 File: cpp.info, Node: Defined, Next: Else, Prev: If, Up: Conditional Syntax
2791 The special operator `defined' is used in `#if' and `#elif' expressions
2792 to test whether a certain name is defined as a macro. `defined NAME'
2793 and `defined (NAME)' are both expressions whose value is 1 if NAME is
2794 defined as a macro at the current point in the program, and 0
2795 otherwise. Thus, `#if defined MACRO' is precisely equivalent to
2798 `defined' is useful when you wish to test more than one macro for
2799 existence at once. For example,
2801 #if defined (__vax__) || defined (__ns16000__)
2803 would succeed if either of the names `__vax__' or `__ns16000__' is
2806 Conditionals written like this:
2808 #if defined BUFSIZE && BUFSIZE >= 1024
2810 can generally be simplified to just `#if BUFSIZE >= 1024', since if
2811 `BUFSIZE' is not defined, it will be interpreted as having the value
2814 If the `defined' operator appears as a result of a macro expansion,
2815 the C standard says the behavior is undefined. GNU cpp treats it as a
2816 genuine `defined' operator and evaluates it normally. It will warn
2817 wherever your code uses this feature if you use the command-line option
2818 `-pedantic', since other compilers may handle it differently.
2821 File: cpp.info, Node: Else, Next: Elif, Prev: Defined, Up: Conditional Syntax
2826 The `#else' directive can be added to a conditional to provide
2827 alternative text to be used if the condition fails. This is what it
2832 #else /* Not EXPRESSION */
2834 #endif /* Not EXPRESSION */
2836 If EXPRESSION is nonzero, the TEXT-IF-TRUE is included and the
2837 TEXT-IF-FALSE is skipped. If EXPRESSION is zero, the opposite happens.
2839 You can use `#else' with `#ifdef' and `#ifndef', too.
2842 File: cpp.info, Node: Elif, Prev: Else, Up: Conditional Syntax
2847 One common case of nested conditionals is used to check for more than
2848 two possible alternatives. For example, you might have
2860 Another conditional directive, `#elif', allows this to be
2861 abbreviated as follows:
2867 #else /* X != 2 and X != 1*/
2869 #endif /* X != 2 and X != 1*/
2871 `#elif' stands for "else if". Like `#else', it goes in the middle
2872 of a conditional group and subdivides it; it does not require a
2873 matching `#endif' of its own. Like `#if', the `#elif' directive
2874 includes an expression to be tested. The text following the `#elif' is
2875 processed only if the original `#if'-condition failed and the `#elif'
2878 More than one `#elif' can go in the same conditional group. Then
2879 the text after each `#elif' is processed only if the `#elif' condition
2880 succeeds after the original `#if' and all previous `#elif' directives
2881 within it have failed.
2883 `#else' is allowed after any number of `#elif' directives, but
2884 `#elif' may not follow `#else'.
2887 File: cpp.info, Node: Deleted Code, Prev: Conditional Syntax, Up: Conditionals
2892 If you replace or delete a part of the program but want to keep the old
2893 code around for future reference, you often cannot simply comment it
2894 out. Block comments do not nest, so the first comment inside the old
2895 code will end the commenting-out. The probable result is a flood of
2898 One way to avoid this problem is to use an always-false conditional
2899 instead. For instance, put `#if 0' before the deleted code and
2900 `#endif' after it. This works even if the code being turned off
2901 contains conditionals, but they must be entire conditionals (balanced
2902 `#if' and `#endif').
2904 Some people use `#ifdef notdef' instead. This is risky, because
2905 `notdef' might be accidentally defined as a macro, and then the
2906 conditional would succeed. `#if 0' can be counted on to fail.
2908 Do not use `#if 0' for comments which are not C code. Use a real
2909 comment, instead. The interior of `#if 0' must consist of complete
2910 tokens; in particular, single-quote characters must balance. Comments
2911 often contain unbalanced single-quote characters (known in English as
2912 apostrophes). These confuse `#if 0'. They don't confuse `/*'.
2915 File: cpp.info, Node: Diagnostics, Next: Line Control, Prev: Conditionals, Up: Top
2920 The directive `#error' causes the preprocessor to report a fatal error.
2921 The tokens forming the rest of the line following `#error' are used as
2924 You would use `#error' inside of a conditional that detects a
2925 combination of parameters which you know the program does not properly
2926 support. For example, if you know that the program will not run
2927 properly on a VAX, you might write
2930 #error "Won't work on VAXen. See comments at get_last_object."
2933 If you have several configuration parameters that must be set up by
2934 the installation in a consistent way, you can use conditionals to detect
2935 an inconsistency and report it with `#error'. For example,
2937 #if !defined(UNALIGNED_INT_ASM_OP) && defined(DWARF2_DEBUGGING_INFO)
2938 #error "DWARF2_DEBUGGING_INFO requires UNALIGNED_INT_ASM_OP."
2941 The directive `#warning' is like `#error', but causes the
2942 preprocessor to issue a warning and continue preprocessing. The tokens
2943 following `#warning' are used as the warning message.
2945 You might use `#warning' in obsolete header files, with a message
2946 directing the user to the header file which should be used instead.
2948 Neither `#error' nor `#warning' macro-expands its argument.
2949 Internal whitespace sequences are each replaced with a single space.
2950 The line must consist of complete tokens. It is wisest to make the
2951 argument of these directives be a single string constant; this avoids
2952 problems with apostrophes and the like.
2955 File: cpp.info, Node: Line Control, Next: Pragmas, Prev: Diagnostics, Up: Top
2960 The C preprocessor informs the C compiler of the location in your source
2961 code where each token came from. Presently, this is just the file name
2962 and line number. All the tokens resulting from macro expansion are
2963 reported as having appeared on the line of the source file where the
2964 outermost macro was used. We intend to be more accurate in the future.
2966 If you write a program which generates source code, such as the
2967 `bison' parser generator, you may want to adjust the preprocessor's
2968 notion of the current file name and line number by hand. Parts of the
2969 output from `bison' are generated from scratch, other parts come from a
2970 standard parser file. The rest are copied verbatim from `bison''s
2971 input. You would like compiler error messages and symbolic debuggers
2972 to be able to refer to `bison''s input file.
2974 `bison' or any such program can arrange this by writing `#line'
2975 directives into the output file. `#line' is a directive that specifies
2976 the original line number and source file name for subsequent input in
2977 the current preprocessor input file. `#line' has three variants:
2980 LINENUM is a non-negative decimal integer constant. It specifies
2981 the line number which should be reported for the following line of
2982 input. Subsequent lines are counted from LINENUM.
2984 `#line LINENUM FILENAME'
2985 LINENUM is the same as for the first form, and has the same
2986 effect. In addition, FILENAME is a string constant. The
2987 following line and all subsequent lines are reported to come from
2988 the file it specifies, until something else happens to change that.
2989 FILENAME is interpreted according to the normal rules for a string
2990 constant: backslash escapes are interpreted. This is different
2993 Previous versions of CPP did not interpret escapes in `#line'; we
2994 have changed it because the standard requires they be interpreted,
2995 and most other compilers do.
2997 `#line ANYTHING ELSE'
2998 ANYTHING ELSE is checked for macro calls, which are expanded. The
2999 result should match one of the above two forms.
3001 `#line' directives alter the results of the `__FILE__' and
3002 `__LINE__' predefined macros from that point on. *Note Standard
3003 Predefined Macros::. They do not have any effect on `#include''s idea
3004 of the directory containing the current file. This is a change from
3005 GCC 2.95. Previously, a file reading
3007 #line 1 "../src/gram.y"
3010 would search for `gram.h' in `../src', then the `-I' chain; the
3011 directory containing the physical source file would not be searched.
3012 In GCC 3.0 and later, the `#include' is not affected by the presence of
3013 a `#line' referring to a different directory.
3015 We made this change because the old behavior caused problems when
3016 generated source files were transported between machines. For instance,
3017 it is common practice to ship generated parsers with a source release,
3018 so that people building the distribution do not need to have yacc or
3019 Bison installed. These files frequently have `#line' directives
3020 referring to the directory tree of the system where the distribution was
3021 created. If GCC tries to search for headers in those directories, the
3022 build is likely to fail.
3024 The new behavior can cause failures too, if the generated file is not
3025 in the same directory as its source and it attempts to include a header
3026 which would be visible searching from the directory containing the
3027 source file. However, this problem is easily solved with an additional
3028 `-I' switch on the command line. The failures caused by the old
3029 semantics could sometimes be corrected only by editing the generated
3030 files, which is difficult and error-prone.
3033 File: cpp.info, Node: Pragmas, Next: Other Directives, Prev: Line Control, Up: Top
3038 The `#pragma' directive is the method specified by the C standard for
3039 providing additional information to the compiler, beyond what is
3040 conveyed in the language itself. Three forms of this directive
3041 (commonly known as "pragmas") are specified by the 1999 C standard. A
3042 C compiler is free to attach any meaning it likes to other pragmas.
3044 GCC has historically preferred to use extensions to the syntax of the
3045 language, such as `__attribute__', for this purpose. However, GCC does
3046 define a few pragmas of its own. These mostly have effects on the
3047 entire translation unit or source file.
3049 In GCC version 3, all GNU-defined, supported pragmas have been given
3050 a `GCC' prefix. This is in line with the `STDC' prefix on all pragmas
3051 defined by C99. For backward compatibility, pragmas which were
3052 recognized by previous versions are still recognized without the `GCC'
3053 prefix, but that usage is deprecated. Some older pragmas are
3054 deprecated in their entirety. They are not recognized with the `GCC'
3055 prefix. *Note Obsolete Features::.
3057 C99 introduces the `_Pragma' operator. This feature addresses a
3058 major problem with `#pragma': being a directive, it cannot be produced
3059 as the result of macro expansion. `_Pragma' is an operator, much like
3060 `sizeof' or `defined', and can be embedded in a macro.
3062 Its syntax is `_Pragma (STRING-LITERAL)', where STRING-LITERAL can
3063 be either a normal or wide-character string literal. It is
3064 destringized, by replacing all `\\' with a single `\' and all `\"' with
3065 a `"'. The result is then processed as if it had appeared as the right
3066 hand side of a `#pragma' directive. For example,
3068 _Pragma ("GCC dependency \"parse.y\"")
3070 has the same effect as `#pragma GCC dependency "parse.y"'. The same
3071 effect could be achieved using macros, for example
3073 #define DO_PRAGMA(x) _Pragma (#x)
3074 DO_PRAGMA (GCC dependency "parse.y")
3076 The standard is unclear on where a `_Pragma' operator can appear.
3077 The preprocessor does not accept it within a preprocessing conditional
3078 directive like `#if'. To be safe, you are probably best keeping it out
3079 of directives other than `#define', and putting it on a line of its own.
3081 This manual documents the pragmas which are meaningful to the
3082 preprocessor itself. Other pragmas are meaningful to the C or C++
3083 compilers. They are documented in the GCC manual.
3085 `#pragma GCC dependency'
3086 `#pragma GCC dependency' allows you to check the relative dates of
3087 the current file and another file. If the other file is more
3088 recent than the current file, a warning is issued. This is useful
3089 if the current file is derived from the other file, and should be
3090 regenerated. The other file is searched for using the normal
3091 include search path. Optional trailing text can be used to give
3092 more information in the warning message.
3094 #pragma GCC dependency "parse.y"
3095 #pragma GCC dependency "/usr/include/time.h" rerun fixincludes
3097 `#pragma GCC poison'
3098 Sometimes, there is an identifier that you want to remove
3099 completely from your program, and make sure that it never creeps
3100 back in. To enforce this, you can "poison" the identifier with
3101 this pragma. `#pragma GCC poison' is followed by a list of
3102 identifiers to poison. If any of those identifiers appears
3103 anywhere in the source after the directive, it is a hard error.
3106 #pragma GCC poison printf sprintf fprintf
3107 sprintf(some_string, "hello");
3109 will produce an error.
3111 If a poisoned identifier appears as part of the expansion of a
3112 macro which was defined before the identifier was poisoned, it
3113 will _not_ cause an error. This lets you poison an identifier
3114 without worrying about system headers defining macros that use it.
3118 #define strrchr rindex
3119 #pragma GCC poison rindex
3120 strrchr(some_string, 'h');
3122 will not produce an error.
3124 `#pragma GCC system_header'
3125 This pragma takes no arguments. It causes the rest of the code in
3126 the current file to be treated as if it came from a system header.
3127 *Note System Headers::.
3131 File: cpp.info, Node: Other Directives, Next: Preprocessor Output, Prev: Pragmas, Up: Top
3136 The `#ident' directive takes one argument, a string constant. On some
3137 systems, that string constant is copied into a special segment of the
3138 object file. On other systems, the directive is ignored. The `#sccs'
3139 directive is a synonym for `#ident'.
3141 These directives are not part of the C standard, but they are not
3142 official GNU extensions either. What historical information we have
3143 been able to find, suggests they originated with System V.
3145 The "null directive" consists of a `#' followed by a newline, with
3146 only whitespace (including comments) in between. A null directive is
3147 understood as a preprocessing directive but has no effect on the
3148 preprocessor output. The primary significance of the existence of the
3149 null directive is that an input line consisting of just a `#' will
3150 produce no output, rather than a line of output containing just a `#'.
3151 Supposedly some old C programs contain such lines.
3154 File: cpp.info, Node: Preprocessor Output, Next: Traditional Mode, Prev: Other Directives, Up: Top
3156 9 Preprocessor Output
3157 *********************
3159 When the C preprocessor is used with the C, C++, or Objective-C
3160 compilers, it is integrated into the compiler and communicates a stream
3161 of binary tokens directly to the compiler's parser. However, it can
3162 also be used in the more conventional standalone mode, where it produces
3165 The output from the C preprocessor looks much like the input, except
3166 that all preprocessing directive lines have been replaced with blank
3167 lines and all comments with spaces. Long runs of blank lines are
3170 The ISO standard specifies that it is implementation defined whether
3171 a preprocessor preserves whitespace between tokens, or replaces it with
3172 e.g. a single space. In GNU CPP, whitespace between tokens is collapsed
3173 to become a single space, with the exception that the first token on a
3174 non-directive line is preceded with sufficient spaces that it appears in
3175 the same column in the preprocessed output that it appeared in the
3176 original source file. This is so the output is easy to read. *Note
3177 Differences from previous versions::. CPP does not insert any
3178 whitespace where there was none in the original source, except where
3179 necessary to prevent an accidental token paste.
3181 Source file name and line number information is conveyed by lines of
3184 # LINENUM FILENAME FLAGS
3186 These are called "linemarkers". They are inserted as needed into the
3187 output (but never within a string or character constant). They mean
3188 that the following line originated in file FILENAME at line LINENUM.
3189 FILENAME will never contain any non-printing characters; they are
3190 replaced with octal escape sequences.
3192 After the file name comes zero or more flags, which are `1', `2',
3193 `3', or `4'. If there are multiple flags, spaces separate them. Here
3194 is what the flags mean:
3197 This indicates the start of a new file.
3200 This indicates returning to a file (after having included another
3204 This indicates that the following text comes from a system header
3205 file, so certain warnings should be suppressed.
3208 This indicates that the following text should be treated as being
3209 wrapped in an implicit `extern "C"' block.
3211 As an extension, the preprocessor accepts linemarkers in
3212 non-assembler input files. They are treated like the corresponding
3213 `#line' directive, (*note Line Control::), except that trailing flags
3214 are permitted, and are interpreted with the meanings described above.
3215 If multiple flags are given, they must be in ascending order.
3217 Some directives may be duplicated in the output of the preprocessor.
3218 These are `#ident' (always), `#pragma' (only if the preprocessor does
3219 not handle the pragma itself), and `#define' and `#undef' (with certain
3220 debugging options). If this happens, the `#' of the directive will
3221 always be in the first column, and there will be no space between the
3222 `#' and the directive name. If macro expansion happens to generate
3223 tokens which might be mistaken for a duplicated directive, a space will
3224 be inserted between the `#' and the directive name.
3227 File: cpp.info, Node: Traditional Mode, Next: Implementation Details, Prev: Preprocessor Output, Up: Top
3232 Traditional (pre-standard) C preprocessing is rather different from the
3233 preprocessing specified by the standard. When GCC is given the
3234 `-traditional-cpp' option, it attempts to emulate a traditional
3237 GCC versions 3.2 and later only support traditional mode semantics in
3238 the preprocessor, and not in the compiler front ends. This chapter
3239 outlines the traditional preprocessor semantics we implemented.
3241 The implementation does not correspond precisely to the behavior of
3242 earlier versions of GCC, nor to any true traditional preprocessor.
3243 After all, inconsistencies among traditional implementations were a
3244 major motivation for C standardization. However, we intend that it
3245 should be compatible with true traditional preprocessors in all ways
3246 that actually matter.
3250 * Traditional lexical analysis::
3251 * Traditional macros::
3252 * Traditional miscellany::
3253 * Traditional warnings::
3256 File: cpp.info, Node: Traditional lexical analysis, Next: Traditional macros, Up: Traditional Mode
3258 10.1 Traditional lexical analysis
3259 =================================
3261 The traditional preprocessor does not decompose its input into tokens
3262 the same way a standards-conforming preprocessor does. The input is
3263 simply treated as a stream of text with minimal internal form.
3265 This implementation does not treat trigraphs (*note trigraphs::)
3266 specially since they were an invention of the standards committee. It
3267 handles arbitrarily-positioned escaped newlines properly and splices
3268 the lines as you would expect; many traditional preprocessors did not
3271 The form of horizontal whitespace in the input file is preserved in
3272 the output. In particular, hard tabs remain hard tabs. This can be
3273 useful if, for example, you are preprocessing a Makefile.
3275 Traditional CPP only recognizes C-style block comments, and treats
3276 the `/*' sequence as introducing a comment only if it lies outside
3277 quoted text. Quoted text is introduced by the usual single and double
3278 quotes, and also by an initial `<' in a `#include' directive.
3280 Traditionally, comments are completely removed and are not replaced
3281 with a space. Since a traditional compiler does its own tokenization
3282 of the output of the preprocessor, this means that comments can
3283 effectively be used as token paste operators. However, comments behave
3284 like separators for text handled by the preprocessor itself, since it
3285 doesn't re-lex its input. For example, in
3289 `foo' and `bar' are distinct identifiers and expanded separately if
3290 they happen to be macros. In other words, this directive is equivalent
3299 Generally speaking, in traditional mode an opening quote need not
3300 have a matching closing quote. In particular, a macro may be defined
3301 with replacement text that contains an unmatched quote. Of course, if
3302 you attempt to compile preprocessed output containing an unmatched quote
3303 you will get a syntax error.
3305 However, all preprocessing directives other than `#define' require
3306 matching quotes. For example:
3308 #define m This macro's fine and has an unmatched quote
3309 "/* This is not a comment. */
3310 /* This is a comment. The following #include directive
3314 Just as for the ISO preprocessor, what would be a closing quote can
3315 be escaped with a backslash to prevent the quoted text from closing.
3318 File: cpp.info, Node: Traditional macros, Next: Traditional miscellany, Prev: Traditional lexical analysis, Up: Traditional Mode
3320 10.2 Traditional macros
3321 =======================
3323 The major difference between traditional and ISO macros is that the
3324 former expand to text rather than to a token sequence. CPP removes all
3325 leading and trailing horizontal whitespace from a macro's replacement
3326 text before storing it, but preserves the form of internal whitespace.
3328 One consequence is that it is legitimate for the replacement text to
3329 contain an unmatched quote (*note Traditional lexical analysis::). An
3330 unclosed string or character constant continues into the text following
3331 the macro call. Similarly, the text at the end of a macro's expansion
3332 can run together with the text after the macro invocation to produce a
3335 Normally comments are removed from the replacement text after the
3336 macro is expanded, but if the `-CC' option is passed on the command
3337 line comments are preserved. (In fact, the current implementation
3338 removes comments even before saving the macro replacement text, but it
3339 careful to do it in such a way that the observed effect is identical
3340 even in the function-like macro case.)
3342 The ISO stringification operator `#' and token paste operator `##'
3343 have no special meaning. As explained later, an effect similar to
3344 these operators can be obtained in a different way. Macro names that
3345 are embedded in quotes, either from the main file or after macro
3346 replacement, do not expand.
3348 CPP replaces an unquoted object-like macro name with its replacement
3349 text, and then rescans it for further macros to replace. Unlike
3350 standard macro expansion, traditional macro expansion has no provision
3351 to prevent recursion. If an object-like macro appears unquoted in its
3352 replacement text, it will be replaced again during the rescan pass, and
3353 so on _ad infinitum_. GCC detects when it is expanding recursive
3354 macros, emits an error message, and continues after the offending macro
3358 #define INC(x) PLUS+x
3362 Function-like macros are similar in form but quite different in
3363 behavior to their ISO counterparts. Their arguments are contained
3364 within parentheses, are comma-separated, and can cross physical lines.
3365 Commas within nested parentheses are not treated as argument
3366 separators. Similarly, a quote in an argument cannot be left unclosed;
3367 a following comma or parenthesis that comes before the closing quote is
3368 treated like any other character. There is no facility for handling
3371 This implementation removes all comments from macro arguments, unless
3372 the `-C' option is given. The form of all other horizontal whitespace
3373 in arguments is preserved, including leading and trailing whitespace.
3378 is treated as an invocation of the macro `f' with a single argument
3379 consisting of a single space. If you want to invoke a function-like
3380 macro that takes no arguments, you must not leave any whitespace
3381 between the parentheses.
3383 If a macro argument crosses a new line, the new line is replaced with
3384 a space when forming the argument. If the previous line contained an
3385 unterminated quote, the following line inherits the quoted state.
3387 Traditional preprocessors replace parameters in the replacement text
3388 with their arguments regardless of whether the parameters are within
3389 quotes or not. This provides a way to stringize arguments. For example
3392 str(/* A comment */some text )
3395 Note that the comment is removed, but that the trailing space is
3396 preserved. Here is an example of using a comment to effect token
3399 #define suffix(x) foo_/**/x
3404 File: cpp.info, Node: Traditional miscellany, Next: Traditional warnings, Prev: Traditional macros, Up: Traditional Mode
3406 10.3 Traditional miscellany
3407 ===========================
3409 Here are some things to be aware of when using the traditional
3412 * Preprocessing directives are recognized only when their leading
3413 `#' appears in the first column. There can be no whitespace
3414 between the beginning of the line and the `#', but whitespace can
3417 * A true traditional C preprocessor does not recognize `#error' or
3418 `#pragma', and may not recognize `#elif'. CPP supports all the
3419 directives in traditional mode that it supports in ISO mode,
3420 including extensions, with the exception that the effects of
3421 `#pragma GCC poison' are undefined.
3423 * __STDC__ is not defined.
3425 * If you use digraphs the behavior is undefined.
3427 * If a line that looks like a directive appears within macro
3428 arguments, the behavior is undefined.
3432 File: cpp.info, Node: Traditional warnings, Prev: Traditional miscellany, Up: Traditional Mode
3434 10.4 Traditional warnings
3435 =========================
3437 You can request warnings about features that did not exist, or worked
3438 differently, in traditional C with the `-Wtraditional' option. GCC
3439 does not warn about features of ISO C which you must use when you are
3440 using a conforming compiler, such as the `#' and `##' operators.
3442 Presently `-Wtraditional' warns about:
3444 * Macro parameters that appear within string literals in the macro
3445 body. In traditional C macro replacement takes place within
3446 string literals, but does not in ISO C.
3448 * In traditional C, some preprocessor directives did not exist.
3449 Traditional preprocessors would only consider a line to be a
3450 directive if the `#' appeared in column 1 on the line. Therefore
3451 `-Wtraditional' warns about directives that traditional C
3452 understands but would ignore because the `#' does not appear as the
3453 first character on the line. It also suggests you hide directives
3454 like `#pragma' not understood by traditional C by indenting them.
3455 Some traditional implementations would not recognize `#elif', so it
3456 suggests avoiding it altogether.
3458 * A function-like macro that appears without an argument list. In
3459 some traditional preprocessors this was an error. In ISO C it
3460 merely means that the macro is not expanded.
3462 * The unary plus operator. This did not exist in traditional C.
3464 * The `U' and `LL' integer constant suffixes, which were not
3465 available in traditional C. (Traditional C does support the `L'
3466 suffix for simple long integer constants.) You are not warned
3467 about uses of these suffixes in macros defined in system headers.
3468 For instance, `UINT_MAX' may well be defined as `4294967295U', but
3469 you will not be warned if you use `UINT_MAX'.
3471 You can usually avoid the warning, and the related warning about
3472 constants which are so large that they are unsigned, by writing the
3473 integer constant in question in hexadecimal, with no U suffix.
3474 Take care, though, because this gives the wrong result in exotic
3478 File: cpp.info, Node: Implementation Details, Next: Invocation, Prev: Traditional Mode, Up: Top
3480 11 Implementation Details
3481 *************************
3483 Here we document details of how the preprocessor's implementation
3484 affects its user-visible behavior. You should try to avoid undue
3485 reliance on behavior described here, as it is possible that it will
3486 change subtly in future implementations.
3488 Also documented here are obsolete features and changes from previous
3493 * Implementation-defined behavior::
3494 * Implementation limits::
3495 * Obsolete Features::
3496 * Differences from previous versions::
3499 File: cpp.info, Node: Implementation-defined behavior, Next: Implementation limits, Up: Implementation Details
3501 11.1 Implementation-defined behavior
3502 ====================================
3504 This is how CPP behaves in all the cases which the C standard describes
3505 as "implementation-defined". This term means that the implementation
3506 is free to do what it likes, but must document its choice and stick to
3509 * The mapping of physical source file multi-byte characters to the
3510 execution character set.
3512 Currently, CPP requires its input to be ASCII or UTF-8. The
3513 execution character set may be controlled by the user, with the
3514 `-fexec-charset' and `-fwide-exec-charset' options.
3516 * Identifier characters.
3518 The C and C++ standards allow identifiers to be composed of `_'
3519 and the alphanumeric characters. C++ and C99 also allow universal
3520 character names, and C99 further permits implementation-defined
3521 characters. GCC currently only permits universal character names
3522 if `-fextended-identifiers' is used, because the implementation of
3523 universal character names in identifiers is experimental.
3525 GCC allows the `$' character in identifiers as an extension for
3526 most targets. This is true regardless of the `std=' switch, since
3527 this extension cannot conflict with standards-conforming programs.
3528 When preprocessing assembler, however, dollars are not identifier
3529 characters by default.
3531 Currently the targets that by default do not permit `$' are AVR,
3532 IP2K, MMIX, MIPS Irix 3, ARM aout, and PowerPC targets for the AIX
3533 and BeOS operating systems.
3535 You can override the default with `-fdollars-in-identifiers' or
3536 `fno-dollars-in-identifiers'. *Note fdollars-in-identifiers::.
3538 * Non-empty sequences of whitespace characters.
3540 In textual output, each whitespace sequence is collapsed to a
3541 single space. For aesthetic reasons, the first token on each
3542 non-directive line of output is preceded with sufficient spaces
3543 that it appears in the same column as it did in the original
3546 * The numeric value of character constants in preprocessor
3549 The preprocessor and compiler interpret character constants in the
3550 same way; i.e. escape sequences such as `\a' are given the values
3551 they would have on the target machine.
3553 The compiler values a multi-character character constant a
3554 character at a time, shifting the previous value left by the
3555 number of bits per target character, and then or-ing in the
3556 bit-pattern of the new character truncated to the width of a
3557 target character. The final bit-pattern is given type `int', and
3558 is therefore signed, regardless of whether single characters are
3559 signed or not (a slight change from versions 3.1 and earlier of
3560 GCC). If there are more characters in the constant than would fit
3561 in the target `int' the compiler issues a warning, and the excess
3562 leading characters are ignored.
3564 For example, `'ab'' for a target with an 8-bit `char' would be
3566 `(int) ((unsigned char) 'a' * 256 + (unsigned char) 'b')', and
3568 `(int) ((unsigned char) '\234' * 256 + (unsigned char) 'a')'.
3570 * Source file inclusion.
3572 For a discussion on how the preprocessor locates header files,
3573 *Note Include Operation::.
3575 * Interpretation of the filename resulting from a macro-expanded
3576 `#include' directive.
3578 *Note Computed Includes::.
3580 * Treatment of a `#pragma' directive that after macro-expansion
3581 results in a standard pragma.
3583 No macro expansion occurs on any `#pragma' directive line, so the
3584 question does not arise.
3586 Note that GCC does not yet implement any of the standard pragmas.
3590 File: cpp.info, Node: Implementation limits, Next: Obsolete Features, Prev: Implementation-defined behavior, Up: Implementation Details
3592 11.2 Implementation limits
3593 ==========================
3595 CPP has a small number of internal limits. This section lists the
3596 limits which the C standard requires to be no lower than some minimum,
3597 and all the others known. It is intended that there should be as few
3598 limits as possible. If you encounter an undocumented or inconvenient
3599 limit, please report that as a bug. *Note Reporting Bugs: (gcc)Bugs.
3601 Where we say something is limited "only by available memory", that
3602 means that internal data structures impose no intrinsic limit, and space
3603 is allocated with `malloc' or equivalent. The actual limit will
3604 therefore depend on many things, such as the size of other things
3605 allocated by the compiler at the same time, the amount of memory
3606 consumed by other processes on the same computer, etc.
3608 * Nesting levels of `#include' files.
3610 We impose an arbitrary limit of 200 levels, to avoid runaway
3611 recursion. The standard requires at least 15 levels.
3613 * Nesting levels of conditional inclusion.
3615 The C standard mandates this be at least 63. CPP is limited only
3616 by available memory.
3618 * Levels of parenthesized expressions within a full expression.
3620 The C standard requires this to be at least 63. In preprocessor
3621 conditional expressions, it is limited only by available memory.
3623 * Significant initial characters in an identifier or macro name.
3625 The preprocessor treats all characters as significant. The C
3626 standard requires only that the first 63 be significant.
3628 * Number of macros simultaneously defined in a single translation
3631 The standard requires at least 4095 be possible. CPP is limited
3632 only by available memory.
3634 * Number of parameters in a macro definition and arguments in a
3637 We allow `USHRT_MAX', which is no smaller than 65,535. The minimum
3638 required by the standard is 127.
3640 * Number of characters on a logical source line.
3642 The C standard requires a minimum of 4096 be permitted. CPP places
3643 no limits on this, but you may get incorrect column numbers
3644 reported in diagnostics for lines longer than 65,535 characters.
3646 * Maximum size of a source file.
3648 The standard does not specify any lower limit on the maximum size
3649 of a source file. GNU cpp maps files into memory, so it is
3650 limited by the available address space. This is generally at
3651 least two gigabytes. Depending on the operating system, the size
3652 of physical memory may or may not be a limitation.
3656 File: cpp.info, Node: Obsolete Features, Next: Differences from previous versions, Prev: Implementation limits, Up: Implementation Details
3658 11.3 Obsolete Features
3659 ======================
3661 CPP has a number of features which are present mainly for compatibility
3662 with older programs. We discourage their use in new code. In some
3663 cases, we plan to remove the feature in a future version of GCC.
3668 * Obsolete once-only headers::
3671 File: cpp.info, Node: Assertions, Next: Obsolete once-only headers, Up: Obsolete Features
3676 "Assertions" are a deprecated alternative to macros in writing
3677 conditionals to test what sort of computer or system the compiled
3678 program will run on. Assertions are usually predefined, but you can
3679 define them with preprocessing directives or command-line options.
3681 Assertions were intended to provide a more systematic way to describe
3682 the compiler's target system. However, in practice they are just as
3683 unpredictable as the system-specific predefined macros. In addition,
3684 they are not part of any standard, and only a few compilers support
3685 them. Therefore, the use of assertions is *less* portable than the use
3686 of system-specific predefined macros. We recommend you do not use them
3689 An assertion looks like this:
3693 PREDICATE must be a single identifier. ANSWER can be any sequence of
3694 tokens; all characters are significant except for leading and trailing
3695 whitespace, and differences in internal whitespace sequences are
3696 ignored. (This is similar to the rules governing macro redefinition.)
3697 Thus, `(x + y)' is different from `(x+y)' but equivalent to
3698 `( x + y )'. Parentheses do not nest inside an answer.
3700 To test an assertion, you write it in an `#if'. For example, this
3701 conditional succeeds if either `vax' or `ns16000' has been asserted as
3702 an answer for `machine'.
3704 #if #machine (vax) || #machine (ns16000)
3706 You can test whether _any_ answer is asserted for a predicate by
3707 omitting the answer in the conditional:
3711 Assertions are made with the `#assert' directive. Its sole argument
3712 is the assertion to make, without the leading `#' that identifies
3713 assertions in conditionals.
3715 #assert PREDICATE (ANSWER)
3717 You may make several assertions with the same predicate and different
3718 answers. Subsequent assertions do not override previous ones for the
3719 same predicate. All the answers for any given predicate are
3720 simultaneously true.
3722 Assertions can be canceled with the `#unassert' directive. It has
3723 the same syntax as `#assert'. In that form it cancels only the answer
3724 which was specified on the `#unassert' line; other answers for that
3725 predicate remain true. You can cancel an entire predicate by leaving
3730 In either form, if no such assertion has been made, `#unassert' has no
3733 You can also make or cancel assertions using command line options.
3737 File: cpp.info, Node: Obsolete once-only headers, Prev: Assertions, Up: Obsolete Features
3739 11.3.2 Obsolete once-only headers
3740 ---------------------------------
3742 CPP supports two more ways of indicating that a header file should be
3743 read only once. Neither one is as portable as a wrapper `#ifndef', and
3744 we recommend you do not use them in new programs.
3746 In the Objective-C language, there is a variant of `#include' called
3747 `#import' which includes a file, but does so at most once. If you use
3748 `#import' instead of `#include', then you don't need the conditionals
3749 inside the header file to prevent multiple inclusion of the contents.
3750 GCC permits the use of `#import' in C and C++ as well as Objective-C.
3751 However, it is not in standard C or C++ and should therefore not be
3752 used by portable programs.
3754 `#import' is not a well designed feature. It requires the users of
3755 a header file to know that it should only be included once. It is much
3756 better for the header file's implementor to write the file so that users
3757 don't need to know this. Using a wrapper `#ifndef' accomplishes this
3760 In the present implementation, a single use of `#import' will
3761 prevent the file from ever being read again, by either `#import' or
3762 `#include'. You should not rely on this; do not use both `#import' and
3763 `#include' to refer to the same header file.
3765 Another way to prevent a header file from being included more than
3766 once is with the `#pragma once' directive. If `#pragma once' is seen
3767 when scanning a header file, that file will never be read again, no
3770 `#pragma once' does not have the problems that `#import' does, but
3771 it is not recognized by all preprocessors, so you cannot rely on it in
3775 File: cpp.info, Node: Differences from previous versions, Prev: Obsolete Features, Up: Implementation Details
3777 11.4 Differences from previous versions
3778 =======================================
3780 This section details behavior which has changed from previous versions
3781 of CPP. We do not plan to change it again in the near future, but we
3782 do not promise not to, either.
3784 The "previous versions" discussed here are 2.95 and before. The
3785 behavior of GCC 3.0 is mostly the same as the behavior of the widely
3786 used 2.96 and 2.97 development snapshots. Where there are differences,
3787 they generally represent bugs in the snapshots.
3791 This option has been deprecated in 4.0. `-iquote' is meant to
3792 replace the need for this option.
3794 * Order of evaluation of `#' and `##' operators
3796 The standard does not specify the order of evaluation of a chain of
3797 `##' operators, nor whether `#' is evaluated before, after, or at
3798 the same time as `##'. You should therefore not write any code
3799 which depends on any specific ordering. It is possible to
3800 guarantee an ordering, if you need one, by suitable use of nested
3803 An example of where this might matter is pasting the arguments `1',
3804 `e' and `-2'. This would be fine for left-to-right pasting, but
3805 right-to-left pasting would produce an invalid token `e-2'.
3807 GCC 3.0 evaluates `#' and `##' at the same time and strictly left
3808 to right. Older versions evaluated all `#' operators first, then
3809 all `##' operators, in an unreliable order.
3811 * The form of whitespace between tokens in preprocessor output
3813 *Note Preprocessor Output::, for the current textual format. This
3814 is also the format used by stringification. Normally, the
3815 preprocessor communicates tokens directly to the compiler's
3816 parser, and whitespace does not come up at all.
3818 Older versions of GCC preserved all whitespace provided by the
3819 user and inserted lots more whitespace of their own, because they
3820 could not accurately predict when extra spaces were needed to
3821 prevent accidental token pasting.
3823 * Optional argument when invoking rest argument macros
3825 As an extension, GCC permits you to omit the variable arguments
3826 entirely when you use a variable argument macro. This is
3827 forbidden by the 1999 C standard, and will provoke a pedantic
3828 warning with GCC 3.0. Previous versions accepted it silently.
3830 * `##' swallowing preceding text in rest argument macros
3832 Formerly, in a macro expansion, if `##' appeared before a variable
3833 arguments parameter, and the set of tokens specified for that
3834 argument in the macro invocation was empty, previous versions of
3835 CPP would back up and remove the preceding sequence of
3836 non-whitespace characters (*not* the preceding token). This
3837 extension is in direct conflict with the 1999 C standard and has
3838 been drastically pared back.
3840 In the current version of the preprocessor, if `##' appears between
3841 a comma and a variable arguments parameter, and the variable
3842 argument is omitted entirely, the comma will be removed from the
3843 expansion. If the variable argument is empty, or the token before
3844 `##' is not a comma, then `##' behaves as a normal token paste.
3846 * `#line' and `#include'
3848 The `#line' directive used to change GCC's notion of the
3849 "directory containing the current file", used by `#include' with a
3850 double-quoted header file name. In 3.0 and later, it does not.
3851 *Note Line Control::, for further explanation.
3855 In GCC 2.95 and previous, the string constant argument to `#line'
3856 was treated the same way as the argument to `#include': backslash
3857 escapes were not honored, and the string ended at the second `"'.
3858 This is not compliant with the C standard. In GCC 3.0, an attempt
3859 was made to correct the behavior, so that the string was treated
3860 as a real string constant, but it turned out to be buggy. In 3.1,
3861 the bugs have been fixed. (We are not fixing the bugs in 3.0
3862 because they affect relatively few people and the fix is quite
3867 File: cpp.info, Node: Invocation, Next: Environment Variables, Prev: Implementation Details, Up: Top
3872 Most often when you use the C preprocessor you will not have to invoke
3873 it explicitly: the C compiler will do so automatically. However, the
3874 preprocessor is sometimes useful on its own. All the options listed
3875 here are also acceptable to the C compiler and have the same meaning,
3876 except that the C compiler has different rules for specifying the output
3879 _Note:_ Whether you use the preprocessor by way of `gcc' or `cpp',
3880 the "compiler driver" is run first. This program's purpose is to
3881 translate your command into invocations of the programs that do the
3882 actual work. Their command line interfaces are similar but not
3883 identical to the documented interface, and may change without notice.
3885 The C preprocessor expects two file names as arguments, INFILE and
3886 OUTFILE. The preprocessor reads INFILE together with any other files
3887 it specifies with `#include'. All the output generated by the combined
3888 input files is written in OUTFILE.
3890 Either INFILE or OUTFILE may be `-', which as INFILE means to read
3891 from standard input and as OUTFILE means to write to standard output.
3892 Also, if either file is omitted, it means the same as if `-' had been
3893 specified for that file.
3895 Unless otherwise noted, or the option ends in `=', all options which
3896 take an argument may have that argument appear either immediately after
3897 the option, or with a space between option and argument: `-Ifoo' and
3898 `-I foo' have the same effect.
3900 Many options have multi-letter names; therefore multiple
3901 single-letter options may _not_ be grouped: `-dM' is very different from
3905 Predefine NAME as a macro, with definition `1'.
3907 `-D NAME=DEFINITION'
3908 The contents of DEFINITION are tokenized and processed as if they
3909 appeared during translation phase three in a `#define' directive.
3910 In particular, the definition will be truncated by embedded
3913 If you are invoking the preprocessor from a shell or shell-like
3914 program you may need to use the shell's quoting syntax to protect
3915 characters such as spaces that have a meaning in the shell syntax.
3917 If you wish to define a function-like macro on the command line,
3918 write its argument list with surrounding parentheses before the
3919 equals sign (if any). Parentheses are meaningful to most shells,
3920 so you will need to quote the option. With `sh' and `csh',
3921 `-D'NAME(ARGS...)=DEFINITION'' works.
3923 `-D' and `-U' options are processed in the order they are given on
3924 the command line. All `-imacros FILE' and `-include FILE' options
3925 are processed after all `-D' and `-U' options.
3928 Cancel any previous definition of NAME, either built in or
3929 provided with a `-D' option.
3932 Do not predefine any system-specific or GCC-specific macros. The
3933 standard predefined macros remain defined. *Note Standard
3934 Predefined Macros::.
3937 Add the directory DIR to the list of directories to be searched
3938 for header files. *Note Search Path::. Directories named by `-I'
3939 are searched before the standard system include directories. If
3940 the directory DIR is a standard system include directory, the
3941 option is ignored to ensure that the default search order for
3942 system directories and the special treatment of system headers are
3943 not defeated (*note System Headers::) . If DIR begins with `=',
3944 then the `=' will be replaced by the sysroot prefix; see
3945 `--sysroot' and `-isysroot'.
3948 Write output to FILE. This is the same as specifying FILE as the
3949 second non-option argument to `cpp'. `gcc' has a different
3950 interpretation of a second non-option argument, so you must use
3951 `-o' to specify the output file.
3954 Turns on all optional warnings which are desirable for normal code.
3955 At present this is `-Wcomment', `-Wtrigraphs', `-Wmultichar' and a
3956 warning about integer promotion causing a change of sign in `#if'
3957 expressions. Note that many of the preprocessor's warnings are on
3958 by default and have no options to control them.
3962 Warn whenever a comment-start sequence `/*' appears in a `/*'
3963 comment, or whenever a backslash-newline appears in a `//' comment.
3964 (Both forms have the same effect.)
3967 Most trigraphs in comments cannot affect the meaning of the
3968 program. However, a trigraph that would form an escaped newline
3969 (`??/' at the end of a line) can, by changing where the comment
3970 begins or ends. Therefore, only trigraphs that would form escaped
3971 newlines produce warnings inside a comment.
3973 This option is implied by `-Wall'. If `-Wall' is not given, this
3974 option is still enabled unless trigraphs are enabled. To get
3975 trigraph conversion without warnings, but get the other `-Wall'
3976 warnings, use `-trigraphs -Wall -Wno-trigraphs'.
3979 Warn about certain constructs that behave differently in
3980 traditional and ISO C. Also warn about ISO C constructs that have
3981 no traditional C equivalent, and problematic constructs which
3982 should be avoided. *Note Traditional Mode::.
3985 Warn the first time `#import' is used.
3988 Warn whenever an identifier which is not a macro is encountered in
3989 an `#if' directive, outside of `defined'. Such identifiers are
3993 Warn about macros defined in the main file that are unused. A
3994 macro is "used" if it is expanded or tested for existence at least
3995 once. The preprocessor will also warn if the macro has not been
3996 used at the time it is redefined or undefined.
3998 Built-in macros, macros defined on the command line, and macros
3999 defined in include files are not warned about.
4001 _Note:_ If a macro is actually used, but only used in skipped
4002 conditional blocks, then CPP will report it as unused. To avoid
4003 the warning in such a case, you might improve the scope of the
4004 macro's definition by, for example, moving it into the first
4005 skipped block. Alternatively, you could provide a dummy use with
4008 #if defined the_macro_causing_the_warning
4012 Warn whenever an `#else' or an `#endif' are followed by text.
4013 This usually happens in code of the form
4021 The second and third `FOO' should be in comments, but often are not
4022 in older programs. This warning is on by default.
4025 Make all warnings into hard errors. Source code which triggers
4026 warnings will be rejected.
4029 Issue warnings for code in system headers. These are normally
4030 unhelpful in finding bugs in your own code, therefore suppressed.
4031 If you are responsible for the system library, you may want to see
4035 Suppress all warnings, including those which GNU CPP issues by
4039 Issue all the mandatory diagnostics listed in the C standard.
4040 Some of them are left out by default, since they trigger
4041 frequently on harmless code.
4044 Issue all the mandatory diagnostics, and make all mandatory
4045 diagnostics into errors. This includes mandatory diagnostics that
4046 GCC issues without `-pedantic' but treats as warnings.
4049 Instead of outputting the result of preprocessing, output a rule
4050 suitable for `make' describing the dependencies of the main source
4051 file. The preprocessor outputs one `make' rule containing the
4052 object file name for that source file, a colon, and the names of
4053 all the included files, including those coming from `-include' or
4054 `-imacros' command line options.
4056 Unless specified explicitly (with `-MT' or `-MQ'), the object file
4057 name consists of the name of the source file with any suffix
4058 replaced with object file suffix and with any leading directory
4059 parts removed. If there are many included files then the rule is
4060 split into several lines using `\'-newline. The rule has no
4063 This option does not suppress the preprocessor's debug output,
4064 such as `-dM'. To avoid mixing such debug output with the
4065 dependency rules you should explicitly specify the dependency
4066 output file with `-MF', or use an environment variable like
4067 `DEPENDENCIES_OUTPUT' (*note Environment Variables::). Debug
4068 output will still be sent to the regular output stream as normal.
4070 Passing `-M' to the driver implies `-E', and suppresses warnings
4071 with an implicit `-w'.
4074 Like `-M' but do not mention header files that are found in system
4075 header directories, nor header files that are included, directly
4076 or indirectly, from such a header.
4078 This implies that the choice of angle brackets or double quotes in
4079 an `#include' directive does not in itself determine whether that
4080 header will appear in `-MM' dependency output. This is a slight
4081 change in semantics from GCC versions 3.0 and earlier.
4084 When used with `-M' or `-MM', specifies a file to write the
4085 dependencies to. If no `-MF' switch is given the preprocessor
4086 sends the rules to the same place it would have sent preprocessed
4089 When used with the driver options `-MD' or `-MMD', `-MF' overrides
4090 the default dependency output file.
4093 In conjunction with an option such as `-M' requesting dependency
4094 generation, `-MG' assumes missing header files are generated files
4095 and adds them to the dependency list without raising an error.
4096 The dependency filename is taken directly from the `#include'
4097 directive without prepending any path. `-MG' also suppresses
4098 preprocessed output, as a missing header file renders this useless.
4100 This feature is used in automatic updating of makefiles.
4103 This option instructs CPP to add a phony target for each dependency
4104 other than the main file, causing each to depend on nothing. These
4105 dummy rules work around errors `make' gives if you remove header
4106 files without updating the `Makefile' to match.
4108 This is typical output:
4110 test.o: test.c test.h
4115 Change the target of the rule emitted by dependency generation. By
4116 default CPP takes the name of the main input file, deletes any
4117 directory components and any file suffix such as `.c', and appends
4118 the platform's usual object suffix. The result is the target.
4120 An `-MT' option will set the target to be exactly the string you
4121 specify. If you want multiple targets, you can specify them as a
4122 single argument to `-MT', or use multiple `-MT' options.
4124 For example, `-MT '$(objpfx)foo.o'' might give
4126 $(objpfx)foo.o: foo.c
4129 Same as `-MT', but it quotes any characters which are special to
4130 Make. `-MQ '$(objpfx)foo.o'' gives
4132 $$(objpfx)foo.o: foo.c
4134 The default target is automatically quoted, as if it were given
4138 `-MD' is equivalent to `-M -MF FILE', except that `-E' is not
4139 implied. The driver determines FILE based on whether an `-o'
4140 option is given. If it is, the driver uses its argument but with
4141 a suffix of `.d', otherwise it takes the name of the input file,
4142 removes any directory components and suffix, and applies a `.d'
4145 If `-MD' is used in conjunction with `-E', any `-o' switch is
4146 understood to specify the dependency output file (*note -MF:
4147 dashMF.), but if used without `-E', each `-o' is understood to
4148 specify a target object file.
4150 Since `-E' is not implied, `-MD' can be used to generate a
4151 dependency output file as a side-effect of the compilation process.
4154 Like `-MD' except mention only user header files, not system
4160 `-x assembler-with-cpp'
4161 Specify the source language: C, C++, Objective-C, or assembly.
4162 This has nothing to do with standards conformance or extensions;
4163 it merely selects which base syntax to expect. If you give none
4164 of these options, cpp will deduce the language from the extension
4165 of the source file: `.c', `.cc', `.m', or `.S'. Some other common
4166 extensions for C++ and assembly are also recognized. If cpp does
4167 not recognize the extension, it will treat the file as C; this is
4168 the most generic mode.
4170 _Note:_ Previous versions of cpp accepted a `-lang' option which
4171 selected both the language and the standards conformance level.
4172 This option has been removed, because it conflicts with the `-l'
4177 Specify the standard to which the code should conform. Currently
4178 CPP knows about C and C++ standards; others may be added in the
4181 STANDARD may be one of:
4184 The ISO C standard from 1990. `c89' is the customary
4185 shorthand for this version of the standard.
4187 The `-ansi' option is equivalent to `-std=c89'.
4190 The 1990 C standard, as amended in 1994.
4196 The revised ISO C standard, published in December 1999.
4197 Before publication, this was known as C9X.
4200 The 1990 C standard plus GNU extensions. This is the default.
4204 The 1999 C standard plus GNU extensions.
4207 The 1998 ISO C++ standard plus amendments.
4210 The same as `-std=c++98' plus GNU extensions. This is the
4211 default for C++ code.
4214 Split the include path. Any directories specified with `-I'
4215 options before `-I-' are searched only for headers requested with
4216 `#include "FILE"'; they are not searched for `#include <FILE>'.
4217 If additional directories are specified with `-I' options after
4218 the `-I-', those directories are searched for all `#include'
4221 In addition, `-I-' inhibits the use of the directory of the current
4222 file directory as the first search directory for `#include "FILE"'.
4223 *Note Search Path::. This option has been deprecated.
4226 Do not search the standard system directories for header files.
4227 Only the directories you have specified with `-I' options (and the
4228 directory of the current file, if appropriate) are searched.
4231 Do not search for header files in the C++-specific standard
4232 directories, but do still search the other standard directories.
4233 (This option is used when building the C++ library.)
4236 Process FILE as if `#include "file"' appeared as the first line of
4237 the primary source file. However, the first directory searched
4238 for FILE is the preprocessor's working directory _instead of_ the
4239 directory containing the main source file. If not found there, it
4240 is searched for in the remainder of the `#include "..."' search
4243 If multiple `-include' options are given, the files are included
4244 in the order they appear on the command line.
4247 Exactly like `-include', except that any output produced by
4248 scanning FILE is thrown away. Macros it defines remain defined.
4249 This allows you to acquire all the macros from a header without
4250 also processing its declarations.
4252 All files specified by `-imacros' are processed before all files
4253 specified by `-include'.
4256 Search DIR for header files, but do it _after_ all directories
4257 specified with `-I' and the standard system directories have been
4258 exhausted. DIR is treated as a system include directory. If DIR
4259 begins with `=', then the `=' will be replaced by the sysroot
4260 prefix; see `--sysroot' and `-isysroot'.
4263 Specify PREFIX as the prefix for subsequent `-iwithprefix'
4264 options. If the prefix represents a directory, you should include
4268 `-iwithprefixbefore DIR'
4269 Append DIR to the prefix specified previously with `-iprefix', and
4270 add the resulting directory to the include search path.
4271 `-iwithprefixbefore' puts it in the same place `-I' would;
4272 `-iwithprefix' puts it where `-idirafter' would.
4275 This option is like the `--sysroot' option, but applies only to
4276 header files. See the `--sysroot' option for more information.
4279 Use DIR as a subdirectory of the directory containing
4280 target-specific C++ headers.
4283 Search DIR for header files, after all directories specified by
4284 `-I' but before the standard system directories. Mark it as a
4285 system directory, so that it gets the same special treatment as is
4286 applied to the standard system directories. *Note System
4287 Headers::. If DIR begins with `=', then the `=' will be replaced
4288 by the sysroot prefix; see `--sysroot' and `-isysroot'.
4291 Search DIR only for header files requested with `#include "FILE"';
4292 they are not searched for `#include <FILE>', before all
4293 directories specified by `-I' and before the standard system
4294 directories. *Note Search Path::. If DIR begins with `=', then
4295 the `=' will be replaced by the sysroot prefix; see `--sysroot'
4299 When preprocessing, handle directives, but do not expand macros.
4301 The option's behavior depends on the `-E' and `-fpreprocessed'
4304 With `-E', preprocessing is limited to the handling of directives
4305 such as `#define', `#ifdef', and `#error'. Other preprocessor
4306 operations, such as macro expansion and trigraph conversion are
4307 not performed. In addition, the `-dD' option is implicitly
4310 With `-fpreprocessed', predefinition of command line and most
4311 builtin macros is disabled. Macros such as `__LINE__', which are
4312 contextually dependent, are handled normally. This enables
4313 compilation of files previously preprocessed with `-E
4316 With both `-E' and `-fpreprocessed', the rules for
4317 `-fpreprocessed' take precedence. This enables full preprocessing
4318 of files previously preprocessed with `-E -fdirectives-only'.
4320 `-fdollars-in-identifiers'
4321 Accept `$' in identifiers. *Note Identifier characters::.
4323 `-fextended-identifiers'
4324 Accept universal character names in identifiers. This option is
4325 experimental; in a future version of GCC, it will be enabled by
4326 default for C99 and C++.
4329 Indicate to the preprocessor that the input file has already been
4330 preprocessed. This suppresses things like macro expansion,
4331 trigraph conversion, escaped newline splicing, and processing of
4332 most directives. The preprocessor still recognizes and removes
4333 comments, so that you can pass a file preprocessed with `-C' to
4334 the compiler without problems. In this mode the integrated
4335 preprocessor is little more than a tokenizer for the front ends.
4337 `-fpreprocessed' is implicit if the input file has one of the
4338 extensions `.i', `.ii' or `.mi'. These are the extensions that
4339 GCC uses for preprocessed files created by `-save-temps'.
4342 Set the distance between tab stops. This helps the preprocessor
4343 report correct column numbers in warnings or errors, even if tabs
4344 appear on the line. If the value is less than 1 or greater than
4345 100, the option is ignored. The default is 8.
4347 `-fexec-charset=CHARSET'
4348 Set the execution character set, used for string and character
4349 constants. The default is UTF-8. CHARSET can be any encoding
4350 supported by the system's `iconv' library routine.
4352 `-fwide-exec-charset=CHARSET'
4353 Set the wide execution character set, used for wide string and
4354 character constants. The default is UTF-32 or UTF-16, whichever
4355 corresponds to the width of `wchar_t'. As with `-fexec-charset',
4356 CHARSET can be any encoding supported by the system's `iconv'
4357 library routine; however, you will have problems with encodings
4358 that do not fit exactly in `wchar_t'.
4360 `-finput-charset=CHARSET'
4361 Set the input character set, used for translation from the
4362 character set of the input file to the source character set used
4363 by GCC. If the locale does not specify, or GCC cannot get this
4364 information from the locale, the default is UTF-8. This can be
4365 overridden by either the locale or this command line option.
4366 Currently the command line option takes precedence if there's a
4367 conflict. CHARSET can be any encoding supported by the system's
4368 `iconv' library routine.
4370 `-fworking-directory'
4371 Enable generation of linemarkers in the preprocessor output that
4372 will let the compiler know the current working directory at the
4373 time of preprocessing. When this option is enabled, the
4374 preprocessor will emit, after the initial linemarker, a second
4375 linemarker with the current working directory followed by two
4376 slashes. GCC will use this directory, when it's present in the
4377 preprocessed input, as the directory emitted as the current
4378 working directory in some debugging information formats. This
4379 option is implicitly enabled if debugging information is enabled,
4380 but this can be inhibited with the negated form
4381 `-fno-working-directory'. If the `-P' flag is present in the
4382 command line, this option has no effect, since no `#line'
4383 directives are emitted whatsoever.
4386 Do not print column numbers in diagnostics. This may be necessary
4387 if diagnostics are being scanned by a program that does not
4388 understand the column numbers, such as `dejagnu'.
4390 `-A PREDICATE=ANSWER'
4391 Make an assertion with the predicate PREDICATE and answer ANSWER.
4392 This form is preferred to the older form `-A PREDICATE(ANSWER)',
4393 which is still supported, because it does not use shell special
4394 characters. *Note Assertions::.
4396 `-A -PREDICATE=ANSWER'
4397 Cancel an assertion with the predicate PREDICATE and answer ANSWER.
4400 CHARS is a sequence of one or more of the following characters,
4401 and must not be preceded by a space. Other characters are
4402 interpreted by the compiler proper, or reserved for future
4403 versions of GCC, and so are silently ignored. If you specify
4404 characters whose behavior conflicts, the result is undefined.
4407 Instead of the normal output, generate a list of `#define'
4408 directives for all the macros defined during the execution of
4409 the preprocessor, including predefined macros. This gives
4410 you a way of finding out what is predefined in your version
4411 of the preprocessor. Assuming you have no file `foo.h', the
4414 touch foo.h; cpp -dM foo.h
4416 will show all the predefined macros.
4418 If you use `-dM' without the `-E' option, `-dM' is
4419 interpreted as a synonym for `-fdump-rtl-mach'. *Note
4420 Debugging Options: (gcc)Debugging Options.
4423 Like `M' except in two respects: it does _not_ include the
4424 predefined macros, and it outputs _both_ the `#define'
4425 directives and the result of preprocessing. Both kinds of
4426 output go to the standard output file.
4429 Like `D', but emit only the macro names, not their expansions.
4432 Output `#include' directives in addition to the result of
4436 Inhibit generation of linemarkers in the output from the
4437 preprocessor. This might be useful when running the preprocessor
4438 on something that is not C code, and will be sent to a program
4439 which might be confused by the linemarkers. *Note Preprocessor
4443 Do not discard comments. All comments are passed through to the
4444 output file, except for comments in processed directives, which
4445 are deleted along with the directive.
4447 You should be prepared for side effects when using `-C'; it causes
4448 the preprocessor to treat comments as tokens in their own right.
4449 For example, comments appearing at the start of what would be a
4450 directive line have the effect of turning that line into an
4451 ordinary source line, since the first token on the line is no
4455 Do not discard comments, including during macro expansion. This is
4456 like `-C', except that comments contained within macros are also
4457 passed through to the output file where the macro is expanded.
4459 In addition to the side-effects of the `-C' option, the `-CC'
4460 option causes all C++-style comments inside a macro to be
4461 converted to C-style comments. This is to prevent later use of
4462 that macro from inadvertently commenting out the remainder of the
4465 The `-CC' option is generally used to support lint comments.
4468 Try to imitate the behavior of old-fashioned C preprocessors, as
4469 opposed to ISO C preprocessors. *Note Traditional Mode::.
4472 Process trigraph sequences. *Note Initial processing::.
4475 Enable special code to work around file systems which only permit
4476 very short file names, such as MS-DOS.
4480 Print text describing all the command line options instead of
4481 preprocessing anything.
4484 Verbose mode. Print out GNU CPP's version number at the beginning
4485 of execution, and report the final form of the include path.
4488 Print the name of each header file used, in addition to other
4489 normal activities. Each name is indented to show how deep in the
4490 `#include' stack it is. Precompiled header files are also
4491 printed, even if they are found to be invalid; an invalid
4492 precompiled header file is printed with `...x' and a valid one
4497 Print out GNU CPP's version number. With one dash, proceed to
4498 preprocess as normal. With two dashes, exit immediately.
4501 File: cpp.info, Node: Environment Variables, Next: GNU Free Documentation License, Prev: Invocation, Up: Top
4503 13 Environment Variables
4504 ************************
4506 This section describes the environment variables that affect how CPP
4507 operates. You can use them to specify directories or prefixes to use
4508 when searching for include files, or to control dependency output.
4510 Note that you can also specify places to search using options such as
4511 `-I', and control dependency output with options like `-M' (*note
4512 Invocation::). These take precedence over environment variables, which
4513 in turn take precedence over the configuration of GCC.
4517 `CPLUS_INCLUDE_PATH'
4519 Each variable's value is a list of directories separated by a
4520 special character, much like `PATH', in which to look for header
4521 files. The special character, `PATH_SEPARATOR', is
4522 target-dependent and determined at GCC build time. For Microsoft
4523 Windows-based targets it is a semicolon, and for almost all other
4524 targets it is a colon.
4526 `CPATH' specifies a list of directories to be searched as if
4527 specified with `-I', but after any paths given with `-I' options
4528 on the command line. This environment variable is used regardless
4529 of which language is being preprocessed.
4531 The remaining environment variables apply only when preprocessing
4532 the particular language indicated. Each specifies a list of
4533 directories to be searched as if specified with `-isystem', but
4534 after any paths given with `-isystem' options on the command line.
4536 In all these variables, an empty element instructs the compiler to
4537 search its current working directory. Empty elements can appear
4538 at the beginning or end of a path. For instance, if the value of
4539 `CPATH' is `:/special/include', that has the same effect as
4540 `-I. -I/special/include'.
4542 See also *Note Search Path::.
4544 `DEPENDENCIES_OUTPUT'
4545 If this variable is set, its value specifies how to output
4546 dependencies for Make based on the non-system header files
4547 processed by the compiler. System header files are ignored in the
4550 The value of `DEPENDENCIES_OUTPUT' can be just a file name, in
4551 which case the Make rules are written to that file, guessing the
4552 target name from the source file name. Or the value can have the
4553 form `FILE TARGET', in which case the rules are written to file
4554 FILE using TARGET as the target name.
4556 In other words, this environment variable is equivalent to
4557 combining the options `-MM' and `-MF' (*note Invocation::), with
4558 an optional `-MT' switch too.
4560 `SUNPRO_DEPENDENCIES'
4561 This variable is the same as `DEPENDENCIES_OUTPUT' (see above),
4562 except that system header files are not ignored, so it implies
4563 `-M' rather than `-MM'. However, the dependence on the main input
4564 file is omitted. *Note Invocation::.
4567 File: cpp.info, Node: GNU Free Documentation License, Next: Index of Directives, Prev: Environment Variables, Up: Top
4569 GNU Free Documentation License
4570 ******************************
4572 Version 1.2, November 2002
4574 Copyright (C) 2000,2001,2002 Free Software Foundation, Inc.
4575 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
4577 Everyone is permitted to copy and distribute verbatim copies
4578 of this license document, but changing it is not allowed.
4582 The purpose of this License is to make a manual, textbook, or other
4583 functional and useful document "free" in the sense of freedom: to
4584 assure everyone the effective freedom to copy and redistribute it,
4585 with or without modifying it, either commercially or
4586 noncommercially. Secondarily, this License preserves for the
4587 author and publisher a way to get credit for their work, while not
4588 being considered responsible for modifications made by others.
4590 This License is a kind of "copyleft", which means that derivative
4591 works of the document must themselves be free in the same sense.
4592 It complements the GNU General Public License, which is a copyleft
4593 license designed for free software.
4595 We have designed this License in order to use it for manuals for
4596 free software, because free software needs free documentation: a
4597 free program should come with manuals providing the same freedoms
4598 that the software does. But this License is not limited to
4599 software manuals; it can be used for any textual work, regardless
4600 of subject matter or whether it is published as a printed book.
4601 We recommend this License principally for works whose purpose is
4602 instruction or reference.
4604 1. APPLICABILITY AND DEFINITIONS
4606 This License applies to any manual or other work, in any medium,
4607 that contains a notice placed by the copyright holder saying it
4608 can be distributed under the terms of this License. Such a notice
4609 grants a world-wide, royalty-free license, unlimited in duration,
4610 to use that work under the conditions stated herein. The
4611 "Document", below, refers to any such manual or work. Any member
4612 of the public is a licensee, and is addressed as "you". You
4613 accept the license if you copy, modify or distribute the work in a
4614 way requiring permission under copyright law.
4616 A "Modified Version" of the Document means any work containing the
4617 Document or a portion of it, either copied verbatim, or with
4618 modifications and/or translated into another language.
4620 A "Secondary Section" is a named appendix or a front-matter section
4621 of the Document that deals exclusively with the relationship of the
4622 publishers or authors of the Document to the Document's overall
4623 subject (or to related matters) and contains nothing that could
4624 fall directly within that overall subject. (Thus, if the Document
4625 is in part a textbook of mathematics, a Secondary Section may not
4626 explain any mathematics.) The relationship could be a matter of
4627 historical connection with the subject or with related matters, or
4628 of legal, commercial, philosophical, ethical or political position
4631 The "Invariant Sections" are certain Secondary Sections whose
4632 titles are designated, as being those of Invariant Sections, in
4633 the notice that says that the Document is released under this
4634 License. If a section does not fit the above definition of
4635 Secondary then it is not allowed to be designated as Invariant.
4636 The Document may contain zero Invariant Sections. If the Document
4637 does not identify any Invariant Sections then there are none.
4639 The "Cover Texts" are certain short passages of text that are
4640 listed, as Front-Cover Texts or Back-Cover Texts, in the notice
4641 that says that the Document is released under this License. A
4642 Front-Cover Text may be at most 5 words, and a Back-Cover Text may
4643 be at most 25 words.
4645 A "Transparent" copy of the Document means a machine-readable copy,
4646 represented in a format whose specification is available to the
4647 general public, that is suitable for revising the document
4648 straightforwardly with generic text editors or (for images
4649 composed of pixels) generic paint programs or (for drawings) some
4650 widely available drawing editor, and that is suitable for input to
4651 text formatters or for automatic translation to a variety of
4652 formats suitable for input to text formatters. A copy made in an
4653 otherwise Transparent file format whose markup, or absence of
4654 markup, has been arranged to thwart or discourage subsequent
4655 modification by readers is not Transparent. An image format is
4656 not Transparent if used for any substantial amount of text. A
4657 copy that is not "Transparent" is called "Opaque".
4659 Examples of suitable formats for Transparent copies include plain
4660 ASCII without markup, Texinfo input format, LaTeX input format,
4661 SGML or XML using a publicly available DTD, and
4662 standard-conforming simple HTML, PostScript or PDF designed for
4663 human modification. Examples of transparent image formats include
4664 PNG, XCF and JPG. Opaque formats include proprietary formats that
4665 can be read and edited only by proprietary word processors, SGML or
4666 XML for which the DTD and/or processing tools are not generally
4667 available, and the machine-generated HTML, PostScript or PDF
4668 produced by some word processors for output purposes only.
4670 The "Title Page" means, for a printed book, the title page itself,
4671 plus such following pages as are needed to hold, legibly, the
4672 material this License requires to appear in the title page. For
4673 works in formats which do not have any title page as such, "Title
4674 Page" means the text near the most prominent appearance of the
4675 work's title, preceding the beginning of the body of the text.
4677 A section "Entitled XYZ" means a named subunit of the Document
4678 whose title either is precisely XYZ or contains XYZ in parentheses
4679 following text that translates XYZ in another language. (Here XYZ
4680 stands for a specific section name mentioned below, such as
4681 "Acknowledgements", "Dedications", "Endorsements", or "History".)
4682 To "Preserve the Title" of such a section when you modify the
4683 Document means that it remains a section "Entitled XYZ" according
4686 The Document may include Warranty Disclaimers next to the notice
4687 which states that this License applies to the Document. These
4688 Warranty Disclaimers are considered to be included by reference in
4689 this License, but only as regards disclaiming warranties: any other
4690 implication that these Warranty Disclaimers may have is void and
4691 has no effect on the meaning of this License.
4695 You may copy and distribute the Document in any medium, either
4696 commercially or noncommercially, provided that this License, the
4697 copyright notices, and the license notice saying this License
4698 applies to the Document are reproduced in all copies, and that you
4699 add no other conditions whatsoever to those of this License. You
4700 may not use technical measures to obstruct or control the reading
4701 or further copying of the copies you make or distribute. However,
4702 you may accept compensation in exchange for copies. If you
4703 distribute a large enough number of copies you must also follow
4704 the conditions in section 3.
4706 You may also lend copies, under the same conditions stated above,
4707 and you may publicly display copies.
4709 3. COPYING IN QUANTITY
4711 If you publish printed copies (or copies in media that commonly
4712 have printed covers) of the Document, numbering more than 100, and
4713 the Document's license notice requires Cover Texts, you must
4714 enclose the copies in covers that carry, clearly and legibly, all
4715 these Cover Texts: Front-Cover Texts on the front cover, and
4716 Back-Cover Texts on the back cover. Both covers must also clearly
4717 and legibly identify you as the publisher of these copies. The
4718 front cover must present the full title with all words of the
4719 title equally prominent and visible. You may add other material
4720 on the covers in addition. Copying with changes limited to the
4721 covers, as long as they preserve the title of the Document and
4722 satisfy these conditions, can be treated as verbatim copying in
4725 If the required texts for either cover are too voluminous to fit
4726 legibly, you should put the first ones listed (as many as fit
4727 reasonably) on the actual cover, and continue the rest onto
4730 If you publish or distribute Opaque copies of the Document
4731 numbering more than 100, you must either include a
4732 machine-readable Transparent copy along with each Opaque copy, or
4733 state in or with each Opaque copy a computer-network location from
4734 which the general network-using public has access to download
4735 using public-standard network protocols a complete Transparent
4736 copy of the Document, free of added material. If you use the
4737 latter option, you must take reasonably prudent steps, when you
4738 begin distribution of Opaque copies in quantity, to ensure that
4739 this Transparent copy will remain thus accessible at the stated
4740 location until at least one year after the last time you
4741 distribute an Opaque copy (directly or through your agents or
4742 retailers) of that edition to the public.
4744 It is requested, but not required, that you contact the authors of
4745 the Document well before redistributing any large number of
4746 copies, to give them a chance to provide you with an updated
4747 version of the Document.
4751 You may copy and distribute a Modified Version of the Document
4752 under the conditions of sections 2 and 3 above, provided that you
4753 release the Modified Version under precisely this License, with
4754 the Modified Version filling the role of the Document, thus
4755 licensing distribution and modification of the Modified Version to
4756 whoever possesses a copy of it. In addition, you must do these
4757 things in the Modified Version:
4759 A. Use in the Title Page (and on the covers, if any) a title
4760 distinct from that of the Document, and from those of
4761 previous versions (which should, if there were any, be listed
4762 in the History section of the Document). You may use the
4763 same title as a previous version if the original publisher of
4764 that version gives permission.
4766 B. List on the Title Page, as authors, one or more persons or
4767 entities responsible for authorship of the modifications in
4768 the Modified Version, together with at least five of the
4769 principal authors of the Document (all of its principal
4770 authors, if it has fewer than five), unless they release you
4771 from this requirement.
4773 C. State on the Title page the name of the publisher of the
4774 Modified Version, as the publisher.
4776 D. Preserve all the copyright notices of the Document.
4778 E. Add an appropriate copyright notice for your modifications
4779 adjacent to the other copyright notices.
4781 F. Include, immediately after the copyright notices, a license
4782 notice giving the public permission to use the Modified
4783 Version under the terms of this License, in the form shown in
4786 G. Preserve in that license notice the full lists of Invariant
4787 Sections and required Cover Texts given in the Document's
4790 H. Include an unaltered copy of this License.
4792 I. Preserve the section Entitled "History", Preserve its Title,
4793 and add to it an item stating at least the title, year, new
4794 authors, and publisher of the Modified Version as given on
4795 the Title Page. If there is no section Entitled "History" in
4796 the Document, create one stating the title, year, authors,
4797 and publisher of the Document as given on its Title Page,
4798 then add an item describing the Modified Version as stated in
4799 the previous sentence.
4801 J. Preserve the network location, if any, given in the Document
4802 for public access to a Transparent copy of the Document, and
4803 likewise the network locations given in the Document for
4804 previous versions it was based on. These may be placed in
4805 the "History" section. You may omit a network location for a
4806 work that was published at least four years before the
4807 Document itself, or if the original publisher of the version
4808 it refers to gives permission.
4810 K. For any section Entitled "Acknowledgements" or "Dedications",
4811 Preserve the Title of the section, and preserve in the
4812 section all the substance and tone of each of the contributor
4813 acknowledgements and/or dedications given therein.
4815 L. Preserve all the Invariant Sections of the Document,
4816 unaltered in their text and in their titles. Section numbers
4817 or the equivalent are not considered part of the section
4820 M. Delete any section Entitled "Endorsements". Such a section
4821 may not be included in the Modified Version.
4823 N. Do not retitle any existing section to be Entitled
4824 "Endorsements" or to conflict in title with any Invariant
4827 O. Preserve any Warranty Disclaimers.
4829 If the Modified Version includes new front-matter sections or
4830 appendices that qualify as Secondary Sections and contain no
4831 material copied from the Document, you may at your option
4832 designate some or all of these sections as invariant. To do this,
4833 add their titles to the list of Invariant Sections in the Modified
4834 Version's license notice. These titles must be distinct from any
4835 other section titles.
4837 You may add a section Entitled "Endorsements", provided it contains
4838 nothing but endorsements of your Modified Version by various
4839 parties--for example, statements of peer review or that the text
4840 has been approved by an organization as the authoritative
4841 definition of a standard.
4843 You may add a passage of up to five words as a Front-Cover Text,
4844 and a passage of up to 25 words as a Back-Cover Text, to the end
4845 of the list of Cover Texts in the Modified Version. Only one
4846 passage of Front-Cover Text and one of Back-Cover Text may be
4847 added by (or through arrangements made by) any one entity. If the
4848 Document already includes a cover text for the same cover,
4849 previously added by you or by arrangement made by the same entity
4850 you are acting on behalf of, you may not add another; but you may
4851 replace the old one, on explicit permission from the previous
4852 publisher that added the old one.
4854 The author(s) and publisher(s) of the Document do not by this
4855 License give permission to use their names for publicity for or to
4856 assert or imply endorsement of any Modified Version.
4858 5. COMBINING DOCUMENTS
4860 You may combine the Document with other documents released under
4861 this License, under the terms defined in section 4 above for
4862 modified versions, provided that you include in the combination
4863 all of the Invariant Sections of all of the original documents,
4864 unmodified, and list them all as Invariant Sections of your
4865 combined work in its license notice, and that you preserve all
4866 their Warranty Disclaimers.
4868 The combined work need only contain one copy of this License, and
4869 multiple identical Invariant Sections may be replaced with a single
4870 copy. If there are multiple Invariant Sections with the same name
4871 but different contents, make the title of each such section unique
4872 by adding at the end of it, in parentheses, the name of the
4873 original author or publisher of that section if known, or else a
4874 unique number. Make the same adjustment to the section titles in
4875 the list of Invariant Sections in the license notice of the
4878 In the combination, you must combine any sections Entitled
4879 "History" in the various original documents, forming one section
4880 Entitled "History"; likewise combine any sections Entitled
4881 "Acknowledgements", and any sections Entitled "Dedications". You
4882 must delete all sections Entitled "Endorsements."
4884 6. COLLECTIONS OF DOCUMENTS
4886 You may make a collection consisting of the Document and other
4887 documents released under this License, and replace the individual
4888 copies of this License in the various documents with a single copy
4889 that is included in the collection, provided that you follow the
4890 rules of this License for verbatim copying of each of the
4891 documents in all other respects.
4893 You may extract a single document from such a collection, and
4894 distribute it individually under this License, provided you insert
4895 a copy of this License into the extracted document, and follow
4896 this License in all other respects regarding verbatim copying of
4899 7. AGGREGATION WITH INDEPENDENT WORKS
4901 A compilation of the Document or its derivatives with other
4902 separate and independent documents or works, in or on a volume of
4903 a storage or distribution medium, is called an "aggregate" if the
4904 copyright resulting from the compilation is not used to limit the
4905 legal rights of the compilation's users beyond what the individual
4906 works permit. When the Document is included in an aggregate, this
4907 License does not apply to the other works in the aggregate which
4908 are not themselves derivative works of the Document.
4910 If the Cover Text requirement of section 3 is applicable to these
4911 copies of the Document, then if the Document is less than one half
4912 of the entire aggregate, the Document's Cover Texts may be placed
4913 on covers that bracket the Document within the aggregate, or the
4914 electronic equivalent of covers if the Document is in electronic
4915 form. Otherwise they must appear on printed covers that bracket
4916 the whole aggregate.
4920 Translation is considered a kind of modification, so you may
4921 distribute translations of the Document under the terms of section
4922 4. Replacing Invariant Sections with translations requires special
4923 permission from their copyright holders, but you may include
4924 translations of some or all Invariant Sections in addition to the
4925 original versions of these Invariant Sections. You may include a
4926 translation of this License, and all the license notices in the
4927 Document, and any Warranty Disclaimers, provided that you also
4928 include the original English version of this License and the
4929 original versions of those notices and disclaimers. In case of a
4930 disagreement between the translation and the original version of
4931 this License or a notice or disclaimer, the original version will
4934 If a section in the Document is Entitled "Acknowledgements",
4935 "Dedications", or "History", the requirement (section 4) to
4936 Preserve its Title (section 1) will typically require changing the
4941 You may not copy, modify, sublicense, or distribute the Document
4942 except as expressly provided for under this License. Any other
4943 attempt to copy, modify, sublicense or distribute the Document is
4944 void, and will automatically terminate your rights under this
4945 License. However, parties who have received copies, or rights,
4946 from you under this License will not have their licenses
4947 terminated so long as such parties remain in full compliance.
4949 10. FUTURE REVISIONS OF THIS LICENSE
4951 The Free Software Foundation may publish new, revised versions of
4952 the GNU Free Documentation License from time to time. Such new
4953 versions will be similar in spirit to the present version, but may
4954 differ in detail to address new problems or concerns. See
4955 `http://www.gnu.org/copyleft/'.
4957 Each version of the License is given a distinguishing version
4958 number. If the Document specifies that a particular numbered
4959 version of this License "or any later version" applies to it, you
4960 have the option of following the terms and conditions either of
4961 that specified version or of any later version that has been
4962 published (not as a draft) by the Free Software Foundation. If
4963 the Document does not specify a version number of this License,
4964 you may choose any version ever published (not as a draft) by the
4965 Free Software Foundation.
4967 ADDENDUM: How to use this License for your documents
4968 ====================================================
4970 To use this License in a document you have written, include a copy of
4971 the License in the document and put the following copyright and license
4972 notices just after the title page:
4974 Copyright (C) YEAR YOUR NAME.
4975 Permission is granted to copy, distribute and/or modify this document
4976 under the terms of the GNU Free Documentation License, Version 1.2
4977 or any later version published by the Free Software Foundation;
4978 with no Invariant Sections, no Front-Cover Texts, and no Back-Cover
4979 Texts. A copy of the license is included in the section entitled ``GNU
4980 Free Documentation License''.
4982 If you have Invariant Sections, Front-Cover Texts and Back-Cover
4983 Texts, replace the "with...Texts." line with this:
4985 with the Invariant Sections being LIST THEIR TITLES, with
4986 the Front-Cover Texts being LIST, and with the Back-Cover Texts
4989 If you have Invariant Sections without Cover Texts, or some other
4990 combination of the three, merge those two alternatives to suit the
4993 If your document contains nontrivial examples of program code, we
4994 recommend releasing these examples in parallel under your choice of
4995 free software license, such as the GNU General Public License, to
4996 permit their use in free software.
4999 File: cpp.info, Node: Index of Directives, Next: Option Index, Prev: GNU Free Documentation License, Up: Top
5007 * #assert: Assertions. (line 41)
5008 * #define: Object-like Macros. (line 11)
5009 * #elif: Elif. (line 6)
5010 * #else: Else. (line 6)
5011 * #endif: Ifdef. (line 6)
5012 * #error: Diagnostics. (line 6)
5013 * #ident: Other Directives. (line 6)
5014 * #if: Conditional Syntax. (line 6)
5015 * #ifdef: Ifdef. (line 6)
5016 * #ifndef: Ifdef. (line 40)
5017 * #import: Obsolete once-only headers.
5019 * #include: Include Syntax. (line 6)
5020 * #include_next: Wrapper Headers. (line 6)
5021 * #line: Line Control. (line 20)
5022 * #pragma GCC dependency: Pragmas. (line 53)
5023 * #pragma GCC poison: Pragmas. (line 65)
5024 * #pragma GCC system_header <1>: Pragmas. (line 92)
5025 * #pragma GCC system_header: System Headers. (line 31)
5026 * #sccs: Other Directives. (line 6)
5027 * #unassert: Assertions. (line 52)
5028 * #undef: Undefining and Redefining Macros.
5030 * #warning: Diagnostics. (line 27)
5033 File: cpp.info, Node: Option Index, Next: Concept Index, Prev: Index of Directives, Up: Top
5038 CPP's command line options and environment variables are indexed here
5039 without any initial `-' or `--'.
5044 * A: Invocation. (line 525)
5045 * ansi: Invocation. (line 311)
5046 * C: Invocation. (line 577)
5047 * C_INCLUDE_PATH: Environment Variables.
5049 * CPATH: Environment Variables.
5051 * CPLUS_INCLUDE_PATH: Environment Variables.
5053 * D: Invocation. (line 39)
5054 * dD: Invocation. (line 557)
5055 * DEPENDENCIES_OUTPUT: Environment Variables.
5057 * dI: Invocation. (line 566)
5058 * dM: Invocation. (line 541)
5059 * dN: Invocation. (line 563)
5060 * fdirectives-only: Invocation. (line 433)
5061 * fdollars-in-identifiers: Invocation. (line 455)
5062 * fexec-charset: Invocation. (line 482)
5063 * fextended-identifiers: Invocation. (line 458)
5064 * finput-charset: Invocation. (line 495)
5065 * fno-show-column: Invocation. (line 520)
5066 * fno-working-directory: Invocation. (line 505)
5067 * fpreprocessed: Invocation. (line 463)
5068 * ftabstop: Invocation. (line 476)
5069 * fwide-exec-charset: Invocation. (line 487)
5070 * fworking-directory: Invocation. (line 505)
5071 * H: Invocation. (line 622)
5072 * help: Invocation. (line 614)
5073 * I: Invocation. (line 71)
5074 * I-: Invocation. (line 348)
5075 * idirafter: Invocation. (line 390)
5076 * imacros: Invocation. (line 381)
5077 * imultilib: Invocation. (line 413)
5078 * include: Invocation. (line 370)
5079 * iprefix: Invocation. (line 397)
5080 * iquote: Invocation. (line 425)
5081 * isysroot: Invocation. (line 409)
5082 * isystem: Invocation. (line 417)
5083 * iwithprefix: Invocation. (line 403)
5084 * iwithprefixbefore: Invocation. (line 403)
5085 * M: Invocation. (line 183)
5086 * MD: Invocation. (line 272)
5087 * MF: Invocation. (line 218)
5088 * MG: Invocation. (line 227)
5089 * MM: Invocation. (line 208)
5090 * MMD: Invocation. (line 288)
5091 * MP: Invocation. (line 237)
5092 * MQ: Invocation. (line 263)
5093 * MT: Invocation. (line 249)
5094 * nostdinc: Invocation. (line 360)
5095 * nostdinc++: Invocation. (line 365)
5096 * o: Invocation. (line 82)
5097 * OBJC_INCLUDE_PATH: Environment Variables.
5099 * P: Invocation. (line 570)
5100 * pedantic: Invocation. (line 173)
5101 * pedantic-errors: Invocation. (line 178)
5102 * remap: Invocation. (line 609)
5103 * std=: Invocation. (line 311)
5104 * SUNPRO_DEPENDENCIES: Environment Variables.
5106 * target-help: Invocation. (line 614)
5107 * traditional-cpp: Invocation. (line 602)
5108 * trigraphs: Invocation. (line 606)
5109 * U: Invocation. (line 62)
5110 * undef: Invocation. (line 66)
5111 * v: Invocation. (line 618)
5112 * version: Invocation. (line 631)
5113 * w: Invocation. (line 169)
5114 * Wall: Invocation. (line 88)
5115 * Wcomment: Invocation. (line 96)
5116 * Wcomments: Invocation. (line 96)
5117 * Wendif-labels: Invocation. (line 146)
5118 * Werror: Invocation. (line 159)
5119 * Wimport: Invocation. (line 119)
5120 * Wsystem-headers: Invocation. (line 163)
5121 * Wtraditional: Invocation. (line 113)
5122 * Wtrigraphs: Invocation. (line 101)
5123 * Wundef: Invocation. (line 122)
5124 * Wunused-macros: Invocation. (line 127)
5125 * x: Invocation. (line 295)
5128 File: cpp.info, Node: Concept Index, Prev: Option Index, Up: Top
5136 * # operator: Stringification. (line 6)
5137 * ## operator: Concatenation. (line 6)
5138 * _Pragma: Pragmas. (line 25)
5139 * alternative tokens: Tokenization. (line 106)
5140 * arguments: Macro Arguments. (line 6)
5141 * arguments in macro definitions: Macro Arguments. (line 6)
5142 * assertions: Assertions. (line 6)
5143 * assertions, canceling: Assertions. (line 52)
5144 * backslash-newline: Initial processing. (line 61)
5145 * block comments: Initial processing. (line 77)
5146 * C++ named operators: C++ Named Operators. (line 6)
5147 * character constants: Tokenization. (line 85)
5148 * character set, execution: Invocation. (line 482)
5149 * character set, input: Invocation. (line 495)
5150 * character set, wide execution: Invocation. (line 487)
5151 * command line: Invocation. (line 6)
5152 * commenting out code: Deleted Code. (line 6)
5153 * comments: Initial processing. (line 77)
5154 * common predefined macros: Common Predefined Macros.
5156 * computed includes: Computed Includes. (line 6)
5157 * concatenation: Concatenation. (line 6)
5158 * conditional group: Ifdef. (line 14)
5159 * conditionals: Conditionals. (line 6)
5160 * continued lines: Initial processing. (line 61)
5161 * controlling macro: Once-Only Headers. (line 35)
5162 * defined: Defined. (line 6)
5163 * dependencies for make as output: Environment Variables.
5165 * dependencies, make: Invocation. (line 183)
5166 * diagnostic: Diagnostics. (line 6)
5167 * differences from previous versions: Differences from previous versions.
5169 * digraphs: Tokenization. (line 106)
5170 * directive line: The preprocessing language.
5172 * directive name: The preprocessing language.
5174 * directives: The preprocessing language.
5176 * empty macro arguments: Macro Arguments. (line 66)
5177 * environment variables: Environment Variables.
5179 * expansion of arguments: Argument Prescan. (line 6)
5180 * FDL, GNU Free Documentation License: GNU Free Documentation License.
5182 * function-like macros: Function-like Macros.
5184 * grouping options: Invocation. (line 34)
5185 * guard macro: Once-Only Headers. (line 35)
5186 * header file: Header Files. (line 6)
5187 * header file names: Tokenization. (line 85)
5188 * identifiers: Tokenization. (line 34)
5189 * implementation limits: Implementation limits.
5191 * implementation-defined behavior: Implementation-defined behavior.
5193 * including just once: Once-Only Headers. (line 6)
5194 * invocation: Invocation. (line 6)
5195 * iso646.h: C++ Named Operators. (line 6)
5196 * line comments: Initial processing. (line 77)
5197 * line control: Line Control. (line 6)
5198 * line endings: Initial processing. (line 14)
5199 * linemarkers: Preprocessor Output. (line 28)
5200 * macro argument expansion: Argument Prescan. (line 6)
5201 * macro arguments and directives: Directives Within Macro Arguments.
5203 * macros in include: Computed Includes. (line 6)
5204 * macros with arguments: Macro Arguments. (line 6)
5205 * macros with variable arguments: Variadic Macros. (line 6)
5206 * make: Invocation. (line 183)
5207 * manifest constants: Object-like Macros. (line 6)
5208 * named operators: C++ Named Operators. (line 6)
5209 * newlines in macro arguments: Newlines in Arguments.
5211 * null directive: Other Directives. (line 15)
5212 * numbers: Tokenization. (line 61)
5213 * object-like macro: Object-like Macros. (line 6)
5214 * options: Invocation. (line 38)
5215 * options, grouping: Invocation. (line 34)
5216 * other tokens: Tokenization. (line 120)
5217 * output format: Preprocessor Output. (line 12)
5218 * overriding a header file: Wrapper Headers. (line 6)
5219 * parentheses in macro bodies: Operator Precedence Problems.
5221 * pitfalls of macros: Macro Pitfalls. (line 6)
5222 * predefined macros: Predefined Macros. (line 6)
5223 * predefined macros, system-specific: System-specific Predefined Macros.
5225 * predicates: Assertions. (line 19)
5226 * preprocessing directives: The preprocessing language.
5228 * preprocessing numbers: Tokenization. (line 61)
5229 * preprocessing tokens: Tokenization. (line 6)
5230 * prescan of macro arguments: Argument Prescan. (line 6)
5231 * problems with macros: Macro Pitfalls. (line 6)
5232 * punctuators: Tokenization. (line 106)
5233 * redefining macros: Undefining and Redefining Macros.
5235 * repeated inclusion: Once-Only Headers. (line 6)
5236 * reporting errors: Diagnostics. (line 6)
5237 * reporting warnings: Diagnostics. (line 6)
5238 * reserved namespace: System-specific Predefined Macros.
5240 * self-reference: Self-Referential Macros.
5242 * semicolons (after macro calls): Swallowing the Semicolon.
5244 * side effects (in macro arguments): Duplication of Side Effects.
5246 * standard predefined macros.: Standard Predefined Macros.
5248 * string constants: Tokenization. (line 85)
5249 * string literals: Tokenization. (line 85)
5250 * stringification: Stringification. (line 6)
5251 * symbolic constants: Object-like Macros. (line 6)
5252 * system header files <1>: System Headers. (line 6)
5253 * system header files: Header Files. (line 13)
5254 * system-specific predefined macros: System-specific Predefined Macros.
5256 * testing predicates: Assertions. (line 30)
5257 * token concatenation: Concatenation. (line 6)
5258 * token pasting: Concatenation. (line 6)
5259 * tokens: Tokenization. (line 6)
5260 * trigraphs: Initial processing. (line 32)
5261 * undefining macros: Undefining and Redefining Macros.
5263 * unsafe macros: Duplication of Side Effects.
5265 * variable number of arguments: Variadic Macros. (line 6)
5266 * variadic macros: Variadic Macros. (line 6)
5267 * wrapper #ifndef: Once-Only Headers. (line 6)
5268 * wrapper headers: Wrapper Headers. (line 6)
5274 Node: Overview
\7f3798
5275 Node: Character sets
\7f6619
5276 Ref: Character sets-Footnote-1
\7f8802
5277 Node: Initial processing
\7f8983
5278 Ref: trigraphs
\7f10542
5279 Node: Tokenization
\7f14744
5280 Ref: Tokenization-Footnote-1
\7f21880
5281 Node: The preprocessing language
\7f21991
5282 Node: Header Files
\7f24869
5283 Node: Include Syntax
\7f26749
5284 Node: Include Operation
\7f28386
5285 Node: Search Path
\7f30234
5286 Node: Once-Only Headers
\7f33424
5287 Node: Computed Includes
\7f35069
5288 Node: Wrapper Headers
\7f38213
5289 Node: System Headers
\7f40639
5290 Node: Macros
\7f42689
5291 Node: Object-like Macros
\7f43830
5292 Node: Function-like Macros
\7f47420
5293 Node: Macro Arguments
\7f49036
5294 Node: Stringification
\7f53181
5295 Node: Concatenation
\7f56387
5296 Node: Variadic Macros
\7f59495
5297 Node: Predefined Macros
\7f64282
5298 Node: Standard Predefined Macros
\7f64870
5299 Node: Common Predefined Macros
\7f70806
5300 Node: System-specific Predefined Macros
\7f83493
5301 Node: C++ Named Operators
\7f85514
5302 Node: Undefining and Redefining Macros
\7f86478
5303 Node: Directives Within Macro Arguments
\7f88582
5304 Node: Macro Pitfalls
\7f90130
5305 Node: Misnesting
\7f90663
5306 Node: Operator Precedence Problems
\7f91775
5307 Node: Swallowing the Semicolon
\7f93641
5308 Node: Duplication of Side Effects
\7f95664
5309 Node: Self-Referential Macros
\7f97847
5310 Node: Argument Prescan
\7f100256
5311 Node: Newlines in Arguments
\7f104010
5312 Node: Conditionals
\7f104961
5313 Node: Conditional Uses
\7f106791
5314 Node: Conditional Syntax
\7f108149
5315 Node: Ifdef
\7f108469
5317 Node: Defined
\7f113934
5320 Node: Deleted Code
\7f117076
5321 Node: Diagnostics
\7f118323
5322 Node: Line Control
\7f119940
5323 Node: Pragmas
\7f123744
5324 Node: Other Directives
\7f128014
5325 Node: Preprocessor Output
\7f129064
5326 Node: Traditional Mode
\7f132265
5327 Node: Traditional lexical analysis
\7f133323
5328 Node: Traditional macros
\7f135826
5329 Node: Traditional miscellany
\7f139628
5330 Node: Traditional warnings
\7f140625
5331 Node: Implementation Details
\7f142822
5332 Node: Implementation-defined behavior
\7f143443
5333 Ref: Identifier characters
\7f144176
5334 Node: Implementation limits
\7f147261
5335 Node: Obsolete Features
\7f149935
5336 Node: Assertions
\7f150390
5337 Node: Obsolete once-only headers
\7f152931
5338 Node: Differences from previous versions
\7f154666
5339 Node: Invocation
\7f158874
5340 Ref: Wtrigraphs
\7f163326
5341 Ref: dashMF
\7f168157
5342 Ref: fdollars-in-identifiers
\7f177540
5343 Node: Environment Variables
\7f185373
5344 Node: GNU Free Documentation License
\7f188339
5345 Node: Index of Directives
\7f210772
5346 Node: Option Index
\7f212701
5347 Node: Concept Index
\7f218885