xref: /netbsd-src/external/gpl3/gcc.old/dist/gcc/doc/cpp.texi (revision bdc22b2e01993381dcefeff2bc9b56ca75a4235c)
1\input texinfo
2@setfilename cpp.info
3@settitle The C Preprocessor
4@setchapternewpage off
5@c @smallbook
6@c @cropmarks
7@c @finalout
8
9@include gcc-common.texi
10
11@copying
12@c man begin COPYRIGHT
13Copyright @copyright{} 1987-2015 Free Software Foundation, Inc.
14
15Permission is granted to copy, distribute and/or modify this document
16under the terms of the GNU Free Documentation License, Version 1.3 or
17any later version published by the Free Software Foundation.  A copy of
18the license is included in the
19@c man end
20section entitled ``GNU Free Documentation License''.
21@ignore
22@c man begin COPYRIGHT
23man page gfdl(7).
24@c man end
25@end ignore
26
27@c man begin COPYRIGHT
28This manual contains no Invariant Sections.  The Front-Cover Texts are
29(a) (see below), and the Back-Cover Texts are (b) (see below).
30
31(a) The FSF's Front-Cover Text is:
32
33     A GNU Manual
34
35(b) The FSF's Back-Cover Text is:
36
37     You have freedom to copy and modify this GNU Manual, like GNU
38     software.  Copies published by the Free Software Foundation raise
39     funds for GNU development.
40@c man end
41@end copying
42
43@c Create a separate index for command line options.
44@defcodeindex op
45@syncodeindex vr op
46
47@c Used in cppopts.texi and cppenv.texi.
48@set cppmanual
49
50@ifinfo
51@dircategory Software development
52@direntry
53* Cpp: (cpp).                  The GNU C preprocessor.
54@end direntry
55@end ifinfo
56
57@titlepage
58@title The C Preprocessor
59@versionsubtitle
60@author Richard M. Stallman, Zachary Weinberg
61@page
62@c There is a fill at the bottom of the page, so we need a filll to
63@c override it.
64@vskip 0pt plus 1filll
65@insertcopying
66@end titlepage
67@contents
68@page
69
70@ifnottex
71@node Top
72@top
73The C preprocessor implements the macro language used to transform C,
74C++, and Objective-C programs before they are compiled.  It can also be
75useful on its own.
76
77@menu
78* Overview::
79* Header Files::
80* Macros::
81* Conditionals::
82* Diagnostics::
83* Line Control::
84* Pragmas::
85* Other Directives::
86* Preprocessor Output::
87* Traditional Mode::
88* Implementation Details::
89* Invocation::
90* Environment Variables::
91* GNU Free Documentation License::
92* Index of Directives::
93* Option Index::
94* Concept Index::
95
96@detailmenu
97 --- The Detailed Node Listing ---
98
99Overview
100
101* Character sets::
102* Initial processing::
103* Tokenization::
104* The preprocessing language::
105
106Header Files
107
108* Include Syntax::
109* Include Operation::
110* Search Path::
111* Once-Only Headers::
112* Alternatives to Wrapper #ifndef::
113* Computed Includes::
114* Wrapper Headers::
115* System Headers::
116
117Macros
118
119* Object-like Macros::
120* Function-like Macros::
121* Macro Arguments::
122* Stringification::
123* Concatenation::
124* Variadic Macros::
125* Predefined Macros::
126* Undefining and Redefining Macros::
127* Directives Within Macro Arguments::
128* Macro Pitfalls::
129
130Predefined Macros
131
132* Standard Predefined Macros::
133* Common Predefined Macros::
134* System-specific Predefined Macros::
135* C++ Named Operators::
136
137Macro Pitfalls
138
139* Misnesting::
140* Operator Precedence Problems::
141* Swallowing the Semicolon::
142* Duplication of Side Effects::
143* Self-Referential Macros::
144* Argument Prescan::
145* Newlines in Arguments::
146
147Conditionals
148
149* Conditional Uses::
150* Conditional Syntax::
151* Deleted Code::
152
153Conditional Syntax
154
155* Ifdef::
156* If::
157* Defined::
158* Else::
159* Elif::
160
161Implementation Details
162
163* Implementation-defined behavior::
164* Implementation limits::
165* Obsolete Features::
166* Differences from previous versions::
167
168Obsolete Features
169
170* Obsolete Features::
171
172@end detailmenu
173@end menu
174
175@insertcopying
176@end ifnottex
177
178@node Overview
179@chapter Overview
180@c man begin DESCRIPTION
181The C preprocessor, often known as @dfn{cpp}, is a @dfn{macro processor}
182that is used automatically by the C compiler to transform your program
183before compilation.  It is called a macro processor because it allows
184you to define @dfn{macros}, which are brief abbreviations for longer
185constructs.
186
187The C preprocessor is intended to be used only with C, C++, and
188Objective-C source code.  In the past, it has been abused as a general
189text processor.  It will choke on input which does not obey C's lexical
190rules.  For example, apostrophes will be interpreted as the beginning of
191character constants, and cause errors.  Also, you cannot rely on it
192preserving characteristics of the input which are not significant to
193C-family languages.  If a Makefile is preprocessed, all the hard tabs
194will be removed, and the Makefile will not work.
195
196Having said that, you can often get away with using cpp on things which
197are not C@.  Other Algol-ish programming languages are often safe
198(Pascal, Ada, etc.) So is assembly, with caution.  @option{-traditional-cpp}
199mode preserves more white space, and is otherwise more permissive.  Many
200of the problems can be avoided by writing C or C++ style comments
201instead of native language comments, and keeping macros simple.
202
203Wherever possible, you should use a preprocessor geared to the language
204you are writing in.  Modern versions of the GNU assembler have macro
205facilities.  Most high level programming languages have their own
206conditional compilation and inclusion mechanism.  If all else fails,
207try a true general text processor, such as GNU M4.
208
209C preprocessors vary in some details.  This manual discusses the GNU C
210preprocessor, which provides a small superset of the features of ISO
211Standard C@.  In its default mode, the GNU C preprocessor does not do a
212few things required by the standard.  These are features which are
213rarely, if ever, used, and may cause surprising changes to the meaning
214of a program which does not expect them.  To get strict ISO Standard C,
215you should use the @option{-std=c90}, @option{-std=c99} or
216@option{-std=c11} options, depending
217on which version of the standard you want.  To get all the mandatory
218diagnostics, you must also use @option{-pedantic}.  @xref{Invocation}.
219
220This manual describes the behavior of the ISO preprocessor.  To
221minimize gratuitous differences, where the ISO preprocessor's
222behavior does not conflict with traditional semantics, the
223traditional preprocessor should behave the same way.  The various
224differences that do exist are detailed in the section @ref{Traditional
225Mode}.
226
227For clarity, unless noted otherwise, references to @samp{CPP} in this
228manual refer to GNU CPP@.
229@c man end
230
231@menu
232* Character sets::
233* Initial processing::
234* Tokenization::
235* The preprocessing language::
236@end menu
237
238@node Character sets
239@section Character sets
240
241Source code character set processing in C and related languages is
242rather complicated.  The C standard discusses two character sets, but
243there are really at least four.
244
245The files input to CPP might be in any character set at all.  CPP's
246very first action, before it even looks for line boundaries, is to
247convert the file into the character set it uses for internal
248processing.  That set is what the C standard calls the @dfn{source}
249character set.  It must be isomorphic with ISO 10646, also known as
250Unicode.  CPP uses the UTF-8 encoding of Unicode.
251
252The character sets of the input files are specified using the
253@option{-finput-charset=} option.
254
255All preprocessing work (the subject of the rest of this manual) is
256carried out in the source character set.  If you request textual
257output from the preprocessor with the @option{-E} option, it will be
258in UTF-8.
259
260After preprocessing is complete, string and character constants are
261converted again, into the @dfn{execution} character set.  This
262character set is under control of the user; the default is UTF-8,
263matching the source character set.  Wide string and character
264constants have their own character set, which is not called out
265specifically in the standard.  Again, it is under control of the user.
266The default is UTF-16 or UTF-32, whichever fits in the target's
267@code{wchar_t} type, in the target machine's byte
268order.@footnote{UTF-16 does not meet the requirements of the C
269standard for a wide character set, but the choice of 16-bit
270@code{wchar_t} is enshrined in some system ABIs so we cannot fix
271this.}  Octal and hexadecimal escape sequences do not undergo
272conversion; @t{'\x12'} has the value 0x12 regardless of the currently
273selected execution character set.  All other escapes are replaced by
274the character in the source character set that they represent, then
275converted to the execution character set, just like unescaped
276characters.
277
278In identifiers, characters outside the ASCII range can only be
279specified with the @samp{\u} and @samp{\U} escapes, not used
280directly.  If strict ISO C90 conformance is specified with an option
281such as @option{-std=c90}, or @option{-fno-extended-identifiers} is
282used, then those escapes are not permitted in identifiers.
283
284@node Initial processing
285@section Initial processing
286
287The preprocessor performs a series of textual transformations on its
288input.  These happen before all other processing.  Conceptually, they
289happen in a rigid order, and the entire file is run through each
290transformation before the next one begins.  CPP actually does them
291all at once, for performance reasons.  These transformations correspond
292roughly to the first three ``phases of translation'' described in the C
293standard.
294
295@enumerate
296@item
297@cindex line endings
298The input file is read into memory and broken into lines.
299
300Different systems use different conventions to indicate the end of a
301line.  GCC accepts the ASCII control sequences @kbd{LF}, @kbd{@w{CR
302LF}} and @kbd{CR} as end-of-line markers.  These are the canonical
303sequences used by Unix, DOS and VMS, and the classic Mac OS (before
304OSX) respectively.  You may therefore safely copy source code written
305on any of those systems to a different one and use it without
306conversion.  (GCC may lose track of the current line number if a file
307doesn't consistently use one convention, as sometimes happens when it
308is edited on computers with different conventions that share a network
309file system.)
310
311If the last line of any input file lacks an end-of-line marker, the end
312of the file is considered to implicitly supply one.  The C standard says
313that this condition provokes undefined behavior, so GCC will emit a
314warning message.
315
316@item
317@cindex trigraphs
318@anchor{trigraphs}If trigraphs are enabled, they are replaced by their
319corresponding single characters.  By default GCC ignores trigraphs,
320but if you request a strictly conforming mode with the @option{-std}
321option, or you specify the @option{-trigraphs} option, then it
322converts them.
323
324These are nine three-character sequences, all starting with @samp{??},
325that are defined by ISO C to stand for single characters.  They permit
326obsolete systems that lack some of C's punctuation to use C@.  For
327example, @samp{??/} stands for @samp{\}, so @t{'??/n'} is a character
328constant for a newline.
329
330Trigraphs are not popular and many compilers implement them
331incorrectly.  Portable code should not rely on trigraphs being either
332converted or ignored.  With @option{-Wtrigraphs} GCC will warn you
333when a trigraph may change the meaning of your program if it were
334converted.  @xref{Wtrigraphs}.
335
336In a string constant, you can prevent a sequence of question marks
337from being confused with a trigraph by inserting a backslash between
338the question marks, or by separating the string literal at the
339trigraph and making use of string literal concatenation.  @t{"(??\?)"}
340is the string @samp{(???)}, not @samp{(?]}.  Traditional C compilers
341do not recognize these idioms.
342
343The nine trigraphs and their replacements are
344
345@smallexample
346Trigraph:       ??(  ??)  ??<  ??>  ??=  ??/  ??'  ??!  ??-
347Replacement:      [    ]    @{    @}    #    \    ^    |    ~
348@end smallexample
349
350@item
351@cindex continued lines
352@cindex backslash-newline
353Continued lines are merged into one long line.
354
355A continued line is a line which ends with a backslash, @samp{\}.  The
356backslash is removed and the following line is joined with the current
357one.  No space is inserted, so you may split a line anywhere, even in
358the middle of a word.  (It is generally more readable to split lines
359only at white space.)
360
361The trailing backslash on a continued line is commonly referred to as a
362@dfn{backslash-newline}.
363
364If there is white space between a backslash and the end of a line, that
365is still a continued line.  However, as this is usually the result of an
366editing mistake, and many compilers will not accept it as a continued
367line, GCC will warn you about it.
368
369@item
370@cindex comments
371@cindex line comments
372@cindex block comments
373All comments are replaced with single spaces.
374
375There are two kinds of comments.  @dfn{Block comments} begin with
376@samp{/*} and continue until the next @samp{*/}.  Block comments do not
377nest:
378
379@smallexample
380/* @r{this is} /* @r{one comment} */ @r{text outside comment}
381@end smallexample
382
383@dfn{Line comments} begin with @samp{//} and continue to the end of the
384current line.  Line comments do not nest either, but it does not matter,
385because they would end in the same place anyway.
386
387@smallexample
388// @r{this is} // @r{one comment}
389@r{text outside comment}
390@end smallexample
391@end enumerate
392
393It is safe to put line comments inside block comments, or vice versa.
394
395@smallexample
396@group
397/* @r{block comment}
398   // @r{contains line comment}
399   @r{yet more comment}
400 */ @r{outside comment}
401
402// @r{line comment} /* @r{contains block comment} */
403@end group
404@end smallexample
405
406But beware of commenting out one end of a block comment with a line
407comment.
408
409@smallexample
410@group
411 // @r{l.c.}  /* @r{block comment begins}
412    @r{oops! this isn't a comment anymore} */
413@end group
414@end smallexample
415
416Comments are not recognized within string literals.
417@t{@w{"/* blah */"}} is the string constant @samp{@w{/* blah */}}, not
418an empty string.
419
420Line comments are not in the 1989 edition of the C standard, but they
421are recognized by GCC as an extension.  In C++ and in the 1999 edition
422of the C standard, they are an official part of the language.
423
424Since these transformations happen before all other processing, you can
425split a line mechanically with backslash-newline anywhere.  You can
426comment out the end of a line.  You can continue a line comment onto the
427next line with backslash-newline.  You can even split @samp{/*},
428@samp{*/}, and @samp{//} onto multiple lines with backslash-newline.
429For example:
430
431@smallexample
432@group
433/\
434*
435*/ # /*
436*/ defi\
437ne FO\
438O 10\
43920
440@end group
441@end smallexample
442
443@noindent
444is equivalent to @code{@w{#define FOO 1020}}.  All these tricks are
445extremely confusing and should not be used in code intended to be
446readable.
447
448There is no way to prevent a backslash at the end of a line from being
449interpreted as a backslash-newline.  This cannot affect any correct
450program, however.
451
452@node Tokenization
453@section Tokenization
454
455@cindex tokens
456@cindex preprocessing tokens
457After the textual transformations are finished, the input file is
458converted into a sequence of @dfn{preprocessing tokens}.  These mostly
459correspond to the syntactic tokens used by the C compiler, but there are
460a few differences.  White space separates tokens; it is not itself a
461token of any kind.  Tokens do not have to be separated by white space,
462but it is often necessary to avoid ambiguities.
463
464When faced with a sequence of characters that has more than one possible
465tokenization, the preprocessor is greedy.  It always makes each token,
466starting from the left, as big as possible before moving on to the next
467token.  For instance, @code{a+++++b} is interpreted as
468@code{@w{a ++ ++ + b}}, not as @code{@w{a ++ + ++ b}}, even though the
469latter tokenization could be part of a valid C program and the former
470could not.
471
472Once the input file is broken into tokens, the token boundaries never
473change, except when the @samp{##} preprocessing operator is used to paste
474tokens together.  @xref{Concatenation}.  For example,
475
476@smallexample
477@group
478#define foo() bar
479foo()baz
480     @expansion{} bar baz
481@emph{not}
482     @expansion{} barbaz
483@end group
484@end smallexample
485
486The compiler does not re-tokenize the preprocessor's output.  Each
487preprocessing token becomes one compiler token.
488
489@cindex identifiers
490Preprocessing tokens fall into five broad classes: identifiers,
491preprocessing numbers, string literals, punctuators, and other.  An
492@dfn{identifier} is the same as an identifier in C: any sequence of
493letters, digits, or underscores, which begins with a letter or
494underscore.  Keywords of C have no significance to the preprocessor;
495they are ordinary identifiers.  You can define a macro whose name is a
496keyword, for instance.  The only identifier which can be considered a
497preprocessing keyword is @code{defined}.  @xref{Defined}.
498
499This is mostly true of other languages which use the C preprocessor.
500However, a few of the keywords of C++ are significant even in the
501preprocessor.  @xref{C++ Named Operators}.
502
503In the 1999 C standard, identifiers may contain letters which are not
504part of the ``basic source character set'', at the implementation's
505discretion (such as accented Latin letters, Greek letters, or Chinese
506ideograms).  This may be done with an extended character set, or the
507@samp{\u} and @samp{\U} escape sequences.  GCC only accepts such
508characters in the @samp{\u} and @samp{\U} forms.
509
510As an extension, GCC treats @samp{$} as a letter.  This is for
511compatibility with some systems, such as VMS, where @samp{$} is commonly
512used in system-defined function and object names.  @samp{$} is not a
513letter in strictly conforming mode, or if you specify the @option{-$}
514option.  @xref{Invocation}.
515
516@cindex numbers
517@cindex preprocessing numbers
518A @dfn{preprocessing number} has a rather bizarre definition.  The
519category includes all the normal integer and floating point constants
520one expects of C, but also a number of other things one might not
521initially recognize as a number.  Formally, preprocessing numbers begin
522with an optional period, a required decimal digit, and then continue
523with any sequence of letters, digits, underscores, periods, and
524exponents.  Exponents are the two-character sequences @samp{e+},
525@samp{e-}, @samp{E+}, @samp{E-}, @samp{p+}, @samp{p-}, @samp{P+}, and
526@samp{P-}.  (The exponents that begin with @samp{p} or @samp{P} are new
527to C99.  They are used for hexadecimal floating-point constants.)
528
529The purpose of this unusual definition is to isolate the preprocessor
530from the full complexity of numeric constants.  It does not have to
531distinguish between lexically valid and invalid floating-point numbers,
532which is complicated.  The definition also permits you to split an
533identifier at any position and get exactly two tokens, which can then be
534pasted back together with the @samp{##} operator.
535
536It's possible for preprocessing numbers to cause programs to be
537misinterpreted.  For example, @code{0xE+12} is a preprocessing number
538which does not translate to any valid numeric constant, therefore a
539syntax error.  It does not mean @code{@w{0xE + 12}}, which is what you
540might have intended.
541
542@cindex string literals
543@cindex string constants
544@cindex character constants
545@cindex header file names
546@c the @: prevents makeinfo from turning '' into ".
547@dfn{String literals} are string constants, character constants, and
548header file names (the argument of @samp{#include}).@footnote{The C
549standard uses the term @dfn{string literal} to refer only to what we are
550calling @dfn{string constants}.}  String constants and character
551constants are straightforward: @t{"@dots{}"} or @t{'@dots{}'}.  In
552either case embedded quotes should be escaped with a backslash:
553@t{'\'@:'} is the character constant for @samp{'}.  There is no limit on
554the length of a character constant, but the value of a character
555constant that contains more than one character is
556implementation-defined.  @xref{Implementation Details}.
557
558Header file names either look like string constants, @t{"@dots{}"}, or are
559written with angle brackets instead, @t{<@dots{}>}.  In either case,
560backslash is an ordinary character.  There is no way to escape the
561closing quote or angle bracket.  The preprocessor looks for the header
562file in different places depending on which form you use.  @xref{Include
563Operation}.
564
565No string literal may extend past the end of a line.  Older versions
566of GCC accepted multi-line string constants.  You may use continued
567lines instead, or string constant concatenation.  @xref{Differences
568from previous versions}.
569
570@cindex punctuators
571@cindex digraphs
572@cindex alternative tokens
573@dfn{Punctuators} are all the usual bits of punctuation which are
574meaningful to C and C++.  All but three of the punctuation characters in
575ASCII are C punctuators.  The exceptions are @samp{@@}, @samp{$}, and
576@samp{`}.  In addition, all the two- and three-character operators are
577punctuators.  There are also six @dfn{digraphs}, which the C++ standard
578calls @dfn{alternative tokens}, which are merely alternate ways to spell
579other punctuators.  This is a second attempt to work around missing
580punctuation in obsolete systems.  It has no negative side effects,
581unlike trigraphs, but does not cover as much ground.  The digraphs and
582their corresponding normal punctuators are:
583
584@smallexample
585Digraph:        <%  %>  <:  :>  %:  %:%:
586Punctuator:      @{   @}   [   ]   #    ##
587@end smallexample
588
589@cindex other tokens
590Any other single character is considered ``other''.  It is passed on to
591the preprocessor's output unmolested.  The C compiler will almost
592certainly reject source code containing ``other'' tokens.  In ASCII, the
593only other characters are @samp{@@}, @samp{$}, @samp{`}, and control
594characters other than NUL (all bits zero).  (Note that @samp{$} is
595normally considered a letter.)  All characters with the high bit set
596(numeric range 0x7F--0xFF) are also ``other'' in the present
597implementation.  This will change when proper support for international
598character sets is added to GCC@.
599
600NUL is a special case because of the high probability that its
601appearance is accidental, and because it may be invisible to the user
602(many terminals do not display NUL at all).  Within comments, NULs are
603silently ignored, just as any other character would be.  In running
604text, NUL is considered white space.  For example, these two directives
605have the same meaning.
606
607@smallexample
608#define X^@@1
609#define X 1
610@end smallexample
611
612@noindent
613(where @samp{^@@} is ASCII NUL)@.  Within string or character constants,
614NULs are preserved.  In the latter two cases the preprocessor emits a
615warning message.
616
617@node The preprocessing language
618@section The preprocessing language
619@cindex directives
620@cindex preprocessing directives
621@cindex directive line
622@cindex directive name
623
624After tokenization, the stream of tokens may simply be passed straight
625to the compiler's parser.  However, if it contains any operations in the
626@dfn{preprocessing language}, it will be transformed first.  This stage
627corresponds roughly to the standard's ``translation phase 4'' and is
628what most people think of as the preprocessor's job.
629
630The preprocessing language consists of @dfn{directives} to be executed
631and @dfn{macros} to be expanded.  Its primary capabilities are:
632
633@itemize @bullet
634@item
635Inclusion of header files.  These are files of declarations that can be
636substituted into your program.
637
638@item
639Macro expansion.  You can define @dfn{macros}, which are abbreviations
640for arbitrary fragments of C code.  The preprocessor will replace the
641macros with their definitions throughout the program.  Some macros are
642automatically defined for you.
643
644@item
645Conditional compilation.  You can include or exclude parts of the
646program according to various conditions.
647
648@item
649Line control.  If you use a program to combine or rearrange source files
650into an intermediate file which is then compiled, you can use line
651control to inform the compiler where each source line originally came
652from.
653
654@item
655Diagnostics.  You can detect problems at compile time and issue errors
656or warnings.
657@end itemize
658
659There are a few more, less useful, features.
660
661Except for expansion of predefined macros, all these operations are
662triggered with @dfn{preprocessing directives}.  Preprocessing directives
663are lines in your program that start with @samp{#}.  Whitespace is
664allowed before and after the @samp{#}.  The @samp{#} is followed by an
665identifier, the @dfn{directive name}.  It specifies the operation to
666perform.  Directives are commonly referred to as @samp{#@var{name}}
667where @var{name} is the directive name.  For example, @samp{#define} is
668the directive that defines a macro.
669
670The @samp{#} which begins a directive cannot come from a macro
671expansion.  Also, the directive name is not macro expanded.  Thus, if
672@code{foo} is defined as a macro expanding to @code{define}, that does
673not make @samp{#foo} a valid preprocessing directive.
674
675The set of valid directive names is fixed.  Programs cannot define new
676preprocessing directives.
677
678Some directives require arguments; these make up the rest of the
679directive line and must be separated from the directive name by
680whitespace.  For example, @samp{#define} must be followed by a macro
681name and the intended expansion of the macro.
682
683A preprocessing directive cannot cover more than one line.  The line
684may, however, be continued with backslash-newline, or by a block comment
685which extends past the end of the line.  In either case, when the
686directive is processed, the continuations have already been merged with
687the first line to make one long line.
688
689@node Header Files
690@chapter Header Files
691
692@cindex header file
693A header file is a file containing C declarations and macro definitions
694(@pxref{Macros}) to be shared between several source files.  You request
695the use of a header file in your program by @dfn{including} it, with the
696C preprocessing directive @samp{#include}.
697
698Header files serve two purposes.
699
700@itemize @bullet
701@item
702@cindex system header files
703System header files declare the interfaces to parts of the operating
704system.  You include them in your program to supply the definitions and
705declarations you need to invoke system calls and libraries.
706
707@item
708Your own header files contain declarations for interfaces between the
709source files of your program.  Each time you have a group of related
710declarations and macro definitions all or most of which are needed in
711several different source files, it is a good idea to create a header
712file for them.
713@end itemize
714
715Including a header file produces the same results as copying the header
716file into each source file that needs it.  Such copying would be
717time-consuming and error-prone.  With a header file, the related
718declarations appear in only one place.  If they need to be changed, they
719can be changed in one place, and programs that include the header file
720will automatically use the new version when next recompiled.  The header
721file eliminates the labor of finding and changing all the copies as well
722as the risk that a failure to find one copy will result in
723inconsistencies within a program.
724
725In C, the usual convention is to give header files names that end with
726@file{.h}.  It is most portable to use only letters, digits, dashes, and
727underscores in header file names, and at most one dot.
728
729@menu
730* Include Syntax::
731* Include Operation::
732* Search Path::
733* Once-Only Headers::
734* Alternatives to Wrapper #ifndef::
735* Computed Includes::
736* Wrapper Headers::
737* System Headers::
738@end menu
739
740@node Include Syntax
741@section Include Syntax
742
743@findex #include
744Both user and system header files are included using the preprocessing
745directive @samp{#include}.  It has two variants:
746
747@table @code
748@item #include <@var{file}>
749This variant is used for system header files.  It searches for a file
750named @var{file} in a standard list of system directories.  You can prepend
751directories to this list with the @option{-I} option (@pxref{Invocation}).
752
753@item #include "@var{file}"
754This variant is used for header files of your own program.  It
755searches for a file named @var{file} first in the directory containing
756the current file, then in the quote directories and then the same
757directories used for @code{<@var{file}>}.  You can prepend directories
758to the list of quote directories with the @option{-iquote} option.
759@end table
760
761The argument of @samp{#include}, whether delimited with quote marks or
762angle brackets, behaves like a string constant in that comments are not
763recognized, and macro names are not expanded.  Thus, @code{@w{#include
764<x/*y>}} specifies inclusion of a system header file named @file{x/*y}.
765
766However, if backslashes occur within @var{file}, they are considered
767ordinary text characters, not escape characters.  None of the character
768escape sequences appropriate to string constants in C are processed.
769Thus, @code{@w{#include "x\n\\y"}} specifies a filename containing three
770backslashes.  (Some systems interpret @samp{\} as a pathname separator.
771All of these also interpret @samp{/} the same way.  It is most portable
772to use only @samp{/}.)
773
774It is an error if there is anything (other than comments) on the line
775after the file name.
776
777@node Include Operation
778@section Include Operation
779
780The @samp{#include} directive works by directing the C preprocessor to
781scan the specified file as input before continuing with the rest of the
782current file.  The output from the preprocessor contains the output
783already generated, followed by the output resulting from the included
784file, followed by the output that comes from the text after the
785@samp{#include} directive.  For example, if you have a header file
786@file{header.h} as follows,
787
788@smallexample
789char *test (void);
790@end smallexample
791
792@noindent
793and a main program called @file{program.c} that uses the header file,
794like this,
795
796@smallexample
797int x;
798#include "header.h"
799
800int
801main (void)
802@{
803  puts (test ());
804@}
805@end smallexample
806
807@noindent
808the compiler will see the same token stream as it would if
809@file{program.c} read
810
811@smallexample
812int x;
813char *test (void);
814
815int
816main (void)
817@{
818  puts (test ());
819@}
820@end smallexample
821
822Included files are not limited to declarations and macro definitions;
823those are merely the typical uses.  Any fragment of a C program can be
824included from another file.  The include file could even contain the
825beginning of a statement that is concluded in the containing file, or
826the end of a statement that was started in the including file.  However,
827an included file must consist of complete tokens.  Comments and string
828literals which have not been closed by the end of an included file are
829invalid.  For error recovery, they are considered to end at the end of
830the file.
831
832To avoid confusion, it is best if header files contain only complete
833syntactic units---function declarations or definitions, type
834declarations, etc.
835
836The line following the @samp{#include} directive is always treated as a
837separate line by the C preprocessor, even if the included file lacks a
838final newline.
839
840@node Search Path
841@section Search Path
842
843GCC looks in several different places for headers.  On a normal Unix
844system, if you do not instruct it otherwise, it will look for headers
845requested with @code{@w{#include <@var{file}>}} in:
846
847@smallexample
848/usr/local/include
849@var{libdir}/gcc/@var{target}/@var{version}/include
850/usr/@var{target}/include
851/usr/include
852@end smallexample
853
854For C++ programs, it will also look in
855@file{@var{libdir}/../include/c++/@var{version}},
856first.  In the above, @var{target} is the canonical name of the system
857GCC was configured to compile code for; often but not always the same as
858the canonical name of the system it runs on.  @var{version} is the
859version of GCC in use.
860
861You can add to this list with the @option{-I@var{dir}} command-line
862option.  All the directories named by @option{-I} are searched, in
863left-to-right order, @emph{before} the default directories.  The only
864exception is when @file{dir} is already searched by default.  In
865this case, the option is ignored and the search order for system
866directories remains unchanged.
867
868Duplicate directories are removed from the quote and bracket search
869chains before the two chains are merged to make the final search chain.
870Thus, it is possible for a directory to occur twice in the final search
871chain if it was specified in both the quote and bracket chains.
872
873You can prevent GCC from searching any of the default directories with
874the @option{-nostdinc} option.  This is useful when you are compiling an
875operating system kernel or some other program that does not use the
876standard C library facilities, or the standard C library itself.
877@option{-I} options are not ignored as described above when
878@option{-nostdinc} is in effect.
879
880GCC looks for headers requested with @code{@w{#include "@var{file}"}}
881first in the directory containing the current file, then in the
882directories as specified by @option{-iquote} options, then in the same
883places it would have looked for a header requested with angle
884brackets.  For example, if @file{/usr/include/sys/stat.h} contains
885@code{@w{#include "types.h"}}, GCC looks for @file{types.h} first in
886@file{/usr/include/sys}, then in its usual search path.
887
888@samp{#line} (@pxref{Line Control}) does not change GCC's idea of the
889directory containing the current file.
890
891You may put @option{-I-} at any point in your list of @option{-I} options.
892This has two effects.  First, directories appearing before the
893@option{-I-} in the list are searched only for headers requested with
894quote marks.  Directories after @option{-I-} are searched for all
895headers.  Second, the directory containing the current file is not
896searched for anything, unless it happens to be one of the directories
897named by an @option{-I} switch.  @option{-I-} is deprecated, @option{-iquote}
898should be used instead.
899
900@option{-I. -I-} is not the same as no @option{-I} options at all, and does
901not cause the same behavior for @samp{<>} includes that @samp{""}
902includes get with no special options.  @option{-I.} searches the
903compiler's current working directory for header files.  That may or may
904not be the same as the directory containing the current file.
905
906If you need to look for headers in a directory named @file{-}, write
907@option{-I./-}.
908
909There are several more ways to adjust the header search path.  They are
910generally less useful.  @xref{Invocation}.
911
912@node Once-Only Headers
913@section Once-Only Headers
914@cindex repeated inclusion
915@cindex including just once
916@cindex wrapper @code{#ifndef}
917
918If a header file happens to be included twice, the compiler will process
919its contents twice.  This is very likely to cause an error, e.g.@: when the
920compiler sees the same structure definition twice.  Even if it does not,
921it will certainly waste time.
922
923The standard way to prevent this is to enclose the entire real contents
924of the file in a conditional, like this:
925
926@smallexample
927@group
928/* File foo.  */
929#ifndef FILE_FOO_SEEN
930#define FILE_FOO_SEEN
931
932@var{the entire file}
933
934#endif /* !FILE_FOO_SEEN */
935@end group
936@end smallexample
937
938This construct is commonly known as a @dfn{wrapper #ifndef}.
939When the header is included again, the conditional will be false,
940because @code{FILE_FOO_SEEN} is defined.  The preprocessor will skip
941over the entire contents of the file, and the compiler will not see it
942twice.
943
944CPP optimizes even further.  It remembers when a header file has a
945wrapper @samp{#ifndef}.  If a subsequent @samp{#include} specifies that
946header, and the macro in the @samp{#ifndef} is still defined, it does
947not bother to rescan the file at all.
948
949You can put comments outside the wrapper.  They will not interfere with
950this optimization.
951
952@cindex controlling macro
953@cindex guard macro
954The macro @code{FILE_FOO_SEEN} is called the @dfn{controlling macro} or
955@dfn{guard macro}.  In a user header file, the macro name should not
956begin with @samp{_}.  In a system header file, it should begin with
957@samp{__} to avoid conflicts with user programs.  In any kind of header
958file, the macro name should contain the name of the file and some
959additional text, to avoid conflicts with other header files.
960
961@node Alternatives to Wrapper #ifndef
962@section Alternatives to Wrapper #ifndef
963
964CPP supports two more ways of indicating that a header file should be
965read only once.  Neither one is as portable as a wrapper @samp{#ifndef}
966and we recommend you do not use them in new programs, with the caveat
967that @samp{#import} is standard practice in Objective-C.
968
969@findex #import
970CPP supports a variant of @samp{#include} called @samp{#import} which
971includes a file, but does so at most once.  If you use @samp{#import}
972instead of @samp{#include}, then you don't need the conditionals
973inside the header file to prevent multiple inclusion of the contents.
974@samp{#import} is standard in Objective-C, but is considered a
975deprecated extension in C and C++.
976
977@samp{#import} is not a well designed feature.  It requires the users of
978a header file to know that it should only be included once.  It is much
979better for the header file's implementor to write the file so that users
980don't need to know this.  Using a wrapper @samp{#ifndef} accomplishes
981this goal.
982
983In the present implementation, a single use of @samp{#import} will
984prevent the file from ever being read again, by either @samp{#import} or
985@samp{#include}.  You should not rely on this; do not use both
986@samp{#import} and @samp{#include} to refer to the same header file.
987
988Another way to prevent a header file from being included more than once
989is with the @samp{#pragma once} directive.  If @samp{#pragma once} is
990seen when scanning a header file, that file will never be read again, no
991matter what.
992
993@samp{#pragma once} does not have the problems that @samp{#import} does,
994but it is not recognized by all preprocessors, so you cannot rely on it
995in a portable program.
996
997@node Computed Includes
998@section Computed Includes
999@cindex computed includes
1000@cindex macros in include
1001
1002Sometimes it is necessary to select one of several different header
1003files to be included into your program.  They might specify
1004configuration parameters to be used on different sorts of operating
1005systems, for instance.  You could do this with a series of conditionals,
1006
1007@smallexample
1008#if SYSTEM_1
1009# include "system_1.h"
1010#elif SYSTEM_2
1011# include "system_2.h"
1012#elif SYSTEM_3
1013@dots{}
1014#endif
1015@end smallexample
1016
1017That rapidly becomes tedious.  Instead, the preprocessor offers the
1018ability to use a macro for the header name.  This is called a
1019@dfn{computed include}.  Instead of writing a header name as the direct
1020argument of @samp{#include}, you simply put a macro name there instead:
1021
1022@smallexample
1023#define SYSTEM_H "system_1.h"
1024@dots{}
1025#include SYSTEM_H
1026@end smallexample
1027
1028@noindent
1029@code{SYSTEM_H} will be expanded, and the preprocessor will look for
1030@file{system_1.h} as if the @samp{#include} had been written that way
1031originally.  @code{SYSTEM_H} could be defined by your Makefile with a
1032@option{-D} option.
1033
1034You must be careful when you define the macro.  @samp{#define} saves
1035tokens, not text.  The preprocessor has no way of knowing that the macro
1036will be used as the argument of @samp{#include}, so it generates
1037ordinary tokens, not a header name.  This is unlikely to cause problems
1038if you use double-quote includes, which are close enough to string
1039constants.  If you use angle brackets, however, you may have trouble.
1040
1041The syntax of a computed include is actually a bit more general than the
1042above.  If the first non-whitespace character after @samp{#include} is
1043not @samp{"} or @samp{<}, then the entire line is macro-expanded
1044like running text would be.
1045
1046If the line expands to a single string constant, the contents of that
1047string constant are the file to be included.  CPP does not re-examine the
1048string for embedded quotes, but neither does it process backslash
1049escapes in the string.  Therefore
1050
1051@smallexample
1052#define HEADER "a\"b"
1053#include HEADER
1054@end smallexample
1055
1056@noindent
1057looks for a file named @file{a\"b}.  CPP searches for the file according
1058to the rules for double-quoted includes.
1059
1060If the line expands to a token stream beginning with a @samp{<} token
1061and including a @samp{>} token, then the tokens between the @samp{<} and
1062the first @samp{>} are combined to form the filename to be included.
1063Any whitespace between tokens is reduced to a single space; then any
1064space after the initial @samp{<} is retained, but a trailing space
1065before the closing @samp{>} is ignored.  CPP searches for the file
1066according to the rules for angle-bracket includes.
1067
1068In either case, if there are any tokens on the line after the file name,
1069an error occurs and the directive is not processed.  It is also an error
1070if the result of expansion does not match either of the two expected
1071forms.
1072
1073These rules are implementation-defined behavior according to the C
1074standard.  To minimize the risk of different compilers interpreting your
1075computed includes differently, we recommend you use only a single
1076object-like macro which expands to a string constant.  This will also
1077minimize confusion for people reading your program.
1078
1079@node Wrapper Headers
1080@section Wrapper Headers
1081@cindex wrapper headers
1082@cindex overriding a header file
1083@findex #include_next
1084
1085Sometimes it is necessary to adjust the contents of a system-provided
1086header file without editing it directly.  GCC's @command{fixincludes}
1087operation does this, for example.  One way to do that would be to create
1088a new header file with the same name and insert it in the search path
1089before the original header.  That works fine as long as you're willing
1090to replace the old header entirely.  But what if you want to refer to
1091the old header from the new one?
1092
1093You cannot simply include the old header with @samp{#include}.  That
1094will start from the beginning, and find your new header again.  If your
1095header is not protected from multiple inclusion (@pxref{Once-Only
1096Headers}), it will recurse infinitely and cause a fatal error.
1097
1098You could include the old header with an absolute pathname:
1099@smallexample
1100#include "/usr/include/old-header.h"
1101@end smallexample
1102@noindent
1103This works, but is not clean; should the system headers ever move, you
1104would have to edit the new headers to match.
1105
1106There is no way to solve this problem within the C standard, but you can
1107use the GNU extension @samp{#include_next}.  It means, ``Include the
1108@emph{next} file with this name''.  This directive works like
1109@samp{#include} except in searching for the specified file: it starts
1110searching the list of header file directories @emph{after} the directory
1111in which the current file was found.
1112
1113Suppose you specify @option{-I /usr/local/include}, and the list of
1114directories to search also includes @file{/usr/include}; and suppose
1115both directories contain @file{signal.h}.  Ordinary @code{@w{#include
1116<signal.h>}} finds the file under @file{/usr/local/include}.  If that
1117file contains @code{@w{#include_next <signal.h>}}, it starts searching
1118after that directory, and finds the file in @file{/usr/include}.
1119
1120@samp{#include_next} does not distinguish between @code{<@var{file}>}
1121and @code{"@var{file}"} inclusion, nor does it check that the file you
1122specify has the same name as the current file.  It simply looks for the
1123file named, starting with the directory in the search path after the one
1124where the current file was found.
1125
1126The use of @samp{#include_next} can lead to great confusion.  We
1127recommend it be used only when there is no other alternative.  In
1128particular, it should not be used in the headers belonging to a specific
1129program; it should be used only to make global corrections along the
1130lines of @command{fixincludes}.
1131
1132@node System Headers
1133@section System Headers
1134@cindex system header files
1135
1136The header files declaring interfaces to the operating system and
1137runtime libraries often cannot be written in strictly conforming C@.
1138Therefore, GCC gives code found in @dfn{system headers} special
1139treatment.  All warnings, other than those generated by @samp{#warning}
1140(@pxref{Diagnostics}), are suppressed while GCC is processing a system
1141header.  Macros defined in a system header are immune to a few warnings
1142wherever they are expanded.  This immunity is granted on an ad-hoc
1143basis, when we find that a warning generates lots of false positives
1144because of code in macros defined in system headers.
1145
1146Normally, only the headers found in specific directories are considered
1147system headers.  These directories are determined when GCC is compiled.
1148There are, however, two ways to make normal headers into system headers.
1149
1150The @option{-isystem} command-line option adds its argument to the list of
1151directories to search for headers, just like @option{-I}.  Any headers
1152found in that directory will be considered system headers.
1153
1154All directories named by @option{-isystem} are searched @emph{after} all
1155directories named by @option{-I}, no matter what their order was on the
1156command line.  If the same directory is named by both @option{-I} and
1157@option{-isystem}, the @option{-I} option is ignored.  GCC provides an
1158informative message when this occurs if @option{-v} is used.
1159
1160The @option{-cxx-isystem} command line option adds its argument to the
1161list of C++ system headers, similar to @option{-isystem} for C headers.
1162
1163@findex #pragma GCC system_header
1164There is also a directive, @code{@w{#pragma GCC system_header}}, which
1165tells GCC to consider the rest of the current include file a system
1166header, no matter where it was found.  Code that comes before the
1167@samp{#pragma} in the file will not be affected.  @code{@w{#pragma GCC
1168system_header}} has no effect in the primary source file.
1169
1170On very old systems, some of the pre-defined system header directories
1171get even more special treatment.  GNU C++ considers code in headers
1172found in those directories to be surrounded by an @code{@w{extern "C"}}
1173block.  There is no way to request this behavior with a @samp{#pragma},
1174or from the command line.
1175
1176@node Macros
1177@chapter Macros
1178
1179A @dfn{macro} is a fragment of code which has been given a name.
1180Whenever the name is used, it is replaced by the contents of the macro.
1181There are two kinds of macros.  They differ mostly in what they look
1182like when they are used.  @dfn{Object-like} macros resemble data objects
1183when used, @dfn{function-like} macros resemble function calls.
1184
1185You may define any valid identifier as a macro, even if it is a C
1186keyword.  The preprocessor does not know anything about keywords.  This
1187can be useful if you wish to hide a keyword such as @code{const} from an
1188older compiler that does not understand it.  However, the preprocessor
1189operator @code{defined} (@pxref{Defined}) can never be defined as a
1190macro, and C++'s named operators (@pxref{C++ Named Operators}) cannot be
1191macros when you are compiling C++.
1192
1193@menu
1194* Object-like Macros::
1195* Function-like Macros::
1196* Macro Arguments::
1197* Stringification::
1198* Concatenation::
1199* Variadic Macros::
1200* Predefined Macros::
1201* Undefining and Redefining Macros::
1202* Directives Within Macro Arguments::
1203* Macro Pitfalls::
1204@end menu
1205
1206@node Object-like Macros
1207@section Object-like Macros
1208@cindex object-like macro
1209@cindex symbolic constants
1210@cindex manifest constants
1211
1212An @dfn{object-like macro} is a simple identifier which will be replaced
1213by a code fragment.  It is called object-like because it looks like a
1214data object in code that uses it.  They are most commonly used to give
1215symbolic names to numeric constants.
1216
1217@findex #define
1218You create macros with the @samp{#define} directive.  @samp{#define} is
1219followed by the name of the macro and then the token sequence it should
1220be an abbreviation for, which is variously referred to as the macro's
1221@dfn{body}, @dfn{expansion} or @dfn{replacement list}.  For example,
1222
1223@smallexample
1224#define BUFFER_SIZE 1024
1225@end smallexample
1226
1227@noindent
1228defines a macro named @code{BUFFER_SIZE} as an abbreviation for the
1229token @code{1024}.  If somewhere after this @samp{#define} directive
1230there comes a C statement of the form
1231
1232@smallexample
1233foo = (char *) malloc (BUFFER_SIZE);
1234@end smallexample
1235
1236@noindent
1237then the C preprocessor will recognize and @dfn{expand} the macro
1238@code{BUFFER_SIZE}.  The C compiler will see the same tokens as it would
1239if you had written
1240
1241@smallexample
1242foo = (char *) malloc (1024);
1243@end smallexample
1244
1245By convention, macro names are written in uppercase.  Programs are
1246easier to read when it is possible to tell at a glance which names are
1247macros.
1248
1249The macro's body ends at the end of the @samp{#define} line.  You may
1250continue the definition onto multiple lines, if necessary, using
1251backslash-newline.  When the macro is expanded, however, it will all
1252come out on one line.  For example,
1253
1254@smallexample
1255#define NUMBERS 1, \
1256                2, \
1257                3
1258int x[] = @{ NUMBERS @};
1259     @expansion{} int x[] = @{ 1, 2, 3 @};
1260@end smallexample
1261
1262@noindent
1263The most common visible consequence of this is surprising line numbers
1264in error messages.
1265
1266There is no restriction on what can go in a macro body provided it
1267decomposes into valid preprocessing tokens.  Parentheses need not
1268balance, and the body need not resemble valid C code.  (If it does not,
1269you may get error messages from the C compiler when you use the macro.)
1270
1271The C preprocessor scans your program sequentially.  Macro definitions
1272take effect at the place you write them.  Therefore, the following input
1273to the C preprocessor
1274
1275@smallexample
1276foo = X;
1277#define X 4
1278bar = X;
1279@end smallexample
1280
1281@noindent
1282produces
1283
1284@smallexample
1285foo = X;
1286bar = 4;
1287@end smallexample
1288
1289When the preprocessor expands a macro name, the macro's expansion
1290replaces the macro invocation, then the expansion is examined for more
1291macros to expand.  For example,
1292
1293@smallexample
1294@group
1295#define TABLESIZE BUFSIZE
1296#define BUFSIZE 1024
1297TABLESIZE
1298     @expansion{} BUFSIZE
1299     @expansion{} 1024
1300@end group
1301@end smallexample
1302
1303@noindent
1304@code{TABLESIZE} is expanded first to produce @code{BUFSIZE}, then that
1305macro is expanded to produce the final result, @code{1024}.
1306
1307Notice that @code{BUFSIZE} was not defined when @code{TABLESIZE} was
1308defined.  The @samp{#define} for @code{TABLESIZE} uses exactly the
1309expansion you specify---in this case, @code{BUFSIZE}---and does not
1310check to see whether it too contains macro names.  Only when you
1311@emph{use} @code{TABLESIZE} is the result of its expansion scanned for
1312more macro names.
1313
1314This makes a difference if you change the definition of @code{BUFSIZE}
1315at some point in the source file.  @code{TABLESIZE}, defined as shown,
1316will always expand using the definition of @code{BUFSIZE} that is
1317currently in effect:
1318
1319@smallexample
1320#define BUFSIZE 1020
1321#define TABLESIZE BUFSIZE
1322#undef BUFSIZE
1323#define BUFSIZE 37
1324@end smallexample
1325
1326@noindent
1327Now @code{TABLESIZE} expands (in two stages) to @code{37}.
1328
1329If the expansion of a macro contains its own name, either directly or
1330via intermediate macros, it is not expanded again when the expansion is
1331examined for more macros.  This prevents infinite recursion.
1332@xref{Self-Referential Macros}, for the precise details.
1333
1334@node Function-like Macros
1335@section Function-like Macros
1336@cindex function-like macros
1337
1338You can also define macros whose use looks like a function call.  These
1339are called @dfn{function-like macros}.  To define a function-like macro,
1340you use the same @samp{#define} directive, but you put a pair of
1341parentheses immediately after the macro name.  For example,
1342
1343@smallexample
1344#define lang_init()  c_init()
1345lang_init()
1346     @expansion{} c_init()
1347@end smallexample
1348
1349A function-like macro is only expanded if its name appears with a pair
1350of parentheses after it.  If you write just the name, it is left alone.
1351This can be useful when you have a function and a macro of the same
1352name, and you wish to use the function sometimes.
1353
1354@smallexample
1355extern void foo(void);
1356#define foo() /* @r{optimized inline version} */
1357@dots{}
1358  foo();
1359  funcptr = foo;
1360@end smallexample
1361
1362Here the call to @code{foo()} will use the macro, but the function
1363pointer will get the address of the real function.  If the macro were to
1364be expanded, it would cause a syntax error.
1365
1366If you put spaces between the macro name and the parentheses in the
1367macro definition, that does not define a function-like macro, it defines
1368an object-like macro whose expansion happens to begin with a pair of
1369parentheses.
1370
1371@smallexample
1372#define lang_init ()    c_init()
1373lang_init()
1374     @expansion{} () c_init()()
1375@end smallexample
1376
1377The first two pairs of parentheses in this expansion come from the
1378macro.  The third is the pair that was originally after the macro
1379invocation.  Since @code{lang_init} is an object-like macro, it does not
1380consume those parentheses.
1381
1382@node Macro Arguments
1383@section Macro Arguments
1384@cindex arguments
1385@cindex macros with arguments
1386@cindex arguments in macro definitions
1387
1388Function-like macros can take @dfn{arguments}, just like true functions.
1389To define a macro that uses arguments, you insert @dfn{parameters}
1390between the pair of parentheses in the macro definition that make the
1391macro function-like.  The parameters must be valid C identifiers,
1392separated by commas and optionally whitespace.
1393
1394To invoke a macro that takes arguments, you write the name of the macro
1395followed by a list of @dfn{actual arguments} in parentheses, separated
1396by commas.  The invocation of the macro need not be restricted to a
1397single logical line---it can cross as many lines in the source file as
1398you wish.  The number of arguments you give must match the number of
1399parameters in the macro definition.  When the macro is expanded, each
1400use of a parameter in its body is replaced by the tokens of the
1401corresponding argument.  (You need not use all of the parameters in the
1402macro body.)
1403
1404As an example, here is a macro that computes the minimum of two numeric
1405values, as it is defined in many C programs, and some uses.
1406
1407@smallexample
1408#define min(X, Y)  ((X) < (Y) ? (X) : (Y))
1409  x = min(a, b);          @expansion{}  x = ((a) < (b) ? (a) : (b));
1410  y = min(1, 2);          @expansion{}  y = ((1) < (2) ? (1) : (2));
1411  z = min(a + 28, *p);    @expansion{}  z = ((a + 28) < (*p) ? (a + 28) : (*p));
1412@end smallexample
1413
1414@noindent
1415(In this small example you can already see several of the dangers of
1416macro arguments.  @xref{Macro Pitfalls}, for detailed explanations.)
1417
1418Leading and trailing whitespace in each argument is dropped, and all
1419whitespace between the tokens of an argument is reduced to a single
1420space.  Parentheses within each argument must balance; a comma within
1421such parentheses does not end the argument.  However, there is no
1422requirement for square brackets or braces to balance, and they do not
1423prevent a comma from separating arguments.  Thus,
1424
1425@smallexample
1426macro (array[x = y, x + 1])
1427@end smallexample
1428
1429@noindent
1430passes two arguments to @code{macro}: @code{array[x = y} and @code{x +
14311]}.  If you want to supply @code{array[x = y, x + 1]} as an argument,
1432you can write it as @code{array[(x = y, x + 1)]}, which is equivalent C
1433code.
1434
1435All arguments to a macro are completely macro-expanded before they are
1436substituted into the macro body.  After substitution, the complete text
1437is scanned again for macros to expand, including the arguments.  This rule
1438may seem strange, but it is carefully designed so you need not worry
1439about whether any function call is actually a macro invocation.  You can
1440run into trouble if you try to be too clever, though.  @xref{Argument
1441Prescan}, for detailed discussion.
1442
1443For example, @code{min (min (a, b), c)} is first expanded to
1444
1445@smallexample
1446  min (((a) < (b) ? (a) : (b)), (c))
1447@end smallexample
1448
1449@noindent
1450and then to
1451
1452@smallexample
1453@group
1454((((a) < (b) ? (a) : (b))) < (c)
1455 ? (((a) < (b) ? (a) : (b)))
1456 : (c))
1457@end group
1458@end smallexample
1459
1460@noindent
1461(Line breaks shown here for clarity would not actually be generated.)
1462
1463@cindex empty macro arguments
1464You can leave macro arguments empty; this is not an error to the
1465preprocessor (but many macros will then expand to invalid code).
1466You cannot leave out arguments entirely; if a macro takes two arguments,
1467there must be exactly one comma at the top level of its argument list.
1468Here are some silly examples using @code{min}:
1469
1470@smallexample
1471min(, b)        @expansion{} ((   ) < (b) ? (   ) : (b))
1472min(a, )        @expansion{} ((a  ) < ( ) ? (a  ) : ( ))
1473min(,)          @expansion{} ((   ) < ( ) ? (   ) : ( ))
1474min((,),)       @expansion{} (((,)) < ( ) ? ((,)) : ( ))
1475
1476min()      @error{} macro "min" requires 2 arguments, but only 1 given
1477min(,,)    @error{} macro "min" passed 3 arguments, but takes just 2
1478@end smallexample
1479
1480Whitespace is not a preprocessing token, so if a macro @code{foo} takes
1481one argument, @code{@w{foo ()}} and @code{@w{foo ( )}} both supply it an
1482empty argument.  Previous GNU preprocessor implementations and
1483documentation were incorrect on this point, insisting that a
1484function-like macro that takes a single argument be passed a space if an
1485empty argument was required.
1486
1487Macro parameters appearing inside string literals are not replaced by
1488their corresponding actual arguments.
1489
1490@smallexample
1491#define foo(x) x, "x"
1492foo(bar)        @expansion{} bar, "x"
1493@end smallexample
1494
1495@node Stringification
1496@section Stringification
1497@cindex stringification
1498@cindex @samp{#} operator
1499
1500Sometimes you may want to convert a macro argument into a string
1501constant.  Parameters are not replaced inside string constants, but you
1502can use the @samp{#} preprocessing operator instead.  When a macro
1503parameter is used with a leading @samp{#}, the preprocessor replaces it
1504with the literal text of the actual argument, converted to a string
1505constant.  Unlike normal parameter replacement, the argument is not
1506macro-expanded first.  This is called @dfn{stringification}.
1507
1508There is no way to combine an argument with surrounding text and
1509stringify it all together.  Instead, you can write a series of adjacent
1510string constants and stringified arguments.  The preprocessor will
1511replace the stringified arguments with string constants.  The C
1512compiler will then combine all the adjacent string constants into one
1513long string.
1514
1515Here is an example of a macro definition that uses stringification:
1516
1517@smallexample
1518@group
1519#define WARN_IF(EXP) \
1520do @{ if (EXP) \
1521        fprintf (stderr, "Warning: " #EXP "\n"); @} \
1522while (0)
1523WARN_IF (x == 0);
1524     @expansion{} do @{ if (x == 0)
1525           fprintf (stderr, "Warning: " "x == 0" "\n"); @} while (0);
1526@end group
1527@end smallexample
1528
1529@noindent
1530The argument for @code{EXP} is substituted once, as-is, into the
1531@code{if} statement, and once, stringified, into the argument to
1532@code{fprintf}.  If @code{x} were a macro, it would be expanded in the
1533@code{if} statement, but not in the string.
1534
1535The @code{do} and @code{while (0)} are a kludge to make it possible to
1536write @code{WARN_IF (@var{arg});}, which the resemblance of
1537@code{WARN_IF} to a function would make C programmers want to do; see
1538@ref{Swallowing the Semicolon}.
1539
1540Stringification in C involves more than putting double-quote characters
1541around the fragment.  The preprocessor backslash-escapes the quotes
1542surrounding embedded string constants, and all backslashes within string and
1543character constants, in order to get a valid C string constant with the
1544proper contents.  Thus, stringifying @code{@w{p = "foo\n";}} results in
1545@t{@w{"p = \"foo\\n\";"}}.  However, backslashes that are not inside string
1546or character constants are not duplicated: @samp{\n} by itself
1547stringifies to @t{"\n"}.
1548
1549All leading and trailing whitespace in text being stringified is
1550ignored.  Any sequence of whitespace in the middle of the text is
1551converted to a single space in the stringified result.  Comments are
1552replaced by whitespace long before stringification happens, so they
1553never appear in stringified text.
1554
1555There is no way to convert a macro argument into a character constant.
1556
1557If you want to stringify the result of expansion of a macro argument,
1558you have to use two levels of macros.
1559
1560@smallexample
1561#define xstr(s) str(s)
1562#define str(s) #s
1563#define foo 4
1564str (foo)
1565     @expansion{} "foo"
1566xstr (foo)
1567     @expansion{} xstr (4)
1568     @expansion{} str (4)
1569     @expansion{} "4"
1570@end smallexample
1571
1572@code{s} is stringified when it is used in @code{str}, so it is not
1573macro-expanded first.  But @code{s} is an ordinary argument to
1574@code{xstr}, so it is completely macro-expanded before @code{xstr}
1575itself is expanded (@pxref{Argument Prescan}).  Therefore, by the time
1576@code{str} gets to its argument, it has already been macro-expanded.
1577
1578@node Concatenation
1579@section Concatenation
1580@cindex concatenation
1581@cindex token pasting
1582@cindex token concatenation
1583@cindex @samp{##} operator
1584
1585It is often useful to merge two tokens into one while expanding macros.
1586This is called @dfn{token pasting} or @dfn{token concatenation}.  The
1587@samp{##} preprocessing operator performs token pasting.  When a macro
1588is expanded, the two tokens on either side of each @samp{##} operator
1589are combined into a single token, which then replaces the @samp{##} and
1590the two original tokens in the macro expansion.  Usually both will be
1591identifiers, or one will be an identifier and the other a preprocessing
1592number.  When pasted, they make a longer identifier.  This isn't the
1593only valid case.  It is also possible to concatenate two numbers (or a
1594number and a name, such as @code{1.5} and @code{e3}) into a number.
1595Also, multi-character operators such as @code{+=} can be formed by
1596token pasting.
1597
1598However, two tokens that don't together form a valid token cannot be
1599pasted together.  For example, you cannot concatenate @code{x} with
1600@code{+} in either order.  If you try, the preprocessor issues a warning
1601and emits the two tokens.  Whether it puts white space between the
1602tokens is undefined.  It is common to find unnecessary uses of @samp{##}
1603in complex macros.  If you get this warning, it is likely that you can
1604simply remove the @samp{##}.
1605
1606Both the tokens combined by @samp{##} could come from the macro body,
1607but you could just as well write them as one token in the first place.
1608Token pasting is most useful when one or both of the tokens comes from a
1609macro argument.  If either of the tokens next to an @samp{##} is a
1610parameter name, it is replaced by its actual argument before @samp{##}
1611executes.  As with stringification, the actual argument is not
1612macro-expanded first.  If the argument is empty, that @samp{##} has no
1613effect.
1614
1615Keep in mind that the C preprocessor converts comments to whitespace
1616before macros are even considered.  Therefore, you cannot create a
1617comment by concatenating @samp{/} and @samp{*}.  You can put as much
1618whitespace between @samp{##} and its operands as you like, including
1619comments, and you can put comments in arguments that will be
1620concatenated.  However, it is an error if @samp{##} appears at either
1621end of a macro body.
1622
1623Consider a C program that interprets named commands.  There probably
1624needs to be a table of commands, perhaps an array of structures declared
1625as follows:
1626
1627@smallexample
1628@group
1629struct command
1630@{
1631  char *name;
1632  void (*function) (void);
1633@};
1634@end group
1635
1636@group
1637struct command commands[] =
1638@{
1639  @{ "quit", quit_command @},
1640  @{ "help", help_command @},
1641  @dots{}
1642@};
1643@end group
1644@end smallexample
1645
1646It would be cleaner not to have to give each command name twice, once in
1647the string constant and once in the function name.  A macro which takes the
1648name of a command as an argument can make this unnecessary.  The string
1649constant can be created with stringification, and the function name by
1650concatenating the argument with @samp{_command}.  Here is how it is done:
1651
1652@smallexample
1653#define COMMAND(NAME)  @{ #NAME, NAME ## _command @}
1654
1655struct command commands[] =
1656@{
1657  COMMAND (quit),
1658  COMMAND (help),
1659  @dots{}
1660@};
1661@end smallexample
1662
1663@node Variadic Macros
1664@section Variadic Macros
1665@cindex variable number of arguments
1666@cindex macros with variable arguments
1667@cindex variadic macros
1668
1669A macro can be declared to accept a variable number of arguments much as
1670a function can.  The syntax for defining the macro is similar to that of
1671a function.  Here is an example:
1672
1673@smallexample
1674#define eprintf(@dots{}) fprintf (stderr, __VA_ARGS__)
1675@end smallexample
1676
1677This kind of macro is called @dfn{variadic}.  When the macro is invoked,
1678all the tokens in its argument list after the last named argument (this
1679macro has none), including any commas, become the @dfn{variable
1680argument}.  This sequence of tokens replaces the identifier
1681@code{@w{__VA_ARGS__}} in the macro body wherever it appears.  Thus, we
1682have this expansion:
1683
1684@smallexample
1685eprintf ("%s:%d: ", input_file, lineno)
1686     @expansion{}  fprintf (stderr, "%s:%d: ", input_file, lineno)
1687@end smallexample
1688
1689The variable argument is completely macro-expanded before it is inserted
1690into the macro expansion, just like an ordinary argument.  You may use
1691the @samp{#} and @samp{##} operators to stringify the variable argument
1692or to paste its leading or trailing token with another token.  (But see
1693below for an important special case for @samp{##}.)
1694
1695If your macro is complicated, you may want a more descriptive name for
1696the variable argument than @code{@w{__VA_ARGS__}}.  CPP permits
1697this, as an extension.  You may write an argument name immediately
1698before the @samp{@dots{}}; that name is used for the variable argument.
1699The @code{eprintf} macro above could be written
1700
1701@smallexample
1702#define eprintf(args@dots{}) fprintf (stderr, args)
1703@end smallexample
1704
1705@noindent
1706using this extension.  You cannot use @code{@w{__VA_ARGS__}} and this
1707extension in the same macro.
1708
1709You can have named arguments as well as variable arguments in a variadic
1710macro.  We could define @code{eprintf} like this, instead:
1711
1712@smallexample
1713#define eprintf(format, @dots{}) fprintf (stderr, format, __VA_ARGS__)
1714@end smallexample
1715
1716@noindent
1717This formulation looks more descriptive, but unfortunately it is less
1718flexible: you must now supply at least one argument after the format
1719string.  In standard C, you cannot omit the comma separating the named
1720argument from the variable arguments.  Furthermore, if you leave the
1721variable argument empty, you will get a syntax error, because
1722there will be an extra comma after the format string.
1723
1724@smallexample
1725eprintf("success!\n", );
1726     @expansion{} fprintf(stderr, "success!\n", );
1727@end smallexample
1728
1729GNU CPP has a pair of extensions which deal with this problem.  First,
1730you are allowed to leave the variable argument out entirely:
1731
1732@smallexample
1733eprintf ("success!\n")
1734     @expansion{} fprintf(stderr, "success!\n", );
1735@end smallexample
1736
1737@noindent
1738Second, the @samp{##} token paste operator has a special meaning when
1739placed between a comma and a variable argument.  If you write
1740
1741@smallexample
1742#define eprintf(format, @dots{}) fprintf (stderr, format, ##__VA_ARGS__)
1743@end smallexample
1744
1745@noindent
1746and the variable argument is left out when the @code{eprintf} macro is
1747used, then the comma before the @samp{##} will be deleted.  This does
1748@emph{not} happen if you pass an empty argument, nor does it happen if
1749the token preceding @samp{##} is anything other than a comma.
1750
1751@smallexample
1752eprintf ("success!\n")
1753     @expansion{} fprintf(stderr, "success!\n");
1754@end smallexample
1755
1756@noindent
1757The above explanation is ambiguous about the case where the only macro
1758parameter is a variable arguments parameter, as it is meaningless to
1759try to distinguish whether no argument at all is an empty argument or
1760a missing argument.  In this case the C99 standard is clear that the
1761comma must remain, however the existing GCC extension used to swallow
1762the comma.  So CPP retains the comma when conforming to a specific C
1763standard, and drops it otherwise.
1764
1765C99 mandates that the only place the identifier @code{@w{__VA_ARGS__}}
1766can appear is in the replacement list of a variadic macro.  It may not
1767be used as a macro name, macro argument name, or within a different type
1768of macro.  It may also be forbidden in open text; the standard is
1769ambiguous.  We recommend you avoid using it except for its defined
1770purpose.
1771
1772Variadic macros are a new feature in C99.  GNU CPP has supported them
1773for a long time, but only with a named variable argument
1774(@samp{args@dots{}}, not @samp{@dots{}} and @code{@w{__VA_ARGS__}}).  If you are
1775concerned with portability to previous versions of GCC, you should use
1776only named variable arguments.  On the other hand, if you are concerned
1777with portability to other conforming implementations of C99, you should
1778use only @code{@w{__VA_ARGS__}}.
1779
1780Previous versions of CPP implemented the comma-deletion extension
1781much more generally.  We have restricted it in this release to minimize
1782the differences from C99.  To get the same effect with both this and
1783previous versions of GCC, the token preceding the special @samp{##} must
1784be a comma, and there must be white space between that comma and
1785whatever comes immediately before it:
1786
1787@smallexample
1788#define eprintf(format, args@dots{}) fprintf (stderr, format , ##args)
1789@end smallexample
1790
1791@noindent
1792@xref{Differences from previous versions}, for the gory details.
1793
1794@node Predefined Macros
1795@section Predefined Macros
1796
1797@cindex predefined macros
1798Several object-like macros are predefined; you use them without
1799supplying their definitions.  They fall into three classes: standard,
1800common, and system-specific.
1801
1802In C++, there is a fourth category, the named operators.  They act like
1803predefined macros, but you cannot undefine them.
1804
1805@menu
1806* Standard Predefined Macros::
1807* Common Predefined Macros::
1808* System-specific Predefined Macros::
1809* C++ Named Operators::
1810@end menu
1811
1812@node Standard Predefined Macros
1813@subsection Standard Predefined Macros
1814@cindex standard predefined macros.
1815
1816The standard predefined macros are specified by the relevant
1817language standards, so they are available with all compilers that
1818implement those standards.  Older compilers may not provide all of
1819them.  Their names all start with double underscores.
1820
1821@table @code
1822@item __FILE__
1823This macro expands to the name of the current input file, in the form of
1824a C string constant.  This is the path by which the preprocessor opened
1825the file, not the short name specified in @samp{#include} or as the
1826input file name argument.  For example,
1827@code{"/usr/local/include/myheader.h"} is a possible expansion of this
1828macro.
1829
1830@item __LINE__
1831This macro expands to the current input line number, in the form of a
1832decimal integer constant.  While we call it a predefined macro, it's
1833a pretty strange macro, since its ``definition'' changes with each
1834new line of source code.
1835@end table
1836
1837@code{__FILE__} and @code{__LINE__} are useful in generating an error
1838message to report an inconsistency detected by the program; the message
1839can state the source line at which the inconsistency was detected.  For
1840example,
1841
1842@smallexample
1843fprintf (stderr, "Internal error: "
1844                 "negative string length "
1845                 "%d at %s, line %d.",
1846         length, __FILE__, __LINE__);
1847@end smallexample
1848
1849An @samp{#include} directive changes the expansions of @code{__FILE__}
1850and @code{__LINE__} to correspond to the included file.  At the end of
1851that file, when processing resumes on the input file that contained
1852the @samp{#include} directive, the expansions of @code{__FILE__} and
1853@code{__LINE__} revert to the values they had before the
1854@samp{#include} (but @code{__LINE__} is then incremented by one as
1855processing moves to the line after the @samp{#include}).
1856
1857A @samp{#line} directive changes @code{__LINE__}, and may change
1858@code{__FILE__} as well.  @xref{Line Control}.
1859
1860C99 introduces @code{__func__}, and GCC has provided @code{__FUNCTION__}
1861for a long time.  Both of these are strings containing the name of the
1862current function (there are slight semantic differences; see the GCC
1863manual).  Neither of them is a macro; the preprocessor does not know the
1864name of the current function.  They tend to be useful in conjunction
1865with @code{__FILE__} and @code{__LINE__}, though.
1866
1867@table @code
1868
1869@item __DATE__
1870This macro expands to a string constant that describes the date on which
1871the preprocessor is being run.  The string constant contains eleven
1872characters and looks like @code{@w{"Feb 12 1996"}}.  If the day of the
1873month is less than 10, it is padded with a space on the left.
1874
1875If GCC cannot determine the current date, it will emit a warning message
1876(once per compilation) and @code{__DATE__} will expand to
1877@code{@w{"??? ?? ????"}}.
1878
1879@item __TIME__
1880This macro expands to a string constant that describes the time at
1881which the preprocessor is being run.  The string constant contains
1882eight characters and looks like @code{"23:59:01"}.
1883
1884If GCC cannot determine the current time, it will emit a warning message
1885(once per compilation) and @code{__TIME__} will expand to
1886@code{"??:??:??"}.
1887
1888@item __STDC__
1889In normal operation, this macro expands to the constant 1, to signify
1890that this compiler conforms to ISO Standard C@.  If GNU CPP is used with
1891a compiler other than GCC, this is not necessarily true; however, the
1892preprocessor always conforms to the standard unless the
1893@option{-traditional-cpp} option is used.
1894
1895This macro is not defined if the @option{-traditional-cpp} option is used.
1896
1897On some hosts, the system compiler uses a different convention, where
1898@code{__STDC__} is normally 0, but is 1 if the user specifies strict
1899conformance to the C Standard.  CPP follows the host convention when
1900processing system header files, but when processing user files
1901@code{__STDC__} is always 1.  This has been reported to cause problems;
1902for instance, some versions of Solaris provide X Windows headers that
1903expect @code{__STDC__} to be either undefined or 1.  @xref{Invocation}.
1904
1905@item __STDC_VERSION__
1906This macro expands to the C Standard's version number, a long integer
1907constant of the form @code{@var{yyyy}@var{mm}L} where @var{yyyy} and
1908@var{mm} are the year and month of the Standard version.  This signifies
1909which version of the C Standard the compiler conforms to.  Like
1910@code{__STDC__}, this is not necessarily accurate for the entire
1911implementation, unless GNU CPP is being used with GCC@.
1912
1913The value @code{199409L} signifies the 1989 C standard as amended in
19141994, which is the current default; the value @code{199901L} signifies
1915the 1999 revision of the C standard.  Support for the 1999 revision is
1916not yet complete.
1917
1918This macro is not defined if the @option{-traditional-cpp} option is
1919used, nor when compiling C++ or Objective-C@.
1920
1921@item __STDC_HOSTED__
1922This macro is defined, with value 1, if the compiler's target is a
1923@dfn{hosted environment}.  A hosted environment has the complete
1924facilities of the standard C library available.
1925
1926@item __cplusplus
1927This macro is defined when the C++ compiler is in use.  You can use
1928@code{__cplusplus} to test whether a header is compiled by a C compiler
1929or a C++ compiler.  This macro is similar to @code{__STDC_VERSION__}, in
1930that it expands to a version number.  Depending on the language standard
1931selected, the value of the macro is @code{199711L}, as mandated by the
19321998 C++ standard; @code{201103L}, per the 2011 C++ standard; an
1933unspecified value strictly larger than @code{201103L} for the experimental
1934languages enabled by @option{-std=c++1y} and @option{-std=gnu++1y}.
1935
1936@item __OBJC__
1937This macro is defined, with value 1, when the Objective-C compiler is in
1938use.  You can use @code{__OBJC__} to test whether a header is compiled
1939by a C compiler or an Objective-C compiler.
1940
1941@item __ASSEMBLER__
1942This macro is defined with value 1 when preprocessing assembly
1943language.
1944
1945@end table
1946
1947@node Common Predefined Macros
1948@subsection Common Predefined Macros
1949@cindex common predefined macros
1950
1951The common predefined macros are GNU C extensions.  They are available
1952with the same meanings regardless of the machine or operating system on
1953which you are using GNU C or GNU Fortran.  Their names all start with
1954double underscores.
1955
1956@table @code
1957
1958@item __COUNTER__
1959This macro expands to sequential integral values starting from 0.  In
1960conjunction with the @code{##} operator, this provides a convenient means to
1961generate unique identifiers.  Care must be taken to ensure that
1962@code{__COUNTER__} is not expanded prior to inclusion of precompiled headers
1963which use it.  Otherwise, the precompiled headers will not be used.
1964
1965@item __GFORTRAN__
1966The GNU Fortran compiler defines this.
1967
1968@item __GNUC__
1969@itemx __GNUC_MINOR__
1970@itemx __GNUC_PATCHLEVEL__
1971These macros are defined by all GNU compilers that use the C
1972preprocessor: C, C++, Objective-C and Fortran.  Their values are the major
1973version, minor version, and patch level of the compiler, as integer
1974constants.  For example, GCC 3.2.1 will define @code{__GNUC__} to 3,
1975@code{__GNUC_MINOR__} to 2, and @code{__GNUC_PATCHLEVEL__} to 1.  These
1976macros are also defined if you invoke the preprocessor directly.
1977
1978@code{__GNUC_PATCHLEVEL__} is new to GCC 3.0; it is also present in the
1979widely-used development snapshots leading up to 3.0 (which identify
1980themselves as GCC 2.96 or 2.97, depending on which snapshot you have).
1981
1982If all you need to know is whether or not your program is being compiled
1983by GCC, or a non-GCC compiler that claims to accept the GNU C dialects,
1984you can simply test @code{__GNUC__}.  If you need to write code
1985which depends on a specific version, you must be more careful.  Each
1986time the minor version is increased, the patch level is reset to zero;
1987each time the major version is increased (which happens rarely), the
1988minor version and patch level are reset.  If you wish to use the
1989predefined macros directly in the conditional, you will need to write it
1990like this:
1991
1992@smallexample
1993/* @r{Test for GCC > 3.2.0} */
1994#if __GNUC__ > 3 || \
1995    (__GNUC__ == 3 && (__GNUC_MINOR__ > 2 || \
1996                       (__GNUC_MINOR__ == 2 && \
1997                        __GNUC_PATCHLEVEL__ > 0))
1998@end smallexample
1999
2000@noindent
2001Another approach is to use the predefined macros to
2002calculate a single number, then compare that against a threshold:
2003
2004@smallexample
2005#define GCC_VERSION (__GNUC__ * 10000 \
2006                     + __GNUC_MINOR__ * 100 \
2007                     + __GNUC_PATCHLEVEL__)
2008@dots{}
2009/* @r{Test for GCC > 3.2.0} */
2010#if GCC_VERSION > 30200
2011@end smallexample
2012
2013@noindent
2014Many people find this form easier to understand.
2015
2016@item __GNUG__
2017The GNU C++ compiler defines this.  Testing it is equivalent to
2018testing @code{@w{(__GNUC__ && __cplusplus)}}.
2019
2020@item __STRICT_ANSI__
2021GCC defines this macro if and only if the @option{-ansi} switch, or a
2022@option{-std} switch specifying strict conformance to some version of ISO C
2023or ISO C++, was specified when GCC was invoked.  It is defined to @samp{1}.
2024This macro exists primarily to direct GNU libc's header files to
2025restrict their definitions to the minimal set found in the 1989 C
2026standard.
2027
2028@item __BASE_FILE__
2029This macro expands to the name of the main input file, in the form
2030of a C string constant.  This is the source file that was specified
2031on the command line of the preprocessor or C compiler.
2032
2033@item __INCLUDE_LEVEL__
2034This macro expands to a decimal integer constant that represents the
2035depth of nesting in include files.  The value of this macro is
2036incremented on every @samp{#include} directive and decremented at the
2037end of every included file.  It starts out at 0, its value within the
2038base file specified on the command line.
2039
2040@item __ELF__
2041This macro is defined if the target uses the ELF object format.
2042
2043@item __VERSION__
2044This macro expands to a string constant which describes the version of
2045the compiler in use.  You should not rely on its contents having any
2046particular form, but it can be counted on to contain at least the
2047release number.
2048
2049@item __OPTIMIZE__
2050@itemx __OPTIMIZE_SIZE__
2051@itemx __NO_INLINE__
2052These macros describe the compilation mode.  @code{__OPTIMIZE__} is
2053defined in all optimizing compilations.  @code{__OPTIMIZE_SIZE__} is
2054defined if the compiler is optimizing for size, not speed.
2055@code{__NO_INLINE__} is defined if no functions will be inlined into
2056their callers (when not optimizing, or when inlining has been
2057specifically disabled by @option{-fno-inline}).
2058
2059These macros cause certain GNU header files to provide optimized
2060definitions, using macros or inline functions, of system library
2061functions.  You should not use these macros in any way unless you make
2062sure that programs will execute with the same effect whether or not they
2063are defined.  If they are defined, their value is 1.
2064
2065@item __GNUC_GNU_INLINE__
2066GCC defines this macro if functions declared @code{inline} will be
2067handled in GCC's traditional gnu90 mode.  Object files will contain
2068externally visible definitions of all functions declared @code{inline}
2069without @code{extern} or @code{static}.  They will not contain any
2070definitions of any functions declared @code{extern inline}.
2071
2072@item __GNUC_STDC_INLINE__
2073GCC defines this macro if functions declared @code{inline} will be
2074handled according to the ISO C99 standard.  Object files will contain
2075externally visible definitions of all functions declared @code{extern
2076inline}.  They will not contain definitions of any functions declared
2077@code{inline} without @code{extern}.
2078
2079If this macro is defined, GCC supports the @code{gnu_inline} function
2080attribute as a way to always get the gnu90 behavior.  Support for
2081this and @code{__GNUC_GNU_INLINE__} was added in GCC 4.1.3.  If
2082neither macro is defined, an older version of GCC is being used:
2083@code{inline} functions will be compiled in gnu90 mode, and the
2084@code{gnu_inline} function attribute will not be recognized.
2085
2086@item __CHAR_UNSIGNED__
2087GCC defines this macro if and only if the data type @code{char} is
2088unsigned on the target machine.  It exists to cause the standard header
2089file @file{limits.h} to work correctly.  You should not use this macro
2090yourself; instead, refer to the standard macros defined in @file{limits.h}.
2091
2092@item __WCHAR_UNSIGNED__
2093Like @code{__CHAR_UNSIGNED__}, this macro is defined if and only if the
2094data type @code{wchar_t} is unsigned and the front-end is in C++ mode.
2095
2096@item __REGISTER_PREFIX__
2097This macro expands to a single token (not a string constant) which is
2098the prefix applied to CPU register names in assembly language for this
2099target.  You can use it to write assembly that is usable in multiple
2100environments.  For example, in the @code{m68k-aout} environment it
2101expands to nothing, but in the @code{m68k-coff} environment it expands
2102to a single @samp{%}.
2103
2104@item __USER_LABEL_PREFIX__
2105This macro expands to a single token which is the prefix applied to
2106user labels (symbols visible to C code) in assembly.  For example, in
2107the @code{m68k-aout} environment it expands to an @samp{_}, but in the
2108@code{m68k-coff} environment it expands to nothing.
2109
2110This macro will have the correct definition even if
2111@option{-f(no-)underscores} is in use, but it will not be correct if
2112target-specific options that adjust this prefix are used (e.g.@: the
2113OSF/rose @option{-mno-underscores} option).
2114
2115@item __SIZE_TYPE__
2116@itemx __PTRDIFF_TYPE__
2117@itemx __WCHAR_TYPE__
2118@itemx __WINT_TYPE__
2119@itemx __INTMAX_TYPE__
2120@itemx __UINTMAX_TYPE__
2121@itemx __SIG_ATOMIC_TYPE__
2122@itemx __INT8_TYPE__
2123@itemx __INT16_TYPE__
2124@itemx __INT32_TYPE__
2125@itemx __INT64_TYPE__
2126@itemx __UINT8_TYPE__
2127@itemx __UINT16_TYPE__
2128@itemx __UINT32_TYPE__
2129@itemx __UINT64_TYPE__
2130@itemx __INT_LEAST8_TYPE__
2131@itemx __INT_LEAST16_TYPE__
2132@itemx __INT_LEAST32_TYPE__
2133@itemx __INT_LEAST64_TYPE__
2134@itemx __UINT_LEAST8_TYPE__
2135@itemx __UINT_LEAST16_TYPE__
2136@itemx __UINT_LEAST32_TYPE__
2137@itemx __UINT_LEAST64_TYPE__
2138@itemx __INT_FAST8_TYPE__
2139@itemx __INT_FAST16_TYPE__
2140@itemx __INT_FAST32_TYPE__
2141@itemx __INT_FAST64_TYPE__
2142@itemx __UINT_FAST8_TYPE__
2143@itemx __UINT_FAST16_TYPE__
2144@itemx __UINT_FAST32_TYPE__
2145@itemx __UINT_FAST64_TYPE__
2146@itemx __INTPTR_TYPE__
2147@itemx __UINTPTR_TYPE__
2148These macros are defined to the correct underlying types for the
2149@code{size_t}, @code{ptrdiff_t}, @code{wchar_t}, @code{wint_t},
2150@code{intmax_t}, @code{uintmax_t}, @code{sig_atomic_t}, @code{int8_t},
2151@code{int16_t}, @code{int32_t}, @code{int64_t}, @code{uint8_t},
2152@code{uint16_t}, @code{uint32_t}, @code{uint64_t},
2153@code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
2154@code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
2155@code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
2156@code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
2157@code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
2158@code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t} typedefs,
2159respectively.  They exist to make the standard header files
2160@file{stddef.h}, @file{stdint.h}, and @file{wchar.h} work correctly.
2161You should not use these macros directly; instead, include the
2162appropriate headers and use the typedefs.  Some of these macros may
2163not be defined on particular systems if GCC does not provide a
2164@file{stdint.h} header on those systems.
2165
2166@item __CHAR_BIT__
2167Defined to the number of bits used in the representation of the
2168@code{char} data type.  It exists to make the standard header given
2169numerical limits work correctly.  You should not use
2170this macro directly; instead, include the appropriate headers.
2171
2172@item __SCHAR_MAX__
2173@itemx __WCHAR_MAX__
2174@itemx __SHRT_MAX__
2175@itemx __INT_MAX__
2176@itemx __LONG_MAX__
2177@itemx __LONG_LONG_MAX__
2178@itemx __WINT_MAX__
2179@itemx __SIZE_MAX__
2180@itemx __PTRDIFF_MAX__
2181@itemx __INTMAX_MAX__
2182@itemx __UINTMAX_MAX__
2183@itemx __SIG_ATOMIC_MAX__
2184@itemx __INT8_MAX__
2185@itemx __INT16_MAX__
2186@itemx __INT32_MAX__
2187@itemx __INT64_MAX__
2188@itemx __UINT8_MAX__
2189@itemx __UINT16_MAX__
2190@itemx __UINT32_MAX__
2191@itemx __UINT64_MAX__
2192@itemx __INT_LEAST8_MAX__
2193@itemx __INT_LEAST16_MAX__
2194@itemx __INT_LEAST32_MAX__
2195@itemx __INT_LEAST64_MAX__
2196@itemx __UINT_LEAST8_MAX__
2197@itemx __UINT_LEAST16_MAX__
2198@itemx __UINT_LEAST32_MAX__
2199@itemx __UINT_LEAST64_MAX__
2200@itemx __INT_FAST8_MAX__
2201@itemx __INT_FAST16_MAX__
2202@itemx __INT_FAST32_MAX__
2203@itemx __INT_FAST64_MAX__
2204@itemx __UINT_FAST8_MAX__
2205@itemx __UINT_FAST16_MAX__
2206@itemx __UINT_FAST32_MAX__
2207@itemx __UINT_FAST64_MAX__
2208@itemx __INTPTR_MAX__
2209@itemx __UINTPTR_MAX__
2210@itemx __WCHAR_MIN__
2211@itemx __WINT_MIN__
2212@itemx __SIG_ATOMIC_MIN__
2213Defined to the maximum value of the @code{signed char}, @code{wchar_t},
2214@code{signed short},
2215@code{signed int}, @code{signed long}, @code{signed long long},
2216@code{wint_t}, @code{size_t}, @code{ptrdiff_t},
2217@code{intmax_t}, @code{uintmax_t}, @code{sig_atomic_t}, @code{int8_t},
2218@code{int16_t}, @code{int32_t}, @code{int64_t}, @code{uint8_t},
2219@code{uint16_t}, @code{uint32_t}, @code{uint64_t},
2220@code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
2221@code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
2222@code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
2223@code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
2224@code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
2225@code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t} types and
2226to the minimum value of the @code{wchar_t}, @code{wint_t}, and
2227@code{sig_atomic_t} types respectively.  They exist to make the
2228standard header given numerical limits work correctly.  You should not
2229use these macros directly; instead, include the appropriate headers.
2230Some of these macros may not be defined on particular systems if GCC
2231does not provide a @file{stdint.h} header on those systems.
2232
2233@item __INT8_C
2234@itemx __INT16_C
2235@itemx __INT32_C
2236@itemx __INT64_C
2237@itemx __UINT8_C
2238@itemx __UINT16_C
2239@itemx __UINT32_C
2240@itemx __UINT64_C
2241@itemx __INTMAX_C
2242@itemx __UINTMAX_C
2243Defined to implementations of the standard @file{stdint.h} macros with
2244the same names without the leading @code{__}.  They exist the make the
2245implementation of that header work correctly.  You should not use
2246these macros directly; instead, include the appropriate headers.  Some
2247of these macros may not be defined on particular systems if GCC does
2248not provide a @file{stdint.h} header on those systems.
2249
2250@item __SIZEOF_INT__
2251@itemx __SIZEOF_LONG__
2252@itemx __SIZEOF_LONG_LONG__
2253@itemx __SIZEOF_SHORT__
2254@itemx __SIZEOF_POINTER__
2255@itemx __SIZEOF_FLOAT__
2256@itemx __SIZEOF_DOUBLE__
2257@itemx __SIZEOF_LONG_DOUBLE__
2258@itemx __SIZEOF_SIZE_T__
2259@itemx __SIZEOF_WCHAR_T__
2260@itemx __SIZEOF_WINT_T__
2261@itemx __SIZEOF_PTRDIFF_T__
2262Defined to the number of bytes of the C standard data types: @code{int},
2263@code{long}, @code{long long}, @code{short}, @code{void *}, @code{float},
2264@code{double}, @code{long double}, @code{size_t}, @code{wchar_t}, @code{wint_t}
2265and @code{ptrdiff_t}.
2266
2267@item __BYTE_ORDER__
2268@itemx __ORDER_LITTLE_ENDIAN__
2269@itemx __ORDER_BIG_ENDIAN__
2270@itemx __ORDER_PDP_ENDIAN__
2271@code{__BYTE_ORDER__} is defined to one of the values
2272@code{__ORDER_LITTLE_ENDIAN__}, @code{__ORDER_BIG_ENDIAN__}, or
2273@code{__ORDER_PDP_ENDIAN__} to reflect the layout of multi-byte and
2274multi-word quantities in memory.  If @code{__BYTE_ORDER__} is equal to
2275@code{__ORDER_LITTLE_ENDIAN__} or @code{__ORDER_BIG_ENDIAN__}, then
2276multi-byte and multi-word quantities are laid out identically: the
2277byte (word) at the lowest address is the least significant or most
2278significant byte (word) of the quantity, respectively.  If
2279@code{__BYTE_ORDER__} is equal to @code{__ORDER_PDP_ENDIAN__}, then
2280bytes in 16-bit words are laid out in a little-endian fashion, whereas
2281the 16-bit subwords of a 32-bit quantity are laid out in big-endian
2282fashion.
2283
2284You should use these macros for testing like this:
2285
2286@smallexample
2287/* @r{Test for a little-endian machine} */
2288#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
2289@end smallexample
2290
2291@item __FLOAT_WORD_ORDER__
2292@code{__FLOAT_WORD_ORDER__} is defined to one of the values
2293@code{__ORDER_LITTLE_ENDIAN__} or @code{__ORDER_BIG_ENDIAN__} to reflect
2294the layout of the words of multi-word floating-point quantities.
2295
2296@item __DEPRECATED
2297This macro is defined, with value 1, when compiling a C++ source file
2298with warnings about deprecated constructs enabled.  These warnings are
2299enabled by default, but can be disabled with @option{-Wno-deprecated}.
2300
2301@item __EXCEPTIONS
2302This macro is defined, with value 1, when compiling a C++ source file
2303with exceptions enabled.  If @option{-fno-exceptions} is used when
2304compiling the file, then this macro is not defined.
2305
2306@item __GXX_RTTI
2307This macro is defined, with value 1, when compiling a C++ source file
2308with runtime type identification enabled.  If @option{-fno-rtti} is
2309used when compiling the file, then this macro is not defined.
2310
2311@item __USING_SJLJ_EXCEPTIONS__
2312This macro is defined, with value 1, if the compiler uses the old
2313mechanism based on @code{setjmp} and @code{longjmp} for exception
2314handling.
2315
2316@item __GXX_EXPERIMENTAL_CXX0X__
2317This macro is defined when compiling a C++ source file with the option
2318@option{-std=c++0x} or @option{-std=gnu++0x}. It indicates that some
2319features likely to be included in C++0x are available. Note that these
2320features are experimental, and may change or be removed in future
2321versions of GCC.
2322
2323@item __GXX_WEAK__
2324This macro is defined when compiling a C++ source file.  It has the
2325value 1 if the compiler will use weak symbols, COMDAT sections, or
2326other similar techniques to collapse symbols with ``vague linkage''
2327that are defined in multiple translation units.  If the compiler will
2328not collapse such symbols, this macro is defined with value 0.  In
2329general, user code should not need to make use of this macro; the
2330purpose of this macro is to ease implementation of the C++ runtime
2331library provided with G++.
2332
2333@item __NEXT_RUNTIME__
2334This macro is defined, with value 1, if (and only if) the NeXT runtime
2335(as in @option{-fnext-runtime}) is in use for Objective-C@.  If the GNU
2336runtime is used, this macro is not defined, so that you can use this
2337macro to determine which runtime (NeXT or GNU) is being used.
2338
2339@item __LP64__
2340@itemx _LP64
2341These macros are defined, with value 1, if (and only if) the compilation
2342is for a target where @code{long int} and pointer both use 64-bits and
2343@code{int} uses 32-bit.
2344
2345@item __SSP__
2346This macro is defined, with value 1, when @option{-fstack-protector} is in
2347use.
2348
2349@item __SSP_ALL__
2350This macro is defined, with value 2, when @option{-fstack-protector-all} is
2351in use.
2352
2353@item __SSP_STRONG__
2354This macro is defined, with value 3, when @option{-fstack-protector-strong} is
2355in use.
2356
2357@item __SSP_EXPLICIT__
2358This macro is defined, with value 4, when @option{-fstack-protector-explicit} is
2359in use.
2360
2361@item __SANITIZE_ADDRESS__
2362This macro is defined, with value 1, when @option{-fsanitize=address}
2363or @option{-fsanitize=kernel-address} are in use.
2364
2365@item __TIMESTAMP__
2366This macro expands to a string constant that describes the date and time
2367of the last modification of the current source file. The string constant
2368contains abbreviated day of the week, month, day of the month, time in
2369hh:mm:ss form, year and looks like @code{@w{"Sun Sep 16 01:03:52 1973"}}.
2370If the day of the month is less than 10, it is padded with a space on the left.
2371
2372If GCC cannot determine the current date, it will emit a warning message
2373(once per compilation) and @code{__TIMESTAMP__} will expand to
2374@code{@w{"??? ??? ?? ??:??:?? ????"}}.
2375
2376@item __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1
2377@itemx __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2
2378@itemx __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4
2379@itemx __GCC_HAVE_SYNC_COMPARE_AND_SWAP_8
2380@itemx __GCC_HAVE_SYNC_COMPARE_AND_SWAP_16
2381These macros are defined when the target processor supports atomic compare
2382and swap operations on operands 1, 2, 4, 8 or 16 bytes in length, respectively.
2383
2384@item __GCC_HAVE_DWARF2_CFI_ASM
2385This macro is defined when the compiler is emitting Dwarf2 CFI directives
2386to the assembler.  When this is defined, it is possible to emit those same
2387directives in inline assembly.
2388
2389@item __FP_FAST_FMA
2390@itemx __FP_FAST_FMAF
2391@itemx __FP_FAST_FMAL
2392These macros are defined with value 1 if the backend supports the
2393@code{fma}, @code{fmaf}, and @code{fmal} builtin functions, so that
2394the include file @file{math.h} can define the macros
2395@code{FP_FAST_FMA}, @code{FP_FAST_FMAF}, and @code{FP_FAST_FMAL}
2396for compatibility with the 1999 C standard.
2397
2398@item __GCC_IEC_559
2399This macro is defined to indicate the intended level of support for
2400IEEE 754 (IEC 60559) floating-point arithmetic.  It expands to a
2401nonnegative integer value.  If 0, it indicates that the combination of
2402the compiler configuration and the command-line options is not
2403intended to support IEEE 754 arithmetic for @code{float} and
2404@code{double} as defined in C99 and C11 Annex F (for example, that the
2405standard rounding modes and exceptions are not supported, or that
2406optimizations are enabled that conflict with IEEE 754 semantics).  If
24071, it indicates that IEEE 754 arithmetic is intended to be supported;
2408this does not mean that all relevant language features are supported
2409by GCC.  If 2 or more, it additionally indicates support for IEEE
2410754-2008 (in particular, that the binary encodings for quiet and
2411signaling NaNs are as specified in IEEE 754-2008).
2412
2413This macro does not indicate the default state of command-line options
2414that control optimizations that C99 and C11 permit to be controlled by
2415standard pragmas, where those standards do not require a particular
2416default state.  It does not indicate whether optimizations respect
2417signaling NaN semantics (the macro for that is
2418@code{__SUPPORT_SNAN__}).  It does not indicate support for decimal
2419floating point or the IEEE 754 binary16 and binary128 types.
2420
2421@item __GCC_IEC_559_COMPLEX
2422This macro is defined to indicate the intended level of support for
2423IEEE 754 (IEC 60559) floating-point arithmetic for complex numbers, as
2424defined in C99 and C11 Annex G.  It expands to a nonnegative integer
2425value.  If 0, it indicates that the combination of the compiler
2426configuration and the command-line options is not intended to support
2427Annex G requirements (for example, because @option{-fcx-limited-range}
2428was used).  If 1 or more, it indicates that it is intended to support
2429those requirements; this does not mean that all relevant language
2430features are supported by GCC.
2431
2432@item __NO_MATH_ERRNO__
2433This macro is defined if @option{-fno-math-errno} is used, or enabled
2434by another option such as @option{-ffast-math} or by default.
2435@end table
2436
2437@node System-specific Predefined Macros
2438@subsection System-specific Predefined Macros
2439
2440@cindex system-specific predefined macros
2441@cindex predefined macros, system-specific
2442@cindex reserved namespace
2443
2444The C preprocessor normally predefines several macros that indicate what
2445type of system and machine is in use.  They are obviously different on
2446each target supported by GCC@.  This manual, being for all systems and
2447machines, cannot tell you what their names are, but you can use
2448@command{cpp -dM} to see them all.  @xref{Invocation}.  All system-specific
2449predefined macros expand to a constant value, so you can test them with
2450either @samp{#ifdef} or @samp{#if}.
2451
2452The C standard requires that all system-specific macros be part of the
2453@dfn{reserved namespace}.  All names which begin with two underscores,
2454or an underscore and a capital letter, are reserved for the compiler and
2455library to use as they wish.  However, historically system-specific
2456macros have had names with no special prefix; for instance, it is common
2457to find @code{unix} defined on Unix systems.  For all such macros, GCC
2458provides a parallel macro with two underscores added at the beginning
2459and the end.  If @code{unix} is defined, @code{__unix__} will be defined
2460too.  There will never be more than two underscores; the parallel of
2461@code{_mips} is @code{__mips__}.
2462
2463When the @option{-ansi} option, or any @option{-std} option that
2464requests strict conformance, is given to the compiler, all the
2465system-specific predefined macros outside the reserved namespace are
2466suppressed.  The parallel macros, inside the reserved namespace, remain
2467defined.
2468
2469We are slowly phasing out all predefined macros which are outside the
2470reserved namespace.  You should never use them in new programs, and we
2471encourage you to correct older code to use the parallel macros whenever
2472you find it.  We don't recommend you use the system-specific macros that
2473are in the reserved namespace, either.  It is better in the long run to
2474check specifically for features you need, using a tool such as
2475@command{autoconf}.
2476
2477@node C++ Named Operators
2478@subsection C++ Named Operators
2479@cindex named operators
2480@cindex C++ named operators
2481@cindex @file{iso646.h}
2482
2483In C++, there are eleven keywords which are simply alternate spellings
2484of operators normally written with punctuation.  These keywords are
2485treated as such even in the preprocessor.  They function as operators in
2486@samp{#if}, and they cannot be defined as macros or poisoned.  In C, you
2487can request that those keywords take their C++ meaning by including
2488@file{iso646.h}.  That header defines each one as a normal object-like
2489macro expanding to the appropriate punctuator.
2490
2491These are the named operators and their corresponding punctuators:
2492
2493@multitable {Named Operator} {Punctuator}
2494@item Named Operator @tab Punctuator
2495@item @code{and}    @tab @code{&&}
2496@item @code{and_eq} @tab @code{&=}
2497@item @code{bitand} @tab @code{&}
2498@item @code{bitor}  @tab @code{|}
2499@item @code{compl}  @tab @code{~}
2500@item @code{not}    @tab @code{!}
2501@item @code{not_eq} @tab @code{!=}
2502@item @code{or}     @tab @code{||}
2503@item @code{or_eq}  @tab @code{|=}
2504@item @code{xor}    @tab @code{^}
2505@item @code{xor_eq} @tab @code{^=}
2506@end multitable
2507
2508@node Undefining and Redefining Macros
2509@section Undefining and Redefining Macros
2510@cindex undefining macros
2511@cindex redefining macros
2512@findex #undef
2513
2514If a macro ceases to be useful, it may be @dfn{undefined} with the
2515@samp{#undef} directive.  @samp{#undef} takes a single argument, the
2516name of the macro to undefine.  You use the bare macro name, even if the
2517macro is function-like.  It is an error if anything appears on the line
2518after the macro name.  @samp{#undef} has no effect if the name is not a
2519macro.
2520
2521@smallexample
2522#define FOO 4
2523x = FOO;        @expansion{} x = 4;
2524#undef FOO
2525x = FOO;        @expansion{} x = FOO;
2526@end smallexample
2527
2528Once a macro has been undefined, that identifier may be @dfn{redefined}
2529as a macro by a subsequent @samp{#define} directive.  The new definition
2530need not have any resemblance to the old definition.
2531
2532However, if an identifier which is currently a macro is redefined, then
2533the new definition must be @dfn{effectively the same} as the old one.
2534Two macro definitions are effectively the same if:
2535@itemize @bullet
2536@item Both are the same type of macro (object- or function-like).
2537@item All the tokens of the replacement list are the same.
2538@item If there are any parameters, they are the same.
2539@item Whitespace appears in the same places in both.  It need not be
2540exactly the same amount of whitespace, though.  Remember that comments
2541count as whitespace.
2542@end itemize
2543
2544@noindent
2545These definitions are effectively the same:
2546@smallexample
2547#define FOUR (2 + 2)
2548#define FOUR         (2    +    2)
2549#define FOUR (2 /* @r{two} */ + 2)
2550@end smallexample
2551@noindent
2552but these are not:
2553@smallexample
2554#define FOUR (2 + 2)
2555#define FOUR ( 2+2 )
2556#define FOUR (2 * 2)
2557#define FOUR(score,and,seven,years,ago) (2 + 2)
2558@end smallexample
2559
2560If a macro is redefined with a definition that is not effectively the
2561same as the old one, the preprocessor issues a warning and changes the
2562macro to use the new definition.  If the new definition is effectively
2563the same, the redefinition is silently ignored.  This allows, for
2564instance, two different headers to define a common macro.  The
2565preprocessor will only complain if the definitions do not match.
2566
2567@node Directives Within Macro Arguments
2568@section Directives Within Macro Arguments
2569@cindex macro arguments and directives
2570
2571Occasionally it is convenient to use preprocessor directives within
2572the arguments of a macro.  The C and C++ standards declare that
2573behavior in these cases is undefined.
2574
2575Versions of CPP prior to 3.2 would reject such constructs with an
2576error message.  This was the only syntactic difference between normal
2577functions and function-like macros, so it seemed attractive to remove
2578this limitation, and people would often be surprised that they could
2579not use macros in this way.  Moreover, sometimes people would use
2580conditional compilation in the argument list to a normal library
2581function like @samp{printf}, only to find that after a library upgrade
2582@samp{printf} had changed to be a function-like macro, and their code
2583would no longer compile.  So from version 3.2 we changed CPP to
2584successfully process arbitrary directives within macro arguments in
2585exactly the same way as it would have processed the directive were the
2586function-like macro invocation not present.
2587
2588If, within a macro invocation, that macro is redefined, then the new
2589definition takes effect in time for argument pre-expansion, but the
2590original definition is still used for argument replacement.  Here is a
2591pathological example:
2592
2593@smallexample
2594#define f(x) x x
2595f (1
2596#undef f
2597#define f 2
2598f)
2599@end smallexample
2600
2601@noindent
2602which expands to
2603
2604@smallexample
26051 2 1 2
2606@end smallexample
2607
2608@noindent
2609with the semantics described above.
2610
2611@node Macro Pitfalls
2612@section Macro Pitfalls
2613@cindex problems with macros
2614@cindex pitfalls of macros
2615
2616In this section we describe some special rules that apply to macros and
2617macro expansion, and point out certain cases in which the rules have
2618counter-intuitive consequences that you must watch out for.
2619
2620@menu
2621* Misnesting::
2622* Operator Precedence Problems::
2623* Swallowing the Semicolon::
2624* Duplication of Side Effects::
2625* Self-Referential Macros::
2626* Argument Prescan::
2627* Newlines in Arguments::
2628@end menu
2629
2630@node Misnesting
2631@subsection Misnesting
2632
2633When a macro is called with arguments, the arguments are substituted
2634into the macro body and the result is checked, together with the rest of
2635the input file, for more macro calls.  It is possible to piece together
2636a macro call coming partially from the macro body and partially from the
2637arguments.  For example,
2638
2639@smallexample
2640#define twice(x) (2*(x))
2641#define call_with_1(x) x(1)
2642call_with_1 (twice)
2643     @expansion{} twice(1)
2644     @expansion{} (2*(1))
2645@end smallexample
2646
2647Macro definitions do not have to have balanced parentheses.  By writing
2648an unbalanced open parenthesis in a macro body, it is possible to create
2649a macro call that begins inside the macro body but ends outside of it.
2650For example,
2651
2652@smallexample
2653#define strange(file) fprintf (file, "%s %d",
2654@dots{}
2655strange(stderr) p, 35)
2656     @expansion{} fprintf (stderr, "%s %d", p, 35)
2657@end smallexample
2658
2659The ability to piece together a macro call can be useful, but the use of
2660unbalanced open parentheses in a macro body is just confusing, and
2661should be avoided.
2662
2663@node Operator Precedence Problems
2664@subsection Operator Precedence Problems
2665@cindex parentheses in macro bodies
2666
2667You may have noticed that in most of the macro definition examples shown
2668above, each occurrence of a macro argument name had parentheses around
2669it.  In addition, another pair of parentheses usually surround the
2670entire macro definition.  Here is why it is best to write macros that
2671way.
2672
2673Suppose you define a macro as follows,
2674
2675@smallexample
2676#define ceil_div(x, y) (x + y - 1) / y
2677@end smallexample
2678
2679@noindent
2680whose purpose is to divide, rounding up.  (One use for this operation is
2681to compute how many @code{int} objects are needed to hold a certain
2682number of @code{char} objects.)  Then suppose it is used as follows:
2683
2684@smallexample
2685a = ceil_div (b & c, sizeof (int));
2686     @expansion{} a = (b & c + sizeof (int) - 1) / sizeof (int);
2687@end smallexample
2688
2689@noindent
2690This does not do what is intended.  The operator-precedence rules of
2691C make it equivalent to this:
2692
2693@smallexample
2694a = (b & (c + sizeof (int) - 1)) / sizeof (int);
2695@end smallexample
2696
2697@noindent
2698What we want is this:
2699
2700@smallexample
2701a = ((b & c) + sizeof (int) - 1)) / sizeof (int);
2702@end smallexample
2703
2704@noindent
2705Defining the macro as
2706
2707@smallexample
2708#define ceil_div(x, y) ((x) + (y) - 1) / (y)
2709@end smallexample
2710
2711@noindent
2712provides the desired result.
2713
2714Unintended grouping can result in another way.  Consider @code{sizeof
2715ceil_div(1, 2)}.  That has the appearance of a C expression that would
2716compute the size of the type of @code{ceil_div (1, 2)}, but in fact it
2717means something very different.  Here is what it expands to:
2718
2719@smallexample
2720sizeof ((1) + (2) - 1) / (2)
2721@end smallexample
2722
2723@noindent
2724This would take the size of an integer and divide it by two.  The
2725precedence rules have put the division outside the @code{sizeof} when it
2726was intended to be inside.
2727
2728Parentheses around the entire macro definition prevent such problems.
2729Here, then, is the recommended way to define @code{ceil_div}:
2730
2731@smallexample
2732#define ceil_div(x, y) (((x) + (y) - 1) / (y))
2733@end smallexample
2734
2735@node Swallowing the Semicolon
2736@subsection Swallowing the Semicolon
2737@cindex semicolons (after macro calls)
2738
2739Often it is desirable to define a macro that expands into a compound
2740statement.  Consider, for example, the following macro, that advances a
2741pointer (the argument @code{p} says where to find it) across whitespace
2742characters:
2743
2744@smallexample
2745#define SKIP_SPACES(p, limit)  \
2746@{ char *lim = (limit);         \
2747  while (p < lim) @{            \
2748    if (*p++ != ' ') @{         \
2749      p--; break; @}@}@}
2750@end smallexample
2751
2752@noindent
2753Here backslash-newline is used to split the macro definition, which must
2754be a single logical line, so that it resembles the way such code would
2755be laid out if not part of a macro definition.
2756
2757A call to this macro might be @code{SKIP_SPACES (p, lim)}.  Strictly
2758speaking, the call expands to a compound statement, which is a complete
2759statement with no need for a semicolon to end it.  However, since it
2760looks like a function call, it minimizes confusion if you can use it
2761like a function call, writing a semicolon afterward, as in
2762@code{SKIP_SPACES (p, lim);}
2763
2764This can cause trouble before @code{else} statements, because the
2765semicolon is actually a null statement.  Suppose you write
2766
2767@smallexample
2768if (*p != 0)
2769  SKIP_SPACES (p, lim);
2770else @dots{}
2771@end smallexample
2772
2773@noindent
2774The presence of two statements---the compound statement and a null
2775statement---in between the @code{if} condition and the @code{else}
2776makes invalid C code.
2777
2778The definition of the macro @code{SKIP_SPACES} can be altered to solve
2779this problem, using a @code{do @dots{} while} statement.  Here is how:
2780
2781@smallexample
2782#define SKIP_SPACES(p, limit)     \
2783do @{ char *lim = (limit);         \
2784     while (p < lim) @{            \
2785       if (*p++ != ' ') @{         \
2786         p--; break; @}@}@}          \
2787while (0)
2788@end smallexample
2789
2790Now @code{SKIP_SPACES (p, lim);} expands into
2791
2792@smallexample
2793do @{@dots{}@} while (0);
2794@end smallexample
2795
2796@noindent
2797which is one statement.  The loop executes exactly once; most compilers
2798generate no extra code for it.
2799
2800@node Duplication of Side Effects
2801@subsection Duplication of Side Effects
2802
2803@cindex side effects (in macro arguments)
2804@cindex unsafe macros
2805Many C programs define a macro @code{min}, for ``minimum'', like this:
2806
2807@smallexample
2808#define min(X, Y)  ((X) < (Y) ? (X) : (Y))
2809@end smallexample
2810
2811When you use this macro with an argument containing a side effect,
2812as shown here,
2813
2814@smallexample
2815next = min (x + y, foo (z));
2816@end smallexample
2817
2818@noindent
2819it expands as follows:
2820
2821@smallexample
2822next = ((x + y) < (foo (z)) ? (x + y) : (foo (z)));
2823@end smallexample
2824
2825@noindent
2826where @code{x + y} has been substituted for @code{X} and @code{foo (z)}
2827for @code{Y}.
2828
2829The function @code{foo} is used only once in the statement as it appears
2830in the program, but the expression @code{foo (z)} has been substituted
2831twice into the macro expansion.  As a result, @code{foo} might be called
2832two times when the statement is executed.  If it has side effects or if
2833it takes a long time to compute, the results might not be what you
2834intended.  We say that @code{min} is an @dfn{unsafe} macro.
2835
2836The best solution to this problem is to define @code{min} in a way that
2837computes the value of @code{foo (z)} only once.  The C language offers
2838no standard way to do this, but it can be done with GNU extensions as
2839follows:
2840
2841@smallexample
2842#define min(X, Y)                \
2843(@{ typeof (X) x_ = (X);          \
2844   typeof (Y) y_ = (Y);          \
2845   (x_ < y_) ? x_ : y_; @})
2846@end smallexample
2847
2848The @samp{(@{ @dots{} @})} notation produces a compound statement that
2849acts as an expression.  Its value is the value of its last statement.
2850This permits us to define local variables and assign each argument to
2851one.  The local variables have underscores after their names to reduce
2852the risk of conflict with an identifier of wider scope (it is impossible
2853to avoid this entirely).  Now each argument is evaluated exactly once.
2854
2855If you do not wish to use GNU C extensions, the only solution is to be
2856careful when @emph{using} the macro @code{min}.  For example, you can
2857calculate the value of @code{foo (z)}, save it in a variable, and use
2858that variable in @code{min}:
2859
2860@smallexample
2861@group
2862#define min(X, Y)  ((X) < (Y) ? (X) : (Y))
2863@dots{}
2864@{
2865  int tem = foo (z);
2866  next = min (x + y, tem);
2867@}
2868@end group
2869@end smallexample
2870
2871@noindent
2872(where we assume that @code{foo} returns type @code{int}).
2873
2874@node Self-Referential Macros
2875@subsection Self-Referential Macros
2876@cindex self-reference
2877
2878A @dfn{self-referential} macro is one whose name appears in its
2879definition.  Recall that all macro definitions are rescanned for more
2880macros to replace.  If the self-reference were considered a use of the
2881macro, it would produce an infinitely large expansion.  To prevent this,
2882the self-reference is not considered a macro call.  It is passed into
2883the preprocessor output unchanged.  Consider an example:
2884
2885@smallexample
2886#define foo (4 + foo)
2887@end smallexample
2888
2889@noindent
2890where @code{foo} is also a variable in your program.
2891
2892Following the ordinary rules, each reference to @code{foo} will expand
2893into @code{(4 + foo)}; then this will be rescanned and will expand into
2894@code{(4 + (4 + foo))}; and so on until the computer runs out of memory.
2895
2896The self-reference rule cuts this process short after one step, at
2897@code{(4 + foo)}.  Therefore, this macro definition has the possibly
2898useful effect of causing the program to add 4 to the value of @code{foo}
2899wherever @code{foo} is referred to.
2900
2901In most cases, it is a bad idea to take advantage of this feature.  A
2902person reading the program who sees that @code{foo} is a variable will
2903not expect that it is a macro as well.  The reader will come across the
2904identifier @code{foo} in the program and think its value should be that
2905of the variable @code{foo}, whereas in fact the value is four greater.
2906
2907One common, useful use of self-reference is to create a macro which
2908expands to itself.  If you write
2909
2910@smallexample
2911#define EPERM EPERM
2912@end smallexample
2913
2914@noindent
2915then the macro @code{EPERM} expands to @code{EPERM}.  Effectively, it is
2916left alone by the preprocessor whenever it's used in running text.  You
2917can tell that it's a macro with @samp{#ifdef}.  You might do this if you
2918want to define numeric constants with an @code{enum}, but have
2919@samp{#ifdef} be true for each constant.
2920
2921If a macro @code{x} expands to use a macro @code{y}, and the expansion of
2922@code{y} refers to the macro @code{x}, that is an @dfn{indirect
2923self-reference} of @code{x}.  @code{x} is not expanded in this case
2924either.  Thus, if we have
2925
2926@smallexample
2927#define x (4 + y)
2928#define y (2 * x)
2929@end smallexample
2930
2931@noindent
2932then @code{x} and @code{y} expand as follows:
2933
2934@smallexample
2935@group
2936x    @expansion{} (4 + y)
2937     @expansion{} (4 + (2 * x))
2938
2939y    @expansion{} (2 * x)
2940     @expansion{} (2 * (4 + y))
2941@end group
2942@end smallexample
2943
2944@noindent
2945Each macro is expanded when it appears in the definition of the other
2946macro, but not when it indirectly appears in its own definition.
2947
2948@node Argument Prescan
2949@subsection Argument Prescan
2950@cindex expansion of arguments
2951@cindex macro argument expansion
2952@cindex prescan of macro arguments
2953
2954Macro arguments are completely macro-expanded before they are
2955substituted into a macro body, unless they are stringified or pasted
2956with other tokens.  After substitution, the entire macro body, including
2957the substituted arguments, is scanned again for macros to be expanded.
2958The result is that the arguments are scanned @emph{twice} to expand
2959macro calls in them.
2960
2961Most of the time, this has no effect.  If the argument contained any
2962macro calls, they are expanded during the first scan.  The result
2963therefore contains no macro calls, so the second scan does not change
2964it.  If the argument were substituted as given, with no prescan, the
2965single remaining scan would find the same macro calls and produce the
2966same results.
2967
2968You might expect the double scan to change the results when a
2969self-referential macro is used in an argument of another macro
2970(@pxref{Self-Referential Macros}): the self-referential macro would be
2971expanded once in the first scan, and a second time in the second scan.
2972However, this is not what happens.  The self-references that do not
2973expand in the first scan are marked so that they will not expand in the
2974second scan either.
2975
2976You might wonder, ``Why mention the prescan, if it makes no difference?
2977And why not skip it and make the preprocessor faster?''  The answer is
2978that the prescan does make a difference in three special cases:
2979
2980@itemize @bullet
2981@item
2982Nested calls to a macro.
2983
2984We say that @dfn{nested} calls to a macro occur when a macro's argument
2985contains a call to that very macro.  For example, if @code{f} is a macro
2986that expects one argument, @code{f (f (1))} is a nested pair of calls to
2987@code{f}.  The desired expansion is made by expanding @code{f (1)} and
2988substituting that into the definition of @code{f}.  The prescan causes
2989the expected result to happen.  Without the prescan, @code{f (1)} itself
2990would be substituted as an argument, and the inner use of @code{f} would
2991appear during the main scan as an indirect self-reference and would not
2992be expanded.
2993
2994@item
2995Macros that call other macros that stringify or concatenate.
2996
2997If an argument is stringified or concatenated, the prescan does not
2998occur.  If you @emph{want} to expand a macro, then stringify or
2999concatenate its expansion, you can do that by causing one macro to call
3000another macro that does the stringification or concatenation.  For
3001instance, if you have
3002
3003@smallexample
3004#define AFTERX(x) X_ ## x
3005#define XAFTERX(x) AFTERX(x)
3006#define TABLESIZE 1024
3007#define BUFSIZE TABLESIZE
3008@end smallexample
3009
3010then @code{AFTERX(BUFSIZE)} expands to @code{X_BUFSIZE}, and
3011@code{XAFTERX(BUFSIZE)} expands to @code{X_1024}.  (Not to
3012@code{X_TABLESIZE}.  Prescan always does a complete expansion.)
3013
3014@item
3015Macros used in arguments, whose expansions contain unshielded commas.
3016
3017This can cause a macro expanded on the second scan to be called with the
3018wrong number of arguments.  Here is an example:
3019
3020@smallexample
3021#define foo  a,b
3022#define bar(x) lose(x)
3023#define lose(x) (1 + (x))
3024@end smallexample
3025
3026We would like @code{bar(foo)} to turn into @code{(1 + (foo))}, which
3027would then turn into @code{(1 + (a,b))}.  Instead, @code{bar(foo)}
3028expands into @code{lose(a,b)}, and you get an error because @code{lose}
3029requires a single argument.  In this case, the problem is easily solved
3030by the same parentheses that ought to be used to prevent misnesting of
3031arithmetic operations:
3032
3033@smallexample
3034#define foo (a,b)
3035@exdent or
3036#define bar(x) lose((x))
3037@end smallexample
3038
3039The extra pair of parentheses prevents the comma in @code{foo}'s
3040definition from being interpreted as an argument separator.
3041
3042@end itemize
3043
3044@node Newlines in Arguments
3045@subsection Newlines in Arguments
3046@cindex newlines in macro arguments
3047
3048The invocation of a function-like macro can extend over many logical
3049lines.  However, in the present implementation, the entire expansion
3050comes out on one line.  Thus line numbers emitted by the compiler or
3051debugger refer to the line the invocation started on, which might be
3052different to the line containing the argument causing the problem.
3053
3054Here is an example illustrating this:
3055
3056@smallexample
3057#define ignore_second_arg(a,b,c) a; c
3058
3059ignore_second_arg (foo (),
3060                   ignored (),
3061                   syntax error);
3062@end smallexample
3063
3064@noindent
3065The syntax error triggered by the tokens @code{syntax error} results in
3066an error message citing line three---the line of ignore_second_arg---
3067even though the problematic code comes from line five.
3068
3069We consider this a bug, and intend to fix it in the near future.
3070
3071@node Conditionals
3072@chapter Conditionals
3073@cindex conditionals
3074
3075A @dfn{conditional} is a directive that instructs the preprocessor to
3076select whether or not to include a chunk of code in the final token
3077stream passed to the compiler.  Preprocessor conditionals can test
3078arithmetic expressions, or whether a name is defined as a macro, or both
3079simultaneously using the special @code{defined} operator.
3080
3081A conditional in the C preprocessor resembles in some ways an @code{if}
3082statement in C, but it is important to understand the difference between
3083them.  The condition in an @code{if} statement is tested during the
3084execution of your program.  Its purpose is to allow your program to
3085behave differently from run to run, depending on the data it is
3086operating on.  The condition in a preprocessing conditional directive is
3087tested when your program is compiled.  Its purpose is to allow different
3088code to be included in the program depending on the situation at the
3089time of compilation.
3090
3091However, the distinction is becoming less clear.  Modern compilers often
3092do test @code{if} statements when a program is compiled, if their
3093conditions are known not to vary at run time, and eliminate code which
3094can never be executed.  If you can count on your compiler to do this,
3095you may find that your program is more readable if you use @code{if}
3096statements with constant conditions (perhaps determined by macros).  Of
3097course, you can only use this to exclude code, not type definitions or
3098other preprocessing directives, and you can only do it if the code
3099remains syntactically valid when it is not to be used.
3100
3101GCC version 3 eliminates this kind of never-executed code even when
3102not optimizing.  Older versions did it only when optimizing.
3103
3104@menu
3105* Conditional Uses::
3106* Conditional Syntax::
3107* Deleted Code::
3108@end menu
3109
3110@node Conditional Uses
3111@section Conditional Uses
3112
3113There are three general reasons to use a conditional.
3114
3115@itemize @bullet
3116@item
3117A program may need to use different code depending on the machine or
3118operating system it is to run on.  In some cases the code for one
3119operating system may be erroneous on another operating system; for
3120example, it might refer to data types or constants that do not exist on
3121the other system.  When this happens, it is not enough to avoid
3122executing the invalid code.  Its mere presence will cause the compiler
3123to reject the program.  With a preprocessing conditional, the offending
3124code can be effectively excised from the program when it is not valid.
3125
3126@item
3127You may want to be able to compile the same source file into two
3128different programs.  One version might make frequent time-consuming
3129consistency checks on its intermediate data, or print the values of
3130those data for debugging, and the other not.
3131
3132@item
3133A conditional whose condition is always false is one way to exclude code
3134from the program but keep it as a sort of comment for future reference.
3135@end itemize
3136
3137Simple programs that do not need system-specific logic or complex
3138debugging hooks generally will not need to use preprocessing
3139conditionals.
3140
3141@node Conditional Syntax
3142@section Conditional Syntax
3143
3144@findex #if
3145A conditional in the C preprocessor begins with a @dfn{conditional
3146directive}: @samp{#if}, @samp{#ifdef} or @samp{#ifndef}.
3147
3148@menu
3149* Ifdef::
3150* If::
3151* Defined::
3152* Else::
3153* Elif::
3154@end menu
3155
3156@node Ifdef
3157@subsection Ifdef
3158@findex #ifdef
3159@findex #endif
3160
3161The simplest sort of conditional is
3162
3163@smallexample
3164@group
3165#ifdef @var{MACRO}
3166
3167@var{controlled text}
3168
3169#endif /* @var{MACRO} */
3170@end group
3171@end smallexample
3172
3173@cindex conditional group
3174This block is called a @dfn{conditional group}.  @var{controlled text}
3175will be included in the output of the preprocessor if and only if
3176@var{MACRO} is defined.  We say that the conditional @dfn{succeeds} if
3177@var{MACRO} is defined, @dfn{fails} if it is not.
3178
3179The @var{controlled text} inside of a conditional can include
3180preprocessing directives.  They are executed only if the conditional
3181succeeds.  You can nest conditional groups inside other conditional
3182groups, but they must be completely nested.  In other words,
3183@samp{#endif} always matches the nearest @samp{#ifdef} (or
3184@samp{#ifndef}, or @samp{#if}).  Also, you cannot start a conditional
3185group in one file and end it in another.
3186
3187Even if a conditional fails, the @var{controlled text} inside it is
3188still run through initial transformations and tokenization.  Therefore,
3189it must all be lexically valid C@.  Normally the only way this matters is
3190that all comments and string literals inside a failing conditional group
3191must still be properly ended.
3192
3193The comment following the @samp{#endif} is not required, but it is a
3194good practice if there is a lot of @var{controlled text}, because it
3195helps people match the @samp{#endif} to the corresponding @samp{#ifdef}.
3196Older programs sometimes put @var{MACRO} directly after the
3197@samp{#endif} without enclosing it in a comment.  This is invalid code
3198according to the C standard.  CPP accepts it with a warning.  It
3199never affects which @samp{#ifndef} the @samp{#endif} matches.
3200
3201@findex #ifndef
3202Sometimes you wish to use some code if a macro is @emph{not} defined.
3203You can do this by writing @samp{#ifndef} instead of @samp{#ifdef}.
3204One common use of @samp{#ifndef} is to include code only the first
3205time a header file is included.  @xref{Once-Only Headers}.
3206
3207Macro definitions can vary between compilations for several reasons.
3208Here are some samples.
3209
3210@itemize @bullet
3211@item
3212Some macros are predefined on each kind of machine
3213(@pxref{System-specific Predefined Macros}).  This allows you to provide
3214code specially tuned for a particular machine.
3215
3216@item
3217System header files define more macros, associated with the features
3218they implement.  You can test these macros with conditionals to avoid
3219using a system feature on a machine where it is not implemented.
3220
3221@item
3222Macros can be defined or undefined with the @option{-D} and @option{-U}
3223command-line options when you compile the program.  You can arrange to
3224compile the same source file into two different programs by choosing a
3225macro name to specify which program you want, writing conditionals to
3226test whether or how this macro is defined, and then controlling the
3227state of the macro with command-line options, perhaps set in the
3228Makefile.  @xref{Invocation}.
3229
3230@item
3231Your program might have a special header file (often called
3232@file{config.h}) that is adjusted when the program is compiled.  It can
3233define or not define macros depending on the features of the system and
3234the desired capabilities of the program.  The adjustment can be
3235automated by a tool such as @command{autoconf}, or done by hand.
3236@end itemize
3237
3238@node If
3239@subsection If
3240
3241The @samp{#if} directive allows you to test the value of an arithmetic
3242expression, rather than the mere existence of one macro.  Its syntax is
3243
3244@smallexample
3245@group
3246#if @var{expression}
3247
3248@var{controlled text}
3249
3250#endif /* @var{expression} */
3251@end group
3252@end smallexample
3253
3254@var{expression} is a C expression of integer type, subject to stringent
3255restrictions.  It may contain
3256
3257@itemize @bullet
3258@item
3259Integer constants.
3260
3261@item
3262Character constants, which are interpreted as they would be in normal
3263code.
3264
3265@item
3266Arithmetic operators for addition, subtraction, multiplication,
3267division, bitwise operations, shifts, comparisons, and logical
3268operations (@code{&&} and @code{||}).  The latter two obey the usual
3269short-circuiting rules of standard C@.
3270
3271@item
3272Macros.  All macros in the expression are expanded before actual
3273computation of the expression's value begins.
3274
3275@item
3276Uses of the @code{defined} operator, which lets you check whether macros
3277are defined in the middle of an @samp{#if}.
3278
3279@item
3280Identifiers that are not macros, which are all considered to be the
3281number zero.  This allows you to write @code{@w{#if MACRO}} instead of
3282@code{@w{#ifdef MACRO}}, if you know that MACRO, when defined, will
3283always have a nonzero value.  Function-like macros used without their
3284function call parentheses are also treated as zero.
3285
3286In some contexts this shortcut is undesirable.  The @option{-Wundef}
3287option causes GCC to warn whenever it encounters an identifier which is
3288not a macro in an @samp{#if}.
3289@end itemize
3290
3291The preprocessor does not know anything about types in the language.
3292Therefore, @code{sizeof} operators are not recognized in @samp{#if}, and
3293neither are @code{enum} constants.  They will be taken as identifiers
3294which are not macros, and replaced by zero.  In the case of
3295@code{sizeof}, this is likely to cause the expression to be invalid.
3296
3297The preprocessor calculates the value of @var{expression}.  It carries
3298out all calculations in the widest integer type known to the compiler;
3299on most machines supported by GCC this is 64 bits.  This is not the same
3300rule as the compiler uses to calculate the value of a constant
3301expression, and may give different results in some cases.  If the value
3302comes out to be nonzero, the @samp{#if} succeeds and the @var{controlled
3303text} is included; otherwise it is skipped.
3304
3305@node Defined
3306@subsection Defined
3307
3308@cindex @code{defined}
3309The special operator @code{defined} is used in @samp{#if} and
3310@samp{#elif} expressions to test whether a certain name is defined as a
3311macro.  @code{defined @var{name}} and @code{defined (@var{name})} are
3312both expressions whose value is 1 if @var{name} is defined as a macro at
3313the current point in the program, and 0 otherwise.  Thus,  @code{@w{#if
3314defined MACRO}} is precisely equivalent to @code{@w{#ifdef MACRO}}.
3315
3316@code{defined} is useful when you wish to test more than one macro for
3317existence at once.  For example,
3318
3319@smallexample
3320#if defined (__vax__) || defined (__ns16000__)
3321@end smallexample
3322
3323@noindent
3324would succeed if either of the names @code{__vax__} or
3325@code{__ns16000__} is defined as a macro.
3326
3327Conditionals written like this:
3328
3329@smallexample
3330#if defined BUFSIZE && BUFSIZE >= 1024
3331@end smallexample
3332
3333@noindent
3334can generally be simplified to just @code{@w{#if BUFSIZE >= 1024}},
3335since if @code{BUFSIZE} is not defined, it will be interpreted as having
3336the value zero.
3337
3338If the @code{defined} operator appears as a result of a macro expansion,
3339the C standard says the behavior is undefined.  GNU cpp treats it as a
3340genuine @code{defined} operator and evaluates it normally.  It will warn
3341wherever your code uses this feature if you use the command-line option
3342@option{-pedantic}, since other compilers may handle it differently.
3343
3344@node Else
3345@subsection Else
3346
3347@findex #else
3348The @samp{#else} directive can be added to a conditional to provide
3349alternative text to be used if the condition fails.  This is what it
3350looks like:
3351
3352@smallexample
3353@group
3354#if @var{expression}
3355@var{text-if-true}
3356#else /* Not @var{expression} */
3357@var{text-if-false}
3358#endif /* Not @var{expression} */
3359@end group
3360@end smallexample
3361
3362@noindent
3363If @var{expression} is nonzero, the @var{text-if-true} is included and
3364the @var{text-if-false} is skipped.  If @var{expression} is zero, the
3365opposite happens.
3366
3367You can use @samp{#else} with @samp{#ifdef} and @samp{#ifndef}, too.
3368
3369@node Elif
3370@subsection Elif
3371
3372@findex #elif
3373One common case of nested conditionals is used to check for more than two
3374possible alternatives.  For example, you might have
3375
3376@smallexample
3377#if X == 1
3378@dots{}
3379#else /* X != 1 */
3380#if X == 2
3381@dots{}
3382#else /* X != 2 */
3383@dots{}
3384#endif /* X != 2 */
3385#endif /* X != 1 */
3386@end smallexample
3387
3388Another conditional directive, @samp{#elif}, allows this to be
3389abbreviated as follows:
3390
3391@smallexample
3392#if X == 1
3393@dots{}
3394#elif X == 2
3395@dots{}
3396#else /* X != 2 and X != 1*/
3397@dots{}
3398#endif /* X != 2 and X != 1*/
3399@end smallexample
3400
3401@samp{#elif} stands for ``else if''.  Like @samp{#else}, it goes in the
3402middle of a conditional group and subdivides it; it does not require a
3403matching @samp{#endif} of its own.  Like @samp{#if}, the @samp{#elif}
3404directive includes an expression to be tested.  The text following the
3405@samp{#elif} is processed only if the original @samp{#if}-condition
3406failed and the @samp{#elif} condition succeeds.
3407
3408More than one @samp{#elif} can go in the same conditional group.  Then
3409the text after each @samp{#elif} is processed only if the @samp{#elif}
3410condition succeeds after the original @samp{#if} and all previous
3411@samp{#elif} directives within it have failed.
3412
3413@samp{#else} is allowed after any number of @samp{#elif} directives, but
3414@samp{#elif} may not follow @samp{#else}.
3415
3416@node Deleted Code
3417@section Deleted Code
3418@cindex commenting out code
3419
3420If you replace or delete a part of the program but want to keep the old
3421code around for future reference, you often cannot simply comment it
3422out.  Block comments do not nest, so the first comment inside the old
3423code will end the commenting-out.  The probable result is a flood of
3424syntax errors.
3425
3426One way to avoid this problem is to use an always-false conditional
3427instead.  For instance, put @code{#if 0} before the deleted code and
3428@code{#endif} after it.  This works even if the code being turned
3429off contains conditionals, but they must be entire conditionals
3430(balanced @samp{#if} and @samp{#endif}).
3431
3432Some people use @code{#ifdef notdef} instead.  This is risky, because
3433@code{notdef} might be accidentally defined as a macro, and then the
3434conditional would succeed.  @code{#if 0} can be counted on to fail.
3435
3436Do not use @code{#if 0} for comments which are not C code.  Use a real
3437comment, instead.  The interior of @code{#if 0} must consist of complete
3438tokens; in particular, single-quote characters must balance.  Comments
3439often contain unbalanced single-quote characters (known in English as
3440apostrophes).  These confuse @code{#if 0}.  They don't confuse
3441@samp{/*}.
3442
3443@node Diagnostics
3444@chapter Diagnostics
3445@cindex diagnostic
3446@cindex reporting errors
3447@cindex reporting warnings
3448
3449@findex #error
3450The directive @samp{#error} causes the preprocessor to report a fatal
3451error.  The tokens forming the rest of the line following @samp{#error}
3452are used as the error message.
3453
3454You would use @samp{#error} inside of a conditional that detects a
3455combination of parameters which you know the program does not properly
3456support.  For example, if you know that the program will not run
3457properly on a VAX, you might write
3458
3459@smallexample
3460@group
3461#ifdef __vax__
3462#error "Won't work on VAXen.  See comments at get_last_object."
3463#endif
3464@end group
3465@end smallexample
3466
3467If you have several configuration parameters that must be set up by
3468the installation in a consistent way, you can use conditionals to detect
3469an inconsistency and report it with @samp{#error}.  For example,
3470
3471@smallexample
3472#if !defined(FOO) && defined(BAR)
3473#error "BAR requires FOO."
3474#endif
3475@end smallexample
3476
3477@findex #warning
3478The directive @samp{#warning} is like @samp{#error}, but causes the
3479preprocessor to issue a warning and continue preprocessing.  The tokens
3480following @samp{#warning} are used as the warning message.
3481
3482You might use @samp{#warning} in obsolete header files, with a message
3483directing the user to the header file which should be used instead.
3484
3485Neither @samp{#error} nor @samp{#warning} macro-expands its argument.
3486Internal whitespace sequences are each replaced with a single space.
3487The line must consist of complete tokens.  It is wisest to make the
3488argument of these directives be a single string constant; this avoids
3489problems with apostrophes and the like.
3490
3491@node Line Control
3492@chapter Line Control
3493@cindex line control
3494
3495The C preprocessor informs the C compiler of the location in your source
3496code where each token came from.  Presently, this is just the file name
3497and line number.  All the tokens resulting from macro expansion are
3498reported as having appeared on the line of the source file where the
3499outermost macro was used.  We intend to be more accurate in the future.
3500
3501If you write a program which generates source code, such as the
3502@command{bison} parser generator, you may want to adjust the preprocessor's
3503notion of the current file name and line number by hand.  Parts of the
3504output from @command{bison} are generated from scratch, other parts come
3505from a standard parser file.  The rest are copied verbatim from
3506@command{bison}'s input.  You would like compiler error messages and
3507symbolic debuggers to be able to refer to @code{bison}'s input file.
3508
3509@findex #line
3510@command{bison} or any such program can arrange this by writing
3511@samp{#line} directives into the output file.  @samp{#line} is a
3512directive that specifies the original line number and source file name
3513for subsequent input in the current preprocessor input file.
3514@samp{#line} has three variants:
3515
3516@table @code
3517@item #line @var{linenum}
3518@var{linenum} is a non-negative decimal integer constant.  It specifies
3519the line number which should be reported for the following line of
3520input.  Subsequent lines are counted from @var{linenum}.
3521
3522@item #line @var{linenum} @var{filename}
3523@var{linenum} is the same as for the first form, and has the same
3524effect.  In addition, @var{filename} is a string constant.  The
3525following line and all subsequent lines are reported to come from the
3526file it specifies, until something else happens to change that.
3527@var{filename} is interpreted according to the normal rules for a string
3528constant: backslash escapes are interpreted.  This is different from
3529@samp{#include}.
3530
3531Previous versions of CPP did not interpret escapes in @samp{#line};
3532we have changed it because the standard requires they be interpreted,
3533and most other compilers do.
3534
3535@item #line @var{anything else}
3536@var{anything else} is checked for macro calls, which are expanded.
3537The result should match one of the above two forms.
3538@end table
3539
3540@samp{#line} directives alter the results of the @code{__FILE__} and
3541@code{__LINE__} predefined macros from that point on.  @xref{Standard
3542Predefined Macros}.  They do not have any effect on @samp{#include}'s
3543idea of the directory containing the current file.  This is a change
3544from GCC 2.95.  Previously, a file reading
3545
3546@smallexample
3547#line 1 "../src/gram.y"
3548#include "gram.h"
3549@end smallexample
3550
3551would search for @file{gram.h} in @file{../src}, then the @option{-I}
3552chain; the directory containing the physical source file would not be
3553searched.  In GCC 3.0 and later, the @samp{#include} is not affected by
3554the presence of a @samp{#line} referring to a different directory.
3555
3556We made this change because the old behavior caused problems when
3557generated source files were transported between machines.  For instance,
3558it is common practice to ship generated parsers with a source release,
3559so that people building the distribution do not need to have yacc or
3560Bison installed.  These files frequently have @samp{#line} directives
3561referring to the directory tree of the system where the distribution was
3562created.  If GCC tries to search for headers in those directories, the
3563build is likely to fail.
3564
3565The new behavior can cause failures too, if the generated file is not
3566in the same directory as its source and it attempts to include a header
3567which would be visible searching from the directory containing the
3568source file.  However, this problem is easily solved with an additional
3569@option{-I} switch on the command line.  The failures caused by the old
3570semantics could sometimes be corrected only by editing the generated
3571files, which is difficult and error-prone.
3572
3573@node Pragmas
3574@chapter Pragmas
3575
3576The @samp{#pragma} directive is the method specified by the C standard
3577for providing additional information to the compiler, beyond what is
3578conveyed in the language itself.  Three forms of this directive
3579(commonly known as @dfn{pragmas}) are specified by the 1999 C standard.
3580A C compiler is free to attach any meaning it likes to other pragmas.
3581
3582GCC has historically preferred to use extensions to the syntax of the
3583language, such as @code{__attribute__}, for this purpose.  However, GCC
3584does define a few pragmas of its own.  These mostly have effects on the
3585entire translation unit or source file.
3586
3587In GCC version 3, all GNU-defined, supported pragmas have been given a
3588@code{GCC} prefix.  This is in line with the @code{STDC} prefix on all
3589pragmas defined by C99.  For backward compatibility, pragmas which were
3590recognized by previous versions are still recognized without the
3591@code{GCC} prefix, but that usage is deprecated.  Some older pragmas are
3592deprecated in their entirety.  They are not recognized with the
3593@code{GCC} prefix.  @xref{Obsolete Features}.
3594
3595@cindex @code{_Pragma}
3596C99 introduces the @code{@w{_Pragma}} operator.  This feature addresses a
3597major problem with @samp{#pragma}: being a directive, it cannot be
3598produced as the result of macro expansion.  @code{@w{_Pragma}} is an
3599operator, much like @code{sizeof} or @code{defined}, and can be embedded
3600in a macro.
3601
3602Its syntax is @code{@w{_Pragma (@var{string-literal})}}, where
3603@var{string-literal} can be either a normal or wide-character string
3604literal.  It is destringized, by replacing all @samp{\\} with a single
3605@samp{\} and all @samp{\"} with a @samp{"}.  The result is then
3606processed as if it had appeared as the right hand side of a
3607@samp{#pragma} directive.  For example,
3608
3609@smallexample
3610_Pragma ("GCC dependency \"parse.y\"")
3611@end smallexample
3612
3613@noindent
3614has the same effect as @code{#pragma GCC dependency "parse.y"}.  The
3615same effect could be achieved using macros, for example
3616
3617@smallexample
3618#define DO_PRAGMA(x) _Pragma (#x)
3619DO_PRAGMA (GCC dependency "parse.y")
3620@end smallexample
3621
3622The standard is unclear on where a @code{_Pragma} operator can appear.
3623The preprocessor does not accept it within a preprocessing conditional
3624directive like @samp{#if}.  To be safe, you are probably best keeping it
3625out of directives other than @samp{#define}, and putting it on a line of
3626its own.
3627
3628This manual documents the pragmas which are meaningful to the
3629preprocessor itself.  Other pragmas are meaningful to the C or C++
3630compilers.  They are documented in the GCC manual.
3631
3632GCC plugins may provide their own pragmas.
3633
3634@ftable @code
3635@item #pragma GCC dependency
3636@code{#pragma GCC dependency} allows you to check the relative dates of
3637the current file and another file.  If the other file is more recent than
3638the current file, a warning is issued.  This is useful if the current
3639file is derived from the other file, and should be regenerated.  The
3640other file is searched for using the normal include search path.
3641Optional trailing text can be used to give more information in the
3642warning message.
3643
3644@smallexample
3645#pragma GCC dependency "parse.y"
3646#pragma GCC dependency "/usr/include/time.h" rerun fixincludes
3647@end smallexample
3648
3649@item #pragma GCC poison
3650Sometimes, there is an identifier that you want to remove completely
3651from your program, and make sure that it never creeps back in.  To
3652enforce this, you can @dfn{poison} the identifier with this pragma.
3653@code{#pragma GCC poison} is followed by a list of identifiers to
3654poison.  If any of those identifiers appears anywhere in the source
3655after the directive, it is a hard error.  For example,
3656
3657@smallexample
3658#pragma GCC poison printf sprintf fprintf
3659sprintf(some_string, "hello");
3660@end smallexample
3661
3662@noindent
3663will produce an error.
3664
3665If a poisoned identifier appears as part of the expansion of a macro
3666which was defined before the identifier was poisoned, it will @emph{not}
3667cause an error.  This lets you poison an identifier without worrying
3668about system headers defining macros that use it.
3669
3670For example,
3671
3672@smallexample
3673#define strrchr rindex
3674#pragma GCC poison rindex
3675strrchr(some_string, 'h');
3676@end smallexample
3677
3678@noindent
3679will not produce an error.
3680
3681@item #pragma GCC system_header
3682This pragma takes no arguments.  It causes the rest of the code in the
3683current file to be treated as if it came from a system header.
3684@xref{System Headers}.
3685
3686@item #pragma GCC warning
3687@itemx #pragma GCC error
3688@code{#pragma GCC warning "message"} causes the preprocessor to issue
3689a warning diagnostic with the text @samp{message}.  The message
3690contained in the pragma must be a single string literal.  Similarly,
3691@code{#pragma GCC error "message"} issues an error message.  Unlike
3692the @samp{#warning} and @samp{#error} directives, these pragmas can be
3693embedded in preprocessor macros using @samp{_Pragma}.
3694
3695@end ftable
3696
3697@node Other Directives
3698@chapter Other Directives
3699
3700@findex #ident
3701@findex #sccs
3702The @samp{#ident} directive takes one argument, a string constant.  On
3703some systems, that string constant is copied into a special segment of
3704the object file.  On other systems, the directive is ignored.  The
3705@samp{#sccs} directive is a synonym for @samp{#ident}.
3706
3707These directives are not part of the C standard, but they are not
3708official GNU extensions either.  What historical information we have
3709been able to find, suggests they originated with System V@.
3710
3711@cindex null directive
3712The @dfn{null directive} consists of a @samp{#} followed by a newline,
3713with only whitespace (including comments) in between.  A null directive
3714is understood as a preprocessing directive but has no effect on the
3715preprocessor output.  The primary significance of the existence of the
3716null directive is that an input line consisting of just a @samp{#} will
3717produce no output, rather than a line of output containing just a
3718@samp{#}.  Supposedly some old C programs contain such lines.
3719
3720@node Preprocessor Output
3721@chapter Preprocessor Output
3722
3723When the C preprocessor is used with the C, C++, or Objective-C
3724compilers, it is integrated into the compiler and communicates a stream
3725of binary tokens directly to the compiler's parser.  However, it can
3726also be used in the more conventional standalone mode, where it produces
3727textual output.
3728@c FIXME: Document the library interface.
3729
3730@cindex output format
3731The output from the C preprocessor looks much like the input, except
3732that all preprocessing directive lines have been replaced with blank
3733lines and all comments with spaces.  Long runs of blank lines are
3734discarded.
3735
3736The ISO standard specifies that it is implementation defined whether a
3737preprocessor preserves whitespace between tokens, or replaces it with
3738e.g.@: a single space.  In GNU CPP, whitespace between tokens is collapsed
3739to become a single space, with the exception that the first token on a
3740non-directive line is preceded with sufficient spaces that it appears in
3741the same column in the preprocessed output that it appeared in the
3742original source file.  This is so the output is easy to read.
3743@xref{Differences from previous versions}.  CPP does not insert any
3744whitespace where there was none in the original source, except where
3745necessary to prevent an accidental token paste.
3746
3747@cindex linemarkers
3748Source file name and line number information is conveyed by lines
3749of the form
3750
3751@smallexample
3752# @var{linenum} @var{filename} @var{flags}
3753@end smallexample
3754
3755@noindent
3756These are called @dfn{linemarkers}.  They are inserted as needed into
3757the output (but never within a string or character constant).  They mean
3758that the following line originated in file @var{filename} at line
3759@var{linenum}.  @var{filename} will never contain any non-printing
3760characters; they are replaced with octal escape sequences.
3761
3762After the file name comes zero or more flags, which are @samp{1},
3763@samp{2}, @samp{3}, or @samp{4}.  If there are multiple flags, spaces
3764separate them.  Here is what the flags mean:
3765
3766@table @samp
3767@item 1
3768This indicates the start of a new file.
3769@item 2
3770This indicates returning to a file (after having included another file).
3771@item 3
3772This indicates that the following text comes from a system header file,
3773so certain warnings should be suppressed.
3774@item 4
3775This indicates that the following text should be treated as being
3776wrapped in an implicit @code{extern "C"} block.
3777@c maybe cross reference NO_IMPLICIT_EXTERN_C
3778@end table
3779
3780As an extension, the preprocessor accepts linemarkers in non-assembler
3781input files.  They are treated like the corresponding @samp{#line}
3782directive, (@pxref{Line Control}), except that trailing flags are
3783permitted, and are interpreted with the meanings described above.  If
3784multiple flags are given, they must be in ascending order.
3785
3786Some directives may be duplicated in the output of the preprocessor.
3787These are @samp{#ident} (always), @samp{#pragma} (only if the
3788preprocessor does not handle the pragma itself), and @samp{#define} and
3789@samp{#undef} (with certain debugging options).  If this happens, the
3790@samp{#} of the directive will always be in the first column, and there
3791will be no space between the @samp{#} and the directive name.  If macro
3792expansion happens to generate tokens which might be mistaken for a
3793duplicated directive, a space will be inserted between the @samp{#} and
3794the directive name.
3795
3796@node Traditional Mode
3797@chapter Traditional Mode
3798
3799Traditional (pre-standard) C preprocessing is rather different from
3800the preprocessing specified by the standard.  When GCC is given the
3801@option{-traditional-cpp} option, it attempts to emulate a traditional
3802preprocessor.
3803
3804GCC versions 3.2 and later only support traditional mode semantics in
3805the preprocessor, and not in the compiler front ends.  This chapter
3806outlines the traditional preprocessor semantics we implemented.
3807
3808The implementation does not correspond precisely to the behavior of
3809earlier versions of GCC, nor to any true traditional preprocessor.
3810After all, inconsistencies among traditional implementations were a
3811major motivation for C standardization.  However, we intend that it
3812should be compatible with true traditional preprocessors in all ways
3813that actually matter.
3814
3815@menu
3816* Traditional lexical analysis::
3817* Traditional macros::
3818* Traditional miscellany::
3819* Traditional warnings::
3820@end menu
3821
3822@node Traditional lexical analysis
3823@section Traditional lexical analysis
3824
3825The traditional preprocessor does not decompose its input into tokens
3826the same way a standards-conforming preprocessor does.  The input is
3827simply treated as a stream of text with minimal internal form.
3828
3829This implementation does not treat trigraphs (@pxref{trigraphs})
3830specially since they were an invention of the standards committee.  It
3831handles arbitrarily-positioned escaped newlines properly and splices
3832the lines as you would expect; many traditional preprocessors did not
3833do this.
3834
3835The form of horizontal whitespace in the input file is preserved in
3836the output.  In particular, hard tabs remain hard tabs.  This can be
3837useful if, for example, you are preprocessing a Makefile.
3838
3839Traditional CPP only recognizes C-style block comments, and treats the
3840@samp{/*} sequence as introducing a comment only if it lies outside
3841quoted text.  Quoted text is introduced by the usual single and double
3842quotes, and also by an initial @samp{<} in a @code{#include}
3843directive.
3844
3845Traditionally, comments are completely removed and are not replaced
3846with a space.  Since a traditional compiler does its own tokenization
3847of the output of the preprocessor, this means that comments can
3848effectively be used as token paste operators.  However, comments
3849behave like separators for text handled by the preprocessor itself,
3850since it doesn't re-lex its input.  For example, in
3851
3852@smallexample
3853#if foo/**/bar
3854@end smallexample
3855
3856@noindent
3857@samp{foo} and @samp{bar} are distinct identifiers and expanded
3858separately if they happen to be macros.  In other words, this
3859directive is equivalent to
3860
3861@smallexample
3862#if foo bar
3863@end smallexample
3864
3865@noindent
3866rather than
3867
3868@smallexample
3869#if foobar
3870@end smallexample
3871
3872Generally speaking, in traditional mode an opening quote need not have
3873a matching closing quote.  In particular, a macro may be defined with
3874replacement text that contains an unmatched quote.  Of course, if you
3875attempt to compile preprocessed output containing an unmatched quote
3876you will get a syntax error.
3877
3878However, all preprocessing directives other than @code{#define}
3879require matching quotes.  For example:
3880
3881@smallexample
3882#define m This macro's fine and has an unmatched quote
3883"/* This is not a comment.  */
3884/* @r{This is a comment.  The following #include directive
3885   is ill-formed.}  */
3886#include <stdio.h
3887@end smallexample
3888
3889Just as for the ISO preprocessor, what would be a closing quote can be
3890escaped with a backslash to prevent the quoted text from closing.
3891
3892@node Traditional macros
3893@section Traditional macros
3894
3895The major difference between traditional and ISO macros is that the
3896former expand to text rather than to a token sequence.  CPP removes
3897all leading and trailing horizontal whitespace from a macro's
3898replacement text before storing it, but preserves the form of internal
3899whitespace.
3900
3901One consequence is that it is legitimate for the replacement text to
3902contain an unmatched quote (@pxref{Traditional lexical analysis}).  An
3903unclosed string or character constant continues into the text
3904following the macro call.  Similarly, the text at the end of a macro's
3905expansion can run together with the text after the macro invocation to
3906produce a single token.
3907
3908Normally comments are removed from the replacement text after the
3909macro is expanded, but if the @option{-CC} option is passed on the
3910command-line comments are preserved.  (In fact, the current
3911implementation removes comments even before saving the macro
3912replacement text, but it careful to do it in such a way that the
3913observed effect is identical even in the function-like macro case.)
3914
3915The ISO stringification operator @samp{#} and token paste operator
3916@samp{##} have no special meaning.  As explained later, an effect
3917similar to these operators can be obtained in a different way.  Macro
3918names that are embedded in quotes, either from the main file or after
3919macro replacement, do not expand.
3920
3921CPP replaces an unquoted object-like macro name with its replacement
3922text, and then rescans it for further macros to replace.  Unlike
3923standard macro expansion, traditional macro expansion has no provision
3924to prevent recursion.  If an object-like macro appears unquoted in its
3925replacement text, it will be replaced again during the rescan pass,
3926and so on @emph{ad infinitum}.  GCC detects when it is expanding
3927recursive macros, emits an error message, and continues after the
3928offending macro invocation.
3929
3930@smallexample
3931#define PLUS +
3932#define INC(x) PLUS+x
3933INC(foo);
3934     @expansion{} ++foo;
3935@end smallexample
3936
3937Function-like macros are similar in form but quite different in
3938behavior to their ISO counterparts.  Their arguments are contained
3939within parentheses, are comma-separated, and can cross physical lines.
3940Commas within nested parentheses are not treated as argument
3941separators.  Similarly, a quote in an argument cannot be left
3942unclosed; a following comma or parenthesis that comes before the
3943closing quote is treated like any other character.  There is no
3944facility for handling variadic macros.
3945
3946This implementation removes all comments from macro arguments, unless
3947the @option{-C} option is given.  The form of all other horizontal
3948whitespace in arguments is preserved, including leading and trailing
3949whitespace.  In particular
3950
3951@smallexample
3952f( )
3953@end smallexample
3954
3955@noindent
3956is treated as an invocation of the macro @samp{f} with a single
3957argument consisting of a single space.  If you want to invoke a
3958function-like macro that takes no arguments, you must not leave any
3959whitespace between the parentheses.
3960
3961If a macro argument crosses a new line, the new line is replaced with
3962a space when forming the argument.  If the previous line contained an
3963unterminated quote, the following line inherits the quoted state.
3964
3965Traditional preprocessors replace parameters in the replacement text
3966with their arguments regardless of whether the parameters are within
3967quotes or not.  This provides a way to stringize arguments.  For
3968example
3969
3970@smallexample
3971#define str(x) "x"
3972str(/* @r{A comment} */some text )
3973     @expansion{} "some text "
3974@end smallexample
3975
3976@noindent
3977Note that the comment is removed, but that the trailing space is
3978preserved.  Here is an example of using a comment to effect token
3979pasting.
3980
3981@smallexample
3982#define suffix(x) foo_/**/x
3983suffix(bar)
3984     @expansion{} foo_bar
3985@end smallexample
3986
3987@node Traditional miscellany
3988@section Traditional miscellany
3989
3990Here are some things to be aware of when using the traditional
3991preprocessor.
3992
3993@itemize @bullet
3994@item
3995Preprocessing directives are recognized only when their leading
3996@samp{#} appears in the first column.  There can be no whitespace
3997between the beginning of the line and the @samp{#}, but whitespace can
3998follow the @samp{#}.
3999
4000@item
4001A true traditional C preprocessor does not recognize @samp{#error} or
4002@samp{#pragma}, and may not recognize @samp{#elif}.  CPP supports all
4003the directives in traditional mode that it supports in ISO mode,
4004including extensions, with the exception that the effects of
4005@samp{#pragma GCC poison} are undefined.
4006
4007@item
4008__STDC__ is not defined.
4009
4010@item
4011If you use digraphs the behavior is undefined.
4012
4013@item
4014If a line that looks like a directive appears within macro arguments,
4015the behavior is undefined.
4016
4017@end itemize
4018
4019@node Traditional warnings
4020@section Traditional warnings
4021You can request warnings about features that did not exist, or worked
4022differently, in traditional C with the @option{-Wtraditional} option.
4023GCC does not warn about features of ISO C which you must use when you
4024are using a conforming compiler, such as the @samp{#} and @samp{##}
4025operators.
4026
4027Presently @option{-Wtraditional} warns about:
4028
4029@itemize @bullet
4030@item
4031Macro parameters that appear within string literals in the macro body.
4032In traditional C macro replacement takes place within string literals,
4033but does not in ISO C@.
4034
4035@item
4036In traditional C, some preprocessor directives did not exist.
4037Traditional preprocessors would only consider a line to be a directive
4038if the @samp{#} appeared in column 1 on the line.  Therefore
4039@option{-Wtraditional} warns about directives that traditional C
4040understands but would ignore because the @samp{#} does not appear as the
4041first character on the line.  It also suggests you hide directives like
4042@samp{#pragma} not understood by traditional C by indenting them.  Some
4043traditional implementations would not recognize @samp{#elif}, so it
4044suggests avoiding it altogether.
4045
4046@item
4047A function-like macro that appears without an argument list.  In some
4048traditional preprocessors this was an error.  In ISO C it merely means
4049that the macro is not expanded.
4050
4051@item
4052The unary plus operator.  This did not exist in traditional C@.
4053
4054@item
4055The @samp{U} and @samp{LL} integer constant suffixes, which were not
4056available in traditional C@.  (Traditional C does support the @samp{L}
4057suffix for simple long integer constants.)  You are not warned about
4058uses of these suffixes in macros defined in system headers.  For
4059instance, @code{UINT_MAX} may well be defined as @code{4294967295U}, but
4060you will not be warned if you use @code{UINT_MAX}.
4061
4062You can usually avoid the warning, and the related warning about
4063constants which are so large that they are unsigned, by writing the
4064integer constant in question in hexadecimal, with no U suffix.  Take
4065care, though, because this gives the wrong result in exotic cases.
4066@end itemize
4067
4068@node Implementation Details
4069@chapter Implementation Details
4070
4071Here we document details of how the preprocessor's implementation
4072affects its user-visible behavior.  You should try to avoid undue
4073reliance on behavior described here, as it is possible that it will
4074change subtly in future implementations.
4075
4076Also documented here are obsolete features and changes from previous
4077versions of CPP@.
4078
4079@menu
4080* Implementation-defined behavior::
4081* Implementation limits::
4082* Obsolete Features::
4083* Differences from previous versions::
4084@end menu
4085
4086@node Implementation-defined behavior
4087@section Implementation-defined behavior
4088@cindex implementation-defined behavior
4089
4090This is how CPP behaves in all the cases which the C standard
4091describes as @dfn{implementation-defined}.  This term means that the
4092implementation is free to do what it likes, but must document its choice
4093and stick to it.
4094@c FIXME: Check the C++ standard for more implementation-defined stuff.
4095
4096@itemize @bullet
4097@need 1000
4098@item The mapping of physical source file multi-byte characters to the
4099execution character set.
4100
4101The input character set can be specified using the
4102@option{-finput-charset} option, while the execution character set may
4103be controlled using the @option{-fexec-charset} and
4104@option{-fwide-exec-charset} options.
4105
4106@item Identifier characters.
4107@anchor{Identifier characters}
4108
4109The C and C++ standards allow identifiers to be composed of @samp{_}
4110and the alphanumeric characters.  C++ and C99 also allow universal
4111character names, and C99 further permits implementation-defined
4112characters.
4113
4114GCC allows the @samp{$} character in identifiers as an extension for
4115most targets.  This is true regardless of the @option{std=} switch,
4116since this extension cannot conflict with standards-conforming
4117programs.  When preprocessing assembler, however, dollars are not
4118identifier characters by default.
4119
4120Currently the targets that by default do not permit @samp{$} are AVR,
4121IP2K, MMIX, MIPS Irix 3, ARM aout, and PowerPC targets for the AIX
4122operating system.
4123
4124You can override the default with @option{-fdollars-in-identifiers} or
4125@option{fno-dollars-in-identifiers}.  @xref{fdollars-in-identifiers}.
4126
4127@item Non-empty sequences of whitespace characters.
4128
4129In textual output, each whitespace sequence is collapsed to a single
4130space.  For aesthetic reasons, the first token on each non-directive
4131line of output is preceded with sufficient spaces that it appears in the
4132same column as it did in the original source file.
4133
4134@item The numeric value of character constants in preprocessor expressions.
4135
4136The preprocessor and compiler interpret character constants in the
4137same way; i.e.@: escape sequences such as @samp{\a} are given the
4138values they would have on the target machine.
4139
4140The compiler evaluates a multi-character character constant a character
4141at a time, shifting the previous value left by the number of bits per
4142target character, and then or-ing in the bit-pattern of the new
4143character truncated to the width of a target character.  The final
4144bit-pattern is given type @code{int}, and is therefore signed,
4145regardless of whether single characters are signed or not (a slight
4146change from versions 3.1 and earlier of GCC)@.  If there are more
4147characters in the constant than would fit in the target @code{int} the
4148compiler issues a warning, and the excess leading characters are
4149ignored.
4150
4151For example, @code{'ab'} for a target with an 8-bit @code{char} would be
4152interpreted as @w{@samp{(int) ((unsigned char) 'a' * 256 + (unsigned char)
4153'b')}}, and @code{'\234a'} as @w{@samp{(int) ((unsigned char) '\234' *
4154256 + (unsigned char) 'a')}}.
4155
4156@item Source file inclusion.
4157
4158For a discussion on how the preprocessor locates header files,
4159@ref{Include Operation}.
4160
4161@item Interpretation of the filename resulting from a macro-expanded
4162@samp{#include} directive.
4163
4164@xref{Computed Includes}.
4165
4166@item Treatment of a @samp{#pragma} directive that after macro-expansion
4167results in a standard pragma.
4168
4169No macro expansion occurs on any @samp{#pragma} directive line, so the
4170question does not arise.
4171
4172Note that GCC does not yet implement any of the standard
4173pragmas.
4174
4175@end itemize
4176
4177@node Implementation limits
4178@section Implementation limits
4179@cindex implementation limits
4180
4181CPP has a small number of internal limits.  This section lists the
4182limits which the C standard requires to be no lower than some minimum,
4183and all the others known.  It is intended that there should be as few limits
4184as possible.  If you encounter an undocumented or inconvenient limit,
4185please report that as a bug.  @xref{Bugs, , Reporting Bugs, gcc, Using
4186the GNU Compiler Collection (GCC)}.
4187
4188Where we say something is limited @dfn{only by available memory}, that
4189means that internal data structures impose no intrinsic limit, and space
4190is allocated with @code{malloc} or equivalent.  The actual limit will
4191therefore depend on many things, such as the size of other things
4192allocated by the compiler at the same time, the amount of memory
4193consumed by other processes on the same computer, etc.
4194
4195@itemize @bullet
4196
4197@item Nesting levels of @samp{#include} files.
4198
4199We impose an arbitrary limit of 200 levels, to avoid runaway recursion.
4200The standard requires at least 15 levels.
4201
4202@item Nesting levels of conditional inclusion.
4203
4204The C standard mandates this be at least 63.  CPP is limited only by
4205available memory.
4206
4207@item Levels of parenthesized expressions within a full expression.
4208
4209The C standard requires this to be at least 63.  In preprocessor
4210conditional expressions, it is limited only by available memory.
4211
4212@item Significant initial characters in an identifier or macro name.
4213
4214The preprocessor treats all characters as significant.  The C standard
4215requires only that the first 63 be significant.
4216
4217@item Number of macros simultaneously defined in a single translation unit.
4218
4219The standard requires at least 4095 be possible.  CPP is limited only
4220by available memory.
4221
4222@item Number of parameters in a macro definition and arguments in a macro call.
4223
4224We allow @code{USHRT_MAX}, which is no smaller than 65,535.  The minimum
4225required by the standard is 127.
4226
4227@item Number of characters on a logical source line.
4228
4229The C standard requires a minimum of 4096 be permitted.  CPP places
4230no limits on this, but you may get incorrect column numbers reported in
4231diagnostics for lines longer than 65,535 characters.
4232
4233@item Maximum size of a source file.
4234
4235The standard does not specify any lower limit on the maximum size of a
4236source file.  GNU cpp maps files into memory, so it is limited by the
4237available address space.  This is generally at least two gigabytes.
4238Depending on the operating system, the size of physical memory may or
4239may not be a limitation.
4240
4241@end itemize
4242
4243@node Obsolete Features
4244@section Obsolete Features
4245
4246CPP has some features which are present mainly for compatibility with
4247older programs.  We discourage their use in new code.  In some cases,
4248we plan to remove the feature in a future version of GCC@.
4249
4250@subsection Assertions
4251@cindex assertions
4252
4253@dfn{Assertions} are a deprecated alternative to macros in writing
4254conditionals to test what sort of computer or system the compiled
4255program will run on.  Assertions are usually predefined, but you can
4256define them with preprocessing directives or command-line options.
4257
4258Assertions were intended to provide a more systematic way to describe
4259the compiler's target system and we added them for compatibility with
4260existing compilers.  In practice they are just as unpredictable as the
4261system-specific predefined macros.  In addition, they are not part of
4262any standard, and only a few compilers support them.
4263Therefore, the use of assertions is @strong{less} portable than the use
4264of system-specific predefined macros.  We recommend you do not use them at
4265all.
4266
4267@cindex predicates
4268An assertion looks like this:
4269
4270@smallexample
4271#@var{predicate} (@var{answer})
4272@end smallexample
4273
4274@noindent
4275@var{predicate} must be a single identifier.  @var{answer} can be any
4276sequence of tokens; all characters are significant except for leading
4277and trailing whitespace, and differences in internal whitespace
4278sequences are ignored.  (This is similar to the rules governing macro
4279redefinition.)  Thus, @code{(x + y)} is different from @code{(x+y)} but
4280equivalent to @code{@w{( x + y )}}.  Parentheses do not nest inside an
4281answer.
4282
4283@cindex testing predicates
4284To test an assertion, you write it in an @samp{#if}.  For example, this
4285conditional succeeds if either @code{vax} or @code{ns16000} has been
4286asserted as an answer for @code{machine}.
4287
4288@smallexample
4289#if #machine (vax) || #machine (ns16000)
4290@end smallexample
4291
4292@noindent
4293You can test whether @emph{any} answer is asserted for a predicate by
4294omitting the answer in the conditional:
4295
4296@smallexample
4297#if #machine
4298@end smallexample
4299
4300@findex #assert
4301Assertions are made with the @samp{#assert} directive.  Its sole
4302argument is the assertion to make, without the leading @samp{#} that
4303identifies assertions in conditionals.
4304
4305@smallexample
4306#assert @var{predicate} (@var{answer})
4307@end smallexample
4308
4309@noindent
4310You may make several assertions with the same predicate and different
4311answers.  Subsequent assertions do not override previous ones for the
4312same predicate.  All the answers for any given predicate are
4313simultaneously true.
4314
4315@cindex assertions, canceling
4316@findex #unassert
4317Assertions can be canceled with the @samp{#unassert} directive.  It
4318has the same syntax as @samp{#assert}.  In that form it cancels only the
4319answer which was specified on the @samp{#unassert} line; other answers
4320for that predicate remain true.  You can cancel an entire predicate by
4321leaving out the answer:
4322
4323@smallexample
4324#unassert @var{predicate}
4325@end smallexample
4326
4327@noindent
4328In either form, if no such assertion has been made, @samp{#unassert} has
4329no effect.
4330
4331You can also make or cancel assertions using command-line options.
4332@xref{Invocation}.
4333
4334@node Differences from previous versions
4335@section Differences from previous versions
4336@cindex differences from previous versions
4337
4338This section details behavior which has changed from previous versions
4339of CPP@.  We do not plan to change it again in the near future, but
4340we do not promise not to, either.
4341
4342The ``previous versions'' discussed here are 2.95 and before.  The
4343behavior of GCC 3.0 is mostly the same as the behavior of the widely
4344used 2.96 and 2.97 development snapshots.  Where there are differences,
4345they generally represent bugs in the snapshots.
4346
4347@itemize @bullet
4348
4349@item -I- deprecated
4350
4351This option has been deprecated in 4.0.  @option{-iquote} is meant to
4352replace the need for this option.
4353
4354@item Order of evaluation of @samp{#} and @samp{##} operators
4355
4356The standard does not specify the order of evaluation of a chain of
4357@samp{##} operators, nor whether @samp{#} is evaluated before, after, or
4358at the same time as @samp{##}.  You should therefore not write any code
4359which depends on any specific ordering.  It is possible to guarantee an
4360ordering, if you need one, by suitable use of nested macros.
4361
4362An example of where this might matter is pasting the arguments @samp{1},
4363@samp{e} and @samp{-2}.  This would be fine for left-to-right pasting,
4364but right-to-left pasting would produce an invalid token @samp{e-2}.
4365
4366GCC 3.0 evaluates @samp{#} and @samp{##} at the same time and strictly
4367left to right.  Older versions evaluated all @samp{#} operators first,
4368then all @samp{##} operators, in an unreliable order.
4369
4370@item The form of whitespace between tokens in preprocessor output
4371
4372@xref{Preprocessor Output}, for the current textual format.  This is
4373also the format used by stringification.  Normally, the preprocessor
4374communicates tokens directly to the compiler's parser, and whitespace
4375does not come up at all.
4376
4377Older versions of GCC preserved all whitespace provided by the user and
4378inserted lots more whitespace of their own, because they could not
4379accurately predict when extra spaces were needed to prevent accidental
4380token pasting.
4381
4382@item Optional argument when invoking rest argument macros
4383
4384As an extension, GCC permits you to omit the variable arguments entirely
4385when you use a variable argument macro.  This is forbidden by the 1999 C
4386standard, and will provoke a pedantic warning with GCC 3.0.  Previous
4387versions accepted it silently.
4388
4389@item @samp{##} swallowing preceding text in rest argument macros
4390
4391Formerly, in a macro expansion, if @samp{##} appeared before a variable
4392arguments parameter, and the set of tokens specified for that argument
4393in the macro invocation was empty, previous versions of CPP would
4394back up and remove the preceding sequence of non-whitespace characters
4395(@strong{not} the preceding token).  This extension is in direct
4396conflict with the 1999 C standard and has been drastically pared back.
4397
4398In the current version of the preprocessor, if @samp{##} appears between
4399a comma and a variable arguments parameter, and the variable argument is
4400omitted entirely, the comma will be removed from the expansion.  If the
4401variable argument is empty, or the token before @samp{##} is not a
4402comma, then @samp{##} behaves as a normal token paste.
4403
4404@item @samp{#line} and @samp{#include}
4405
4406The @samp{#line} directive used to change GCC's notion of the
4407``directory containing the current file'', used by @samp{#include} with
4408a double-quoted header file name.  In 3.0 and later, it does not.
4409@xref{Line Control}, for further explanation.
4410
4411@item Syntax of @samp{#line}
4412
4413In GCC 2.95 and previous, the string constant argument to @samp{#line}
4414was treated the same way as the argument to @samp{#include}: backslash
4415escapes were not honored, and the string ended at the second @samp{"}.
4416This is not compliant with the C standard.  In GCC 3.0, an attempt was
4417made to correct the behavior, so that the string was treated as a real
4418string constant, but it turned out to be buggy.  In 3.1, the bugs have
4419been fixed.  (We are not fixing the bugs in 3.0 because they affect
4420relatively few people and the fix is quite invasive.)
4421
4422@end itemize
4423
4424@node Invocation
4425@chapter Invocation
4426@cindex invocation
4427@cindex command line
4428
4429Most often when you use the C preprocessor you will not have to invoke it
4430explicitly: the C compiler will do so automatically.  However, the
4431preprocessor is sometimes useful on its own.  All the options listed
4432here are also acceptable to the C compiler and have the same meaning,
4433except that the C compiler has different rules for specifying the output
4434file.
4435
4436@emph{Note:} Whether you use the preprocessor by way of @command{gcc}
4437or @command{cpp}, the @dfn{compiler driver} is run first.  This
4438program's purpose is to translate your command into invocations of the
4439programs that do the actual work.  Their command-line interfaces are
4440similar but not identical to the documented interface, and may change
4441without notice.
4442
4443@ignore
4444@c man begin SYNOPSIS
4445cpp [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
4446    [@option{-I}@var{dir}@dots{}] [@option{-iquote}@var{dir}@dots{}]
4447    [@option{-iremap}@var{src}:@var{dst}]
4448    [@option{-W}@var{warn}@dots{}]
4449    [@option{-M}|@option{-MM}] [@option{-MG}] [@option{-MF} @var{filename}]
4450    [@option{-MP}] [@option{-MQ} @var{target}@dots{}]
4451    [@option{-MT} @var{target}@dots{}]
4452    [@option{-P}] [@option{-fno-working-directory}]
4453    [@option{-x} @var{language}] [@option{-std=}@var{standard}]
4454    @var{infile} @var{outfile}
4455
4456Only the most useful options are listed here; see below for the remainder.
4457@c man end
4458@c man begin SEEALSO
4459gpl(7), gfdl(7), fsf-funding(7),
4460gcc(1), as(1), ld(1), and the Info entries for @file{cpp}, @file{gcc}, and
4461@file{binutils}.
4462@c man end
4463@end ignore
4464
4465@c man begin OPTIONS
4466The C preprocessor expects two file names as arguments, @var{infile} and
4467@var{outfile}.  The preprocessor reads @var{infile} together with any
4468other files it specifies with @samp{#include}.  All the output generated
4469by the combined input files is written in @var{outfile}.
4470
4471Either @var{infile} or @var{outfile} may be @option{-}, which as
4472@var{infile} means to read from standard input and as @var{outfile}
4473means to write to standard output.  Also, if either file is omitted, it
4474means the same as if @option{-} had been specified for that file.
4475
4476Unless otherwise noted, or the option ends in @samp{=}, all options
4477which take an argument may have that argument appear either immediately
4478after the option, or with a space between option and argument:
4479@option{-Ifoo} and @option{-I foo} have the same effect.
4480
4481@cindex grouping options
4482@cindex options, grouping
4483Many options have multi-letter names; therefore multiple single-letter
4484options may @emph{not} be grouped: @option{-dM} is very different from
4485@w{@samp{-d -M}}.
4486
4487@cindex options
4488@include cppopts.texi
4489@c man end
4490
4491@node Environment Variables
4492@chapter Environment Variables
4493@cindex environment variables
4494@c man begin ENVIRONMENT
4495
4496This section describes the environment variables that affect how CPP
4497operates.  You can use them to specify directories or prefixes to use
4498when searching for include files, or to control dependency output.
4499
4500Note that you can also specify places to search using options such as
4501@option{-I}, and control dependency output with options like
4502@option{-M} (@pxref{Invocation}).  These take precedence over
4503environment variables, which in turn take precedence over the
4504configuration of GCC@.
4505
4506@include cppenv.texi
4507@c man end
4508
4509@page
4510@include fdl.texi
4511
4512@page
4513@node Index of Directives
4514@unnumbered Index of Directives
4515@printindex fn
4516
4517@node Option Index
4518@unnumbered Option Index
4519@noindent
4520CPP's command-line options and environment variables are indexed here
4521without any initial @samp{-} or @samp{--}.
4522@printindex op
4523
4524@page
4525@node Concept Index
4526@unnumbered Concept Index
4527@printindex cp
4528
4529@bye
4530