cpp.texi revision 90075
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@macro copyrightnotice 10@c man begin COPYRIGHT 11Copyright @copyright{} 1987, 1989, 1991, 1992, 1993, 1994, 1995, 1996, 121997, 1998, 1999, 2000, 2001 13Free 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 1.1 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@end macro 27 28@macro covertexts 29@c man begin COPYRIGHT 30This manual contains no Invariant Sections. The Front-Cover Texts are 31(a) (see below), and the Back-Cover Texts are (b) (see below). 32 33(a) The FSF's Front-Cover Text is: 34 35 A GNU Manual 36 37(b) The FSF's Back-Cover Text is: 38 39 You have freedom to copy and modify this GNU Manual, like GNU 40 software. Copies published by the Free Software Foundation raise 41 funds for GNU development. 42@c man end 43@end macro 44 45@macro gcctabopt{body} 46@code{\body\} 47@end macro 48 49@ifinfo 50@dircategory Programming 51@direntry 52* Cpp: (cpp). The GNU C preprocessor. 53@end direntry 54@end ifinfo 55 56@titlepage 57@title The C Preprocessor 58@subtitle Last revised April 2001 59@subtitle for GCC version 3 60@author Richard M. Stallman 61@author Zachary Weinberg 62@page 63@c There is a fill at the bottom of the page, so we need a filll to 64@c override it. 65@vskip 0pt plus 1filll 66@copyrightnotice{} 67@covertexts{} 68@end titlepage 69@contents 70@page 71 72@node Top 73@top 74The C preprocessor implements the macro language used to transform C, 75C++, and Objective-C programs before they are compiled. It can also be 76useful on its own. 77 78@menu 79* Overview:: 80* Header Files:: 81* Macros:: 82* Conditionals:: 83* Diagnostics:: 84* Line Control:: 85* Pragmas:: 86* Other Directives:: 87* Preprocessor Output:: 88* Traditional Mode:: 89* Implementation Details:: 90* Invocation:: 91* GNU Free Documentation License:: 92* Index of Directives:: 93* Concept Index:: 94 95@detailmenu 96 --- The Detailed Node Listing --- 97 98Overview 99 100* Initial processing:: 101* Tokenization:: 102* The preprocessing language:: 103 104Header Files 105 106* Include Syntax:: 107* Include Operation:: 108* Search Path:: 109* Once-Only Headers:: 110* Computed Includes:: 111* Wrapper Headers:: 112* System Headers:: 113 114Macros 115 116* Object-like Macros:: 117* Function-like Macros:: 118* Macro Arguments:: 119* Stringification:: 120* Concatenation:: 121* Variadic Macros:: 122* Predefined Macros:: 123* Undefining and Redefining Macros:: 124* Macro Pitfalls:: 125 126Predefined Macros 127 128* Standard Predefined Macros:: 129* Common Predefined Macros:: 130* System-specific Predefined Macros:: 131* C++ Named Operators:: 132 133Macro Pitfalls 134 135* Misnesting:: 136* Operator Precedence Problems:: 137* Swallowing the Semicolon:: 138* Duplication of Side Effects:: 139* Self-Referential Macros:: 140* Argument Prescan:: 141* Newlines in Arguments:: 142 143Conditionals 144 145* Conditional Uses:: 146* Conditional Syntax:: 147* Deleted Code:: 148 149Conditional Syntax 150 151* Ifdef:: 152* If:: 153* Defined:: 154* Else:: 155* Elif:: 156 157Implementation Details 158 159* Implementation-defined behavior:: 160* Implementation limits:: 161* Obsolete Features:: 162* Differences from previous versions:: 163 164Obsolete Features 165 166* Assertions:: 167* Obsolete once-only headers:: 168* Miscellaneous obsolete features:: 169 170@end detailmenu 171@end menu 172 173@ifnottex 174@copyrightnotice{} 175@covertexts{} 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} 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=c89} or @option{-std=c99} options, depending 216on which version of the standard you want. To get all the mandatory 217diagnostics, you must also use @option{-pedantic}. @xref{Invocation}. 218@c man end 219 220@menu 221* Initial processing:: 222* Tokenization:: 223* The preprocessing language:: 224@end menu 225 226@node Initial processing 227@section Initial processing 228 229The preprocessor performs a series of textual transformations on its 230input. These happen before all other processing. Conceptually, they 231happen in a rigid order, and the entire file is run through each 232transformation before the next one begins. GNU CPP actually does them 233all at once, for performance reasons. These transformations correspond 234roughly to the first three ``phases of translation'' described in the C 235standard. 236 237@enumerate 238@item 239@cindex character sets 240@cindex line endings 241The input file is read into memory and broken into lines. 242 243GNU CPP expects its input to be a text file, that is, an unstructured 244stream of ASCII characters, with some characters indicating the end of a 245line of text. Extended ASCII character sets, such as ISO Latin-1 or 246Unicode encoded in UTF-8, are also acceptable. Character sets that are 247not strict supersets of seven-bit ASCII will not work. We plan to add 248complete support for international character sets in a future release. 249 250Different systems use different conventions to indicate the end of a 251line. GCC accepts the ASCII control sequences @kbd{LF}, @kbd{@w{CR 252LF}}, @kbd{CR}, and @kbd{@w{LF CR}} as end-of-line markers. The first 253three are the canonical sequences used by Unix, DOS and VMS, and the 254classic Mac OS (before OSX) respectively. You may therefore safely copy 255source code written on any of those systems to a different one and use 256it without conversion. (GCC may lose track of the current line number 257if a file doesn't consistently use one convention, as sometimes happens 258when it is edited on computers with different conventions that share a 259network file system.) @kbd{@w{LF CR}} is included because it has been 260reported as an end-of-line marker under exotic conditions. 261 262If the last line of any input file lacks an end-of-line marker, the end 263of the file is considered to implicitly supply one. The C standard says 264that this condition provokes undefined behavior, so GCC will emit a 265warning message. 266 267@item 268@cindex trigraphs 269If trigraphs are enabled, they are replaced by their corresponding 270single characters. 271 272These are nine three-character sequences, all starting with @samp{??}, 273that are defined by ISO C to stand for single characters. They permit 274obsolete systems that lack some of C's punctuation to use C@. For 275example, @samp{??/} stands for @samp{\}, so @t{'??/n'} is a character 276constant for a newline. By default, GCC ignores trigraphs, but if you 277request a strictly conforming mode with the @option{-std} option, then 278it converts them. 279 280Trigraphs are not popular and many compilers implement them incorrectly. 281Portable code should not rely on trigraphs being either converted or 282ignored. If you use the @option{-Wall} or @option{-Wtrigraphs} options, 283GCC will warn you when a trigraph would change the meaning of your 284program if it were converted. 285 286In a string constant, you can prevent a sequence of question marks from 287being confused with a trigraph by inserting a backslash between the 288question marks. @t{"(??\?)"} is the string @samp{(???)}, not 289@samp{(?]}. Traditional C compilers do not recognize this idiom. 290 291The nine trigraphs and their replacements are 292 293@example 294Trigraph: ??( ??) ??< ??> ??= ??/ ??' ??! ??- 295Replacement: [ ] @{ @} # \ ^ | ~ 296@end example 297 298@item 299@cindex continued lines 300@cindex backslash-newline 301Continued lines are merged into one long line. 302 303A continued line is a line which ends with a backslash, @samp{\}. The 304backslash is removed and the following line is joined with the current 305one. No space is inserted, so you may split a line anywhere, even in 306the middle of a word. (It is generally more readable to split lines 307only at white space.) 308 309The trailing backslash on a continued line is commonly referred to as a 310@dfn{backslash-newline}. 311 312If there is white space between a backslash and the end of a line, that 313is still a continued line. However, as this is usually the result of an 314editing mistake, and many compilers will not accept it as a continued 315line, GCC will warn you about it. 316 317@item 318@cindex comments 319@cindex line comments 320@cindex block comments 321All comments are replaced with single spaces. 322 323There are two kinds of comments. @dfn{Block comments} begin with 324@samp{/*} and continue until the next @samp{*/}. Block comments do not 325nest: 326 327@example 328/* @r{this is} /* @r{one comment} */ @r{text outside comment} 329@end example 330 331@dfn{Line comments} begin with @samp{//} and continue to the end of the 332current line. Line comments do not nest either, but it does not matter, 333because they would end in the same place anyway. 334 335@example 336// @r{this is} // @r{one comment} 337@r{text outside comment} 338@end example 339@end enumerate 340 341It is safe to put line comments inside block comments, or vice versa. 342 343@example 344@group 345/* @r{block comment} 346 // @r{contains line comment} 347 @r{yet more comment} 348 */ @r{outside comment} 349 350// @r{line comment} /* @r{contains block comment} */ 351@end group 352@end example 353 354But beware of commenting out one end of a block comment with a line 355comment. 356 357@example 358@group 359 // @r{l.c.} /* @r{block comment begins} 360 @r{oops! this isn't a comment anymore} */ 361@end group 362@end example 363 364Comments are not recognized within string literals. @t{@w{"/* blah 365*/"}} is the string constant @samp{@w{/* blah */}}, not an empty string. 366 367Line comments are not in the 1989 edition of the C standard, but they 368are recognized by GCC as an extension. In C++ and in the 1999 edition 369of the C standard, they are an official part of the language. 370 371Since these transformations happen before all other processing, you can 372split a line mechanically with backslash-newline anywhere. You can 373comment out the end of a line. You can continue a line comment onto the 374next line with backslash-newline. You can even split @samp{/*}, 375@samp{*/}, and @samp{//} onto multiple lines with backslash-newline. 376For example: 377 378@example 379@group 380/\ 381* 382*/ # /* 383*/ defi\ 384ne FO\ 385O 10\ 38620 387@end group 388@end example 389 390@noindent 391is equivalent to @code{@w{#define FOO 1020}}. All these tricks are 392extremely confusing and should not be used in code intended to be 393readable. 394 395There is no way to prevent a backslash at the end of a line from being 396interpreted as a backslash-newline. 397 398@example 399"foo\\ 400bar" 401@end example 402 403@noindent 404is equivalent to @code{"foo\bar"}, not to @code{"foo\\bar"}. To avoid 405having to worry about this, do not use the deprecated GNU extension 406which permits multi-line strings. Instead, use string literal 407concatenation: 408 409@example 410 "foo\\" 411 "bar" 412@end example 413 414@noindent 415Your program will be more portable this way, too. 416 417@node Tokenization 418@section Tokenization 419 420@cindex tokens 421@cindex preprocessing tokens 422After the textual transformations are finished, the input file is 423converted into a sequence of @dfn{preprocessing tokens}. These mostly 424correspond to the syntactic tokens used by the C compiler, but there are 425a few differences. White space separates tokens; it is not itself a 426token of any kind. Tokens do not have to be separated by white space, 427but it is often necessary to avoid ambiguities. 428 429When faced with a sequence of characters that has more than one possible 430tokenization, the preprocessor is greedy. It always makes each token, 431starting from the left, as big as possible before moving on to the next 432token. For instance, @code{a+++++b} is interpreted as 433@code{@w{a ++ ++ + b}}, not as @code{@w{a ++ + ++ b}}, even though the 434latter tokenization could be part of a valid C program and the former 435could not. 436 437Once the input file is broken into tokens, the token boundaries never 438change, except when the @samp{##} preprocessing operator is used to paste 439tokens together. @xref{Concatenation}. For example, 440 441@example 442@group 443#define foo() bar 444foo()baz 445 @expansion{} bar baz 446@emph{not} 447 @expansion{} barbaz 448@end group 449@end example 450 451The compiler does not re-tokenize the preprocessor's output. Each 452preprocessing token becomes one compiler token. 453 454@cindex identifiers 455Preprocessing tokens fall into five broad classes: identifiers, 456preprocessing numbers, string literals, punctuators, and other. An 457@dfn{identifier} is the same as an identifier in C: any sequence of 458letters, digits, or underscores, which begins with a letter or 459underscore. Keywords of C have no significance to the preprocessor; 460they are ordinary identifiers. You can define a macro whose name is a 461keyword, for instance. The only identifier which can be considered a 462preprocessing keyword is @code{defined}. @xref{Defined}. 463 464This is mostly true of other languages which use the C preprocessor. 465However, a few of the keywords of C++ are significant even in the 466preprocessor. @xref{C++ Named Operators}. 467 468In the 1999 C standard, identifiers may contain letters which are not 469part of the ``basic source character set,'' at the implementation's 470discretion (such as accented Latin letters, Greek letters, or Chinese 471ideograms). This may be done with an extended character set, or the 472@samp{\u} and @samp{\U} escape sequences. GCC does not presently 473implement either feature in the preprocessor or the compiler. 474 475As an extension, GCC treats @samp{$} as a letter. This is for 476compatibility with some systems, such as VMS, where @samp{$} is commonly 477used in system-defined function and object names. @samp{$} is not a 478letter in strictly conforming mode, or if you specify the @option{-$} 479option. @xref{Invocation}. 480 481@cindex numbers 482@cindex preprocessing numbers 483A @dfn{preprocessing number} has a rather bizarre definition. The 484category includes all the normal integer and floating point constants 485one expects of C, but also a number of other things one might not 486initially recognize as a number. Formally, preprocessing numbers begin 487with an optional period, a required decimal digit, and then continue 488with any sequence of letters, digits, underscores, periods, and 489exponents. Exponents are the two-character sequences @samp{e+}, 490@samp{e-}, @samp{E+}, @samp{E-}, @samp{p+}, @samp{p-}, @samp{P+}, and 491@samp{P-}. (The exponents that begin with @samp{p} or @samp{P} are new 492to C99. They are used for hexadecimal floating-point constants.) 493 494The purpose of this unusual definition is to isolate the preprocessor 495from the full complexity of numeric constants. It does not have to 496distinguish between lexically valid and invalid floating-point numbers, 497which is complicated. The definition also permits you to split an 498identifier at any position and get exactly two tokens, which can then be 499pasted back together with the @samp{##} operator. 500 501It's possible for preprocessing numbers to cause programs to be 502misinterpreted. For example, @code{0xE+12} is a preprocessing number 503which does not translate to any valid numeric constant, therefore a 504syntax error. It does not mean @code{@w{0xE + 12}}, which is what you 505might have intended. 506 507@cindex string literals 508@cindex string constants 509@cindex character constants 510@cindex header file names 511@c the @: prevents makeinfo from turning '' into ". 512@dfn{String literals} are string constants, character constants, and 513header file names (the argument of @samp{#include}).@footnote{The C 514standard uses the term @dfn{string literal} to refer only to what we are 515calling @dfn{string constants}.} String constants and character 516constants are straightforward: @t{"@dots{}"} or @t{'@dots{}'}. In 517either case embedded quotes should be escaped with a backslash: 518@t{'\'@:'} is the character constant for @samp{'}. There is no limit on 519the length of a character constant, but the value of a character 520constant that contains more than one character is 521implementation-defined. @xref{Implementation Details}. 522 523Header file names either look like string constants, @t{"@dots{}"}, or are 524written with angle brackets instead, @t{<@dots{}>}. In either case, 525backslash is an ordinary character. There is no way to escape the 526closing quote or angle bracket. The preprocessor looks for the header 527file in different places depending on which form you use. @xref{Include 528Operation}. 529 530In standard C, no string literal may extend past the end of a line. GNU 531CPP accepts multi-line string constants, but not multi-line character 532constants or header file names. This extension is deprecated and will 533be removed in GCC 3.1. You may use continued lines instead, or string 534constant concatenation. @xref{Differences from previous versions}. 535 536@cindex punctuators 537@cindex digraphs 538@cindex alternative tokens 539@dfn{Punctuators} are all the usual bits of punctuation which are 540meaningful to C and C++. All but three of the punctuation characters in 541ASCII are C punctuators. The exceptions are @samp{@@}, @samp{$}, and 542@samp{`}. In addition, all the two- and three-character operators are 543punctuators. There are also six @dfn{digraphs}, which the C++ standard 544calls @dfn{alternative tokens}, which are merely alternate ways to spell 545other punctuators. This is a second attempt to work around missing 546punctuation in obsolete systems. It has no negative side effects, 547unlike trigraphs, but does not cover as much ground. The digraphs and 548their corresponding normal punctuators are: 549 550@example 551Digraph: <% %> <: :> %: %:%: 552Punctuator: @{ @} [ ] # ## 553@end example 554 555@cindex other tokens 556Any other single character is considered ``other.'' It is passed on to 557the preprocessor's output unmolested. The C compiler will almost 558certainly reject source code containing ``other'' tokens. In ASCII, the 559only other characters are @samp{@@}, @samp{$}, @samp{`}, and control 560characters other than NUL (all bits zero). (Note that @samp{$} is 561normally considered a letter.) All characters with the high bit set 562(numeric range 0x7F--0xFF) are also ``other'' in the present 563implementation. This will change when proper support for international 564character sets is added to GCC@. 565 566NUL is a special case because of the high probability that its 567appearance is accidental, and because it may be invisible to the user 568(many terminals do not display NUL at all). Within comments, NULs are 569silently ignored, just as any other character would be. In running 570text, NUL is considered white space. For example, these two directives 571have the same meaning. 572 573@example 574#define X^@@1 575#define X 1 576@end example 577 578@noindent 579(where @samp{^@@} is ASCII NUL)@. Within string or character constants, 580NULs are preserved. In the latter two cases the preprocessor emits a 581warning message. 582 583@node The preprocessing language 584@section The preprocessing language 585@cindex directives 586@cindex preprocessing directives 587@cindex directive line 588@cindex directive name 589 590After tokenization, the stream of tokens may simply be passed straight 591to the compiler's parser. However, if it contains any operations in the 592@dfn{preprocessing language}, it will be transformed first. This stage 593corresponds roughly to the standard's ``translation phase 4'' and is 594what most people think of as the preprocessor's job. 595 596The preprocessing language consists of @dfn{directives} to be executed 597and @dfn{macros} to be expanded. Its primary capabilities are: 598 599@itemize @bullet 600@item 601Inclusion of header files. These are files of declarations that can be 602substituted into your program. 603 604@item 605Macro expansion. You can define @dfn{macros}, which are abbreviations 606for arbitrary fragments of C code. The preprocessor will replace the 607macros with their definitions throughout the program. Some macros are 608automatically defined for you. 609 610@item 611Conditional compilation. You can include or exclude parts of the 612program according to various conditions. 613 614@item 615Line control. If you use a program to combine or rearrange source files 616into an intermediate file which is then compiled, you can use line 617control to inform the compiler where each source line originally came 618from. 619 620@item 621Diagnostics. You can detect problems at compile time and issue errors 622or warnings. 623@end itemize 624 625There are a few more, less useful, features. 626 627Except for expansion of predefined macros, all these operations are 628triggered with @dfn{preprocessing directives}. Preprocessing directives 629are lines in your program that start with @samp{#}. Whitespace is 630allowed before and after the @samp{#}. The @samp{#} is followed by an 631identifier, the @dfn{directive name}. It specifies the operation to 632perform. Directives are commonly referred to as @samp{#@var{name}} 633where @var{name} is the directive name. For example, @samp{#define} is 634the directive that defines a macro. 635 636The @samp{#} which begins a directive cannot come from a macro 637expansion. Also, the directive name is not macro expanded. Thus, if 638@code{foo} is defined as a macro expanding to @code{define}, that does 639not make @samp{#foo} a valid preprocessing directive. 640 641The set of valid directive names is fixed. Programs cannot define new 642preprocessing directives. 643 644Some directives require arguments; these make up the rest of the 645directive line and must be separated from the directive name by 646whitespace. For example, @samp{#define} must be followed by a macro 647name and the intended expansion of the macro. 648 649A preprocessing directive cannot cover more than one line. The line 650may, however, be continued with backslash-newline, or by a block comment 651which extends past the end of the line. In either case, when the 652directive is processed, the continuations have already been merged with 653the first line to make one long line. 654 655@node Header Files 656@chapter Header Files 657 658@cindex header file 659A header file is a file containing C declarations and macro definitions 660(@pxref{Macros}) to be shared between several source files. You request 661the use of a header file in your program by @dfn{including} it, with the 662C preprocessing directive @samp{#include}. 663 664Header files serve two purposes. 665 666@itemize @bullet 667@item 668@cindex system header files 669System header files declare the interfaces to parts of the operating 670system. You include them in your program to supply the definitions and 671declarations you need to invoke system calls and libraries. 672 673@item 674Your own header files contain declarations for interfaces between the 675source files of your program. Each time you have a group of related 676declarations and macro definitions all or most of which are needed in 677several different source files, it is a good idea to create a header 678file for them. 679@end itemize 680 681Including a header file produces the same results as copying the header 682file into each source file that needs it. Such copying would be 683time-consuming and error-prone. With a header file, the related 684declarations appear in only one place. If they need to be changed, they 685can be changed in one place, and programs that include the header file 686will automatically use the new version when next recompiled. The header 687file eliminates the labor of finding and changing all the copies as well 688as the risk that a failure to find one copy will result in 689inconsistencies within a program. 690 691In C, the usual convention is to give header files names that end with 692@file{.h}. It is most portable to use only letters, digits, dashes, and 693underscores in header file names, and at most one dot. 694 695@menu 696* Include Syntax:: 697* Include Operation:: 698* Search Path:: 699* Once-Only Headers:: 700* Computed Includes:: 701* Wrapper Headers:: 702* System Headers:: 703@end menu 704 705@node Include Syntax 706@section Include Syntax 707 708@findex #include 709Both user and system header files are included using the preprocessing 710directive @samp{#include}. It has two variants: 711 712@table @code 713@item #include <@var{file}> 714This variant is used for system header files. It searches for a file 715named @var{file} in a standard list of system directories. You can prepend 716directories to this list with the @option{-I} option (@pxref{Invocation}). 717 718@item #include "@var{file}" 719This variant is used for header files of your own program. It searches 720for a file named @var{file} first in the directory containing the 721current file, then in the same directories used for @code{<@var{file}>}. 722@end table 723 724The argument of @samp{#include}, whether delimited with quote marks or 725angle brackets, behaves like a string constant in that comments are not 726recognized, and macro names are not expanded. Thus, @code{@w{#include 727<x/*y>}} specifies inclusion of a system header file named @file{x/*y}. 728 729However, if backslashes occur within @var{file}, they are considered 730ordinary text characters, not escape characters. None of the character 731escape sequences appropriate to string constants in C are processed. 732Thus, @code{@w{#include "x\n\\y"}} specifies a filename containing three 733backslashes. (Some systems interpret @samp{\} as a pathname separator. 734All of these also interpret @samp{/} the same way. It is most portable 735to use only @samp{/}.) 736 737It is an error if there is anything (other than comments) on the line 738after the file name. 739 740@node Include Operation 741@section Include Operation 742 743The @samp{#include} directive works by directing the C preprocessor to 744scan the specified file as input before continuing with the rest of the 745current file. The output from the preprocessor contains the output 746already generated, followed by the output resulting from the included 747file, followed by the output that comes from the text after the 748@samp{#include} directive. For example, if you have a header file 749@file{header.h} as follows, 750 751@example 752char *test (void); 753@end example 754 755@noindent 756and a main program called @file{program.c} that uses the header file, 757like this, 758 759@example 760int x; 761#include "header.h" 762 763int 764main (void) 765@{ 766 puts (test ()); 767@} 768@end example 769 770@noindent 771the compiler will see the same token stream as it would if 772@file{program.c} read 773 774@example 775int x; 776char *test (void); 777 778int 779main (void) 780@{ 781 puts (test ()); 782@} 783@end example 784 785Included files are not limited to declarations and macro definitions; 786those are merely the typical uses. Any fragment of a C program can be 787included from another file. The include file could even contain the 788beginning of a statement that is concluded in the containing file, or 789the end of a statement that was started in the including file. However, 790a comment or a string or character constant may not start in the 791included file and finish in the including file. An unterminated 792comment, string constant or character constant in an included file is 793considered to end (with an error message) at the end of the file. 794 795To avoid confusion, it is best if header files contain only complete 796syntactic units---function declarations or definitions, type 797declarations, etc. 798 799The line following the @samp{#include} directive is always treated as a 800separate line by the C preprocessor, even if the included file lacks a 801final newline. 802 803@node Search Path 804@section Search Path 805 806GCC looks in several different places for headers. On a normal Unix 807system, if you do not instruct it otherwise, it will look for headers 808requested with @code{@w{#include <@var{file}>}} in: 809 810@example 811/usr/local/include 812/usr/lib/gcc-lib/@var{target}/@var{version}/include 813/usr/@var{target}/include 814/usr/include 815@end example 816 817For C++ programs, it will also look in @file{/usr/include/g++-v3}, 818first. In the above, @var{target} is the canonical name of the system 819GCC was configured to compile code for; often but not always the same as 820the canonical name of the system it runs on. @var{version} is the 821version of GCC in use. 822 823You can add to this list with the @option{-I@var{dir}} command line 824option. All the directories named by @option{-I} are searched, in 825left-to-right order, @emph{before} the default directories. You can 826also prevent GCC from searching any of the default directories with the 827@option{-nostdinc} option. This is useful when you are compiling an 828operating system kernel or some other program that does not use the 829standard C library facilities, or the standard C library itself. 830 831GCC looks for headers requested with @code{@w{#include "@var{file}"}} 832first in the directory containing the current file, then in the same 833places it would have looked for a header requested with angle brackets. 834For example, if @file{/usr/include/sys/stat.h} contains 835@code{@w{#include "types.h"}}, GCC looks for @file{types.h} first in 836@file{/usr/include/sys}, then in its usual search path. 837 838If you name a search directory with @option{-I@var{dir}} that is also a 839system include directory, the @option{-I} wins; the directory will be 840searched according to the @option{-I} ordering, and it will not be 841treated as a system include directory. GCC will warn you when a system 842include directory is hidden in this way. 843 844@samp{#line} (@pxref{Line Control}) does not change GCC's idea of the 845directory containing the current file. 846 847You may put @option{-I-} at any point in your list of @option{-I} options. 848This has two effects. First, directories appearing before the 849@option{-I-} in the list are searched only for headers requested with 850quote marks. Directories after @option{-I-} are searched for all 851headers. Second, the directory containing the current file is not 852searched for anything, unless it happens to be one of the directories 853named by an @option{-I} switch. 854 855@option{-I. -I-} is not the same as no @option{-I} options at all, and does 856not cause the same behavior for @samp{<>} includes that @samp{""} 857includes get with no special options. @option{-I.} searches the 858compiler's current working directory for header files. That may or may 859not be the same as the directory containing the current file. 860 861If you need to look for headers in a directory named @file{-}, write 862@option{-I./-}. 863 864There are several more ways to adjust the header search path. They are 865generally less useful. @xref{Invocation}. 866 867@node Once-Only Headers 868@section Once-Only Headers 869@cindex repeated inclusion 870@cindex including just once 871@cindex wrapper @code{#ifndef} 872 873If a header file happens to be included twice, the compiler will process 874its contents twice. This is very likely to cause an error, e.g.@: when the 875compiler sees the same structure definition twice. Even if it does not, 876it will certainly waste time. 877 878The standard way to prevent this is to enclose the entire real contents 879of the file in a conditional, like this: 880 881@example 882@group 883/* File foo. */ 884#ifndef FILE_FOO_SEEN 885#define FILE_FOO_SEEN 886 887@var{the entire file} 888 889#endif /* !FILE_FOO_SEEN */ 890@end group 891@end example 892 893This construct is commonly known as a @dfn{wrapper #ifndef}. 894When the header is included again, the conditional will be false, 895because @code{FILE_FOO_SEEN} is defined. The preprocessor will skip 896over the entire contents of the file, and the compiler will not see it 897twice. 898 899GNU CPP optimizes even further. It remembers when a header file has a 900wrapper @samp{#ifndef}. If a subsequent @samp{#include} specifies that 901header, and the macro in the @samp{#ifndef} is still defined, it does 902not bother to rescan the file at all. 903 904You can put comments outside the wrapper. They will not interfere with 905this optimization. 906 907@cindex controlling macro 908@cindex guard macro 909The macro @code{FILE_FOO_SEEN} is called the @dfn{controlling macro} or 910@dfn{guard macro}. In a user header file, the macro name should not 911begin with @samp{_}. In a system header file, it should begin with 912@samp{__} to avoid conflicts with user programs. In any kind of header 913file, the macro name should contain the name of the file and some 914additional text, to avoid conflicts with other header files. 915 916@node Computed Includes 917@section Computed Includes 918@cindex computed includes 919@cindex macros in include 920 921Sometimes it is necessary to select one of several different header 922files to be included into your program. They might specify 923configuration parameters to be used on different sorts of operating 924systems, for instance. You could do this with a series of conditionals, 925 926@example 927#if SYSTEM_1 928# include "system_1.h" 929#elif SYSTEM_2 930# include "system_2.h" 931#elif SYSTEM_3 932@dots{} 933#endif 934@end example 935 936That rapidly becomes tedious. Instead, the preprocessor offers the 937ability to use a macro for the header name. This is called a 938@dfn{computed include}. Instead of writing a header name as the direct 939argument of @samp{#include}, you simply put a macro name there instead: 940 941@example 942#define SYSTEM_H "system_1.h" 943@dots{} 944#include SYSTEM_H 945@end example 946 947@noindent 948@code{SYSTEM_H} will be expanded, and the preprocessor will look for 949@file{system_1.h} as if the @samp{#include} had been written that way 950originally. @code{SYSTEM_H} could be defined by your Makefile with a 951@option{-D} option. 952 953You must be careful when you define the macro. @samp{#define} saves 954tokens, not text. The preprocessor has no way of knowing that the macro 955will be used as the argument of @samp{#include}, so it generates 956ordinary tokens, not a header name. This is unlikely to cause problems 957if you use double-quote includes, which are close enough to string 958constants. If you use angle brackets, however, you may have trouble. 959 960The syntax of a computed include is actually a bit more general than the 961above. If the first non-whitespace character after @samp{#include} is 962not @samp{"} or @samp{<}, then the entire line is macro-expanded 963like running text would be. 964 965If the line expands to a single string constant, the contents of that 966string constant are the file to be included. CPP does not re-examine the 967string for embedded quotes, but neither does it process backslash 968escapes in the string. Therefore 969 970@example 971#define HEADER "a\"b" 972#include HEADER 973@end example 974 975@noindent 976looks for a file named @file{a\"b}. CPP searches for the file according 977to the rules for double-quoted includes. 978 979If the line expands to a token stream beginning with a @samp{<} token 980and including a @samp{>} token, then the tokens between the @samp{<} and 981the first @samp{>} are combined to form the filename to be included. 982Any whitespace between tokens is reduced to a single space; then any 983space after the initial @samp{<} is retained, but a trailing space 984before the closing @samp{>} is ignored. CPP searches for the file 985according to the rules for angle-bracket includes. 986 987In either case, if there are any tokens on the line after the file name, 988an error occurs and the directive is not processed. It is also an error 989if the result of expansion does not match either of the two expected 990forms. 991 992These rules are implementation-defined behavior according to the C 993standard. To minimize the risk of different compilers interpreting your 994computed includes differently, we recommend you use only a single 995object-like macro which expands to a string constant. This will also 996minimize confusion for people reading your program. 997 998@node Wrapper Headers 999@section Wrapper Headers 1000@cindex wrapper headers 1001@cindex overriding a header file 1002@findex #include_next 1003 1004Sometimes it is necessary to adjust the contents of a system-provided 1005header file without editing it directly. GCC's @command{fixincludes} 1006operation does this, for example. One way to do that would be to create 1007a new header file with the same name and insert it in the search path 1008before the original header. That works fine as long as you're willing 1009to replace the old header entirely. But what if you want to refer to 1010the old header from the new one? 1011 1012You cannot simply include the old header with @samp{#include}. That 1013will start from the beginning, and find your new header again. If your 1014header is not protected from multiple inclusion (@pxref{Once-Only 1015Headers}), it will recurse infinitely and cause a fatal error. 1016 1017You could include the old header with an absolute pathname: 1018@example 1019#include "/usr/include/old-header.h" 1020@end example 1021@noindent 1022This works, but is not clean; should the system headers ever move, you 1023would have to edit the new headers to match. 1024 1025There is no way to solve this problem within the C standard, but you can 1026use the GNU extension @samp{#include_next}. It means, ``Include the 1027@emph{next} file with this name.'' This directive works like 1028@samp{#include} except in searching for the specified file: it starts 1029searching the list of header file directories @emph{after} the directory 1030in which the current file was found. 1031 1032Suppose you specify @option{-I /usr/local/include}, and the list of 1033directories to search also includes @file{/usr/include}; and suppose 1034both directories contain @file{signal.h}. Ordinary @code{@w{#include 1035<signal.h>}} finds the file under @file{/usr/local/include}. If that 1036file contains @code{@w{#include_next <signal.h>}}, it starts searching 1037after that directory, and finds the file in @file{/usr/include}. 1038 1039@samp{#include_next} does not distinguish between @code{<@var{file}>} 1040and @code{"@var{file}"} inclusion, nor does it check that the file you 1041specify has the same name as the current file. It simply looks for the 1042file named, starting with the directory in the search path after the one 1043where the current file was found. 1044 1045The use of @samp{#include_next} can lead to great confusion. We 1046recommend it be used only when there is no other alternative. In 1047particular, it should not be used in the headers belonging to a specific 1048program; it should be used only to make global corrections along the 1049lines of @command{fixincludes}. 1050 1051@node System Headers 1052@section System Headers 1053@cindex system header files 1054 1055The header files declaring interfaces to the operating system and 1056runtime libraries often cannot be written in strictly conforming C@. 1057Therefore, GCC gives code found in @dfn{system headers} special 1058treatment. All warnings, other than those generated by @samp{#warning} 1059(@pxref{Diagnostics}), are suppressed while GCC is processing a system 1060header. Macros defined in a system header are immune to a few warnings 1061wherever they are expanded. This immunity is granted on an ad-hoc 1062basis, when we find that a warning generates lots of false positives 1063because of code in macros defined in system headers. 1064 1065Normally, only the headers found in specific directories are considered 1066system headers. These directories are determined when GCC is compiled. 1067There are, however, two ways to make normal headers into system headers. 1068 1069The @option{-isystem} command line option adds its argument to the list of 1070directories to search for headers, just like @option{-I}. Any headers 1071found in that directory will be considered system headers. 1072 1073All directories named by @option{-isystem} are searched @emph{after} all 1074directories named by @option{-I}, no matter what their order was on the 1075command line. If the same directory is named by both @option{-I} and 1076@option{-isystem}, @option{-I} wins; it is as if the @option{-isystem} option 1077had never been specified at all. GCC warns you when this happens. 1078 1079@findex #pragma GCC system_header 1080There is also a directive, @code{@w{#pragma GCC system_header}}, which 1081tells GCC to consider the rest of the current include file a system 1082header, no matter where it was found. Code that comes before the 1083@samp{#pragma} in the file will not be affected. @code{@w{#pragma GCC 1084system_header}} has no effect in the primary source file. 1085 1086On very old systems, some of the pre-defined system header directories 1087get even more special treatment. GNU C++ considers code in headers 1088found in those directories to be surrounded by an @code{@w{extern "C"}} 1089block. There is no way to request this behavior with a @samp{#pragma}, 1090or from the command line. 1091 1092@node Macros 1093@chapter Macros 1094 1095A @dfn{macro} is a fragment of code which has been given a name. 1096Whenever the name is used, it is replaced by the contents of the macro. 1097There are two kinds of macros. They differ mostly in what they look 1098like when they are used. @dfn{Object-like} macros resemble data objects 1099when used, @dfn{function-like} macros resemble function calls. 1100 1101You may define any valid identifier as a macro, even if it is a C 1102keyword. The preprocessor does not know anything about keywords. This 1103can be useful if you wish to hide a keyword such as @code{const} from an 1104older compiler that does not understand it. However, the preprocessor 1105operator @code{defined} (@pxref{Defined}) can never be defined as a 1106macro, and C++'s named operators (@pxref{C++ Named Operators}) cannot be 1107macros when you are compiling C++. 1108 1109@menu 1110* Object-like Macros:: 1111* Function-like Macros:: 1112* Macro Arguments:: 1113* Stringification:: 1114* Concatenation:: 1115* Variadic Macros:: 1116* Predefined Macros:: 1117* Undefining and Redefining Macros:: 1118* Macro Pitfalls:: 1119@end menu 1120 1121@node Object-like Macros 1122@section Object-like Macros 1123@cindex object-like macro 1124@cindex symbolic constants 1125@cindex manifest constants 1126 1127An @dfn{object-like macro} is a simple identifier which will be replaced 1128by a code fragment. It is called object-like because it looks like a 1129data object in code that uses it. They are most commonly used to give 1130symbolic names to numeric constants. 1131 1132@findex #define 1133You create macros with the @samp{#define} directive. @samp{#define} is 1134followed by the name of the macro and then the token sequence it should 1135be an abbreviation for, which is variously referred to as the macro's 1136@dfn{body}, @dfn{expansion} or @dfn{replacement list}. For example, 1137 1138@example 1139#define BUFFER_SIZE 1024 1140@end example 1141 1142@noindent 1143defines a macro named @code{BUFFER_SIZE} as an abbreviation for the 1144token @code{1024}. If somewhere after this @samp{#define} directive 1145there comes a C statement of the form 1146 1147@example 1148foo = (char *) malloc (BUFFER_SIZE); 1149@end example 1150 1151@noindent 1152then the C preprocessor will recognize and @dfn{expand} the macro 1153@code{BUFFER_SIZE}. The C compiler will see the same tokens as it would 1154if you had written 1155 1156@example 1157foo = (char *) malloc (1024); 1158@end example 1159 1160By convention, macro names are written in upper case. Programs are 1161easier to read when it is possible to tell at a glance which names are 1162macros. 1163 1164The macro's body ends at the end of the @samp{#define} line. You may 1165continue the definition onto multiple lines, if necessary, using 1166backslash-newline. When the macro is expanded, however, it will all 1167come out on one line. For example, 1168 1169@example 1170#define NUMBERS 1, \ 1171 2, \ 1172 3 1173int x[] = @{ NUMBERS @}; 1174 @expansion{} int x[] = @{ 1, 2, 3 @}; 1175@end example 1176 1177@noindent 1178The most common visible consequence of this is surprising line numbers 1179in error messages. 1180 1181There is no restriction on what can go in a macro body provided it 1182decomposes into valid preprocessing tokens. Parentheses need not 1183balance, and the body need not resemble valid C code. (If it does not, 1184you may get error messages from the C compiler when you use the macro.) 1185 1186The C preprocessor scans your program sequentially. Macro definitions 1187take effect at the place you write them. Therefore, the following input 1188to the C preprocessor 1189 1190@example 1191foo = X; 1192#define X 4 1193bar = X; 1194@end example 1195 1196@noindent 1197produces 1198 1199@example 1200foo = X; 1201bar = 4; 1202@end example 1203 1204When the preprocessor expands a macro name, the macro's expansion 1205replaces the macro invocation, then the expansion is examined for more 1206macros to expand. For example, 1207 1208@example 1209@group 1210#define TABLESIZE BUFSIZE 1211#define BUFSIZE 1024 1212TABLESIZE 1213 @expansion{} BUFSIZE 1214 @expansion{} 1024 1215@end group 1216@end example 1217 1218@noindent 1219@code{TABLESIZE} is expanded first to produce @code{BUFSIZE}, then that 1220macro is expanded to produce the final result, @code{1024}. 1221 1222Notice that @code{BUFSIZE} was not defined when @code{TABLESIZE} was 1223defined. The @samp{#define} for @code{TABLESIZE} uses exactly the 1224expansion you specify---in this case, @code{BUFSIZE}---and does not 1225check to see whether it too contains macro names. Only when you 1226@emph{use} @code{TABLESIZE} is the result of its expansion scanned for 1227more macro names. 1228 1229This makes a difference if you change the definition of @code{BUFSIZE} 1230at some point in the source file. @code{TABLESIZE}, defined as shown, 1231will always expand using the definition of @code{BUFSIZE} that is 1232currently in effect: 1233 1234@example 1235#define BUFSIZE 1020 1236#define TABLESIZE BUFSIZE 1237#undef BUFSIZE 1238#define BUFSIZE 37 1239@end example 1240 1241@noindent 1242Now @code{TABLESIZE} expands (in two stages) to @code{37}. 1243 1244If the expansion of a macro contains its own name, either directly or 1245via intermediate macros, it is not expanded again when the expansion is 1246examined for more macros. This prevents infinite recursion. 1247@xref{Self-Referential Macros}, for the precise details. 1248 1249@node Function-like Macros 1250@section Function-like Macros 1251@cindex function-like macros 1252 1253You can also define macros whose use looks like a function call. These 1254are called @dfn{function-like macros}. To define a function-like macro, 1255you use the same @samp{#define} directive, but you put a pair of 1256parentheses immediately after the macro name. For example, 1257 1258@example 1259#define lang_init() c_init() 1260lang_init() 1261 @expansion{} c_init() 1262@end example 1263 1264A function-like macro is only expanded if its name appears with a pair 1265of parentheses after it. If you write just the name, it is left alone. 1266This can be useful when you have a function and a macro of the same 1267name, and you wish to use the function sometimes. 1268 1269@example 1270extern void foo(void); 1271#define foo() /* optimized inline version */ 1272@dots{} 1273 foo(); 1274 funcptr = foo; 1275@end example 1276 1277Here the call to @code{foo()} will use the macro, but the function 1278pointer will get the address of the real function. If the macro were to 1279be expanded, it would cause a syntax error. 1280 1281If you put spaces between the macro name and the parentheses in the 1282macro definition, that does not define a function-like macro, it defines 1283an object-like macro whose expansion happens to begin with a pair of 1284parentheses. 1285 1286@example 1287#define lang_init () c_init() 1288lang_init() 1289 @expansion{} () c_init()() 1290@end example 1291 1292The first two pairs of parentheses in this expansion come from the 1293macro. The third is the pair that was originally after the macro 1294invocation. Since @code{lang_init} is an object-like macro, it does not 1295consume those parentheses. 1296 1297@node Macro Arguments 1298@section Macro Arguments 1299@cindex arguments 1300@cindex macros with arguments 1301@cindex arguments in macro definitions 1302 1303Function-like macros can take @dfn{arguments}, just like true functions. 1304To define a macro that uses arguments, you insert @dfn{parameters} 1305between the pair of parentheses in the macro definition that make the 1306macro function-like. The parameters must be valid C identifiers, 1307separated by commas and optionally whitespace. 1308 1309To invoke a macro that takes arguments, you write the name of the macro 1310followed by a list of @dfn{actual arguments} in parentheses, separated 1311by commas. The invocation of the macro need not be restricted to a 1312single logical line---it can cross as many lines in the source file as 1313you wish. The number of arguments you give must match the number of 1314parameters in the macro definition. When the macro is expanded, each 1315use of a parameter in its body is replaced by the tokens of the 1316corresponding argument. (You need not use all of the parameters in the 1317macro body.) 1318 1319As an example, here is a macro that computes the minimum of two numeric 1320values, as it is defined in many C programs, and some uses. 1321 1322@example 1323#define min(X, Y) ((X) < (Y) ? (X) : (Y)) 1324 x = min(a, b); @expansion{} x = ((a) < (b) ? (a) : (b)); 1325 y = min(1, 2); @expansion{} y = ((1) < (2) ? (1) : (2)); 1326 z = min(a + 28, *p); @expansion{} z = ((a + 28) < (*p) ? (a + 28) : (*p)); 1327@end example 1328 1329@noindent 1330(In this small example you can already see several of the dangers of 1331macro arguments. @xref{Macro Pitfalls}, for detailed explanations.) 1332 1333Leading and trailing whitespace in each argument is dropped, and all 1334whitespace between the tokens of an argument is reduced to a single 1335space. Parentheses within each argument must balance; a comma within 1336such parentheses does not end the argument. However, there is no 1337requirement for square brackets or braces to balance, and they do not 1338prevent a comma from separating arguments. Thus, 1339 1340@example 1341macro (array[x = y, x + 1]) 1342@end example 1343 1344@noindent 1345passes two arguments to @code{macro}: @code{array[x = y} and @code{x + 13461]}. If you want to supply @code{array[x = y, x + 1]} as an argument, 1347you can write it as @code{array[(x = y, x + 1)]}, which is equivalent C 1348code. 1349 1350All arguments to a macro are completely macro-expanded before they are 1351substituted into the macro body. After substitution, the complete text 1352is scanned again for macros to expand, including the arguments. This rule 1353may seem strange, but it is carefully designed so you need not worry 1354about whether any function call is actually a macro invocation. You can 1355run into trouble if you try to be too clever, though. @xref{Argument 1356Prescan}, for detailed discussion. 1357 1358For example, @code{min (min (a, b), c)} is first expanded to 1359 1360@example 1361 min (((a) < (b) ? (a) : (b)), (c)) 1362@end example 1363 1364@noindent 1365and then to 1366 1367@example 1368@group 1369((((a) < (b) ? (a) : (b))) < (c) 1370 ? (((a) < (b) ? (a) : (b))) 1371 : (c)) 1372@end group 1373@end example 1374 1375@noindent 1376(Line breaks shown here for clarity would not actually be generated.) 1377 1378@cindex empty macro arguments 1379You can leave macro arguments empty; this is not an error to the 1380preprocessor (but many macros will then expand to invalid code). 1381You cannot leave out arguments entirely; if a macro takes two arguments, 1382there must be exactly one comma at the top level of its argument list. 1383Here are some silly examples using @code{min}: 1384 1385@example 1386min(, b) @expansion{} (( ) < (b) ? ( ) : (b)) 1387min(a, ) @expansion{} ((a ) < ( ) ? (a ) : ( )) 1388min(,) @expansion{} (( ) < ( ) ? ( ) : ( )) 1389min((,),) @expansion{} (((,)) < ( ) ? ((,)) : ( )) 1390 1391min() @error{} macro "min" requires 2 arguments, but only 1 given 1392min(,,) @error{} macro "min" passed 3 arguments, but takes just 2 1393@end example 1394 1395Whitespace is not a preprocessing token, so if a macro @code{foo} takes 1396one argument, @code{@w{foo ()}} and @code{@w{foo ( )}} both supply it an 1397empty argument. Previous GNU preprocessor implementations and 1398documentation were incorrect on this point, insisting that a 1399function-like macro that takes a single argument be passed a space if an 1400empty argument was required. 1401 1402Macro parameters appearing inside string literals are not replaced by 1403their corresponding actual arguments. 1404 1405@example 1406#define foo(x) x, "x" 1407foo(bar) @expansion{} bar, "x" 1408@end example 1409 1410@node Stringification 1411@section Stringification 1412@cindex stringification 1413@cindex @samp{#} operator 1414 1415Sometimes you may want to convert a macro argument into a string 1416constant. Parameters are not replaced inside string constants, but you 1417can use the @samp{#} preprocessing operator instead. When a macro 1418parameter is used with a leading @samp{#}, the preprocessor replaces it 1419with the literal text of the actual argument, converted to a string 1420constant. Unlike normal parameter replacement, the argument is not 1421macro-expanded first. This is called @dfn{stringification}. 1422 1423There is no way to combine an argument with surrounding text and 1424stringify it all together. Instead, you can write a series of adjacent 1425string constants and stringified arguments. The preprocessor will 1426replace the stringified arguments with string constants. The C 1427compiler will then combine all the adjacent string constants into one 1428long string. 1429 1430Here is an example of a macro definition that uses stringification: 1431 1432@example 1433@group 1434#define WARN_IF(EXP) \ 1435do @{ if (EXP) \ 1436 fprintf (stderr, "Warning: " #EXP "\n"); @} \ 1437while (0) 1438WARN_IF (x == 0); 1439 @expansion{} do @{ if (x == 0) 1440 fprintf (stderr, "Warning: " "x == 0" "\n"); @} while (0); 1441@end group 1442@end example 1443 1444@noindent 1445The argument for @code{EXP} is substituted once, as-is, into the 1446@code{if} statement, and once, stringified, into the argument to 1447@code{fprintf}. If @code{x} were a macro, it would be expanded in the 1448@code{if} statement, but not in the string. 1449 1450The @code{do} and @code{while (0)} are a kludge to make it possible to 1451write @code{WARN_IF (@var{arg});}, which the resemblance of 1452@code{WARN_IF} to a function would make C programmers want to do; see 1453@ref{Swallowing the Semicolon}. 1454 1455Stringification in C involves more than putting double-quote characters 1456around the fragment. The preprocessor backslash-escapes the quotes 1457surrounding embedded string constants, and all backslashes within string and 1458character constants, in order to get a valid C string constant with the 1459proper contents. Thus, stringifying @code{@w{p = "foo\n";}} results in 1460@t{@w{"p = \"foo\\n\";"}}. However, backslashes that are not inside string 1461or character constants are not duplicated: @samp{\n} by itself 1462stringifies to @t{"\n"}. 1463 1464All leading and trailing whitespace in text being stringified is 1465ignored. Any sequence of whitespace in the middle of the text is 1466converted to a single space in the stringified result. Comments are 1467replaced by whitespace long before stringification happens, so they 1468never appear in stringified text. 1469 1470There is no way to convert a macro argument into a character constant. 1471 1472If you want to stringify the result of expansion of a macro argument, 1473you have to use two levels of macros. 1474 1475@example 1476#define xstr(s) str(s) 1477#define str(s) #s 1478#define foo 4 1479str (foo) 1480 @expansion{} "foo" 1481xstr (foo) 1482 @expansion{} xstr (4) 1483 @expansion{} str (4) 1484 @expansion{} "4" 1485@end example 1486 1487@code{s} is stringified when it is used in @code{str}, so it is not 1488macro-expanded first. But @code{s} is an ordinary argument to 1489@code{xstr}, so it is completely macro-expanded before @code{xstr} 1490itself is expanded (@pxref{Argument Prescan}). Therefore, by the time 1491@code{str} gets to its argument, it has already been macro-expanded. 1492 1493@node Concatenation 1494@section Concatenation 1495@cindex concatenation 1496@cindex token pasting 1497@cindex token concatenation 1498@cindex @samp{##} operator 1499 1500It is often useful to merge two tokens into one while expanding macros. 1501This is called @dfn{token pasting} or @dfn{token concatenation}. The 1502@samp{##} preprocessing operator performs token pasting. When a macro 1503is expanded, the two tokens on either side of each @samp{##} operator 1504are combined into a single token, which then replaces the @samp{##} and 1505the two original tokens in the macro expansion. Usually both will be 1506identifiers, or one will be an identifier and the other a preprocessing 1507number. When pasted, they make a longer identifier. This isn't the 1508only valid case. It is also possible to concatenate two numbers (or a 1509number and a name, such as @code{1.5} and @code{e3}) into a number. 1510Also, multi-character operators such as @code{+=} can be formed by 1511token pasting. 1512 1513However, two tokens that don't together form a valid token cannot be 1514pasted together. For example, you cannot concatenate @code{x} with 1515@code{+} in either order. If you try, the preprocessor issues a warning 1516and emits the two tokens. Whether it puts white space between the 1517tokens is undefined. It is common to find unnecessary uses of @samp{##} 1518in complex macros. If you get this warning, it is likely that you can 1519simply remove the @samp{##}. 1520 1521Both the tokens combined by @samp{##} could come from the macro body, 1522but you could just as well write them as one token in the first place. 1523Token pasting is most useful when one or both of the tokens comes from a 1524macro argument. If either of the tokens next to an @samp{##} is a 1525parameter name, it is replaced by its actual argument before @samp{##} 1526executes. As with stringification, the actual argument is not 1527macro-expanded first. If the argument is empty, that @samp{##} has no 1528effect. 1529 1530Keep in mind that the C preprocessor converts comments to whitespace 1531before macros are even considered. Therefore, you cannot create a 1532comment by concatenating @samp{/} and @samp{*}. You can put as much 1533whitespace between @samp{##} and its operands as you like, including 1534comments, and you can put comments in arguments that will be 1535concatenated. However, it is an error if @samp{##} appears at either 1536end of a macro body. 1537 1538Consider a C program that interprets named commands. There probably 1539needs to be a table of commands, perhaps an array of structures declared 1540as follows: 1541 1542@example 1543@group 1544struct command 1545@{ 1546 char *name; 1547 void (*function) (void); 1548@}; 1549@end group 1550 1551@group 1552struct command commands[] = 1553@{ 1554 @{ "quit", quit_command @}, 1555 @{ "help", help_command @}, 1556 @dots{} 1557@}; 1558@end group 1559@end example 1560 1561It would be cleaner not to have to give each command name twice, once in 1562the string constant and once in the function name. A macro which takes the 1563name of a command as an argument can make this unnecessary. The string 1564constant can be created with stringification, and the function name by 1565concatenating the argument with @samp{_command}. Here is how it is done: 1566 1567@example 1568#define COMMAND(NAME) @{ #NAME, NAME ## _command @} 1569 1570struct command commands[] = 1571@{ 1572 COMMAND (quit), 1573 COMMAND (help), 1574 @dots{} 1575@}; 1576@end example 1577 1578@node Variadic Macros 1579@section Variadic Macros 1580@cindex variable number of arguments 1581@cindex macros with variable arguments 1582@cindex variadic macros 1583 1584A macro can be declared to accept a variable number of arguments much as 1585a function can. The syntax for defining the macro is similar to that of 1586a function. Here is an example: 1587 1588@example 1589#define eprintf(@dots{}) fprintf (stderr, __VA_ARGS__) 1590@end example 1591 1592This kind of macro is called @dfn{variadic}. When the macro is invoked, 1593all the tokens in its argument list after the last named argument (this 1594macro has none), including any commas, become the @dfn{variable 1595argument}. This sequence of tokens replaces the identifier 1596@code{@w{__VA_ARGS__}} in the macro body wherever it appears. Thus, we 1597have this expansion: 1598 1599@example 1600eprintf ("%s:%d: ", input_file, lineno) 1601 @expansion{} fprintf (stderr, "%s:%d: ", input_file, lineno) 1602@end example 1603 1604The variable argument is completely macro-expanded before it is inserted 1605into the macro expansion, just like an ordinary argument. You may use 1606the @samp{#} and @samp{##} operators to stringify the variable argument 1607or to paste its leading or trailing token with another token. (But see 1608below for an important special case for @samp{##}.) 1609 1610If your macro is complicated, you may want a more descriptive name for 1611the variable argument than @code{@w{__VA_ARGS__}}. GNU CPP permits 1612this, as an extension. You may write an argument name immediately 1613before the @samp{@dots{}}; that name is used for the variable argument. 1614The @code{eprintf} macro above could be written 1615 1616@example 1617#define eprintf(args@dots{}) fprintf (stderr, args) 1618@end example 1619 1620@noindent 1621using this extension. You cannot use @code{__VA_ARGS__} and this 1622extension in the same macro. 1623 1624You can have named arguments as well as variable arguments in a variadic 1625macro. We could define @code{eprintf} like this, instead: 1626 1627@example 1628#define eprintf(format, @dots{}) fprintf (stderr, format, __VA_ARGS__) 1629@end example 1630 1631@noindent 1632This formulation looks more descriptive, but unfortunately it is less 1633flexible: you must now supply at least one argument after the format 1634string. In standard C, you cannot omit the comma separating the named 1635argument from the variable arguments. Furthermore, if you leave the 1636variable argument empty, you will get a syntax error, because 1637there will be an extra comma after the format string. 1638 1639@example 1640eprintf("success!\n", ); 1641 @expansion{} fprintf(stderr, "success!\n", ); 1642@end example 1643 1644GNU CPP has a pair of extensions which deal with this problem. First, 1645you are allowed to leave the variable argument out entirely: 1646 1647@example 1648eprintf ("success!\n") 1649 @expansion{} fprintf(stderr, "success!\n", ); 1650@end example 1651 1652@noindent 1653Second, the @samp{##} token paste operator has a special meaning when 1654placed between a comma and a variable argument. If you write 1655 1656@example 1657#define eprintf(format, @dots{}) fprintf (stderr, format, ##__VA_ARGS__) 1658@end example 1659 1660@noindent 1661and the variable argument is left out when the @code{eprintf} macro is 1662used, then the comma before the @samp{##} will be deleted. This does 1663@emph{not} happen if you pass an empty argument, nor does it happen if 1664the token preceding @samp{##} is anything other than a comma. 1665 1666@example 1667eprintf ("success!\n") 1668 @expansion{} fprintf(stderr, "success!\n"); 1669@end example 1670 1671C99 mandates that the only place the identifier @code{@w{__VA_ARGS__}} 1672can appear is in the replacement list of a variadic macro. It may not 1673be used as a macro name, macro argument name, or within a different type 1674of macro. It may also be forbidden in open text; the standard is 1675ambiguous. We recommend you avoid using it except for its defined 1676purpose. 1677 1678Variadic macros are a new feature in C99. GNU CPP has supported them 1679for a long time, but only with a named variable argument 1680(@samp{args@dots{}}, not @samp{@dots{}} and @code{@w{__VA_ARGS__}}). If you are 1681concerned with portability to previous versions of GCC, you should use 1682only named variable arguments. On the other hand, if you are concerned 1683with portability to other conforming implementations of C99, you should 1684use only @code{@w{__VA_ARGS__}}. 1685 1686Previous versions of GNU CPP implemented the comma-deletion extension 1687much more generally. We have restricted it in this release to minimize 1688the differences from C99. To get the same effect with both this and 1689previous versions of GCC, the token preceding the special @samp{##} must 1690be a comma, and there must be white space between that comma and 1691whatever comes immediately before it: 1692 1693@example 1694#define eprintf(format, args@dots{}) fprintf (stderr, format , ##args) 1695@end example 1696 1697@noindent 1698@xref{Differences from previous versions}, for the gory details. 1699 1700@node Predefined Macros 1701@section Predefined Macros 1702 1703@cindex predefined macros 1704Several object-like macros are predefined; you use them without 1705supplying their definitions. They fall into three classes: standard, 1706common, and system-specific. 1707 1708In C++, there is a fourth category, the named operators. They act like 1709predefined macros, but you cannot undefine them. 1710 1711@menu 1712* Standard Predefined Macros:: 1713* Common Predefined Macros:: 1714* System-specific Predefined Macros:: 1715* C++ Named Operators:: 1716@end menu 1717 1718@node Standard Predefined Macros 1719@subsection Standard Predefined Macros 1720@cindex standard predefined macros. 1721 1722The standard predefined macros are specified by the C and/or C++ 1723language standards, so they are available with all compilers that 1724implement those standards. Older compilers may not provide all of 1725them. Their names all start with double underscores. 1726 1727@table @code 1728@item __FILE__ 1729This macro expands to the name of the current input file, in the form of 1730a C string constant. This is the path by which the preprocessor opened 1731the file, not the short name specified in @samp{#include} or as the 1732input file name argument. For example, 1733@code{"/usr/local/include/myheader.h"} is a possible expansion of this 1734macro. 1735 1736@item __LINE__ 1737This macro expands to the current input line number, in the form of a 1738decimal integer constant. While we call it a predefined macro, it's 1739a pretty strange macro, since its ``definition'' changes with each 1740new line of source code. 1741@end table 1742 1743@code{__FILE__} and @code{__LINE__} are useful in generating an error 1744message to report an inconsistency detected by the program; the message 1745can state the source line at which the inconsistency was detected. For 1746example, 1747 1748@example 1749fprintf (stderr, "Internal error: " 1750 "negative string length " 1751 "%d at %s, line %d.", 1752 length, __FILE__, __LINE__); 1753@end example 1754 1755An @samp{#include} directive changes the expansions of @code{__FILE__} 1756and @code{__LINE__} to correspond to the included file. At the end of 1757that file, when processing resumes on the input file that contained 1758the @samp{#include} directive, the expansions of @code{__FILE__} and 1759@code{__LINE__} revert to the values they had before the 1760@samp{#include} (but @code{__LINE__} is then incremented by one as 1761processing moves to the line after the @samp{#include}). 1762 1763A @samp{#line} directive changes @code{__LINE__}, and may change 1764@code{__FILE__} as well. @xref{Line Control}. 1765 1766C99 introduces @code{__func__}, and GCC has provided @code{__FUNCTION__} 1767for a long time. Both of these are strings containing the name of the 1768current function (there are slight semantic differences; see the GCC 1769manual). Neither of them is a macro; the preprocessor does not know the 1770name of the current function. They tend to be useful in conjunction 1771with @code{__FILE__} and @code{__LINE__}, though. 1772 1773@table @code 1774 1775@item __DATE__ 1776This macro expands to a string constant that describes the date on which 1777the preprocessor is being run. The string constant contains eleven 1778characters and looks like @code{@w{"Feb 12 1996"}}. If the day of the 1779month is less than 10, it is padded with a space on the left. 1780 1781@item __TIME__ 1782This macro expands to a string constant that describes the time at 1783which the preprocessor is being run. The string constant contains 1784eight characters and looks like @code{"23:59:01"}. 1785 1786@item __STDC__ 1787In normal operation, this macro expands to the constant 1, to signify 1788that this compiler conforms to ISO Standard C@. If GNU CPP is used with 1789a compiler other than GCC, this is not necessarily true; however, the 1790preprocessor always conforms to the standard, unless the 1791@option{-traditional} option is used. 1792 1793This macro is not defined if the @option{-traditional} option is used. 1794 1795On some hosts, the system compiler uses a different convention, where 1796@code{__STDC__} is normally 0, but is 1 if the user specifies strict 1797conformance to the C Standard. GNU CPP follows the host convention when 1798processing system header files, but when processing user files 1799@code{__STDC__} is always 1. This has been reported to cause problems; 1800for instance, some versions of Solaris provide X Windows headers that 1801expect @code{__STDC__} to be either undefined or 1. You may be able to 1802work around this sort of problem by using an @option{-I} option to 1803cancel treatment of those headers as system headers. @xref{Invocation}. 1804 1805@item __STDC_VERSION__ 1806This macro expands to the C Standard's version number, a long integer 1807constant of the form @code{@var{yyyy}@var{mm}L} where @var{yyyy} and 1808@var{mm} are the year and month of the Standard version. This signifies 1809which version of the C Standard the compiler conforms to. Like 1810@code{__STDC__}, this is not necessarily accurate for the entire 1811implementation, unless GNU CPP is being used with GCC@. 1812 1813The value @code{199409L} signifies the 1989 C standard as amended in 18141994, which is the current default; the value @code{199901L} signifies 1815the 1999 revision of the C standard. Support for the 1999 revision is 1816not yet complete. 1817 1818This macro is not defined if the @option{-traditional} option is used, nor 1819when compiling C++ or Objective-C@. 1820 1821@item __STDC_HOSTED__ 1822This macro is defined, with value 1, if the compiler's target is a 1823@dfn{hosted environment}. A hosted environment has the complete 1824facilities of the standard C library available. 1825 1826@item __cplusplus 1827This macro is defined when the C++ compiler is in use. You can use 1828@code{__cplusplus} to test whether a header is compiled by a C compiler 1829or a C++ compiler. This macro is similar to @code{__STDC_VERSION__}, in 1830that it expands to a version number. A fully conforming implementation 1831of the 1998 C++ standard will define this macro to @code{199711L}. The 1832GNU C++ compiler is not yet fully conforming, so it uses @code{1} 1833instead. We hope to complete our implementation in the near future. 1834 1835@end table 1836 1837@node Common Predefined Macros 1838@subsection Common Predefined Macros 1839@cindex common predefined macros 1840 1841The common predefined macros are GNU C extensions. They are available 1842with the same meanings regardless of the machine or operating system on 1843which you are using GNU C@. Their names all start with double 1844underscores. 1845 1846@table @code 1847 1848@item __GNUC__ 1849@itemx __GNUC_MINOR__ 1850@itemx __GNUC_PATCHLEVEL__ 1851These macros are defined by all GNU compilers that use the C 1852preprocessor: C, C++, and Objective-C@. Their values are the major 1853version, minor version, and patch level of the compiler, as integer 1854constants. For example, GCC 3.2.1 will define @code{__GNUC__} to 3, 1855@code{__GNUC_MINOR__} to 2, and @code{__GNUC_PATCHLEVEL__} to 1. They 1856are defined only when the entire compiler is in use; if you invoke the 1857preprocessor directly, they are not defined. 1858 1859@code{__GNUC_PATCHLEVEL__} is new to GCC 3.0; it is also present in the 1860widely-used development snapshots leading up to 3.0 (which identify 1861themselves as GCC 2.96 or 2.97, depending on which snapshot you have). 1862 1863If all you need to know is whether or not your program is being compiled 1864by GCC, you can simply test @code{__GNUC__}. If you need to write code 1865which depends on a specific version, you must be more careful. Each 1866time the minor version is increased, the patch level is reset to zero; 1867each time the major version is increased (which happens rarely), the 1868minor version and patch level are reset. If you wish to use the 1869predefined macros directly in the conditional, you will need to write it 1870like this: 1871 1872@example 1873/* @r{Test for GCC > 3.2.0} */ 1874#if __GNUC__ > 3 || \ 1875 (__GNUC__ == 3 && (__GNUC_MINOR__ > 2 || \ 1876 (__GNUC_MINOR__ == 2 && \ 1877 __GNUC_PATCHLEVEL__ > 0)) 1878@end example 1879 1880@noindent 1881Another approach is to use the predefined macros to 1882calculate a single number, then compare that against a threshold: 1883 1884@example 1885#define GCC_VERSION (__GNUC__ * 10000 \ 1886 + __GNUC_MINOR__ * 100 \ 1887 + __GNUC_PATCHLEVEL__) 1888@dots{} 1889/* @r{Test for GCC > 3.2.0} */ 1890#if GCC_VERSION > 30200 1891@end example 1892 1893@noindent 1894Many people find this form easier to understand. 1895 1896@item __OBJC__ 1897This macro is defined, with value 1, when the Objective-C compiler is in 1898use. You can use @code{__OBJC__} to test whether a header is compiled 1899by a C compiler or a Objective-C compiler. 1900 1901@item __GNUG__ 1902The GNU C++ compiler defines this. Testing it is equivalent to 1903testing @code{@w{(__GNUC__ && __cplusplus)}}. 1904 1905@item __STRICT_ANSI__ 1906GCC defines this macro if and only if the @option{-ansi} switch, or a 1907@option{-std} switch specifying strict conformance to some version of ISO C, 1908was specified when GCC was invoked. It is defined to @samp{1}. 1909This macro exists primarily to direct GNU libc's header files to 1910restrict their definitions to the minimal set found in the 1989 C 1911standard. 1912 1913@item __BASE_FILE__ 1914This macro expands to the name of the main input file, in the form 1915of a C string constant. This is the source file that was specified 1916on the command line of the preprocessor or C compiler. 1917 1918@item __INCLUDE_LEVEL__ 1919This macro expands to a decimal integer constant that represents the 1920depth of nesting in include files. The value of this macro is 1921incremented on every @samp{#include} directive and decremented at the 1922end of every included file. It starts out at 0, it's value within the 1923base file specified on the command line. 1924 1925@item __VERSION__ 1926This macro expands to a string constant which describes the version of 1927the compiler in use. You should not rely on its contents having any 1928particular form, but it can be counted on to contain at least the 1929release number. 1930 1931@item __OPTIMIZE__ 1932@itemx __OPTIMIZE_SIZE__ 1933@itemx __NO_INLINE__ 1934These macros describe the compilation mode. @code{__OPTIMIZE__} is 1935defined in all optimizing compilations. @code{__OPTIMIZE_SIZE__} is 1936defined if the compiler is optimizing for size, not speed. 1937@code{__NO_INLINE__} is defined if no functions will be inlined into 1938their callers (when not optimizing, or when inlining has been 1939specifically disabled by @option{-fno-inline}). 1940 1941These macros cause certain GNU header files to provide optimized 1942definitions, using macros or inline functions, of system library 1943functions. You should not use these macros in any way unless you make 1944sure that programs will execute with the same effect whether or not they 1945are defined. If they are defined, their value is 1. 1946 1947@item __CHAR_UNSIGNED__ 1948GCC defines this macro if and only if the data type @code{char} is 1949unsigned on the target machine. It exists to cause the standard header 1950file @file{limits.h} to work correctly. You should not use this macro 1951yourself; instead, refer to the standard macros defined in @file{limits.h}. 1952 1953@item __REGISTER_PREFIX__ 1954This macro expands to a single token (not a string constant) which is 1955the prefix applied to CPU register names in assembly language for this 1956target. You can use it to write assembly that is usable in multiple 1957environments. For example, in the @code{m68k-aout} environment it 1958expands to nothing, but in the @code{m68k-coff} environment it expands 1959to a single @samp{%}. 1960 1961@item __USER_LABEL_PREFIX__ 1962This macro expands to a single token which is the prefix applied to 1963user labels (symbols visible to C code) in assembly. For example, in 1964the @code{m68k-aout} environment it expands to an @samp{_}, but in the 1965@code{m68k-coff} environment it expands to nothing. 1966 1967This macro will have the correct definition even if 1968@option{-f(no-)underscores} is in use, but it will not be correct if 1969target-specific options that adjust this prefix are used (e.g.@: the 1970OSF/rose @option{-mno-underscores} option). 1971 1972@item __SIZE_TYPE__ 1973@itemx __PTRDIFF_TYPE__ 1974@itemx __WCHAR_TYPE__ 1975@itemx __WINT_TYPE__ 1976These macros are defined to the correct underlying types for the 1977@code{size_t}, @code{ptrdiff_t}, @code{wchar_t}, and @code{wint_t} 1978typedefs, respectively. They exist to make the standard header files 1979@file{stddef.h} and @file{wchar.h} work correctly. You should not use 1980these macros directly; instead, include the appropriate headers and use 1981the typedefs. 1982 1983@item __USING_SJLJ_EXCEPTIONS__ 1984This macro is defined, with value 1, if the compiler uses the old 1985mechanism based on @code{setjmp} and @code{longjmp} for exception 1986handling. 1987@end table 1988 1989@node System-specific Predefined Macros 1990@subsection System-specific Predefined Macros 1991 1992@cindex system-specific predefined macros 1993@cindex predefined macros, system-specific 1994@cindex reserved namespace 1995 1996The C preprocessor normally predefines several macros that indicate what 1997type of system and machine is in use. They are obviously different on 1998each target supported by GCC@. This manual, being for all systems and 1999machines, cannot tell you what their names are, but you can use 2000@command{cpp -dM} to see them all. @xref{Invocation}. All system-specific 2001predefined macros expand to the constant 1, so you can test them with 2002either @samp{#ifdef} or @samp{#if}. 2003 2004The C standard requires that all system-specific macros be part of the 2005@dfn{reserved namespace}. All names which begin with two underscores, 2006or an underscore and a capital letter, are reserved for the compiler and 2007library to use as they wish. However, historically system-specific 2008macros have had names with no special prefix; for instance, it is common 2009to find @code{unix} defined on Unix systems. For all such macros, GCC 2010provides a parallel macro with two underscores added at the beginning 2011and the end. If @code{unix} is defined, @code{__unix__} will be defined 2012too. There will never be more than two underscores; the parallel of 2013@code{_mips} is @code{__mips__}. 2014 2015When the @option{-ansi} option, or any @option{-std} option that 2016requests strict conformance, is given to the compiler, all the 2017system-specific predefined macros outside the reserved namespace are 2018suppressed. The parallel macros, inside the reserved namespace, remain 2019defined. 2020 2021We are slowly phasing out all predefined macros which are outside the 2022reserved namespace. You should never use them in new programs, and we 2023encourage you to correct older code to use the parallel macros whenever 2024you find it. We don't recommend you use the system-specific macros that 2025are in the reserved namespace, either. It is better in the long run to 2026check specifically for features you need, using a tool such as 2027@command{autoconf}. 2028 2029@node C++ Named Operators 2030@subsection C++ Named Operators 2031@cindex named operators 2032@cindex C++ named operators 2033@cindex iso646.h 2034 2035In C++, there are eleven keywords which are simply alternate spellings 2036of operators normally written with punctuation. These keywords are 2037treated as such even in the preprocessor. They function as operators in 2038@samp{#if}, and they cannot be defined as macros or poisoned. In C, you 2039can request that those keywords take their C++ meaning by including 2040@file{iso646.h}. That header defines each one as a normal object-like 2041macro expanding to the appropriate punctuator. 2042 2043These are the named operators and their corresponding punctuators: 2044 2045@multitable {Named Operator} {Punctuator} 2046@item Named Operator @tab Punctuator 2047@item @code{and} @tab @code{&&} 2048@item @code{and_eq} @tab @code{&=} 2049@item @code{bitand} @tab @code{&} 2050@item @code{bitor} @tab @code{|} 2051@item @code{compl} @tab @code{~} 2052@item @code{not} @tab @code{!} 2053@item @code{not_eq} @tab @code{!=} 2054@item @code{or} @tab @code{||} 2055@item @code{or_eq} @tab @code{|=} 2056@item @code{xor} @tab @code{^} 2057@item @code{xor_eq} @tab @code{^=} 2058@end multitable 2059 2060@node Undefining and Redefining Macros 2061@section Undefining and Redefining Macros 2062@cindex undefining macros 2063@cindex redefining macros 2064@findex #undef 2065 2066If a macro ceases to be useful, it may be @dfn{undefined} with the 2067@samp{#undef} directive. @samp{#undef} takes a single argument, the 2068name of the macro to undefine. You use the bare macro name, even if the 2069macro is function-like. It is an error if anything appears on the line 2070after the macro name. @samp{#undef} has no effect if the name is not a 2071macro. 2072 2073@example 2074#define FOO 4 2075x = FOO; @expansion{} x = 4; 2076#undef FOO 2077x = FOO; @expansion{} x = FOO; 2078@end example 2079 2080Once a macro has been undefined, that identifier may be @dfn{redefined} 2081as a macro by a subsequent @samp{#define} directive. The new definition 2082need not have any resemblance to the old definition. 2083 2084However, if an identifier which is currently a macro is redefined, then 2085the new definition must be @dfn{effectively the same} as the old one. 2086Two macro definitions are effectively the same if: 2087@itemize @bullet 2088@item Both are the same type of macro (object- or function-like). 2089@item All the tokens of the replacement list are the same. 2090@item If there are any parameters, they are the same. 2091@item Whitespace appears in the same places in both. It need not be 2092exactly the same amount of whitespace, though. Remember that comments 2093count as whitespace. 2094@end itemize 2095 2096@noindent 2097These definitions are effectively the same: 2098@example 2099#define FOUR (2 + 2) 2100#define FOUR (2 + 2) 2101#define FOUR (2 /* two */ + 2) 2102@end example 2103@noindent 2104but these are not: 2105@example 2106#define FOUR (2 + 2) 2107#define FOUR ( 2+2 ) 2108#define FOUR (2 * 2) 2109#define FOUR(score,and,seven,years,ago) (2 + 2) 2110@end example 2111 2112If a macro is redefined with a definition that is not effectively the 2113same as the old one, the preprocessor issues a warning and changes the 2114macro to use the new definition. If the new definition is effectively 2115the same, the redefinition is silently ignored. This allows, for 2116instance, two different headers to define a common macro. The 2117preprocessor will only complain if the definitions do not match. 2118 2119@node Macro Pitfalls 2120@section Macro Pitfalls 2121@cindex problems with macros 2122@cindex pitfalls of macros 2123 2124In this section we describe some special rules that apply to macros and 2125macro expansion, and point out certain cases in which the rules have 2126counter-intuitive consequences that you must watch out for. 2127 2128@menu 2129* Misnesting:: 2130* Operator Precedence Problems:: 2131* Swallowing the Semicolon:: 2132* Duplication of Side Effects:: 2133* Self-Referential Macros:: 2134* Argument Prescan:: 2135* Newlines in Arguments:: 2136@end menu 2137 2138@node Misnesting 2139@subsection Misnesting 2140 2141When a macro is called with arguments, the arguments are substituted 2142into the macro body and the result is checked, together with the rest of 2143the input file, for more macro calls. It is possible to piece together 2144a macro call coming partially from the macro body and partially from the 2145arguments. For example, 2146 2147@example 2148#define twice(x) (2*(x)) 2149#define call_with_1(x) x(1) 2150call_with_1 (twice) 2151 @expansion{} twice(1) 2152 @expansion{} (2*(1)) 2153@end example 2154 2155Macro definitions do not have to have balanced parentheses. By writing 2156an unbalanced open parenthesis in a macro body, it is possible to create 2157a macro call that begins inside the macro body but ends outside of it. 2158For example, 2159 2160@example 2161#define strange(file) fprintf (file, "%s %d", 2162@dots{} 2163strange(stderr) p, 35) 2164 @expansion{} fprintf (stderr, "%s %d", p, 35) 2165@end example 2166 2167The ability to piece together a macro call can be useful, but the use of 2168unbalanced open parentheses in a macro body is just confusing, and 2169should be avoided. 2170 2171@node Operator Precedence Problems 2172@subsection Operator Precedence Problems 2173@cindex parentheses in macro bodies 2174 2175You may have noticed that in most of the macro definition examples shown 2176above, each occurrence of a macro argument name had parentheses around 2177it. In addition, another pair of parentheses usually surround the 2178entire macro definition. Here is why it is best to write macros that 2179way. 2180 2181Suppose you define a macro as follows, 2182 2183@example 2184#define ceil_div(x, y) (x + y - 1) / y 2185@end example 2186 2187@noindent 2188whose purpose is to divide, rounding up. (One use for this operation is 2189to compute how many @code{int} objects are needed to hold a certain 2190number of @code{char} objects.) Then suppose it is used as follows: 2191 2192@example 2193a = ceil_div (b & c, sizeof (int)); 2194 @expansion{} a = (b & c + sizeof (int) - 1) / sizeof (int); 2195@end example 2196 2197@noindent 2198This does not do what is intended. The operator-precedence rules of 2199C make it equivalent to this: 2200 2201@example 2202a = (b & (c + sizeof (int) - 1)) / sizeof (int); 2203@end example 2204 2205@noindent 2206What we want is this: 2207 2208@example 2209a = ((b & c) + sizeof (int) - 1)) / sizeof (int); 2210@end example 2211 2212@noindent 2213Defining the macro as 2214 2215@example 2216#define ceil_div(x, y) ((x) + (y) - 1) / (y) 2217@end example 2218 2219@noindent 2220provides the desired result. 2221 2222Unintended grouping can result in another way. Consider @code{sizeof 2223ceil_div(1, 2)}. That has the appearance of a C expression that would 2224compute the size of the type of @code{ceil_div (1, 2)}, but in fact it 2225means something very different. Here is what it expands to: 2226 2227@example 2228sizeof ((1) + (2) - 1) / (2) 2229@end example 2230 2231@noindent 2232This would take the size of an integer and divide it by two. The 2233precedence rules have put the division outside the @code{sizeof} when it 2234was intended to be inside. 2235 2236Parentheses around the entire macro definition prevent such problems. 2237Here, then, is the recommended way to define @code{ceil_div}: 2238 2239@example 2240#define ceil_div(x, y) (((x) + (y) - 1) / (y)) 2241@end example 2242 2243@node Swallowing the Semicolon 2244@subsection Swallowing the Semicolon 2245@cindex semicolons (after macro calls) 2246 2247Often it is desirable to define a macro that expands into a compound 2248statement. Consider, for example, the following macro, that advances a 2249pointer (the argument @code{p} says where to find it) across whitespace 2250characters: 2251 2252@example 2253#define SKIP_SPACES(p, limit) \ 2254@{ char *lim = (limit); \ 2255 while (p < lim) @{ \ 2256 if (*p++ != ' ') @{ \ 2257 p--; break; @}@}@} 2258@end example 2259 2260@noindent 2261Here backslash-newline is used to split the macro definition, which must 2262be a single logical line, so that it resembles the way such code would 2263be laid out if not part of a macro definition. 2264 2265A call to this macro might be @code{SKIP_SPACES (p, lim)}. Strictly 2266speaking, the call expands to a compound statement, which is a complete 2267statement with no need for a semicolon to end it. However, since it 2268looks like a function call, it minimizes confusion if you can use it 2269like a function call, writing a semicolon afterward, as in 2270@code{SKIP_SPACES (p, lim);} 2271 2272This can cause trouble before @code{else} statements, because the 2273semicolon is actually a null statement. Suppose you write 2274 2275@example 2276if (*p != 0) 2277 SKIP_SPACES (p, lim); 2278else @dots{} 2279@end example 2280 2281@noindent 2282The presence of two statements---the compound statement and a null 2283statement---in between the @code{if} condition and the @code{else} 2284makes invalid C code. 2285 2286The definition of the macro @code{SKIP_SPACES} can be altered to solve 2287this problem, using a @code{do @dots{} while} statement. Here is how: 2288 2289@example 2290#define SKIP_SPACES(p, limit) \ 2291do @{ char *lim = (limit); \ 2292 while (p < lim) @{ \ 2293 if (*p++ != ' ') @{ \ 2294 p--; break; @}@}@} \ 2295while (0) 2296@end example 2297 2298Now @code{SKIP_SPACES (p, lim);} expands into 2299 2300@example 2301do @{@dots{}@} while (0); 2302@end example 2303 2304@noindent 2305which is one statement. The loop executes exactly once; most compilers 2306generate no extra code for it. 2307 2308@node Duplication of Side Effects 2309@subsection Duplication of Side Effects 2310 2311@cindex side effects (in macro arguments) 2312@cindex unsafe macros 2313Many C programs define a macro @code{min}, for ``minimum'', like this: 2314 2315@example 2316#define min(X, Y) ((X) < (Y) ? (X) : (Y)) 2317@end example 2318 2319When you use this macro with an argument containing a side effect, 2320as shown here, 2321 2322@example 2323next = min (x + y, foo (z)); 2324@end example 2325 2326@noindent 2327it expands as follows: 2328 2329@example 2330next = ((x + y) < (foo (z)) ? (x + y) : (foo (z))); 2331@end example 2332 2333@noindent 2334where @code{x + y} has been substituted for @code{X} and @code{foo (z)} 2335for @code{Y}. 2336 2337The function @code{foo} is used only once in the statement as it appears 2338in the program, but the expression @code{foo (z)} has been substituted 2339twice into the macro expansion. As a result, @code{foo} might be called 2340two times when the statement is executed. If it has side effects or if 2341it takes a long time to compute, the results might not be what you 2342intended. We say that @code{min} is an @dfn{unsafe} macro. 2343 2344The best solution to this problem is to define @code{min} in a way that 2345computes the value of @code{foo (z)} only once. The C language offers 2346no standard way to do this, but it can be done with GNU extensions as 2347follows: 2348 2349@example 2350#define min(X, Y) \ 2351(@{ typeof (X) x_ = (X); \ 2352 typeof (Y) y_ = (Y); \ 2353 (x_ < y_) ? x_ : y_; @}) 2354@end example 2355 2356The @samp{(@{ @dots{} @})} notation produces a compound statement that 2357acts as an expression. Its value is the value of its last statement. 2358This permits us to define local variables and assign each argument to 2359one. The local variables have underscores after their names to reduce 2360the risk of conflict with an identifier of wider scope (it is impossible 2361to avoid this entirely). Now each argument is evaluated exactly once. 2362 2363If you do not wish to use GNU C extensions, the only solution is to be 2364careful when @emph{using} the macro @code{min}. For example, you can 2365calculate the value of @code{foo (z)}, save it in a variable, and use 2366that variable in @code{min}: 2367 2368@example 2369@group 2370#define min(X, Y) ((X) < (Y) ? (X) : (Y)) 2371@dots{} 2372@{ 2373 int tem = foo (z); 2374 next = min (x + y, tem); 2375@} 2376@end group 2377@end example 2378 2379@noindent 2380(where we assume that @code{foo} returns type @code{int}). 2381 2382@node Self-Referential Macros 2383@subsection Self-Referential Macros 2384@cindex self-reference 2385 2386A @dfn{self-referential} macro is one whose name appears in its 2387definition. Recall that all macro definitions are rescanned for more 2388macros to replace. If the self-reference were considered a use of the 2389macro, it would produce an infinitely large expansion. To prevent this, 2390the self-reference is not considered a macro call. It is passed into 2391the preprocessor output unchanged. Let's consider an example: 2392 2393@example 2394#define foo (4 + foo) 2395@end example 2396 2397@noindent 2398where @code{foo} is also a variable in your program. 2399 2400Following the ordinary rules, each reference to @code{foo} will expand 2401into @code{(4 + foo)}; then this will be rescanned and will expand into 2402@code{(4 + (4 + foo))}; and so on until the computer runs out of memory. 2403 2404The self-reference rule cuts this process short after one step, at 2405@code{(4 + foo)}. Therefore, this macro definition has the possibly 2406useful effect of causing the program to add 4 to the value of @code{foo} 2407wherever @code{foo} is referred to. 2408 2409In most cases, it is a bad idea to take advantage of this feature. A 2410person reading the program who sees that @code{foo} is a variable will 2411not expect that it is a macro as well. The reader will come across the 2412identifier @code{foo} in the program and think its value should be that 2413of the variable @code{foo}, whereas in fact the value is four greater. 2414 2415One common, useful use of self-reference is to create a macro which 2416expands to itself. If you write 2417 2418@example 2419#define EPERM EPERM 2420@end example 2421 2422@noindent 2423then the macro @code{EPERM} expands to @code{EPERM}. Effectively, it is 2424left alone by the preprocessor whenever it's used in running text. You 2425can tell that it's a macro with @samp{#ifdef}. You might do this if you 2426want to define numeric constants with an @code{enum}, but have 2427@samp{#ifdef} be true for each constant. 2428 2429If a macro @code{x} expands to use a macro @code{y}, and the expansion of 2430@code{y} refers to the macro @code{x}, that is an @dfn{indirect 2431self-reference} of @code{x}. @code{x} is not expanded in this case 2432either. Thus, if we have 2433 2434@example 2435#define x (4 + y) 2436#define y (2 * x) 2437@end example 2438 2439@noindent 2440then @code{x} and @code{y} expand as follows: 2441 2442@example 2443@group 2444x @expansion{} (4 + y) 2445 @expansion{} (4 + (2 * x)) 2446 2447y @expansion{} (2 * x) 2448 @expansion{} (2 * (4 + y)) 2449@end group 2450@end example 2451 2452@noindent 2453Each macro is expanded when it appears in the definition of the other 2454macro, but not when it indirectly appears in its own definition. 2455 2456@node Argument Prescan 2457@subsection Argument Prescan 2458@cindex expansion of arguments 2459@cindex macro argument expansion 2460@cindex prescan of macro arguments 2461 2462Macro arguments are completely macro-expanded before they are 2463substituted into a macro body, unless they are stringified or pasted 2464with other tokens. After substitution, the entire macro body, including 2465the substituted arguments, is scanned again for macros to be expanded. 2466The result is that the arguments are scanned @emph{twice} to expand 2467macro calls in them. 2468 2469Most of the time, this has no effect. If the argument contained any 2470macro calls, they are expanded during the first scan. The result 2471therefore contains no macro calls, so the second scan does not change 2472it. If the argument were substituted as given, with no prescan, the 2473single remaining scan would find the same macro calls and produce the 2474same results. 2475 2476You might expect the double scan to change the results when a 2477self-referential macro is used in an argument of another macro 2478(@pxref{Self-Referential Macros}): the self-referential macro would be 2479expanded once in the first scan, and a second time in the second scan. 2480However, this is not what happens. The self-references that do not 2481expand in the first scan are marked so that they will not expand in the 2482second scan either. 2483 2484You might wonder, ``Why mention the prescan, if it makes no difference? 2485And why not skip it and make the preprocessor faster?'' The answer is 2486that the prescan does make a difference in three special cases: 2487 2488@itemize @bullet 2489@item 2490Nested calls to a macro. 2491 2492We say that @dfn{nested} calls to a macro occur when a macro's argument 2493contains a call to that very macro. For example, if @code{f} is a macro 2494that expects one argument, @code{f (f (1))} is a nested pair of calls to 2495@code{f}. The desired expansion is made by expanding @code{f (1)} and 2496substituting that into the definition of @code{f}. The prescan causes 2497the expected result to happen. Without the prescan, @code{f (1)} itself 2498would be substituted as an argument, and the inner use of @code{f} would 2499appear during the main scan as an indirect self-reference and would not 2500be expanded. 2501 2502@item 2503Macros that call other macros that stringify or concatenate. 2504 2505If an argument is stringified or concatenated, the prescan does not 2506occur. If you @emph{want} to expand a macro, then stringify or 2507concatenate its expansion, you can do that by causing one macro to call 2508another macro that does the stringification or concatenation. For 2509instance, if you have 2510 2511@example 2512#define AFTERX(x) X_ ## x 2513#define XAFTERX(x) AFTERX(x) 2514#define TABLESIZE 1024 2515#define BUFSIZE TABLESIZE 2516@end example 2517 2518then @code{AFTERX(BUFSIZE)} expands to @code{X_BUFSIZE}, and 2519@code{XAFTERX(BUFSIZE)} expands to @code{X_1024}. (Not to 2520@code{X_TABLESIZE}. Prescan always does a complete expansion.) 2521 2522@item 2523Macros used in arguments, whose expansions contain unshielded commas. 2524 2525This can cause a macro expanded on the second scan to be called with the 2526wrong number of arguments. Here is an example: 2527 2528@example 2529#define foo a,b 2530#define bar(x) lose(x) 2531#define lose(x) (1 + (x)) 2532@end example 2533 2534We would like @code{bar(foo)} to turn into @code{(1 + (foo))}, which 2535would then turn into @code{(1 + (a,b))}. Instead, @code{bar(foo)} 2536expands into @code{lose(a,b)}, and you get an error because @code{lose} 2537requires a single argument. In this case, the problem is easily solved 2538by the same parentheses that ought to be used to prevent misnesting of 2539arithmetic operations: 2540 2541@example 2542#define foo (a,b) 2543@exdent or 2544#define bar(x) lose((x)) 2545@end example 2546 2547The extra pair of parentheses prevents the comma in @code{foo}'s 2548definition from being interpreted as an argument separator. 2549 2550@end itemize 2551 2552@node Newlines in Arguments 2553@subsection Newlines in Arguments 2554@cindex newlines in macro arguments 2555 2556The invocation of a function-like macro can extend over many logical 2557lines. However, in the present implementation, the entire expansion 2558comes out on one line. Thus line numbers emitted by the compiler or 2559debugger refer to the line the invocation started on, which might be 2560different to the line containing the argument causing the problem. 2561 2562Here is an example illustrating this: 2563 2564@example 2565#define ignore_second_arg(a,b,c) a; c 2566 2567ignore_second_arg (foo (), 2568 ignored (), 2569 syntax error); 2570@end example 2571 2572@noindent 2573The syntax error triggered by the tokens @code{syntax error} results in 2574an error message citing line three---the line of ignore_second_arg--- 2575even though the problematic code comes from line five. 2576 2577We consider this a bug, and intend to fix it in the near future. 2578 2579@node Conditionals 2580@chapter Conditionals 2581@cindex conditionals 2582 2583A @dfn{conditional} is a directive that instructs the preprocessor to 2584select whether or not to include a chunk of code in the final token 2585stream passed to the compiler. Preprocessor conditionals can test 2586arithmetic expressions, or whether a name is defined as a macro, or both 2587simultaneously using the special @code{defined} operator. 2588 2589A conditional in the C preprocessor resembles in some ways an @code{if} 2590statement in C, but it is important to understand the difference between 2591them. The condition in an @code{if} statement is tested during the 2592execution of your program. Its purpose is to allow your program to 2593behave differently from run to run, depending on the data it is 2594operating on. The condition in a preprocessing conditional directive is 2595tested when your program is compiled. Its purpose is to allow different 2596code to be included in the program depending on the situation at the 2597time of compilation. 2598 2599However, the distinction is becoming less clear. Modern compilers often 2600do test @code{if} statements when a program is compiled, if their 2601conditions are known not to vary at run time, and eliminate code which 2602can never be executed. If you can count on your compiler to do this, 2603you may find that your program is more readable if you use @code{if} 2604statements with constant conditions (perhaps determined by macros). Of 2605course, you can only use this to exclude code, not type definitions or 2606other preprocessing directives, and you can only do it if the code 2607remains syntactically valid when it is not to be used. 2608 2609GCC version 3 eliminates this kind of never-executed code even when 2610not optimizing. Older versions did it only when optimizing. 2611 2612@menu 2613* Conditional Uses:: 2614* Conditional Syntax:: 2615* Deleted Code:: 2616@end menu 2617 2618@node Conditional Uses 2619@section Conditional Uses 2620 2621There are three general reasons to use a conditional. 2622 2623@itemize @bullet 2624@item 2625A program may need to use different code depending on the machine or 2626operating system it is to run on. In some cases the code for one 2627operating system may be erroneous on another operating system; for 2628example, it might refer to data types or constants that do not exist on 2629the other system. When this happens, it is not enough to avoid 2630executing the invalid code. Its mere presence will cause the compiler 2631to reject the program. With a preprocessing conditional, the offending 2632code can be effectively excised from the program when it is not valid. 2633 2634@item 2635You may want to be able to compile the same source file into two 2636different programs. One version might make frequent time-consuming 2637consistency checks on its intermediate data, or print the values of 2638those data for debugging, and the other not. 2639 2640@item 2641A conditional whose condition is always false is one way to exclude code 2642from the program but keep it as a sort of comment for future reference. 2643@end itemize 2644 2645Simple programs that do not need system-specific logic or complex 2646debugging hooks generally will not need to use preprocessing 2647conditionals. 2648 2649@node Conditional Syntax 2650@section Conditional Syntax 2651 2652@findex #if 2653A conditional in the C preprocessor begins with a @dfn{conditional 2654directive}: @samp{#if}, @samp{#ifdef} or @samp{#ifndef}. 2655 2656@menu 2657* Ifdef:: 2658* If:: 2659* Defined:: 2660* Else:: 2661* Elif:: 2662@end menu 2663 2664@node Ifdef 2665@subsection Ifdef 2666@findex #ifdef 2667@findex #endif 2668 2669The simplest sort of conditional is 2670 2671@example 2672@group 2673#ifdef @var{MACRO} 2674 2675@var{controlled text} 2676 2677#endif /* @var{MACRO} */ 2678@end group 2679@end example 2680 2681@cindex conditional group 2682This block is called a @dfn{conditional group}. @var{controlled text} 2683will be included in the output of the preprocessor if and only if 2684@var{MACRO} is defined. We say that the conditional @dfn{succeeds} if 2685@var{MACRO} is defined, @dfn{fails} if it is not. 2686 2687The @var{controlled text} inside of a conditional can include 2688preprocessing directives. They are executed only if the conditional 2689succeeds. You can nest conditional groups inside other conditional 2690groups, but they must be completely nested. In other words, 2691@samp{#endif} always matches the nearest @samp{#ifdef} (or 2692@samp{#ifndef}, or @samp{#if}). Also, you cannot start a conditional 2693group in one file and end it in another. 2694 2695Even if a conditional fails, the @var{controlled text} inside it is 2696still run through initial transformations and tokenization. Therefore, 2697it must all be lexically valid C@. Normally the only way this matters is 2698that all comments and string literals inside a failing conditional group 2699must still be properly ended. 2700 2701The comment following the @samp{#endif} is not required, but it is a 2702good practice if there is a lot of @var{controlled text}, because it 2703helps people match the @samp{#endif} to the corresponding @samp{#ifdef}. 2704Older programs sometimes put @var{MACRO} directly after the 2705@samp{#endif} without enclosing it in a comment. This is invalid code 2706according to the C standard. GNU CPP accepts it with a warning. It 2707never affects which @samp{#ifndef} the @samp{#endif} matches. 2708 2709@findex #ifndef 2710Sometimes you wish to use some code if a macro is @emph{not} defined. 2711You can do this by writing @samp{#ifndef} instead of @samp{#ifdef}. 2712One common use of @samp{#ifndef} is to include code only the first 2713time a header file is included. @xref{Once-Only Headers}. 2714 2715Macro definitions can vary between compilations for several reasons. 2716Here are some samples. 2717 2718@itemize @bullet 2719@item 2720Some macros are predefined on each kind of machine 2721(@pxref{System-specific Predefined Macros}). This allows you to provide 2722code specially tuned for a particular machine. 2723 2724@item 2725System header files define more macros, associated with the features 2726they implement. You can test these macros with conditionals to avoid 2727using a system feature on a machine where it is not implemented. 2728 2729@item 2730Macros can be defined or undefined with the @option{-D} and @option{-U} 2731command line options when you compile the program. You can arrange to 2732compile the same source file into two different programs by choosing a 2733macro name to specify which program you want, writing conditionals to 2734test whether or how this macro is defined, and then controlling the 2735state of the macro with command line options, perhaps set in the 2736Makefile. @xref{Invocation}. 2737 2738@item 2739Your program might have a special header file (often called 2740@file{config.h}) that is adjusted when the program is compiled. It can 2741define or not define macros depending on the features of the system and 2742the desired capabilities of the program. The adjustment can be 2743automated by a tool such as @command{autoconf}, or done by hand. 2744@end itemize 2745 2746@node If 2747@subsection If 2748 2749The @samp{#if} directive allows you to test the value of an arithmetic 2750expression, rather than the mere existence of one macro. Its syntax is 2751 2752@example 2753@group 2754#if @var{expression} 2755 2756@var{controlled text} 2757 2758#endif /* @var{expression} */ 2759@end group 2760@end example 2761 2762@var{expression} is a C expression of integer type, subject to stringent 2763restrictions. It may contain 2764 2765@itemize @bullet 2766@item 2767Integer constants. 2768 2769@item 2770Character constants, which are interpreted as they would be in normal 2771code. 2772 2773@item 2774Arithmetic operators for addition, subtraction, multiplication, 2775division, bitwise operations, shifts, comparisons, and logical 2776operations (@code{&&} and @code{||}). The latter two obey the usual 2777short-circuiting rules of standard C@. 2778 2779@item 2780Macros. All macros in the expression are expanded before actual 2781computation of the expression's value begins. 2782 2783@item 2784Uses of the @code{defined} operator, which lets you check whether macros 2785are defined in the middle of an @samp{#if}. 2786 2787@item 2788Identifiers that are not macros, which are all considered to be the 2789number zero. This allows you to write @code{@w{#if MACRO}} instead of 2790@code{@w{#ifdef MACRO}}, if you know that MACRO, when defined, will 2791always have a nonzero value. Function-like macros used without their 2792function call parentheses are also treated as zero. 2793 2794In some contexts this shortcut is undesirable. The @option{-Wundef} 2795option causes GCC to warn whenever it encounters an identifier which is 2796not a macro in an @samp{#if}. 2797@end itemize 2798 2799The preprocessor does not know anything about types in the language. 2800Therefore, @code{sizeof} operators are not recognized in @samp{#if}, and 2801neither are @code{enum} constants. They will be taken as identifiers 2802which are not macros, and replaced by zero. In the case of 2803@code{sizeof}, this is likely to cause the expression to be invalid. 2804 2805The preprocessor calculates the value of @var{expression}. It carries 2806out all calculations in the widest integer type known to the compiler; 2807on most machines supported by GCC this is 64 bits. This is not the same 2808rule as the compiler uses to calculate the value of a constant 2809expression, and may give different results in some cases. If the value 2810comes out to be nonzero, the @samp{#if} succeeds and the @var{controlled 2811text} is included; otherwise it is skipped. 2812 2813If @var{expression} is not correctly formed, GCC issues an error and 2814treats the conditional as having failed. 2815 2816@node Defined 2817@subsection Defined 2818 2819@cindex @code{defined} 2820The special operator @code{defined} is used in @samp{#if} and 2821@samp{#elif} expressions to test whether a certain name is defined as a 2822macro. @code{defined @var{name}} and @code{defined (@var{name})} are 2823both expressions whose value is 1 if @var{name} is defined as a macro at 2824the current point in the program, and 0 otherwise. Thus, @code{@w{#if 2825defined MACRO}} is precisely equivalent to @code{@w{#ifdef MACRO}}. 2826 2827@code{defined} is useful when you wish to test more than one macro for 2828existence at once. For example, 2829 2830@example 2831#if defined (__vax__) || defined (__ns16000__) 2832@end example 2833 2834@noindent 2835would succeed if either of the names @code{__vax__} or 2836@code{__ns16000__} is defined as a macro. 2837 2838Conditionals written like this: 2839 2840@example 2841#if defined BUFSIZE && BUFSIZE >= 1024 2842@end example 2843 2844@noindent 2845can generally be simplified to just @code{@w{#if BUFSIZE >= 1024}}, 2846since if @code{BUFSIZE} is not defined, it will be interpreted as having 2847the value zero. 2848 2849If the @code{defined} operator appears as a result of a macro expansion, 2850the C standard says the behavior is undefined. GNU cpp treats it as a 2851genuine @code{defined} operator and evaluates it normally. It will warn 2852wherever your code uses this feature if you use the command-line option 2853@option{-pedantic}, since other compilers may handle it differently. 2854 2855@node Else 2856@subsection Else 2857 2858@findex #else 2859The @samp{#else} directive can be added to a conditional to provide 2860alternative text to be used if the condition fails. This is what it 2861looks like: 2862 2863@example 2864@group 2865#if @var{expression} 2866@var{text-if-true} 2867#else /* Not @var{expression} */ 2868@var{text-if-false} 2869#endif /* Not @var{expression} */ 2870@end group 2871@end example 2872 2873@noindent 2874If @var{expression} is nonzero, the @var{text-if-true} is included and 2875the @var{text-if-false} is skipped. If @var{expression} is zero, the 2876opposite happens. 2877 2878You can use @samp{#else} with @samp{#ifdef} and @samp{#ifndef}, too. 2879 2880@node Elif 2881@subsection Elif 2882 2883@findex #elif 2884One common case of nested conditionals is used to check for more than two 2885possible alternatives. For example, you might have 2886 2887@example 2888#if X == 1 2889@dots{} 2890#else /* X != 1 */ 2891#if X == 2 2892@dots{} 2893#else /* X != 2 */ 2894@dots{} 2895#endif /* X != 2 */ 2896#endif /* X != 1 */ 2897@end example 2898 2899Another conditional directive, @samp{#elif}, allows this to be 2900abbreviated as follows: 2901 2902@example 2903#if X == 1 2904@dots{} 2905#elif X == 2 2906@dots{} 2907#else /* X != 2 and X != 1*/ 2908@dots{} 2909#endif /* X != 2 and X != 1*/ 2910@end example 2911 2912@samp{#elif} stands for ``else if''. Like @samp{#else}, it goes in the 2913middle of a conditional group and subdivides it; it does not require a 2914matching @samp{#endif} of its own. Like @samp{#if}, the @samp{#elif} 2915directive includes an expression to be tested. The text following the 2916@samp{#elif} is processed only if the original @samp{#if}-condition 2917failed and the @samp{#elif} condition succeeds. 2918 2919More than one @samp{#elif} can go in the same conditional group. Then 2920the text after each @samp{#elif} is processed only if the @samp{#elif} 2921condition succeeds after the original @samp{#if} and all previous 2922@samp{#elif} directives within it have failed. 2923 2924@samp{#else} is allowed after any number of @samp{#elif} directives, but 2925@samp{#elif} may not follow @samp{#else}. 2926 2927@node Deleted Code 2928@section Deleted Code 2929@cindex commenting out code 2930 2931If you replace or delete a part of the program but want to keep the old 2932code around for future reference, you often cannot simply comment it 2933out. Block comments do not nest, so the first comment inside the old 2934code will end the commenting-out. The probable result is a flood of 2935syntax errors. 2936 2937One way to avoid this problem is to use an always-false conditional 2938instead. For instance, put @code{#if 0} before the deleted code and 2939@code{#endif} after it. This works even if the code being turned 2940off contains conditionals, but they must be entire conditionals 2941(balanced @samp{#if} and @samp{#endif}). 2942 2943Some people use @code{#ifdef notdef} instead. This is risky, because 2944@code{notdef} might be accidentally defined as a macro, and then the 2945conditional would succeed. @code{#if 0} can be counted on to fail. 2946 2947Do not use @code{#if 0} for comments which are not C code. Use a real 2948comment, instead. The interior of @code{#if 0} must consist of complete 2949tokens; in particular, single-quote characters must balance. Comments 2950often contain unbalanced single-quote characters (known in English as 2951apostrophes). These confuse @code{#if 0}. They don't confuse 2952@samp{/*}. 2953 2954@node Diagnostics 2955@chapter Diagnostics 2956@cindex diagnostic 2957@cindex reporting errors 2958@cindex reporting warnings 2959 2960@findex #error 2961The directive @samp{#error} causes the preprocessor to report a fatal 2962error. The tokens forming the rest of the line following @samp{#error} 2963are used as the error message. 2964 2965You would use @samp{#error} inside of a conditional that detects a 2966combination of parameters which you know the program does not properly 2967support. For example, if you know that the program will not run 2968properly on a VAX, you might write 2969 2970@example 2971@group 2972#ifdef __vax__ 2973#error "Won't work on VAXen. See comments at get_last_object." 2974#endif 2975@end group 2976@end example 2977 2978If you have several configuration parameters that must be set up by 2979the installation in a consistent way, you can use conditionals to detect 2980an inconsistency and report it with @samp{#error}. For example, 2981 2982@example 2983#if !defined(UNALIGNED_INT_ASM_OP) && defined(DWARF2_DEBUGGING_INFO) 2984#error "DWARF2_DEBUGGING_INFO requires UNALIGNED_INT_ASM_OP." 2985#endif 2986@end example 2987 2988@findex #warning 2989The directive @samp{#warning} is like @samp{#error}, but causes the 2990preprocessor to issue a warning and continue preprocessing. The tokens 2991following @samp{#warning} are used as the warning message. 2992 2993You might use @samp{#warning} in obsolete header files, with a message 2994directing the user to the header file which should be used instead. 2995 2996Neither @samp{#error} nor @samp{#warning} macro-expands its argument. 2997Internal whitespace sequences are each replaced with a single space. 2998The line must consist of complete tokens. It is wisest to make the 2999argument of these directives be a single string constant; this avoids 3000problems with apostrophes and the like. 3001 3002@node Line Control 3003@chapter Line Control 3004@cindex line control 3005 3006The C preprocessor informs the C compiler of the location in your source 3007code where each token came from. Presently, this is just the file name 3008and line number. All the tokens resulting from macro expansion are 3009reported as having appeared on the line of the source file where the 3010outermost macro was used. We intend to be more accurate in the future. 3011 3012If you write a program which generates source code, such as the 3013@command{bison} parser generator, you may want to adjust the preprocessor's 3014notion of the current file name and line number by hand. Parts of the 3015output from @command{bison} are generated from scratch, other parts come 3016from a standard parser file. The rest are copied verbatim from 3017@command{bison}'s input. You would like compiler error messages and 3018symbolic debuggers to be able to refer to @code{bison}'s input file. 3019 3020@findex #line 3021@command{bison} or any such program can arrange this by writing 3022@samp{#line} directives into the output file. @samp{#line} is a 3023directive that specifies the original line number and source file name 3024for subsequent input in the current preprocessor input file. 3025@samp{#line} has three variants: 3026 3027@table @code 3028@item #line @var{linenum} 3029@var{linenum} is a non-negative decimal integer constant. It specifies 3030the line number which should be reported for the following line of 3031input. Subsequent lines are counted from @var{linenum}. 3032 3033@item #line @var{linenum} @var{filename} 3034@var{linenum} is the same as for the first form, and has the same 3035effect. In addition, @var{filename} is a string constant. The 3036following line and all subsequent lines are reported to come from the 3037file it specifies, until something else happens to change that. 3038 3039@item #line @var{anything else} 3040@var{anything else} is checked for macro calls, which are expanded. 3041The result should match one of the above two forms. 3042@end table 3043 3044@samp{#line} directives alter the results of the @code{__FILE__} and 3045@code{__LINE__} predefined macros from that point on. @xref{Standard 3046Predefined Macros}. They do not have any effect on @samp{#include}'s 3047idea of the directory containing the current file. 3048 3049@node Pragmas 3050@chapter Pragmas 3051 3052The @samp{#pragma} directive is the method specified by the C standard 3053for providing additional information to the compiler, beyond what is 3054conveyed in the language itself. Three forms of this directive 3055(commonly known as @dfn{pragmas}) are specified by the 1999 C standard. 3056A C compiler is free to attach any meaning it likes to other pragmas. 3057 3058GCC has historically preferred to use extensions to the syntax of the 3059language, such as @code{__attribute__}, for this purpose. However, GCC 3060does define a few pragmas of its own. These mostly have effects on the 3061entire translation unit or source file. 3062 3063In GCC version 3, all GNU-defined, supported pragmas have been given a 3064@code{GCC} prefix. This is in line with the @code{STDC} prefix on all 3065pragmas defined by C99. For backward compatibility, pragmas which were 3066recognized by previous versions are still recognized without the 3067@code{GCC} prefix, but that usage is deprecated. Some older pragmas are 3068deprecated in their entirety. They are not recognized with the 3069@code{GCC} prefix. @xref{Obsolete Features}. 3070 3071@cindex @code{_Pragma} 3072C99 introduces the @code{@w{_Pragma}} operator. This feature addresses a 3073major problem with @samp{#pragma}: being a directive, it cannot be 3074produced as the result of macro expansion. @code{@w{_Pragma}} is an 3075operator, much like @code{sizeof} or @code{defined}, and can be embedded 3076in a macro. 3077 3078Its syntax is @code{@w{_Pragma (@var{string-literal})}}, where 3079@var{string-literal} can be either a normal or wide-character string 3080literal. It is destringized, by replacing all @samp{\\} with a single 3081@samp{\} and all @samp{\"} with a @samp{"}. The result is then 3082processed as if it had appeared as the right hand side of a 3083@samp{#pragma} directive. For example, 3084 3085@example 3086_Pragma ("GCC dependency \"parse.y\"") 3087@end example 3088 3089@noindent 3090has the same effect as @code{#pragma GCC dependency "parse.y"}. The 3091same effect could be achieved using macros, for example 3092 3093@example 3094#define DO_PRAGMA(x) _Pragma (#x) 3095DO_PRAGMA (GCC dependency "parse.y") 3096@end example 3097 3098The standard is unclear on where a @code{_Pragma} operator can appear. 3099The preprocessor does not accept it within a preprocessing conditional 3100directive like @samp{#if}. To be safe, you are probably best keeping it 3101out of directives other than @samp{#define}, and putting it on a line of 3102its own. 3103 3104This manual documents the pragmas which are meaningful to the 3105preprocessor itself. Other pragmas are meaningful to the C or C++ 3106compilers. They are documented in the GCC manual. 3107 3108@ftable @code 3109@item #pragma GCC dependency 3110@code{#pragma GCC dependency} allows you to check the relative dates of 3111the current file and another file. If the other file is more recent than 3112the current file, a warning is issued. This is useful if the current 3113file is derived from the other file, and should be regenerated. The 3114other file is searched for using the normal include search path. 3115Optional trailing text can be used to give more information in the 3116warning message. 3117 3118@example 3119#pragma GCC dependency "parse.y" 3120#pragma GCC dependency "/usr/include/time.h" rerun fixincludes 3121@end example 3122 3123@item #pragma GCC poison 3124Sometimes, there is an identifier that you want to remove completely 3125from your program, and make sure that it never creeps back in. To 3126enforce this, you can @dfn{poison} the identifier with this pragma. 3127@code{#pragma GCC poison} is followed by a list of identifiers to 3128poison. If any of those identifiers appears anywhere in the source 3129after the directive, it is a hard error. For example, 3130 3131@example 3132#pragma GCC poison printf sprintf fprintf 3133sprintf(some_string, "hello"); 3134@end example 3135 3136@noindent 3137will produce an error. 3138 3139If a poisoned identifier appears as part of the expansion of a macro 3140which was defined before the identifier was poisoned, it will @emph{not} 3141cause an error. This lets you poison an identifier without worrying 3142about system headers defining macros that use it. 3143 3144For example, 3145 3146@example 3147#define strrchr rindex 3148#pragma GCC poison rindex 3149strrchr(some_string, 'h'); 3150@end example 3151 3152@noindent 3153will not produce an error. 3154 3155@item #pragma GCC system_header 3156This pragma takes no arguments. It causes the rest of the code in the 3157current file to be treated as if it came from a system header. 3158@xref{System Headers}. 3159 3160@end ftable 3161 3162@node Other Directives 3163@chapter Other Directives 3164 3165@findex #ident 3166The @samp{#ident} directive takes one argument, a string constant. On 3167some systems, that string constant is copied into a special segment of 3168the object file. On other systems, the directive is ignored. 3169 3170This directive is not part of the C standard, but it is not an official 3171GNU extension either. We believe it came from System V@. 3172 3173@findex #sccs 3174The @samp{#sccs} directive is recognized on some systems, because it 3175appears in their header files. It is a very old, obscure, extension 3176which we did not invent, and we have been unable to find any 3177documentation of what it should do, so GCC simply ignores it. 3178 3179@cindex null directive 3180The @dfn{null directive} consists of a @samp{#} followed by a newline, 3181with only whitespace (including comments) in between. A null directive 3182is understood as a preprocessing directive but has no effect on the 3183preprocessor output. The primary significance of the existence of the 3184null directive is that an input line consisting of just a @samp{#} will 3185produce no output, rather than a line of output containing just a 3186@samp{#}. Supposedly some old C programs contain such lines. 3187 3188@node Preprocessor Output 3189@chapter Preprocessor Output 3190 3191When the C preprocessor is used with the C, C++, or Objective-C 3192compilers, it is integrated into the compiler and communicates a stream 3193of binary tokens directly to the compiler's parser. However, it can 3194also be used in the more conventional standalone mode, where it produces 3195textual output. 3196@c FIXME: Document the library interface. 3197 3198@cindex output format 3199The output from the C preprocessor looks much like the input, except 3200that all preprocessing directive lines have been replaced with blank 3201lines and all comments with spaces. Long runs of blank lines are 3202discarded. 3203 3204The ISO standard specifies that it is implementation defined whether a 3205preprocessor preserves whitespace between tokens, or replaces it with 3206e.g.@: a single space. In GNU CPP, whitespace between tokens is collapsed 3207to become a single space, with the exception that the first token on a 3208non-directive line is preceded with sufficient spaces that it appears in 3209the same column in the preprocessed output that it appeared in the 3210original source file. This is so the output is easy to read. 3211@xref{Differences from previous versions}. CPP does not insert any 3212whitespace where there was none in the original source, except where 3213necessary to prevent an accidental token paste. 3214 3215@cindex linemarkers 3216Source file name and line number information is conveyed by lines 3217of the form 3218 3219@example 3220# @var{linenum} @var{filename} @var{flags} 3221@end example 3222 3223@noindent 3224These are called @dfn{linemarkers}. They are inserted as needed into 3225the output (but never within a string or character constant). They mean 3226that the following line originated in file @var{filename} at line 3227@var{linenum}. 3228 3229After the file name comes zero or more flags, which are @samp{1}, 3230@samp{2}, @samp{3}, or @samp{4}. If there are multiple flags, spaces 3231separate them. Here is what the flags mean: 3232 3233@table @samp 3234@item 1 3235This indicates the start of a new file. 3236@item 2 3237This indicates returning to a file (after having included another file). 3238@item 3 3239This indicates that the following text comes from a system header file, 3240so certain warnings should be suppressed. 3241@item 4 3242This indicates that the following text should be treated as being 3243wrapped in an implicit @code{extern "C"} block. 3244@c maybe cross reference NO_IMPLICIT_EXTERN_C 3245@end table 3246 3247As an extension, the preprocessor accepts linemarkers in non-assembler 3248input files. They are treated like the corresponding @samp{#line} 3249directive, (@pxref{Line Control}), except that trailing flags are 3250permitted, and are interpreted with the meanings described above. If 3251multiple flags are given, they must be in ascending order. 3252 3253Some directives may be duplicated in the output of the preprocessor. 3254These are @samp{#ident} (always), @samp{#pragma} (only if the 3255preprocessor does not handle the pragma itself), and @samp{#define} and 3256@samp{#undef} (with certain debugging options). If this happens, the 3257@samp{#} of the directive will always be in the first column, and there 3258will be no space between the @samp{#} and the directive name. If macro 3259expansion happens to generate tokens which might be mistaken for a 3260duplicated directive, a space will be inserted between the @samp{#} and 3261the directive name. 3262 3263@node Traditional Mode 3264@chapter Traditional Mode 3265 3266Traditional (pre-standard) C preprocessing is rather different from 3267the preprocessing specified by the standard. When GCC is given the 3268@option{-traditional} option, it attempts to emulate a traditional 3269preprocessor. We do not guarantee that GCC's behavior under 3270@option{-traditional} matches any pre-standard preprocessor exactly. 3271 3272Traditional mode exists only for backward compatibility. We have no 3273plans to augment it in any way nor will we change it except to fix 3274catastrophic bugs. You should be aware that modern C libraries often 3275have header files which are incompatible with traditional mode. 3276 3277This is a list of the differences. It may not be complete, and may not 3278correspond exactly to the behavior of either GCC or a true traditional 3279preprocessor. 3280 3281@itemize @bullet 3282@item 3283Traditional macro expansion pays no attention to single-quote or 3284double-quote characters; macro argument symbols are replaced by the 3285argument values even when they appear within apparent string or 3286character constants. 3287 3288@item 3289Traditionally, it is permissible for a macro expansion to end in the 3290middle of a string or character constant. The constant continues into 3291the text surrounding the macro call. 3292 3293@item 3294However, the end of the line terminates a string or character constant, 3295with no error. (This is a kluge. Traditional mode is commonly used to 3296preprocess things which are not C, and have a different comment syntax. 3297Single apostrophes often appear in comments. This kluge prevents the 3298traditional preprocessor from issuing errors on such comments.) 3299 3300@item 3301Preprocessing directives are recognized in traditional C only when their 3302leading @samp{#} appears in the first column. There can be no 3303whitespace between the beginning of the line and the @samp{#}. 3304 3305@item 3306In traditional C, a comment is equivalent to no text at all. (In ISO 3307C, a comment counts as whitespace.) It can be used sort of the same way 3308that @samp{##} is used in ISO C, to paste macro arguments together. 3309 3310@item 3311Traditional C does not have the concept of a preprocessing number. 3312 3313@item 3314A macro is not suppressed within its own definition, in traditional C@. 3315Thus, any macro that is used recursively inevitably causes an error. 3316 3317@item 3318The @samp{#} and @samp{##} operators are not available in traditional 3319C@. 3320 3321@item 3322In traditional C, the text at the end of a macro expansion can run 3323together with the text after the macro call, to produce a single token. 3324This is impossible in ISO C@. 3325 3326@item 3327None of the GNU extensions to the preprocessor are available in 3328traditional mode, with the exception of a partial implementation of 3329assertions, and those may be removed in the future. 3330 3331@item 3332A true traditional C preprocessor does not recognize @samp{#elif}, 3333@samp{#error}, or @samp{#pragma}. GCC supports @samp{#elif} and 3334@samp{#error} even in traditional mode, but not @samp{#pragma}. 3335 3336@item 3337Traditional mode is text-based, not token-based, and comments are 3338stripped after macro expansion. Therefore, @samp{/**/} can be used to 3339paste tokens together provided that there is no whitespace between it 3340and the tokens to be pasted. 3341 3342@item 3343Traditional mode preserves the amount and form of whitespace provided by 3344the user. Hard tabs remain hard tabs. This can be useful, e.g.@: if you 3345are preprocessing a Makefile (which we do not encourage). 3346@end itemize 3347 3348You can request warnings about features that did not exist, or worked 3349differently, in traditional C with the @option{-Wtraditional} option. 3350This works only if you do @emph{not} specify @option{-traditional}. GCC 3351does not warn about features of ISO C which you must use when you are 3352using a conforming compiler, such as the @samp{#} and @samp{##} 3353operators. 3354 3355Presently @option{-Wtraditional} warns about: 3356 3357@itemize @bullet 3358@item 3359Macro parameters that appear within string literals in the macro body. 3360In traditional C macro replacement takes place within string literals, 3361but does not in ISO C@. 3362 3363@item 3364In traditional C, some preprocessor directives did not exist. 3365Traditional preprocessors would only consider a line to be a directive 3366if the @samp{#} appeared in column 1 on the line. Therefore 3367@option{-Wtraditional} warns about directives that traditional C 3368understands but would ignore because the @samp{#} does not appear as the 3369first character on the line. It also suggests you hide directives like 3370@samp{#pragma} not understood by traditional C by indenting them. Some 3371traditional implementations would not recognize @samp{#elif}, so it 3372suggests avoiding it altogether. 3373 3374@item 3375A function-like macro that appears without an argument list. In 3376traditional C this was an error. In ISO C it merely means that the 3377macro is not expanded. 3378 3379@item 3380The unary plus operator. This did not exist in traditional C@. 3381 3382@item 3383The @samp{U} and @samp{LL} integer constant suffixes, which were not 3384available in traditional C@. (Traditional C does support the @samp{L} 3385suffix for simple long integer constants.) You are not warned about 3386uses of these suffixes in macros defined in system headers. For 3387instance, @code{UINT_MAX} may well be defined as @code{4294967295U}, but 3388you will not be warned if you use @code{UINT_MAX}. 3389 3390You can usually avoid the warning, and the related warning about 3391constants which are so large that they are unsigned, by writing the 3392integer constant in question in hexadecimal, with no U suffix. Take 3393care, though, because this gives the wrong result in exotic cases. 3394@end itemize 3395 3396@node Implementation Details 3397@chapter Implementation Details 3398 3399Here we document details of how the preprocessor's implementation 3400affects its user-visible behavior. You should try to avoid undue 3401reliance on behavior described here, as it is possible that it will 3402change subtly in future implementations. 3403 3404Also documented here are obsolete features and changes from previous 3405versions of GNU CPP@. 3406 3407@menu 3408* Implementation-defined behavior:: 3409* Implementation limits:: 3410* Obsolete Features:: 3411* Differences from previous versions:: 3412@end menu 3413 3414@node Implementation-defined behavior 3415@section Implementation-defined behavior 3416@cindex implementation-defined behavior 3417 3418This is how GNU CPP behaves in all the cases which the C standard 3419describes as @dfn{implementation-defined}. This term means that the 3420implementation is free to do what it likes, but must document its choice 3421and stick to it. 3422@c FIXME: Check the C++ standard for more implementation-defined stuff. 3423 3424@itemize @bullet 3425@need 1000 3426@item The mapping of physical source file multi-byte characters to the 3427execution character set. 3428 3429Currently, GNU cpp only supports character sets that are strict supersets 3430of ASCII, and performs no translation of characters. 3431 3432@item Non-empty sequences of whitespace characters. 3433 3434In textual output, each whitespace sequence is collapsed to a single 3435space. For aesthetic reasons, the first token on each non-directive 3436line of output is preceded with sufficient spaces that it appears in the 3437same column as it did in the original source file. 3438 3439@item The numeric value of character constants in preprocessor expressions. 3440 3441The preprocessor and compiler interpret character constants in the same 3442way; escape sequences such as @samp{\a} are given the values they would 3443have on the target machine. 3444 3445Multi-character character constants are interpreted a character at a 3446time, shifting the previous result left by the number of bits per 3447character on the host, and adding the new character. For example, 'ab' 3448on an 8-bit host would be interpreted as @w{'a' * 256 + 'b'}. If there 3449are more characters in the constant than can fit in the widest native 3450integer type on the host, usually a @code{long}, the excess characters 3451are ignored and a diagnostic is given. 3452 3453@item Source file inclusion. 3454 3455For a discussion on how the preprocessor locates header files, 3456@ref{Include Operation}. 3457 3458@item Interpretation of the filename resulting from a macro-expanded 3459@samp{#include} directive. 3460 3461@xref{Computed Includes}. 3462 3463@item Treatment of a @samp{#pragma} directive that after macro-expansion 3464results in a standard pragma. 3465 3466No macro expansion occurs on any @samp{#pragma} directive line, so the 3467question does not arise. 3468 3469Note that GCC does not yet implement any of the standard 3470pragmas. 3471 3472@end itemize 3473 3474@node Implementation limits 3475@section Implementation limits 3476@cindex implementation limits 3477 3478GNU CPP has a small number of internal limits. This section lists the 3479limits which the C standard requires to be no lower than some minimum, 3480and all the others we are aware of. We intend there to be as few limits 3481as possible. If you encounter an undocumented or inconvenient limit, 3482please report that to us as a bug. (See the section on reporting bugs in 3483the GCC manual.) 3484 3485Where we say something is limited @dfn{only by available memory}, that 3486means that internal data structures impose no intrinsic limit, and space 3487is allocated with @code{malloc} or equivalent. The actual limit will 3488therefore depend on many things, such as the size of other things 3489allocated by the compiler at the same time, the amount of memory 3490consumed by other processes on the same computer, etc. 3491 3492@itemize @bullet 3493 3494@item Nesting levels of @samp{#include} files. 3495 3496We impose an arbitrary limit of 200 levels, to avoid runaway recursion. 3497The standard requires at least 15 levels. 3498 3499@item Nesting levels of conditional inclusion. 3500 3501The C standard mandates this be at least 63. GNU CPP is limited only by 3502available memory. 3503 3504@item Levels of parenthesised expressions within a full expression. 3505 3506The C standard requires this to be at least 63. In preprocessor 3507conditional expressions, it is limited only by available memory. 3508 3509@item Significant initial characters in an identifier or macro name. 3510 3511The preprocessor treats all characters as significant. The C standard 3512requires only that the first 63 be significant. 3513 3514@item Number of macros simultaneously defined in a single translation unit. 3515 3516The standard requires at least 4095 be possible. GNU CPP is limited only 3517by available memory. 3518 3519@item Number of parameters in a macro definition and arguments in a macro call. 3520 3521We allow @code{USHRT_MAX}, which is no smaller than 65,535. The minimum 3522required by the standard is 127. 3523 3524@item Number of characters on a logical source line. 3525 3526The C standard requires a minimum of 4096 be permitted. GNU CPP places 3527no limits on this, but you may get incorrect column numbers reported in 3528diagnostics for lines longer than 65,535 characters. 3529 3530@item Maximum size of a source file. 3531 3532The standard does not specify any lower limit on the maximum size of a 3533source file. GNU cpp maps files into memory, so it is limited by the 3534available address space. This is generally at least two gigabytes. 3535Depending on the operating system, the size of physical memory may or 3536may not be a limitation. 3537 3538@end itemize 3539 3540@node Obsolete Features 3541@section Obsolete Features 3542 3543GNU CPP has a number of features which are present mainly for 3544compatibility with older programs. We discourage their use in new code. 3545In some cases, we plan to remove the feature in a future version of GCC@. 3546 3547@menu 3548* Assertions:: 3549* Obsolete once-only headers:: 3550* Miscellaneous obsolete features:: 3551@end menu 3552 3553@node Assertions 3554@subsection Assertions 3555@cindex assertions 3556 3557@dfn{Assertions} are a deprecated alternative to macros in writing 3558conditionals to test what sort of computer or system the compiled 3559program will run on. Assertions are usually predefined, but you can 3560define them with preprocessing directives or command-line options. 3561 3562Assertions were intended to provide a more systematic way to describe 3563the compiler's target system. However, in practice they are just as 3564unpredictable as the system-specific predefined macros. In addition, they 3565are not part of any standard, and only a few compilers support them. 3566Therefore, the use of assertions is @strong{less} portable than the use 3567of system-specific predefined macros. We recommend you do not use them at 3568all. 3569 3570@cindex predicates 3571An assertion looks like this: 3572 3573@example 3574#@var{predicate} (@var{answer}) 3575@end example 3576 3577@noindent 3578@var{predicate} must be a single identifier. @var{answer} can be any 3579sequence of tokens; all characters are significant except for leading 3580and trailing whitespace, and differences in internal whitespace 3581sequences are ignored. (This is similar to the rules governing macro 3582redefinition.) Thus, @code{(x + y)} is different from @code{(x+y)} but 3583equivalent to @code{@w{( x + y )}}. Parentheses do not nest inside an 3584answer. 3585 3586@cindex testing predicates 3587To test an assertion, you write it in an @samp{#if}. For example, this 3588conditional succeeds if either @code{vax} or @code{ns16000} has been 3589asserted as an answer for @code{machine}. 3590 3591@example 3592#if #machine (vax) || #machine (ns16000) 3593@end example 3594 3595@noindent 3596You can test whether @emph{any} answer is asserted for a predicate by 3597omitting the answer in the conditional: 3598 3599@example 3600#if #machine 3601@end example 3602 3603@findex #assert 3604Assertions are made with the @samp{#assert} directive. Its sole 3605argument is the assertion to make, without the leading @samp{#} that 3606identifies assertions in conditionals. 3607 3608@example 3609#assert @var{predicate} (@var{answer}) 3610@end example 3611 3612@noindent 3613You may make several assertions with the same predicate and different 3614answers. Subsequent assertions do not override previous ones for the 3615same predicate. All the answers for any given predicate are 3616simultaneously true. 3617 3618@cindex assertions, cancelling 3619@findex #unassert 3620Assertions can be cancelled with the @samp{#unassert} directive. It 3621has the same syntax as @samp{#assert}. In that form it cancels only the 3622answer which was specified on the @samp{#unassert} line; other answers 3623for that predicate remain true. You can cancel an entire predicate by 3624leaving out the answer: 3625 3626@example 3627#unassert @var{predicate} 3628@end example 3629 3630@noindent 3631In either form, if no such assertion has been made, @samp{#unassert} has 3632no effect. 3633 3634You can also make or cancel assertions using command line options. 3635@xref{Invocation}. 3636 3637@node Obsolete once-only headers 3638@subsection Obsolete once-only headers 3639 3640GNU CPP supports two more ways of indicating that a header file should be 3641read only once. Neither one is as portable as a wrapper @samp{#ifndef}, 3642and we recommend you do not use them in new programs. 3643 3644@findex #import 3645In the Objective-C language, there is a variant of @samp{#include} 3646called @samp{#import} which includes a file, but does so at most once. 3647If you use @samp{#import} instead of @samp{#include}, then you don't 3648need the conditionals inside the header file to prevent multiple 3649inclusion of the contents. GCC permits the use of @samp{#import} in C 3650and C++ as well as Objective-C@. However, it is not in standard C or C++ 3651and should therefore not be used by portable programs. 3652 3653@samp{#import} is not a well designed feature. It requires the users of 3654a header file to know that it should only be included once. It is much 3655better for the header file's implementor to write the file so that users 3656don't need to know this. Using a wrapper @samp{#ifndef} accomplishes 3657this goal. 3658 3659In the present implementation, a single use of @samp{#import} will 3660prevent the file from ever being read again, by either @samp{#import} or 3661@samp{#include}. You should not rely on this; do not use both 3662@samp{#import} and @samp{#include} to refer to the same header file. 3663 3664Another way to prevent a header file from being included more than once 3665is with the @samp{#pragma once} directive. If @samp{#pragma once} is 3666seen when scanning a header file, that file will never be read again, no 3667matter what. 3668 3669@samp{#pragma once} does not have the problems that @samp{#import} does, 3670but it is not recognized by all preprocessors, so you cannot rely on it 3671in a portable program. 3672 3673@node Miscellaneous obsolete features 3674@subsection Miscellaneous obsolete features 3675 3676Here are a few more obsolete features. 3677 3678@itemize @bullet 3679@cindex invalid token paste 3680@item Attempting to paste two tokens which together do not form a valid 3681preprocessing token. 3682 3683The preprocessor currently warns about this and outputs the two tokens 3684adjacently, which is probably the behavior the programmer intends. It 3685may not work in future, though. 3686 3687Most of the time, when you get this warning, you will find that @samp{##} 3688is being used superstitiously, to guard against whitespace appearing 3689between two tokens. It is almost always safe to delete the @samp{##}. 3690 3691@cindex pragma poison 3692@item @code{#pragma poison} 3693 3694This is the same as @code{#pragma GCC poison}. The version without the 3695@code{GCC} prefix is deprecated. @xref{Pragmas}. 3696 3697@cindex multi-line string constants 3698@item Multi-line string constants 3699 3700GCC currently allows a string constant to extend across multiple logical 3701lines of the source file. This extension is deprecated and will be 3702removed in a future version of GCC@. Such string constants are already 3703rejected in all directives apart from @samp{#define}. 3704 3705Instead, make use of ISO C concatenation of adjacent string literals, or 3706use @samp{\n} followed by a backslash-newline. 3707 3708@end itemize 3709 3710@node Differences from previous versions 3711@section Differences from previous versions 3712@cindex differences from previous versions 3713 3714This section details behavior which has changed from previous versions 3715of GNU CPP@. We do not plan to change it again in the near future, but 3716we do not promise not to, either. 3717 3718The ``previous versions'' discussed here are 2.95 and before. The 3719behavior of GCC 3.0 is mostly the same as the behavior of the widely 3720used 2.96 and 2.97 development snapshots. Where there are differences, 3721they generally represent bugs in the snapshots. 3722 3723@itemize @bullet 3724 3725@item Order of evaluation of @samp{#} and @samp{##} operators 3726 3727The standard does not specify the order of evaluation of a chain of 3728@samp{##} operators, nor whether @samp{#} is evaluated before, after, or 3729at the same time as @samp{##}. You should therefore not write any code 3730which depends on any specific ordering. It is possible to guarantee an 3731ordering, if you need one, by suitable use of nested macros. 3732 3733An example of where this might matter is pasting the arguments @samp{1}, 3734@samp{e} and @samp{-2}. This would be fine for left-to-right pasting, 3735but right-to-left pasting would produce an invalid token @samp{e-2}. 3736 3737GCC 3.0 evaluates @samp{#} and @samp{##} at the same time and strictly 3738left to right. Older versions evaluated all @samp{#} operators first, 3739then all @samp{##} operators, in an unreliable order. 3740 3741@item The form of whitespace betwen tokens in preprocessor output 3742 3743@xref{Preprocessor Output}, for the current textual format. This is 3744also the format used by stringification. Normally, the preprocessor 3745communicates tokens directly to the compiler's parser, and whitespace 3746does not come up at all. 3747 3748Older versions of GCC preserved all whitespace provided by the user and 3749inserted lots more whitespace of their own, because they could not 3750accurately predict when extra spaces were needed to prevent accidental 3751token pasting. 3752 3753@item Optional argument when invoking rest argument macros 3754 3755As an extension, GCC permits you to omit the variable arguments entirely 3756when you use a variable argument macro. This is forbidden by the 1999 C 3757standard, and will provoke a pedantic warning with GCC 3.0. Previous 3758versions accepted it silently. 3759 3760@item @samp{##} swallowing preceding text in rest argument macros 3761 3762Formerly, in a macro expansion, if @samp{##} appeared before a variable 3763arguments parameter, and the set of tokens specified for that argument 3764in the macro invocation was empty, previous versions of GNU CPP would 3765back up and remove the preceding sequence of non-whitespace characters 3766(@strong{not} the preceding token). This extension is in direct 3767conflict with the 1999 C standard and has been drastically pared back. 3768 3769In the current version of the preprocessor, if @samp{##} appears between 3770a comma and a variable arguments parameter, and the variable argument is 3771omitted entirely, the comma will be removed from the expansion. If the 3772variable argument is empty, or the token before @samp{##} is not a 3773comma, then @samp{##} behaves as a normal token paste. 3774 3775@item Traditional mode and GNU extensions 3776 3777Traditional mode used to be implemented in the same program as normal 3778preprocessing. Therefore, all the GNU extensions to the preprocessor 3779were still available in traditional mode. It is now a separate program 3780and does not implement any of the GNU extensions, except for a partial 3781implementation of assertions. Even those may be removed in a future 3782release. 3783@end itemize 3784 3785@node Invocation 3786@chapter Invocation 3787@cindex invocation 3788@cindex command line 3789 3790Most often when you use the C preprocessor you will not have to invoke it 3791explicitly: the C compiler will do so automatically. However, the 3792preprocessor is sometimes useful on its own. All the options listed 3793here are also acceptable to the C compiler and have the same meaning, 3794except that the C compiler has different rules for specifying the output 3795file. 3796 3797@strong{Note:} Whether you use the preprocessor by way of @command{gcc} 3798or @command{cpp}, the @dfn{compiler driver} is run first. This 3799program's purpose is to translate your command into invocations of the 3800programs that do the actual work. Their command line interfaces are 3801similar but not identical to the documented interface, and may change 3802without notice. 3803 3804@ignore 3805@c man begin SYNOPSIS 3806cpp [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}] 3807 [@option{-I}@var{dir}@dots{}] [@option{-W}@var{warn}@dots{}] 3808 [@option{-M}|@option{-MM}] [@option{-MG}] [@option{-MF} @var{filename}] 3809 [@option{-MP}] [@option{-MQ} @var{target}@dots{}] [@option{-MT} @var{target}@dots{}] 3810 [@option{-x} @var{language}] [@option{-std=}@var{standard}] 3811 @var{infile} @var{outfile} 3812 3813Only the most useful options are listed here; see below for the remainder. 3814@c man end 3815@c man begin SEEALSO 3816gpl(7), gfdl(7), fsf-funding(7), 3817gcc(1), as(1), ld(1), and the Info entries for @file{cpp}, @file{gcc}, and 3818@file{binutils}. 3819@c man end 3820@end ignore 3821 3822@c man begin OPTIONS 3823The C preprocessor expects two file names as arguments, @var{infile} and 3824@var{outfile}. The preprocessor reads @var{infile} together with any 3825other files it specifies with @samp{#include}. All the output generated 3826by the combined input files is written in @var{outfile}. 3827 3828Either @var{infile} or @var{outfile} may be @option{-}, which as 3829@var{infile} means to read from standard input and as @var{outfile} 3830means to write to standard output. Also, if either file is omitted, it 3831means the same as if @option{-} had been specified for that file. 3832 3833Unless otherwise noted, or the option ends in @samp{=}, all options 3834which take an argument may have that argument appear either immediately 3835after the option, or with a space between option and argument: 3836@option{-Ifoo} and @option{-I foo} have the same effect. 3837 3838@cindex grouping options 3839@cindex options, grouping 3840Many options have multi-letter names; therefore multiple single-letter 3841options may @emph{not} be grouped: @option{-dM} is very different from 3842@w{@samp{-d -M}}. 3843 3844@cindex options 3845@table @gcctabopt 3846@item -D @var{name} 3847Predefine @var{name} as a macro, with definition @code{1}. 3848 3849@item -D @var{name}=@var{definition} 3850Predefine @var{name} as a macro, with definition @var{definition}. 3851There are no restrictions on the contents of @var{definition}, but if 3852you are invoking the preprocessor from a shell or shell-like program you 3853may need to use the shell's quoting syntax to protect characters such as 3854spaces that have a meaning in the shell syntax. If you use more than 3855one @option{-D} for the same @var{name}, the rightmost definition takes 3856effect. 3857 3858If you wish to define a function-like macro on the command line, write 3859its argument list with surrounding parentheses before the equals sign 3860(if any). Parentheses are meaningful to most shells, so you will need 3861to quote the option. With @command{sh} and @command{csh}, 3862@option{-D'@var{name}(@var{args@dots{}})=@var{definition}'} works. 3863 3864@item -U @var{name} 3865Cancel any previous definition of @var{name}, either built in or 3866provided with a @option{-D} option. 3867 3868All @option{-imacros @var{file}} and @option{-include @var{file}} options 3869are processed after all @option{-D} and @option{-U} options. 3870 3871@item -undef 3872Do not predefine any system-specific macros. The common predefined 3873macros remain defined. 3874 3875@item -I @var{dir} 3876Add the directory @var{dir} to the list of directories to be searched 3877for header files. @xref{Search Path}. Directories named by @option{-I} 3878are searched before the standard system include directories. 3879 3880It is dangerous to specify a standard system include directory in an 3881@option{-I} option. This defeats the special treatment of system 3882headers (@pxref{System Headers}). It can also defeat the repairs to 3883buggy system headers which GCC makes when it is installed. 3884 3885@item -o @var{file} 3886Write output to @var{file}. This is the same as specifying @var{file} 3887as the second non-option argument to @command{cpp}. @command{gcc} has a 3888different interpretation of a second non-option argument, so you must 3889use @option{-o} to specify the output file. 3890 3891@item -Wall 3892Turns on all optional warnings which are desirable for normal code. At 3893present this is @option{-Wcomment} and @option{-Wtrigraphs}. Note that 3894many of the preprocessor's warnings are on by default and have no 3895options to control them. 3896 3897@item -Wcomment 3898@itemx -Wcomments 3899Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*} 3900comment, or whenever a backslash-newline appears in a @samp{//} comment. 3901(Both forms have the same effect.) 3902 3903@item -Wtrigraphs 3904Warn if any trigraphs are encountered. This option used to take effect 3905only if @option{-trigraphs} was also specified, but now works 3906independently. Warnings are not given for trigraphs within comments, as 3907they do not affect the meaning of the program. 3908 3909@item -Wtraditional 3910Warn about certain constructs that behave differently in traditional and 3911ISO C@. Also warn about ISO C constructs that have no traditional C 3912equivalent, and problematic constructs which should be avoided. 3913@xref{Traditional Mode}. 3914 3915@item -Wimport 3916Warn the first time @samp{#import} is used. 3917 3918@item -Wundef 3919Warn whenever an identifier which is not a macro is encountered in an 3920@samp{#if} directive, outside of @samp{defined}. Such identifiers are 3921replaced with zero. 3922 3923@item -Werror 3924Make all warnings into hard errors. Source code which triggers warnings 3925will be rejected. 3926 3927@item -Wsystem-headers 3928Issue warnings for code in system headers. These are normally unhelpful 3929in finding bugs in your own code, therefore suppressed. If you are 3930responsible for the system library, you may want to see them. 3931 3932@item -w 3933Suppress all warnings, including those which GNU CPP issues by default. 3934 3935@item -pedantic 3936Issue all the mandatory diagnostics listed in the C standard. Some of 3937them are left out by default, since they trigger frequently on harmless 3938code. 3939 3940@item -pedantic-errors 3941Issue all the mandatory diagnostics, and make all mandatory diagnostics 3942into errors. This includes mandatory diagnostics that GCC issues 3943without @samp{-pedantic} but treats as warnings. 3944 3945@item -M 3946Instead of outputting the result of preprocessing, output a rule 3947suitable for @command{make} describing the dependencies of the main 3948source file. The preprocessor outputs one @command{make} rule containing 3949the object file name for that source file, a colon, and the names of all 3950the included files, including those coming from @option{-include} or 3951@option{-imacros} command line options. 3952 3953Unless specified explicitly (with @option{-MT} or @option{-MQ}), the 3954object file name consists of the basename of the source file with any 3955suffix replaced with object file suffix. If there are many included 3956files then the rule is split into several lines using @samp{\}-newline. 3957The rule has no commands. 3958 3959@item -MM 3960Like @option{-M}, but mention only the files included with @code{@w{#include 3961"@var{file}"}} or with @option{-include} or @option{-imacros} command line 3962options. System header files included with @code{@w{#include <@var{file}>}} 3963are omitted. 3964 3965@item -MF @var{file} 3966When used with @option{-M} or @option{-MM}, specifies a file to write the 3967dependencies to. This allows the preprocessor to write the preprocessed 3968file to stdout normally. If no @option{-MF} switch is given, CPP sends 3969the rules to stdout and suppresses normal preprocessed output. 3970 3971@item -MG 3972When used with @option{-M} or @option{-MM}, @option{-MG} says to treat missing 3973header files as generated files and assume they live in the same 3974directory as the source file. It suppresses preprocessed output, as a 3975missing header file is ordinarily an error. 3976 3977This feature is used in automatic updating of makefiles. 3978 3979@item -MP 3980This option instructs CPP to add a phony target for each dependency 3981other than the main file, causing each to depend on nothing. These 3982dummy rules work around errors @command{make} gives if you remove header 3983files without updating the @file{Makefile} to match. 3984 3985This is typical output: 3986 3987@example 3988test.o: test.c test.h 3989 3990test.h: 3991@end example 3992 3993@item -MT @var{target} 3994 3995Change the target of the rule emitted by dependency generation. By 3996default CPP takes the name of the main input file, including any path, 3997deletes any file suffix such as @samp{.c}, and appends the platform's 3998usual object suffix. The result is the target. 3999 4000An @option{-MT} option will set the target to be exactly the string you 4001specify. If you want multiple targets, you can specify them as a single 4002argument to @option{-MT}, or use multiple @option{-MT} options. 4003 4004For example, @option{@w{-MT '$(objpfx)foo.o'}} might give 4005 4006@example 4007$(objpfx)foo.o: foo.c 4008@end example 4009 4010@item -MQ @var{target} 4011 4012Same as @option{-MT}, but it quotes any characters which are special to 4013Make. @option{@w{-MQ '$(objpfx)foo.o'}} gives 4014 4015@example 4016$$(objpfx)foo.o: foo.c 4017@end example 4018 4019The default target is automatically quoted, as if it were given with 4020@option{-MQ}. 4021 4022@item -MD @var{file} 4023@itemx -MMD @var{file} 4024@option{-MD @var{file}} is equivalent to @option{-M -MF @var{file}}, and 4025@option{-MMD @var{file}} is equivalent to @option{-MM -MF @var{file}}. 4026 4027Due to limitations in the compiler driver, you must use these switches 4028when you want to generate a dependency file as a side-effect of normal 4029compilation. 4030 4031@item -x c 4032@itemx -x c++ 4033@itemx -x objective-c 4034@itemx -x assembler-with-cpp 4035Specify the source language: C, C++, Objective-C, or assembly. This has 4036nothing to do with standards conformance or extensions; it merely 4037selects which base syntax to expect. If you give none of these options, 4038cpp will deduce the language from the extension of the source file: 4039@samp{.c}, @samp{.cc}, @samp{.m}, or @samp{.S}. Some other common 4040extensions for C++ and assembly are also recognized. If cpp does not 4041recognize the extension, it will treat the file as C; this is the most 4042generic mode. 4043 4044@strong{Note:} Previous versions of cpp accepted a @option{-lang} option 4045which selected both the language and the standards conformance level. 4046This option has been removed, because it conflicts with the @option{-l} 4047option. 4048 4049@item -std=@var{standard} 4050@itemx -ansi 4051Specify the standard to which the code should conform. Currently cpp 4052only knows about the standards for C; other language standards will be 4053added in the future. 4054 4055@var{standard} 4056may be one of: 4057@table @code 4058@item iso9899:1990 4059@itemx c89 4060The ISO C standard from 1990. @samp{c89} is the customary shorthand for 4061this version of the standard. 4062 4063The @option{-ansi} option is equivalent to @option{-std=c89}. 4064 4065@item iso9899:199409 4066The 1990 C standard, as amended in 1994. 4067 4068@item iso9899:1999 4069@itemx c99 4070@itemx iso9899:199x 4071@itemx c9x 4072The revised ISO C standard, published in December 1999. Before 4073publication, this was known as C9X@. 4074 4075@item gnu89 4076The 1990 C standard plus GNU extensions. This is the default. 4077 4078@item gnu99 4079@itemx gnu9x 4080The 1999 C standard plus GNU extensions. 4081@end table 4082 4083@item -I- 4084Split the include path. Any directories specified with @option{-I} 4085options before @option{-I-} are searched only for headers requested with 4086@code{@w{#include "@var{file}"}}; they are not searched for 4087@code{@w{#include <@var{file}>}}. If additional directories are 4088specified with @option{-I} options after the @option{-I-}, those 4089directories are searched for all @samp{#include} directives. 4090 4091In addition, @option{-I-} inhibits the use of the directory of the current 4092file directory as the first search directory for @code{@w{#include 4093"@var{file}"}}. @xref{Search Path}. 4094 4095@item -nostdinc 4096Do not search the standard system directories for header files. 4097Only the directories you have specified with @option{-I} options 4098(and the directory of the current file, if appropriate) are searched. 4099 4100@item -nostdinc++ 4101Do not search for header files in the C++-specific standard directories, 4102but do still search the other standard directories. (This option is 4103used when building the C++ library.) 4104 4105@item -include @var{file} 4106 4107Process @var{file} as if @code{#include "file"} appeared as the first 4108line of the primary source file. However, the first directory searched 4109for @var{file} is the preprocessor's working directory @emph{instead of} 4110the directory containing the main source file. If not found there, it 4111is searched for in the remainder of the @code{#include "@dots{}"} search 4112chain as normal. 4113 4114If multiple @option{-include} options are given, the files are included 4115in the order they appear on the command line. 4116 4117@item -imacros @var{file} 4118 4119Exactly like @option{-include}, except that any output produced by 4120scanning @var{file} is thrown away. Macros it defines remain defined. 4121This allows you to acquire all the macros from a header without also 4122processing its declarations. 4123 4124All files specified by @option{-imacros} are processed before all files 4125specified by @option{-include}. 4126 4127@item -idirafter @var{dir} 4128Search @var{dir} for header files, but do it @emph{after} all 4129directories specified with @option{-I} and the standard system directories 4130have been exhausted. @var{dir} is treated as a system include directory. 4131 4132@item -iprefix @var{prefix} 4133Specify @var{prefix} as the prefix for subsequent @option{-iwithprefix} 4134options. If the prefix represents a directory, you should include the 4135final @samp{/}. 4136 4137@item -iwithprefix @var{dir} 4138@itemx -iwithprefixbefore @var{dir} 4139 4140Append @var{dir} to the prefix specified previously with 4141@option{-iprefix}, and add the resulting directory to the include search 4142path. @option{-iwithprefixbefore} puts it in the same place @option{-I} 4143would; @option{-iwithprefix} puts it where @option{-idirafter} would. 4144 4145Use of these options is discouraged. 4146 4147@item -isystem @var{dir} 4148Search @var{dir} for header files, after all directories specified by 4149@option{-I} but before the standard system directories. Mark it 4150as a system directory, so that it gets the same special treatment as 4151is applied to the standard system directories. @xref{System Headers}. 4152 4153@item -fpreprocessed 4154Indicate to the preprocessor that the input file has already been 4155preprocessed. This suppresses things like macro expansion, trigraph 4156conversion, escaped newline splicing, and processing of most directives. 4157The preprocessor still recognizes and removes comments, so that you can 4158pass a file preprocessed with @option{-C} to the compiler without 4159problems. In this mode the integrated preprocessor is little more than 4160a tokenizer for the front ends. 4161 4162@option{-fpreprocessed} is implicit if the input file has one of the 4163extensions @samp{.i}, @samp{.ii} or @samp{.mi}. These are the 4164extensions that GCC uses for preprocessed files created by 4165@option{-save-temps}. 4166 4167@item -ftabstop=@var{width} 4168Set the distance between tab stops. This helps the preprocessor report 4169correct column numbers in warnings or errors, even if tabs appear on the 4170line. If the value is less than 1 or greater than 100, the option is 4171ignored. The default is 8. 4172 4173@item -fno-show-column 4174Do not print column numbers in diagnostics. This may be necessary if 4175diagnostics are being scanned by a program that does not understand the 4176column numbers, such as @command{dejagnu}. 4177 4178@item -A @var{predicate}=@var{answer} 4179Make an assertion with the predicate @var{predicate} and answer 4180@var{answer}. This form is preferred to the older form @option{-A 4181@var{predicate}(@var{answer})}, which is still supported, because 4182it does not use shell special characters. @xref{Assertions}. 4183 4184@item -A -@var{predicate}=@var{answer} 4185Cancel an assertion with the predicate @var{predicate} and answer 4186@var{answer}. 4187 4188@item -A- 4189Cancel all predefined assertions and all assertions preceding it on 4190the command line. Also, undefine all predefined macros and all 4191macros preceding it on the command line. (This is a historical wart and 4192may change in the future.) 4193 4194@item -dCHARS 4195@var{CHARS} is a sequence of one or more of the following characters, 4196and must not be preceded by a space. Other characters are interpreted 4197by the compiler proper, or reserved for future versions of GCC, and so 4198are silently ignored. If you specify characters whose behavior 4199conflicts, the result is undefined. 4200 4201@table @samp 4202@item M 4203Instead of the normal output, generate a list of @samp{#define} 4204directives for all the macros defined during the execution of the 4205preprocessor, including predefined macros. This gives you a way of 4206finding out what is predefined in your version of the preprocessor. 4207Assuming you have no file @file{foo.h}, the command 4208 4209@example 4210touch foo.h; cpp -dM foo.h 4211@end example 4212 4213@noindent 4214will show all the predefined macros. 4215 4216@item D 4217Like @samp{M} except in two respects: it does @emph{not} include the 4218predefined macros, and it outputs @emph{both} the @samp{#define} 4219directives and the result of preprocessing. Both kinds of output go to 4220the standard output file. 4221 4222@item N 4223Like @samp{D}, but emit only the macro names, not their expansions. 4224 4225@item I 4226Output @samp{#include} directives in addition to the result of 4227preprocessing. 4228@end table 4229 4230@item -P 4231Inhibit generation of linemarkers in the output from the preprocessor. 4232This might be useful when running the preprocessor on something that is 4233not C code, and will be sent to a program which might be confused by the 4234linemarkers. @xref{Preprocessor Output}. 4235 4236@item -C 4237Do not discard comments. All comments are passed through to the output 4238file, except for comments in processed directives, which are deleted 4239along with the directive. 4240 4241You should be prepared for side effects when using @option{-C}; it 4242causes the preprocessor to treat comments as tokens in their own right. 4243For example, comments appearing at the start of what would be a 4244directive line have the effect of turning that line into an ordinary 4245source line, since the first token on the line is no longer a @samp{#}. 4246 4247@item -gcc 4248Define the macros @sc{__gnuc__}, @sc{__gnuc_minor__} and 4249@sc{__gnuc_patchlevel__}. These are defined automatically when you use 4250@command{gcc -E}; you can turn them off in that case with 4251@option{-no-gcc}. 4252 4253@item -traditional 4254Try to imitate the behavior of old-fashioned C, as opposed to ISO 4255C@. @xref{Traditional Mode}. 4256 4257@item -trigraphs 4258Process trigraph sequences. @xref{Initial processing}. 4259 4260@item -remap 4261Enable special code to work around file systems which only permit very 4262short file names, such as MS-DOS@. 4263 4264@item -$ 4265Forbid the use of @samp{$} in identifiers. The C standard allows 4266implementations to define extra characters that can appear in 4267identifiers. By default GNU CPP permits @samp{$}, a common extension. 4268 4269@item -h 4270@itemx --help 4271@itemx --target-help 4272Print text describing all the command line options instead of 4273preprocessing anything. 4274 4275@item -v 4276Verbose mode. Print out GNU CPP's version number at the beginning of 4277execution, and report the final form of the include path. 4278 4279@item -H 4280Print the name of each header file used, in addition to other normal 4281activities. Each name is indented to show how deep in the 4282@samp{#include} stack it is. 4283 4284@item -version 4285@itemx --version 4286Print out GNU CPP's version number. With one dash, proceed to 4287preprocess as normal. With two dashes, exit immediately. 4288@end table 4289@c man end 4290 4291@include fdl.texi 4292 4293@page 4294@node Index of Directives 4295@unnumbered Index of Directives 4296@printindex fn 4297 4298@node Concept Index 4299@unnumbered Concept Index 4300@printindex cp 4301 4302@bye 4303