1@c Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2004 2@c Free Software Foundation, Inc. 3@c This is part of the G77 manual. 4@c For copying conditions, see the file g77.texi. 5 6@ignore 7@c man begin COPYRIGHT 8Copyright @copyright{} 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2004 9Free Software Foundation, Inc. 10 11Permission is granted to copy, distribute and/or modify this document 12under the terms of the GNU Free Documentation License, Version 1.2 or 13any later version published by the Free Software Foundation; with the 14Invariant Sections being ``GNU General Public License'' and ``Funding 15Free Software'', the Front-Cover texts being (a) (see below), and with 16the Back-Cover Texts being (b) (see below). A copy of the license is 17included in the gfdl(7) man page. 18 19(a) The FSF's Front-Cover Text is: 20 21 A GNU Manual 22 23(b) The FSF's Back-Cover Text is: 24 25 You have freedom to copy and modify this GNU Manual, like GNU 26 software. Copies published by the Free Software Foundation raise 27 funds for GNU development. 28@c man end 29@c Set file name and title for the man page. 30@setfilename g77 31@settitle GNU project Fortran 77 compiler. 32@c man begin SYNOPSIS 33g77 [@option{-c}|@option{-S}|@option{-E}] 34 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}] 35 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}] 36 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}] 37 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}] 38 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}] 39 [@option{-o} @var{outfile}] @var{infile}@dots{} 40 41Only the most useful options are listed here; see below for the 42remainder. 43@c man end 44@c man begin SEEALSO 45gpl(7), gfdl(7), fsf-funding(7), 46cpp(1), gcov(1), gcc(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1) 47and the Info entries for @file{gcc}, @file{cpp}, @file{g77}, @file{as}, 48@file{ld}, @file{binutils} and @file{gdb}. 49@c man end 50@c man begin BUGS 51For instructions on reporting bugs, see 52@w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug} 53script to report bugs is recommended. 54@c man end 55@c man begin AUTHOR 56See the Info entry for @command{g77} for contributors to GCC and G77@. 57@c man end 58@end ignore 59 60@node Invoking G77 61@chapter GNU Fortran Command Options 62@cindex GNU Fortran command options 63@cindex command options 64@cindex options, GNU Fortran command 65 66@c man begin DESCRIPTION 67 68The @command{g77} command supports all the options supported by the 69@command{gcc} command. 70@xref{Invoking GCC,,GCC Command Options,gcc,Using the GNU Compiler 71Collection (GCC)}, for information 72on the non-Fortran-specific aspects of the @command{gcc} command (and, 73therefore, the @command{g77} command). 74 75@cindex options, negative forms 76@cindex negative forms of options 77All @command{gcc} and @command{g77} options 78are accepted both by @command{g77} and by @command{gcc} 79(as well as any other drivers built at the same time, 80such as @command{g++}), 81since adding @command{g77} to the @command{gcc} distribution 82enables acceptance of @command{g77} options 83by all of the relevant drivers. 84 85In some cases, options have positive and negative forms; 86the negative form of @option{-ffoo} would be @option{-fno-foo}. 87This manual documents only one of these two forms, whichever 88one is not the default. 89 90@c man end 91 92@menu 93* Option Summary:: Brief list of all @command{g77} options, 94 without explanations. 95* Overall Options:: Controlling the kind of output: 96 an executable, object files, assembler files, 97 or preprocessed source. 98* Shorthand Options:: Options that are shorthand for other options. 99* Fortran Dialect Options:: Controlling the variant of Fortran language 100 compiled. 101* Warning Options:: How picky should the compiler be? 102* Debugging Options:: Symbol tables, measurements, and debugging dumps. 103* Optimize Options:: How much optimization? 104* Preprocessor Options:: Controlling header files and macro definitions. 105 Also, getting dependency information for Make. 106* Directory Options:: Where to find header files and libraries. 107 Where to find the compiler executable files. 108* Code Gen Options:: Specifying conventions for function calls, data layout 109 and register usage. 110* Environment Variables:: Env vars that affect GNU Fortran. 111@end menu 112 113@node Option Summary 114@section Option Summary 115 116@c man begin OPTIONS 117 118Here is a summary of all the options specific to GNU Fortran, grouped 119by type. Explanations are in the following sections. 120 121@table @emph 122@item Overall Options 123@xref{Overall Options,,Options Controlling the Kind of Output}. 124@gccoptlist{ 125-fversion -fset-g77-defaults -fno-silent} 126 127@item Shorthand Options 128@xref{Shorthand Options}. 129@gccoptlist{ 130-ff66 -fno-f66 -ff77 -fno-f77 -fno-ugly} 131 132@item Fortran Language Options 133@xref{Fortran Dialect Options,,Options Controlling Fortran Dialect}. 134@gccoptlist{ 135-ffree-form -fno-fixed-form -ff90 @gol 136-fvxt -fdollar-ok -fno-backslash @gol 137-fno-ugly-args -fno-ugly-assign -fno-ugly-assumed @gol 138-fugly-comma -fugly-complex -fugly-init -fugly-logint @gol 139-fonetrip -ftypeless-boz @gol 140-fintrin-case-initcap -fintrin-case-upper @gol 141-fintrin-case-lower -fintrin-case-any @gol 142-fmatch-case-initcap -fmatch-case-upper @gol 143-fmatch-case-lower -fmatch-case-any @gol 144-fsource-case-upper -fsource-case-lower @gol 145-fsource-case-preserve @gol 146-fsymbol-case-initcap -fsymbol-case-upper @gol 147-fsymbol-case-lower -fsymbol-case-any @gol 148-fcase-strict-upper -fcase-strict-lower @gol 149-fcase-initcap -fcase-upper -fcase-lower -fcase-preserve @gol 150-ff2c-intrinsics-delete -ff2c-intrinsics-hide @gol 151-ff2c-intrinsics-disable -ff2c-intrinsics-enable @gol 152-fbadu77-intrinsics-delete -fbadu77-intrinsics-hide @gol 153-fbadu77-intrinsics-disable -fbadu77-intrinsics-enable @gol 154-ff90-intrinsics-delete -ff90-intrinsics-hide @gol 155-ff90-intrinsics-disable -ff90-intrinsics-enable @gol 156-fgnu-intrinsics-delete -fgnu-intrinsics-hide @gol 157-fgnu-intrinsics-disable -fgnu-intrinsics-enable @gol 158-fmil-intrinsics-delete -fmil-intrinsics-hide @gol 159-fmil-intrinsics-disable -fmil-intrinsics-enable @gol 160-funix-intrinsics-delete -funix-intrinsics-hide @gol 161-funix-intrinsics-disable -funix-intrinsics-enable @gol 162-fvxt-intrinsics-delete -fvxt-intrinsics-hide @gol 163-fvxt-intrinsics-disable -fvxt-intrinsics-enable @gol 164-ffixed-line-length-@var{n} -ffixed-line-length-none} 165 166@item Warning Options 167@xref{Warning Options,,Options to Request or Suppress Warnings}. 168@gccoptlist{ 169-fsyntax-only -pedantic -pedantic-errors -fpedantic @gol 170-w -Wno-globals -Wimplicit -Wunused -Wuninitialized @gol 171-Wall -Wsurprising @gol 172-Werror -W} 173 174@item Debugging Options 175@xref{Debugging Options,,Options for Debugging Your Program or GCC}. 176@gccoptlist{ 177-g} 178 179@item Optimization Options 180@xref{Optimize Options,,Options that Control Optimization}. 181@gccoptlist{ 182-malign-double @gol 183-ffloat-store -fforce-mem -fforce-addr -fno-inline @gol 184-ffast-math -fstrength-reduce -frerun-cse-after-loop @gol 185-funsafe-math-optimizations -ffinite-math-only -fno-trapping-math @gol 186-fexpensive-optimizations -fdelayed-branch @gol 187-fschedule-insns -fschedule-insn2 -fcaller-saves @gol 188-funroll-loops -funroll-all-loops @gol 189-fno-move-all-movables -fno-reduce-all-givs @gol 190-fno-rerun-loop-opt} 191 192@item Directory Options 193@xref{Directory Options,,Options for Directory Search}. 194@gccoptlist{ 195-I@var{dir} -I-} 196 197@item Code Generation Options 198@xref{Code Gen Options,,Options for Code Generation Conventions}. 199@gccoptlist{ 200-fno-automatic -finit-local-zero -fno-f2c @gol 201-ff2c-library -fno-underscoring -fno-ident @gol 202-fpcc-struct-return -freg-struct-return @gol 203-fshort-double -fno-common -fpack-struct @gol 204-fzeros -fno-second-underscore @gol 205-femulate-complex @gol 206-falias-check -fargument-alias @gol 207-fargument-noalias -fno-argument-noalias-global @gol 208-fno-globals -fflatten-arrays @gol 209-fbounds-check -ffortran-bounds-check} 210@end table 211 212@c man end 213 214@menu 215* Overall Options:: Controlling the kind of output: 216 an executable, object files, assembler files, 217 or preprocessed source. 218* Shorthand Options:: Options that are shorthand for other options. 219* Fortran Dialect Options:: Controlling the variant of Fortran language 220 compiled. 221* Warning Options:: How picky should the compiler be? 222* Debugging Options:: Symbol tables, measurements, and debugging dumps. 223* Optimize Options:: How much optimization? 224* Preprocessor Options:: Controlling header files and macro definitions. 225 Also, getting dependency information for Make. 226* Directory Options:: Where to find header files and libraries. 227 Where to find the compiler executable files. 228* Code Gen Options:: Specifying conventions for function calls, data layout 229 and register usage. 230@end menu 231 232@node Overall Options 233@section Options Controlling the Kind of Output 234@cindex overall options 235@cindex options, overall 236 237@c man begin OPTIONS 238 239Compilation can involve as many as four stages: preprocessing, code 240generation (often what is really meant by the term ``compilation''), 241assembly, and linking, always in that order. The first three 242stages apply to an individual source file, and end by producing an 243object file; linking combines all the object files (those newly 244compiled, and those specified as input) into an executable file. 245 246@cindex file name suffix 247@cindex suffixes, file name 248@cindex file name extension 249@cindex extensions, file name 250@cindex file type 251@cindex types, file 252For any given input file, the file name suffix determines what kind of 253program is contained in the file---that is, the language in which the 254program is written is generally indicated by the suffix. 255Suffixes specific to GNU Fortran are listed below. 256@xref{Overall Options,,Options Controlling the Kind of 257Output,gcc,Using the GNU Compiler Collection (GCC)}, for 258information on suffixes recognized by GNU CC. 259 260@table @gcctabopt 261@cindex .f filename suffix 262@cindex .for filename suffix 263@cindex .FOR filename suffix 264@item @var{file}.f 265@item @var{file}.for 266@item @var{file}.FOR 267Fortran source code that should not be preprocessed. 268 269Such source code cannot contain any preprocessor directives, such 270as @code{#include}, @code{#define}, @code{#if}, and so on. 271 272You can force @samp{.f} files to be preprocessed by @command{cpp} by using 273@option{-x f77-cpp-input}. 274@xref{LEX}. 275 276@cindex preprocessor 277@cindex C preprocessor 278@cindex cpp preprocessor 279@cindex Fortran preprocessor 280@cindex cpp program 281@cindex programs, cpp 282@cindex .F filename suffix 283@cindex .fpp filename suffix 284@cindex .FPP filename suffix 285@item @var{file}.F 286@item @var{file}.fpp 287@item @var{file}.FPP 288Fortran source code that must be preprocessed (by the C preprocessor 289@command{cpp}, which is part of GNU CC). 290 291Note that preprocessing is not extended to the contents of 292files included by the @code{INCLUDE} directive---the @code{#include} 293preprocessor directive must be used instead. 294 295@cindex Ratfor preprocessor 296@cindex programs, @command{ratfor} 297@cindex @samp{.r} filename suffix 298@cindex @command{ratfor} 299@item @var{file}.r 300Ratfor source code, which must be preprocessed by the @command{ratfor} 301command, which is available separately (as it is not yet part of the GNU 302Fortran distribution). 303A public domain version in C is at 304@uref{http://sepwww.stanford.edu/sep/prof/ratfor.shar.2}. 305@end table 306 307UNIX users typically use the @file{@var{file}.f} and @file{@var{file}.F} 308nomenclature. 309Users of other operating systems, especially those that cannot 310distinguish upper-case 311letters from lower-case letters in their file names, typically use 312the @file{@var{file}.for} and @file{@var{file}.fpp} nomenclature. 313 314@cindex #define 315@cindex #include 316@cindex #if 317Use of the preprocessor @command{cpp} allows use of C-like 318constructs such as @code{#define} and @code{#include}, but can 319lead to unexpected, even mistaken, results due to Fortran's source file 320format. 321It is recommended that use of the C preprocessor 322be limited to @code{#include} and, in 323conjunction with @code{#define}, only @code{#if} and related directives, 324thus avoiding in-line macro expansion entirely. 325This recommendation applies especially 326when using the traditional fixed source form. 327With free source form, 328fewer unexpected transformations are likely to happen, but use of 329constructs such as Hollerith and character constants can nevertheless 330present problems, especially when these are continued across multiple 331source lines. 332These problems result, primarily, from differences between the way 333such constants are interpreted by the C preprocessor and by a Fortran 334compiler. 335 336Another example of a problem that results from using the C preprocessor 337is that a Fortran comment line that happens to contain any 338characters ``interesting'' to the C preprocessor, 339such as a backslash at the end of the line, 340is not recognized by the preprocessor as a comment line, 341so instead of being passed through ``raw'', 342the line is edited according to the rules for the preprocessor. 343For example, the backslash at the end of the line is removed, 344along with the subsequent newline, resulting in the next 345line being effectively commented out---unfortunate if that 346line is a non-comment line of important code! 347 348@emph{Note:} The @option{-traditional} and @option{-undef} flags are supplied 349to @command{cpp} by default, to help avoid unpleasant surprises. 350@xref{Preprocessor Options,,Options Controlling the Preprocessor, 351gcc,Using the GNU Compiler Collection (GCC)}. 352This means that ANSI C preprocessor features (such as the @samp{#} 353operator) aren't available, and only variables in the C reserved 354namespace (generally, names with a leading underscore) are liable to 355substitution by C predefines. 356Thus, if you want to do system-specific 357tests, use, for example, @samp{#ifdef __linux__} rather than @samp{#ifdef linux}. 358Use the @option{-v} option to see exactly how the preprocessor is invoked. 359 360@cindex /* 361Unfortunately, the @option{-traditional} flag will not avoid an error from 362anything that @command{cpp} sees as an unterminated C comment, such as: 363@smallexample 364C Some Fortran compilers accept /* as starting 365C an inline comment. 366@end smallexample 367@xref{Trailing Comment}. 368 369The following options that affect overall processing are recognized 370by the @command{g77} and @command{gcc} commands in a GNU Fortran installation: 371 372@table @gcctabopt 373@cindex -fversion option 374@cindex options, -fversion 375@cindex printing version information 376@cindex version information, printing 377@cindex consistency checks 378@cindex internal consistency checks 379@cindex checks, of internal consistency 380@item -fversion 381Ensure that the @command{g77} version of the compiler phase is reported, 382if run, 383and, starting in @code{egcs} version 1.1, 384that internal consistency checks in the @file{f771} program are run. 385 386This option is supplied automatically when @option{-v} or @option{--verbose} 387is specified as a command-line option for @command{g77} or @command{gcc} 388and when the resulting commands compile Fortran source files. 389 390In GCC 3.1, this is changed back to the behavior @command{gcc} displays 391for @samp{.c} files. 392 393@cindex -fset-g77-defaults option 394@cindex options, -fset-g77-defaults 395@item -fset-g77-defaults 396@emph{Version info:} 397This option was obsolete as of @code{egcs} 398version 1.1. 399The effect is instead achieved 400by the @code{lang_init_options} routine 401in @file{gcc/gcc/f/com.c}. 402 403@cindex consistency checks 404@cindex internal consistency checks 405@cindex checks, of internal consistency 406Set up whatever @command{gcc} options are to apply to Fortran 407compilations, and avoid running internal consistency checks 408that might take some time. 409 410This option is supplied automatically when compiling Fortran code 411via the @command{g77} or @command{gcc} command. 412The description of this option is provided so that users seeing 413it in the output of, say, @samp{g77 -v} understand why it is 414there. 415 416@cindex modifying @command{g77} 417@cindex @command{g77}, modifying 418Also, developers who run @code{f771} directly might want to specify it 419by hand to get the same defaults as they would running @code{f771} 420via @command{g77} or @command{gcc} 421However, such developers should, after linking a new @code{f771} 422executable, invoke it without this option once, 423e.g. via @kbd{./f771 -quiet < /dev/null}, 424to ensure that they have not introduced any 425internal inconsistencies (such as in the table of 426intrinsics) before proceeding---@command{g77} will crash 427with a diagnostic if it detects an inconsistency. 428 429@cindex -fno-silent option 430@cindex options, -fno-silent 431@cindex f2c compatibility 432@cindex compatibility, f2c 433@cindex status, compilation 434@cindex compilation, status 435@cindex reporting compilation status 436@cindex printing compilation status 437@item -fno-silent 438Print (to @code{stderr}) the names of the program units as 439they are compiled, in a form similar to that used by popular 440UNIX @command{f77} implementations and @command{f2c} 441@end table 442 443@xref{Overall Options,,Options Controlling the Kind of Output, 444gcc,Using the GNU Compiler Collection (GCC)}, for information 445on more options that control the overall operation of the @command{gcc} command 446(and, by extension, the @command{g77} command). 447 448@node Shorthand Options 449@section Shorthand Options 450@cindex shorthand options 451@cindex options, shorthand 452@cindex macro options 453@cindex options, macro 454 455The following options serve as ``shorthand'' 456for other options accepted by the compiler: 457 458@table @gcctabopt 459@cindex -fugly option 460@cindex options, -fugly 461@item -fugly 462@cindex ugly features 463@cindex features, ugly 464@emph{Note:} This option is no longer supported. 465The information, below, is provided to aid 466in the conversion of old scripts. 467 468Specify that certain ``ugly'' constructs are to be quietly accepted. 469Same as: 470 471@smallexample 472-fugly-args -fugly-assign -fugly-assumed 473-fugly-comma -fugly-complex -fugly-init 474-fugly-logint 475@end smallexample 476 477These constructs are considered inappropriate to use in new 478or well-maintained portable Fortran code, but widely used 479in old code. 480@xref{Distensions}, for more information. 481 482@cindex -fno-ugly option 483@cindex options, -fno-ugly 484@item -fno-ugly 485@cindex ugly features 486@cindex features, ugly 487Specify that all ``ugly'' constructs are to be noisily rejected. 488Same as: 489 490@smallexample 491-fno-ugly-args -fno-ugly-assign -fno-ugly-assumed 492-fno-ugly-comma -fno-ugly-complex -fno-ugly-init 493-fno-ugly-logint 494@end smallexample 495 496@xref{Distensions}, for more information. 497 498@cindex -ff66 option 499@cindex options, -ff66 500@item -ff66 501@cindex FORTRAN 66 502@cindex compatibility, FORTRAN 66 503Specify that the program is written in idiomatic FORTRAN 66. 504Same as @samp{-fonetrip -fugly-assumed}. 505 506The @option{-fno-f66} option is the inverse of @option{-ff66}. 507As such, it is the same as @samp{-fno-onetrip -fno-ugly-assumed}. 508 509The meaning of this option is likely to be refined as future 510versions of @command{g77} provide more compatibility with other 511existing and obsolete Fortran implementations. 512 513@cindex -ff77 option 514@cindex options, -ff77 515@item -ff77 516@cindex UNIX f77 517@cindex f2c compatibility 518@cindex compatibility, f2c 519@cindex f77 compatibility 520@cindex compatibility, f77 521Specify that the program is written in idiomatic UNIX FORTRAN 77 522and/or the dialect accepted by the @command{f2c} product. 523Same as @samp{-fbackslash -fno-typeless-boz}. 524 525The meaning of this option is likely to be refined as future 526versions of @command{g77} provide more compatibility with other 527existing and obsolete Fortran implementations. 528 529@cindex -fno-f77 option 530@cindex options, -fno-f77 531@item -fno-f77 532@cindex UNIX f77 533The @option{-fno-f77} option is @emph{not} the inverse 534of @option{-ff77}. 535It specifies that the program is not written in idiomatic UNIX 536FORTRAN 77 or @command{f2c} but in a more widely portable dialect. 537@option{-fno-f77} is the same as @option{-fno-backslash}. 538 539The meaning of this option is likely to be refined as future 540versions of @command{g77} provide more compatibility with other 541existing and obsolete Fortran implementations. 542@end table 543 544@node Fortran Dialect Options 545@section Options Controlling Fortran Dialect 546@cindex dialect options 547@cindex language, dialect options 548@cindex options, dialect 549 550The following options control the dialect of Fortran 551that the compiler accepts: 552 553@table @gcctabopt 554@cindex -ffree-form option 555@cindex options, -ffree-form 556@cindex -fno-fixed-form option 557@cindex options, -fno-fixed-form 558@cindex source file format 559@cindex free form 560@cindex fixed form 561@cindex Fortran 90, features 562@item -ffree-form 563@item -fno-fixed-form 564Specify that the source file is written in free form 565(introduced in Fortran 90) instead of the more-traditional fixed form. 566 567@cindex -ff90 option 568@cindex options, -ff90 569@cindex Fortran 90, features 570@item -ff90 571Allow certain Fortran-90 constructs. 572 573This option controls whether certain 574Fortran 90 constructs are recognized. 575(Other Fortran 90 constructs 576might or might not be recognized depending on other options such as 577@option{-fvxt}, @option{-ff90-intrinsics-enable}, and the 578current level of support for Fortran 90.) 579 580@xref{Fortran 90}, for more information. 581 582@cindex -fvxt option 583@cindex options, -fvxt 584@item -fvxt 585@cindex Fortran 90, features 586@cindex VXT extensions 587Specify the treatment of certain constructs that have different 588meanings depending on whether the code is written in 589GNU Fortran (based on FORTRAN 77 and akin to Fortran 90) 590or VXT Fortran (more like VAX FORTRAN). 591 592The default is @option{-fno-vxt}. 593@option{-fvxt} specifies that the VXT Fortran interpretations 594for those constructs are to be chosen. 595 596@xref{VXT Fortran}, for more information. 597 598@cindex -fdollar-ok option 599@cindex options, -fdollar-ok 600@item -fdollar-ok 601@cindex dollar sign 602@cindex symbol names 603@cindex character set 604Allow @samp{$} as a valid character in a symbol name. 605 606@cindex -fno-backslash option 607@cindex options, -fno-backslash 608@item -fno-backslash 609@cindex backslash 610@cindex character constants 611@cindex Hollerith constants 612Specify that @samp{\} is not to be specially interpreted in character 613and Hollerith constants a la C and many UNIX Fortran compilers. 614 615For example, with @option{-fbackslash} in effect, @samp{A\nB} specifies 616three characters, with the second one being newline. 617With @option{-fno-backslash}, it specifies four characters, 618@samp{A}, @samp{\}, @samp{n}, and @samp{B}. 619 620Note that @command{g77} implements a fairly general form of backslash 621processing that is incompatible with the narrower forms supported 622by some other compilers. 623For example, @samp{'A\003B'} is a three-character string in @command{g77} 624whereas other compilers that support backslash might not support 625the three-octal-digit form, and thus treat that string as longer 626than three characters. 627 628@xref{Backslash in Constants}, for 629information on why @option{-fbackslash} is the default 630instead of @option{-fno-backslash}. 631 632@cindex -fno-ugly-args option 633@cindex options, -fno-ugly-args 634@item -fno-ugly-args 635Disallow passing Hollerith and typeless constants as actual 636arguments (for example, @samp{CALL FOO(4HABCD)}). 637 638@xref{Ugly Implicit Argument Conversion}, for more information. 639 640@cindex -fugly-assign option 641@cindex options, -fugly-assign 642@item -fugly-assign 643Use the same storage for a given variable regardless of 644whether it is used to hold an assigned-statement label 645(as in @samp{ASSIGN 10 TO I}) or used to hold numeric data 646(as in @samp{I = 3}). 647 648@xref{Ugly Assigned Labels}, for more information. 649 650@cindex -fugly-assumed option 651@cindex options, -fugly-assumed 652@item -fugly-assumed 653Assume any dummy array with a final dimension specified as @samp{1} 654is really an assumed-size array, as if @samp{*} had been specified 655for the final dimension instead of @samp{1}. 656 657For example, @samp{DIMENSION X(1)} is treated as if it 658had read @samp{DIMENSION X(*)}. 659 660@xref{Ugly Assumed-Size Arrays}, for more information. 661 662@cindex -fugly-comma option 663@cindex options, -fugly-comma 664@item -fugly-comma 665In an external-procedure invocation, 666treat a trailing comma in the argument list 667as specification of a trailing null argument, 668and treat an empty argument list 669as specification of a single null argument. 670 671For example, @samp{CALL FOO(,)} is treated as 672@samp{CALL FOO(%VAL(0), %VAL(0))}. 673That is, @emph{two} null arguments are specified 674by the procedure call when @option{-fugly-comma} is in force. 675And @samp{F = FUNC()} is treated as @samp{F = FUNC(%VAL(0))}. 676 677The default behavior, @option{-fno-ugly-comma}, is to ignore 678a single trailing comma in an argument list. 679So, by default, @samp{CALL FOO(X,)} is treated 680exactly the same as @samp{CALL FOO(X)}. 681 682@xref{Ugly Null Arguments}, for more information. 683 684@cindex -fugly-complex option 685@cindex options, -fugly-complex 686@item -fugly-complex 687Do not complain about @samp{REAL(@var{expr})} or 688@samp{AIMAG(@var{expr})} when @var{expr} is a @code{COMPLEX} 689type other than @code{COMPLEX(KIND=1)}---usually 690this is used to permit @code{COMPLEX(KIND=2)} 691(@code{DOUBLE COMPLEX}) operands. 692 693The @option{-ff90} option controls the interpretation 694of this construct. 695 696@xref{Ugly Complex Part Extraction}, for more information. 697 698@cindex -fno-ugly-init option 699@cindex options, -fno-ugly-init 700@item -fno-ugly-init 701Disallow use of Hollerith and typeless constants as initial 702values (in @code{PARAMETER} and @code{DATA} statements), and 703use of character constants to 704initialize numeric types and vice versa. 705 706For example, @samp{DATA I/'F'/, CHRVAR/65/, J/4HABCD/} is disallowed by 707@option{-fno-ugly-init}. 708 709@xref{Ugly Conversion of Initializers}, for more information. 710 711@cindex -fugly-logint option 712@cindex options, -fugly-logint 713@item -fugly-logint 714Treat @code{INTEGER} and @code{LOGICAL} variables and 715expressions as potential stand-ins for each other. 716 717For example, automatic conversion between @code{INTEGER} and 718@code{LOGICAL} is enabled, for many contexts, via this option. 719 720@xref{Ugly Integer Conversions}, for more information. 721 722@cindex -fonetrip option 723@cindex options, -fonetrip 724@item -fonetrip 725@cindex FORTRAN 66 726@cindex @code{DO} loops, one-trip 727@cindex one-trip @code{DO} loops 728@cindex @code{DO} loops, zero-trip 729@cindex zero-trip @code{DO} loops 730@cindex compatibility, FORTRAN 66 731Executable iterative @code{DO} loops are to be executed at 732least once each time they are reached. 733 734ANSI FORTRAN 77 and more recent versions of the Fortran standard 735specify that the body of an iterative @code{DO} loop is not executed 736if the number of iterations calculated from the parameters of the 737loop is less than 1. 738(For example, @samp{DO 10 I = 1, 0}.) 739Such a loop is called a @dfn{zero-trip loop}. 740 741Prior to ANSI FORTRAN 77, many compilers implemented @code{DO} loops 742such that the body of a loop would be executed at least once, even 743if the iteration count was zero. 744Fortran code written assuming this behavior is said to require 745@dfn{one-trip loops}. 746For example, some code written to the FORTRAN 66 standard 747expects this behavior from its @code{DO} loops, although that 748standard did not specify this behavior. 749 750The @option{-fonetrip} option specifies that the source file(s) being 751compiled require one-trip loops. 752 753This option affects only those loops specified by the (iterative) @code{DO} 754statement and by implied-@code{DO} lists in I/O statements. 755Loops specified by implied-@code{DO} lists in @code{DATA} and 756specification (non-executable) statements are not affected. 757 758@cindex -ftypeless-boz option 759@cindex options, -ftypeless-boz 760@cindex prefix-radix constants 761@cindex constants, prefix-radix 762@cindex constants, types 763@cindex types, constants 764@item -ftypeless-boz 765Specifies that prefix-radix non-decimal constants, such as 766@samp{Z'ABCD'}, are typeless instead of @code{INTEGER(KIND=1)}. 767 768You can test for yourself whether a particular compiler treats 769the prefix form as @code{INTEGER(KIND=1)} or typeless by running the 770following program: 771 772@smallexample 773EQUIVALENCE (I, R) 774R = Z'ABCD1234' 775J = Z'ABCD1234' 776IF (J .EQ. I) PRINT *, 'Prefix form is TYPELESS' 777IF (J .NE. I) PRINT *, 'Prefix form is INTEGER' 778END 779@end smallexample 780 781Reports indicate that many compilers process this form as 782@code{INTEGER(KIND=1)}, though a few as typeless, and at least one 783based on a command-line option specifying some kind of 784compatibility. 785 786@cindex -fintrin-case-initcap option 787@cindex options, -fintrin-case-initcap 788@item -fintrin-case-initcap 789@cindex -fintrin-case-upper option 790@cindex options, -fintrin-case-upper 791@item -fintrin-case-upper 792@cindex -fintrin-case-lower option 793@cindex options, -fintrin-case-lower 794@item -fintrin-case-lower 795@cindex -fintrin-case-any option 796@cindex options, -fintrin-case-any 797@item -fintrin-case-any 798Specify expected case for intrinsic names. 799@option{-fintrin-case-lower} is the default. 800 801@cindex -fmatch-case-initcap option 802@cindex options, -fmatch-case-initcap 803@item -fmatch-case-initcap 804@cindex -fmatch-case-upper option 805@cindex options, -fmatch-case-upper 806@item -fmatch-case-upper 807@cindex -fmatch-case-lower option 808@cindex options, -fmatch-case-lower 809@item -fmatch-case-lower 810@cindex -fmatch-case-any option 811@cindex options, -fmatch-case-any 812@item -fmatch-case-any 813Specify expected case for keywords. 814@option{-fmatch-case-lower} is the default. 815 816@cindex -fsource-case-upper option 817@cindex options, -fsource-case-upper 818@item -fsource-case-upper 819@cindex -fsource-case-lower option 820@cindex options, -fsource-case-lower 821@item -fsource-case-lower 822@cindex -fsource-case-preserve option 823@cindex options, -fsource-case-preserve 824@item -fsource-case-preserve 825Specify whether source text other than character and Hollerith constants 826is to be translated to uppercase, to lowercase, or preserved as is. 827@option{-fsource-case-lower} is the default. 828 829@cindex -fsymbol-case-initcap option 830@cindex options, -fsymbol-case-initcap 831@item -fsymbol-case-initcap 832@cindex -fsymbol-case-upper option 833@cindex options, -fsymbol-case-upper 834@item -fsymbol-case-upper 835@cindex -fsymbol-case-lower option 836@cindex options, -fsymbol-case-lower 837@item -fsymbol-case-lower 838@cindex -fsymbol-case-any option 839@cindex options, -fsymbol-case-any 840@item -fsymbol-case-any 841Specify valid cases for user-defined symbol names. 842@option{-fsymbol-case-any} is the default. 843 844@cindex -fcase-strict-upper option 845@cindex options, -fcase-strict-upper 846@item -fcase-strict-upper 847Same as @samp{-fintrin-case-upper -fmatch-case-upper -fsource-case-preserve 848-fsymbol-case-upper}. 849(Requires all pertinent source to be in uppercase.) 850 851@cindex -fcase-strict-lower option 852@cindex options, -fcase-strict-lower 853@item -fcase-strict-lower 854Same as @samp{-fintrin-case-lower -fmatch-case-lower -fsource-case-preserve 855-fsymbol-case-lower}. 856(Requires all pertinent source to be in lowercase.) 857 858@cindex -fcase-initcap option 859@cindex options, -fcase-initcap 860@item -fcase-initcap 861Same as @samp{-fintrin-case-initcap -fmatch-case-initcap -fsource-case-preserve 862-fsymbol-case-initcap}. 863(Requires all pertinent source to be in initial capitals, 864as in @samp{Print *,SqRt(Value)}.) 865 866@cindex -fcase-upper option 867@cindex options, -fcase-upper 868@item -fcase-upper 869Same as @samp{-fintrin-case-any -fmatch-case-any -fsource-case-upper 870-fsymbol-case-any}. 871(Maps all pertinent source to uppercase.) 872 873@cindex -fcase-lower option 874@cindex options, -fcase-lower 875@item -fcase-lower 876Same as @samp{-fintrin-case-any -fmatch-case-any -fsource-case-lower 877-fsymbol-case-any}. 878(Maps all pertinent source to lowercase.) 879 880@cindex -fcase-preserve option 881@cindex options, -fcase-preserve 882@item -fcase-preserve 883Same as @samp{-fintrin-case-any -fmatch-case-any -fsource-case-preserve 884-fsymbol-case-any}. 885(Preserves all case in user-defined symbols, 886while allowing any-case matching of intrinsics and keywords. 887For example, @samp{call Foo(i,I)} would pass two @emph{different} 888variables named @samp{i} and @samp{I} to a procedure named @samp{Foo}.) 889 890@cindex -fbadu77-intrinsics-delete option 891@cindex options, -fbadu77-intrinsics-delete 892@item -fbadu77-intrinsics-delete 893@cindex -fbadu77-intrinsics-hide option 894@cindex options, -fbadu77-intrinsics-hide 895@item -fbadu77-intrinsics-hide 896@cindex -fbadu77-intrinsics-disable option 897@cindex options, -fbadu77-intrinsics-disable 898@item -fbadu77-intrinsics-disable 899@cindex -fbadu77-intrinsics-enable option 900@cindex options, -fbadu77-intrinsics-enable 901@item -fbadu77-intrinsics-enable 902@cindex @code{badu77} intrinsics 903@cindex intrinsics, @code{badu77} 904Specify status of UNIX intrinsics having inappropriate forms. 905@option{-fbadu77-intrinsics-enable} is the default. 906@xref{Intrinsic Groups}. 907 908@cindex -ff2c-intrinsics-delete option 909@cindex options, -ff2c-intrinsics-delete 910@item -ff2c-intrinsics-delete 911@cindex -ff2c-intrinsics-hide option 912@cindex options, -ff2c-intrinsics-hide 913@item -ff2c-intrinsics-hide 914@cindex -ff2c-intrinsics-disable option 915@cindex options, -ff2c-intrinsics-disable 916@item -ff2c-intrinsics-disable 917@cindex -ff2c-intrinsics-enable option 918@cindex options, -ff2c-intrinsics-enable 919@item -ff2c-intrinsics-enable 920@cindex @command{f2c} intrinsics 921@cindex intrinsics, @command{f2c} 922Specify status of f2c-specific intrinsics. 923@option{-ff2c-intrinsics-enable} is the default. 924@xref{Intrinsic Groups}. 925 926@cindex -ff90-intrinsics-delete option 927@cindex options, -ff90-intrinsics-delete 928@item -ff90-intrinsics-delete 929@cindex -ff90-intrinsics-hide option 930@cindex options, -ff90-intrinsics-hide 931@item -ff90-intrinsics-hide 932@cindex -ff90-intrinsics-disable option 933@cindex options, -ff90-intrinsics-disable 934@item -ff90-intrinsics-disable 935@cindex -ff90-intrinsics-enable option 936@cindex options, -ff90-intrinsics-enable 937@item -ff90-intrinsics-enable 938@cindex Fortran 90, intrinsics 939@cindex intrinsics, Fortran 90 940Specify status of F90-specific intrinsics. 941@option{-ff90-intrinsics-enable} is the default. 942@xref{Intrinsic Groups}. 943 944@cindex -fgnu-intrinsics-delete option 945@cindex options, -fgnu-intrinsics-delete 946@item -fgnu-intrinsics-delete 947@cindex -fgnu-intrinsics-hide option 948@cindex options, -fgnu-intrinsics-hide 949@item -fgnu-intrinsics-hide 950@cindex -fgnu-intrinsics-disable option 951@cindex options, -fgnu-intrinsics-disable 952@item -fgnu-intrinsics-disable 953@cindex -fgnu-intrinsics-enable option 954@cindex options, -fgnu-intrinsics-enable 955@item -fgnu-intrinsics-enable 956@cindex Digital Fortran features 957@cindex @code{COMPLEX} intrinsics 958@cindex intrinsics, @code{COMPLEX} 959Specify status of Digital's COMPLEX-related intrinsics. 960@option{-fgnu-intrinsics-enable} is the default. 961@xref{Intrinsic Groups}. 962 963@cindex -fmil-intrinsics-delete option 964@cindex options, -fmil-intrinsics-delete 965@item -fmil-intrinsics-delete 966@cindex -fmil-intrinsics-hide option 967@cindex options, -fmil-intrinsics-hide 968@item -fmil-intrinsics-hide 969@cindex -fmil-intrinsics-disable option 970@cindex options, -fmil-intrinsics-disable 971@item -fmil-intrinsics-disable 972@cindex -fmil-intrinsics-enable option 973@cindex options, -fmil-intrinsics-enable 974@item -fmil-intrinsics-enable 975@cindex MIL-STD 1753 976@cindex intrinsics, MIL-STD 1753 977Specify status of MIL-STD-1753-specific intrinsics. 978@option{-fmil-intrinsics-enable} is the default. 979@xref{Intrinsic Groups}. 980 981@cindex -funix-intrinsics-delete option 982@cindex options, -funix-intrinsics-delete 983@item -funix-intrinsics-delete 984@cindex -funix-intrinsics-hide option 985@cindex options, -funix-intrinsics-hide 986@item -funix-intrinsics-hide 987@cindex -funix-intrinsics-disable option 988@cindex options, -funix-intrinsics-disable 989@item -funix-intrinsics-disable 990@cindex -funix-intrinsics-enable option 991@cindex options, -funix-intrinsics-enable 992@item -funix-intrinsics-enable 993@cindex UNIX intrinsics 994@cindex intrinsics, UNIX 995Specify status of UNIX intrinsics. 996@option{-funix-intrinsics-enable} is the default. 997@xref{Intrinsic Groups}. 998 999@cindex -fvxt-intrinsics-delete option 1000@cindex options, -fvxt-intrinsics-delete 1001@item -fvxt-intrinsics-delete 1002@cindex -fvxt-intrinsics-hide option 1003@cindex options, -fvxt-intrinsics-hide 1004@item -fvxt-intrinsics-hide 1005@cindex -fvxt-intrinsics-disable option 1006@cindex options, -fvxt-intrinsics-disable 1007@item -fvxt-intrinsics-disable 1008@cindex -fvxt-intrinsics-enable option 1009@cindex options, -fvxt-intrinsics-enable 1010@item -fvxt-intrinsics-enable 1011@cindex VXT intrinsics 1012@cindex intrinsics, VXT 1013Specify status of VXT intrinsics. 1014@option{-fvxt-intrinsics-enable} is the default. 1015@xref{Intrinsic Groups}. 1016 1017@cindex -ffixed-line-length-@var{n} option 1018@cindex options, -ffixed-line-length-@var{n} 1019@item -ffixed-line-length-@var{n} 1020@cindex source file format 1021@cindex lines, length 1022@cindex length of source lines 1023@cindex fixed form 1024@cindex limits, lengths of source lines 1025Set column after which characters are ignored in typical fixed-form 1026lines in the source file, and through which spaces are assumed (as 1027if padded to that length) after the ends of short fixed-form lines. 1028 1029@cindex card image 1030@cindex extended-source option 1031Popular values for @var{n} include 72 (the 1032standard and the default), 80 (card image), and 132 (corresponds 1033to ``extended-source'' options in some popular compilers). 1034@var{n} may be @samp{none}, meaning that the entire line is meaningful 1035and that continued character constants never have implicit spaces appended 1036to them to fill out the line. 1037@option{-ffixed-line-length-0} means the same thing as 1038@option{-ffixed-line-length-none}. 1039 1040@xref{Source Form}, for more information. 1041@end table 1042 1043@node Warning Options 1044@section Options to Request or Suppress Warnings 1045@cindex options, warnings 1046@cindex warnings, suppressing 1047@cindex messages, warning 1048@cindex suppressing warnings 1049 1050Warnings are diagnostic messages that report constructions which 1051are not inherently erroneous but which are risky or suggest there 1052might have been an error. 1053 1054You can request many specific warnings with options beginning @option{-W}, 1055for example @option{-Wimplicit} to request warnings on implicit 1056declarations. Each of these specific warning options also has a 1057negative form beginning @option{-Wno-} to turn off warnings; 1058for example, @option{-Wno-implicit}. This manual lists only one of the 1059two forms, whichever is not the default. 1060 1061These options control the amount and kinds of warnings produced by GNU 1062Fortran: 1063 1064@table @gcctabopt 1065@cindex syntax checking 1066@cindex -fsyntax-only option 1067@cindex options, -fsyntax-only 1068@item -fsyntax-only 1069Check the code for syntax errors, but don't do anything beyond that. 1070 1071@cindex -pedantic option 1072@cindex options, -pedantic 1073@item -pedantic 1074Issue warnings for uses of extensions to ANSI FORTRAN 77. 1075@option{-pedantic} also applies to C-language constructs where they 1076occur in GNU Fortran source files, such as use of @samp{\e} in a 1077character constant within a directive like @samp{#include}. 1078 1079Valid ANSI FORTRAN 77 programs should compile properly with or without 1080this option. 1081However, without this option, certain GNU extensions and traditional 1082Fortran features are supported as well. 1083With this option, many of them are rejected. 1084 1085Some users try to use @option{-pedantic} to check programs for strict ANSI 1086conformance. 1087They soon find that it does not do quite what they want---it finds some 1088non-ANSI practices, but not all. 1089However, improvements to @command{g77} in this area are welcome. 1090 1091@cindex -pedantic-errors option 1092@cindex options, -pedantic-errors 1093@item -pedantic-errors 1094Like @option{-pedantic}, except that errors are produced rather than 1095warnings. 1096 1097@cindex -fpedantic option 1098@cindex options, -fpedantic 1099@item -fpedantic 1100Like @option{-pedantic}, but applies only to Fortran constructs. 1101 1102@cindex -w option 1103@cindex options, -w 1104@item -w 1105Inhibit all warning messages. 1106 1107@cindex -Wno-globals option 1108@cindex options, -Wno-globals 1109@item -Wno-globals 1110@cindex global names, warning 1111@cindex warnings, global names 1112Inhibit warnings about use of a name as both a global name 1113(a subroutine, function, or block data program unit, or a 1114common block) and implicitly as the name of an intrinsic 1115in a source file. 1116 1117Also inhibit warnings about inconsistent invocations and/or 1118definitions of global procedures (function and subroutines). 1119Such inconsistencies include different numbers of arguments 1120and different types of arguments. 1121 1122@cindex -Wimplicit option 1123@cindex options, -Wimplicit 1124@item -Wimplicit 1125@cindex implicit declaration, warning 1126@cindex warnings, implicit declaration 1127@cindex -u option 1128@cindex /WARNINGS=DECLARATIONS switch 1129@cindex IMPLICIT NONE, similar effect 1130@cindex effecting IMPLICIT NONE 1131Warn whenever a variable, array, or function is implicitly 1132declared. 1133Has an effect similar to using the @code{IMPLICIT NONE} statement 1134in every program unit. 1135(Some Fortran compilers provide this feature by an option 1136named @option{-u} or @samp{/WARNINGS=DECLARATIONS}.) 1137 1138@cindex -Wunused option 1139@cindex options, -Wunused 1140@item -Wunused 1141@cindex unused variables 1142@cindex variables, unused 1143Warn whenever a variable is unused aside from its declaration. 1144 1145@cindex -Wuninitialized option 1146@cindex options, -Wuninitialized 1147@item -Wuninitialized 1148@cindex uninitialized variables 1149@cindex variables, uninitialized 1150Warn whenever an automatic variable is used without first being initialized. 1151 1152These warnings are possible only in optimizing compilation, 1153because they require data-flow information that is computed only 1154when optimizing. If you don't specify @option{-O}, you simply won't 1155get these warnings. 1156 1157These warnings occur only for variables that are candidates for 1158register allocation. Therefore, they do not occur for a variable 1159@c that is declared @code{VOLATILE}, or 1160whose address is taken, or whose size 1161is other than 1, 2, 4 or 8 bytes. Also, they do not occur for 1162arrays, even when they are in registers. 1163 1164Note that there might be no warning about a variable that is used only 1165to compute a value that itself is never used, because such 1166computations may be deleted by data-flow analysis before the warnings 1167are printed. 1168 1169These warnings are made optional because GNU Fortran is not smart 1170enough to see all the reasons why the code might be correct 1171despite appearing to have an error. Here is one example of how 1172this can happen: 1173 1174@example 1175SUBROUTINE DISPAT(J) 1176IF (J.EQ.1) I=1 1177IF (J.EQ.2) I=4 1178IF (J.EQ.3) I=5 1179CALL FOO(I) 1180END 1181@end example 1182 1183@noindent 1184If the value of @code{J} is always 1, 2 or 3, then @code{I} is 1185always initialized, but GNU Fortran doesn't know this. Here is 1186another common case: 1187 1188@example 1189SUBROUTINE MAYBE(FLAG) 1190LOGICAL FLAG 1191IF (FLAG) VALUE = 9.4 1192@dots{} 1193IF (FLAG) PRINT *, VALUE 1194END 1195@end example 1196 1197@noindent 1198This has no bug because @code{VALUE} is used only if it is set. 1199 1200@cindex -Wall option 1201@cindex options, -Wall 1202@item -Wall 1203@cindex all warnings 1204@cindex warnings, all 1205The @option{-Wunused} and @option{-Wuninitialized} options combined. 1206These are all the 1207options which pertain to usage that we recommend avoiding and that we 1208believe is easy to avoid. 1209(As more warnings are added to @command{g77} some might 1210be added to the list enabled by @option{-Wall}.) 1211@end table 1212 1213The remaining @option{-W@dots{}} options are not implied by @option{-Wall} 1214because they warn about constructions that we consider reasonable to 1215use, on occasion, in clean programs. 1216 1217@table @gcctabopt 1218@c @item -W 1219@c Print extra warning messages for these events: 1220@c 1221@c @itemize @bullet 1222@c @item 1223@c If @option{-Wall} or @option{-Wunused} is also specified, warn about unused 1224@c arguments. 1225@c 1226@c @end itemize 1227@c 1228@cindex -Wsurprising option 1229@cindex options, -Wsurprising 1230@item -Wsurprising 1231Warn about ``suspicious'' constructs that are interpreted 1232by the compiler in a way that might well be surprising to 1233someone reading the code. 1234These differences can result in subtle, compiler-dependent 1235(even machine-dependent) behavioral differences. 1236The constructs warned about include: 1237 1238@itemize @bullet 1239@item 1240Expressions having two arithmetic operators in a row, such 1241as @samp{X*-Y}. 1242Such a construct is nonstandard, and can produce 1243unexpected results in more complicated situations such 1244as @samp{X**-Y*Z}. 1245@command{g77} along with many other compilers, interprets 1246this example differently than many programmers, and a few 1247other compilers. 1248Specifically, @command{g77} interprets @samp{X**-Y*Z} as 1249@samp{(X**(-Y))*Z}, while others might think it should 1250be interpreted as @samp{X**(-(Y*Z))}. 1251 1252A revealing example is the constant expression @samp{2**-2*1.}, 1253which @command{g77} evaluates to .25, while others might evaluate 1254it to 0., the difference resulting from the way precedence affects 1255type promotion. 1256 1257(The @option{-fpedantic} option also warns about expressions 1258having two arithmetic operators in a row.) 1259 1260@item 1261Expressions with a unary minus followed by an operand and then 1262a binary operator other than plus or minus. 1263For example, @samp{-2**2} produces a warning, because 1264the precedence is @samp{-(2**2)}, yielding -4, not 1265@samp{(-2)**2}, which yields 4, and which might represent 1266what a programmer expects. 1267 1268An example of an expression producing different results 1269in a surprising way is @samp{-I*S}, where @var{I} holds 1270the value @samp{-2147483648} and @var{S} holds @samp{0.5}. 1271On many systems, negating @var{I} results in the same 1272value, not a positive number, because it is already the 1273lower bound of what an @code{INTEGER(KIND=1)} variable can hold. 1274So, the expression evaluates to a positive number, while 1275the ``expected'' interpretation, @samp{(-I)*S}, would 1276evaluate to a negative number. 1277 1278Even cases such as @samp{-I*J} produce warnings, 1279even though, in most configurations and situations, 1280there is no computational difference between the 1281results of the two interpretations---the purpose 1282of this warning is to warn about differing interpretations 1283and encourage a better style of coding, not to identify 1284only those places where bugs might exist in the user's 1285code. 1286 1287@cindex DO statement 1288@cindex statements, DO 1289@item 1290@code{DO} loops with @code{DO} variables that are not 1291of integral type---that is, using @code{REAL} 1292variables as loop control variables. 1293Although such loops can be written to work in the 1294``obvious'' way, the way @command{g77} is required by the 1295Fortran standard to interpret such code is likely to 1296be quite different from the way many programmers expect. 1297(This is true of all @code{DO} loops, but the differences 1298are pronounced for non-integral loop control variables.) 1299 1300@xref{Loops}, for more information. 1301@end itemize 1302 1303@cindex -Werror option 1304@cindex options, -Werror 1305@item -Werror 1306Make all warnings into errors. 1307 1308@cindex -W option 1309@cindex options, -W 1310@item -W 1311@cindex extra warnings 1312@cindex warnings, extra 1313Turns on ``extra warnings'' and, if optimization is specified 1314via @option{-O}, the @option{-Wuninitialized} option. 1315(This might change in future versions of @command{g77} 1316 1317``Extra warnings'' are issued for: 1318 1319@itemize @bullet 1320@item 1321@cindex unused parameters 1322@cindex parameters, unused 1323@cindex unused arguments 1324@cindex arguments, unused 1325@cindex unused dummies 1326@cindex dummies, unused 1327Unused parameters to a procedure (when @option{-Wunused} also is 1328specified). 1329 1330@item 1331@cindex overflow 1332Overflows involving floating-point constants (not available 1333for certain configurations). 1334@end itemize 1335@end table 1336 1337@xref{Warning Options,,Options to Request or Suppress Warnings, 1338gcc,Using the GNU Compiler Collection (GCC)}, for information on more 1339options offered 1340by the GBE shared by @command{g77} @command{gcc} and other GNU compilers. 1341 1342Some of these have no effect when compiling programs written in Fortran: 1343 1344@table @gcctabopt 1345@cindex -Wcomment option 1346@cindex options, -Wcomment 1347@item -Wcomment 1348@cindex -Wformat option 1349@cindex options, -Wformat 1350@item -Wformat 1351@cindex -Wparentheses option 1352@cindex options, -Wparentheses 1353@item -Wparentheses 1354@cindex -Wswitch option 1355@cindex options, -Wswitch 1356@item -Wswitch 1357@cindex -Wswitch-default option 1358@cindex options, -Wswitch-default 1359@item -Wswitch-default 1360@cindex -Wswitch-enum option 1361@cindex options, -Wswitch-enum 1362@item -Wswitch-enum 1363@cindex -Wtraditional option 1364@cindex options, -Wtraditional 1365@item -Wtraditional 1366@cindex -Wshadow option 1367@cindex options, -Wshadow 1368@item -Wshadow 1369@cindex -Wid-clash-@var{len} option 1370@cindex options, -Wid-clash-@var{len} 1371@item -Wid-clash-@var{len} 1372@cindex -Wlarger-than-@var{len} option 1373@cindex options, -Wlarger-than-@var{len} 1374@item -Wlarger-than-@var{len} 1375@cindex -Wconversion option 1376@cindex options, -Wconversion 1377@item -Wconversion 1378@cindex -Waggregate-return option 1379@cindex options, -Waggregate-return 1380@item -Waggregate-return 1381@cindex -Wredundant-decls option 1382@cindex options, -Wredundant-decls 1383@item -Wredundant-decls 1384@cindex unsupported warnings 1385@cindex warnings, unsupported 1386These options all could have some relevant meaning for 1387GNU Fortran programs, but are not yet supported. 1388@end table 1389 1390@node Debugging Options 1391@section Options for Debugging Your Program or GNU Fortran 1392@cindex options, debugging 1393@cindex debugging information options 1394 1395GNU Fortran has various special options that are used for debugging 1396either your program or @command{g77} 1397 1398@table @gcctabopt 1399@cindex -g option 1400@cindex options, -g 1401@item -g 1402Produce debugging information in the operating system's native format 1403(stabs, COFF, XCOFF, or DWARF). GDB can work with this debugging 1404information. 1405 1406A sample debugging session looks like this (note the use of the breakpoint): 1407@smallexample 1408$ cat gdb.f 1409 PROGRAM PROG 1410 DIMENSION A(10) 1411 DATA A /1.,2.,3.,4.,5.,6.,7.,8.,9.,10./ 1412 A(5) = 4. 1413 PRINT*,A 1414 END 1415$ g77 -g -O gdb.f 1416$ gdb a.out 1417... 1418(gdb) break MAIN__ 1419Breakpoint 1 at 0x8048e96: file gdb.f, line 4. 1420(gdb) run 1421Starting program: /home/toon/g77-bugs/./a.out 1422Breakpoint 1, MAIN__ () at gdb.f:4 14234 A(5) = 4. 1424Current language: auto; currently fortran 1425(gdb) print a(5) 1426$1 = 5 1427(gdb) step 14285 PRINT*,A 1429(gdb) print a(5) 1430$2 = 4 1431... 1432@end smallexample 1433One could also add the setting of the breakpoint and the first run command 1434to the file @file{.gdbinit} in the current directory, to simplify the debugging 1435session. 1436@end table 1437 1438@xref{Debugging Options,,Options for Debugging Your Program or GCC, 1439gcc,Using the GNU Compiler Collection (GCC)}, for more information on 1440debugging options. 1441 1442@node Optimize Options 1443@section Options That Control Optimization 1444@cindex optimize options 1445@cindex options, optimization 1446 1447Most Fortran users will want to use no optimization when 1448developing and testing programs, and use @option{-O} or @option{-O2} when 1449compiling programs for late-cycle testing and for production use. 1450However, note that certain diagnostics---such as for uninitialized 1451variables---depend on the flow analysis done by @option{-O}, i.e.@: you 1452must use @option{-O} or @option{-O2} to get such diagnostics. 1453 1454The following flags have particular applicability when 1455compiling Fortran programs: 1456 1457@table @gcctabopt 1458@cindex -malign-double option 1459@cindex options, -malign-double 1460@item -malign-double 1461(Intel x86 architecture only.) 1462 1463Noticeably improves performance of @command{g77} programs making 1464heavy use of @code{REAL(KIND=2)} (@code{DOUBLE PRECISION}) data 1465on some systems. 1466In particular, systems using Pentium, Pentium Pro, 586, and 1467686 implementations 1468of the i386 architecture execute programs faster when 1469@code{REAL(KIND=2)} (@code{DOUBLE PRECISION}) data are 1470aligned on 64-bit boundaries 1471in memory. 1472 1473This option can, at least, make benchmark results more consistent 1474across various system configurations, versions of the program, 1475and data sets. 1476 1477@emph{Note:} The warning in the @command{gcc} documentation about 1478this option does not apply, generally speaking, to Fortran 1479code compiled by @command{g77} 1480 1481@xref{Aligned Data}, for more information on alignment issues. 1482 1483@emph{Also also note:} The negative form of @option{-malign-double} 1484is @option{-mno-align-double}, not @option{-benign-double}. 1485 1486@cindex -ffloat-store option 1487@cindex options, -ffloat-store 1488@item -ffloat-store 1489@cindex IEEE 754 conformance 1490@cindex conformance, IEEE 754 1491@cindex floating-point, precision 1492Might help a Fortran program that depends on exact IEEE conformance on 1493some machines, but might slow down a program that doesn't. 1494 1495This option is effective when the floating-point unit is set to work in 1496IEEE 854 `extended precision'---as it typically is on x86 and m68k GNU 1497systems---rather than IEEE 754 double precision. @option{-ffloat-store} 1498tries to remove the extra precision by spilling data from floating-point 1499registers into memory and this typically involves a big performance 1500hit. However, it doesn't affect intermediate results, so that it is 1501only partially effective. `Excess precision' is avoided in code like: 1502@smallexample 1503a = b + c 1504d = a * e 1505@end smallexample 1506but not in code like: 1507@smallexample 1508 d = (b + c) * e 1509@end smallexample 1510 1511For another, potentially better, way of controlling the precision, 1512see @ref{Floating-point precision}. 1513 1514@cindex -fforce-mem option 1515@cindex options, -fforce-mem 1516@item -fforce-mem 1517@cindex -fforce-addr option 1518@cindex options, -fforce-addr 1519@item -fforce-addr 1520@cindex loops, speeding up 1521@cindex speed, of loops 1522Might improve optimization of loops. 1523 1524@cindex -fno-inline option 1525@cindex options, -fno-inline 1526@item -fno-inline 1527@cindex in-line code 1528@cindex compilation, in-line 1529@c DL: Only relevant for -O3? TM: No, statement functions are 1530@c inlined even at -O1. 1531Don't compile statement functions inline. 1532Might reduce the size of a program unit---which might be at 1533expense of some speed (though it should compile faster). 1534Note that if you are not optimizing, no functions can be expanded inline. 1535 1536@cindex -ffast-math option 1537@cindex options, -ffast-math 1538@item -ffast-math 1539@cindex IEEE 754 conformance 1540@cindex conformance, IEEE 754 1541Might allow some programs designed to not be too dependent 1542on IEEE behavior for floating-point to run faster, or die trying. 1543Sets @option{-funsafe-math-optimizations}, @option{-ffinite-math-only}, 1544and @option{-fno-trapping-math}. 1545 1546@cindex -funsafe-math-optimizations option 1547@cindex options, -funsafe-math-optimizations 1548@item -funsafe-math-optimizations 1549Allow optimizations that may be give incorrect results 1550for certain IEEE inputs. 1551 1552@cindex -ffinite-math-only option 1553@cindex options, -ffinite-math-only 1554@item -ffinite-math-only 1555Allow optimizations for floating-point arithmetic that assume 1556that arguments and results are not NaNs or +-Infs. 1557 1558This option should never be turned on by any @option{-O} option since 1559it can result in incorrect output for programs which depend on 1560an exact implementation of IEEE or ISO rules/specifications. 1561 1562The default is @option{-fno-finite-math-only}. 1563 1564@cindex -fno-trapping-math option 1565@cindex options, -fno-trapping-math 1566@item -fno-trapping-math 1567Allow the compiler to assume that floating-point arithmetic 1568will not generate traps on any inputs. This is useful, for 1569example, when running a program using IEEE "non-stop" 1570floating-point arithmetic. 1571 1572@cindex -fstrength-reduce option 1573@cindex options, -fstrength-reduce 1574@item -fstrength-reduce 1575@cindex loops, speeding up 1576@cindex speed, of loops 1577@c DL: normally defaulted? 1578Might make some loops run faster. 1579 1580@cindex -frerun-cse-after-loop option 1581@cindex options, -frerun-cse-after-loop 1582@item -frerun-cse-after-loop 1583@cindex -fexpensive-optimizations option 1584@cindex options, -fexpensive-optimizations 1585@c DL: This is -O2? 1586@item -fexpensive-optimizations 1587@cindex -fdelayed-branch option 1588@cindex options, -fdelayed-branch 1589@item -fdelayed-branch 1590@cindex -fschedule-insns option 1591@cindex options, -fschedule-insns 1592@item -fschedule-insns 1593@cindex -fschedule-insns2 option 1594@cindex options, -fschedule-insns2 1595@item -fschedule-insns2 1596@cindex -fcaller-saves option 1597@cindex options, -fcaller-saves 1598@item -fcaller-saves 1599Might improve performance on some code. 1600 1601@cindex -funroll-loops option 1602@cindex options, -funroll-loops 1603@item -funroll-loops 1604@cindex loops, unrolling 1605@cindex unrolling loops 1606@cindex loops, optimizing 1607@cindex indexed (iterative) @code{DO} 1608@cindex iterative @code{DO} 1609@c DL: fixme: Craig doesn't like `indexed' but f95 doesn't seem to 1610@c provide a suitable term 1611@c CB: I've decided on `iterative', for the time being, and changed 1612@c my previous, rather bizarre, use of `imperative' to that 1613@c (though `precomputed-trip' would be a more precise adjective) 1614Typically improves performance on code using iterative @code{DO} loops by 1615unrolling them and is probably generally appropriate for Fortran, though 1616it is not turned on at any optimization level. 1617Note that outer loop unrolling isn't done specifically; decisions about 1618whether to unroll a loop are made on the basis of its instruction count. 1619 1620@c DL: Fixme: This should obviously go somewhere else... 1621Also, no `loop discovery'@footnote{@dfn{loop discovery} refers to the 1622process by which a compiler, or indeed any reader of a program, 1623determines which portions of the program are more likely to be executed 1624repeatedly as it is being run. Such discovery typically is done early 1625when compiling using optimization techniques, so the ``discovered'' 1626loops get more attention---and more run-time resources, such as 1627registers---from the compiler. It is easy to ``discover'' loops that are 1628constructed out of looping constructs in the language 1629(such as Fortran's @code{DO}). For some programs, ``discovering'' loops 1630constructed out of lower-level constructs (such as @code{IF} and 1631@code{GOTO}) can lead to generation of more optimal code 1632than otherwise.} is done, so only loops written with @code{DO} 1633benefit from loop optimizations, including---but not limited 1634to---unrolling. Loops written with @code{IF} and @code{GOTO} are not 1635currently recognized as such. This option unrolls only iterative 1636@code{DO} loops, not @code{DO WHILE} loops. 1637 1638@cindex -funroll-all-loops option 1639@cindex options, -funroll-all-loops 1640@cindex DO WHILE 1641@item -funroll-all-loops 1642@c DL: Check my understanding of -funroll-all-loops v. -funroll-loops is correct. 1643Probably improves performance on code using @code{DO WHILE} loops by 1644unrolling them in addition to iterative @code{DO} loops. In the absence 1645of @code{DO WHILE}, this option is equivalent to @option{-funroll-loops} 1646but possibly slower. 1647 1648@item -fno-move-all-movables 1649@cindex -fno-move-all-movables option 1650@cindex options, -fno-move-all-movables 1651@item -fno-reduce-all-givs 1652@cindex -fno-reduce-all-givs option 1653@cindex options, -fno-reduce-all-givs 1654@item -fno-rerun-loop-opt 1655@cindex -fno-rerun-loop-opt option 1656@cindex options, -fno-rerun-loop-opt 1657In general, the optimizations enabled with these options will lead to 1658faster code being generated by GNU Fortran; hence they are enabled by default 1659when issuing the @command{g77} command. 1660 1661@option{-fmove-all-movables} and @option{-freduce-all-givs} will enable 1662loop optimization to move all loop-invariant index computations in nested 1663loops over multi-rank array dummy arguments out of these loops. 1664 1665@option{-frerun-loop-opt} will move offset calculations resulting 1666from the fact that Fortran arrays by default have a lower bound of 1 1667out of the loops. 1668 1669These three options are intended to be removed someday, once 1670loop optimization is sufficiently advanced to perform all those 1671transformations without help from these options. 1672@end table 1673 1674@xref{Optimize Options,,Options That Control Optimization, 1675gcc,Using the GNU Compiler Collection (GCC)}, for more information on options 1676to optimize the generated machine code. 1677 1678@node Preprocessor Options 1679@section Options Controlling the Preprocessor 1680@cindex preprocessor options 1681@cindex options, preprocessor 1682@cindex cpp program 1683@cindex programs, cpp 1684 1685These options control the C preprocessor, which is run on each C source 1686file before actual compilation. 1687 1688@xref{Preprocessor Options,,Options Controlling the Preprocessor, 1689gcc,Using the GNU Compiler Collection (GCC)}, for information on C 1690preprocessor options. 1691 1692@cindex INCLUDE directive 1693@cindex directive, INCLUDE 1694Some of these options also affect how @command{g77} processes the 1695@code{INCLUDE} directive. 1696Since this directive is processed even when preprocessing 1697is not requested, it is not described in this section. 1698@xref{Directory Options,,Options for Directory Search}, for 1699information on how @command{g77} processes the @code{INCLUDE} directive. 1700 1701However, the @code{INCLUDE} directive does not apply 1702preprocessing to the contents of the included file itself. 1703 1704Therefore, any file that contains preprocessor directives 1705(such as @code{#include}, @code{#define}, and @code{#if}) 1706must be included via the @code{#include} directive, not 1707via the @code{INCLUDE} directive. 1708Therefore, any file containing preprocessor directives, 1709if included, is necessarily included by a file that itself 1710contains preprocessor directives. 1711 1712@node Directory Options 1713@section Options for Directory Search 1714@cindex directory, options 1715@cindex options, directory search 1716@cindex search path 1717 1718These options affect how the @command{cpp} preprocessor searches 1719for files specified via the @code{#include} directive. 1720Therefore, when compiling Fortran programs, they are meaningful 1721when the preprocessor is used. 1722 1723@cindex INCLUDE directive 1724@cindex directive, INCLUDE 1725Some of these options also affect how @command{g77} searches 1726for files specified via the @code{INCLUDE} directive, 1727although files included by that directive are not, 1728themselves, preprocessed. 1729These options are: 1730 1731@table @gcctabopt 1732@cindex -I- option 1733@cindex options, -I- 1734@item -I- 1735@cindex -Idir option 1736@cindex options, -Idir 1737@item -I@var{dir} 1738@cindex directory, search paths for inclusion 1739@cindex inclusion, directory search paths for 1740@cindex search paths, for included files 1741@cindex paths, search 1742These affect interpretation of the @code{INCLUDE} directive 1743(as well as of the @code{#include} directive of the @command{cpp} 1744preprocessor). 1745 1746Note that @option{-I@var{dir}} must be specified @emph{without} any 1747spaces between @option{-I} and the directory name---that is, 1748@option{-Ifoo/bar} is valid, but @option{-I foo/bar} 1749is rejected by the @command{g77} compiler (though the preprocessor supports 1750the latter form). 1751@c this is due to toplev.c's inflexible option processing 1752Also note that the general behavior of @option{-I} and 1753@code{INCLUDE} is pretty much the same as of @option{-I} with 1754@code{#include} in the @command{cpp} preprocessor, with regard to 1755looking for @file{header.gcc} files and other such things. 1756 1757@xref{Directory Options,,Options for Directory Search, 1758gcc,Using the GNU Compiler Collection (GCC)}, for information on the 1759@option{-I} option. 1760@end table 1761 1762@node Code Gen Options 1763@section Options for Code Generation Conventions 1764@cindex code generation, conventions 1765@cindex options, code generation 1766@cindex run-time, options 1767 1768These machine-independent options control the interface conventions 1769used in code generation. 1770 1771Most of them have both positive and negative forms; the negative form 1772of @option{-ffoo} would be @option{-fno-foo}. In the table below, only 1773one of the forms is listed---the one which is not the default. You 1774can figure out the other form by either removing @option{no-} or adding 1775it. 1776 1777@table @gcctabopt 1778@cindex -fno-automatic option 1779@cindex options, -fno-automatic 1780@item -fno-automatic 1781@cindex SAVE statement 1782@cindex statements, SAVE 1783Treat each program unit as if the @code{SAVE} statement was specified 1784for every local variable and array referenced in it. 1785Does not affect common blocks. 1786(Some Fortran compilers provide this option under 1787the name @option{-static}.) 1788 1789@cindex -finit-local-zero option 1790@cindex options, -finit-local-zero 1791@item -finit-local-zero 1792@cindex DATA statement 1793@cindex statements, DATA 1794@cindex initialization, of local variables 1795@cindex variables, initialization of 1796@cindex uninitialized variables 1797@cindex variables, uninitialized 1798Specify that variables and arrays that are local to a program unit 1799(not in a common block and not passed as an argument) are to be initialized 1800to binary zeros. 1801 1802Since there is a run-time penalty for initialization of variables 1803that are not given the @code{SAVE} attribute, it might be a 1804good idea to also use @option{-fno-automatic} with @option{-finit-local-zero}. 1805 1806@cindex -fno-f2c option 1807@cindex options, -fno-f2c 1808@item -fno-f2c 1809@cindex @command{f2c} compatibility 1810@cindex compatibility, @command{f2c} 1811Do not generate code designed to be compatible with code generated 1812by @command{f2c} use the GNU calling conventions instead. 1813 1814The @command{f2c} calling conventions require functions that return 1815type @code{REAL(KIND=1)} to actually return the C type @code{double}, 1816and functions that return type @code{COMPLEX} to return the 1817values via an extra argument in the calling sequence that points 1818to where to store the return value. 1819Under the GNU calling conventions, such functions simply return 1820their results as they would in GNU C---@code{REAL(KIND=1)} functions 1821return the C type @code{float}, and @code{COMPLEX} functions 1822return the GNU C type @code{complex} (or its @code{struct} 1823equivalent). 1824 1825This does not affect the generation of code that interfaces with the 1826@code{libg2c} library. 1827 1828However, because the @code{libg2c} library uses @command{f2c} 1829calling conventions, @command{g77} rejects attempts to pass 1830intrinsics implemented by routines in this library as actual 1831arguments when @option{-fno-f2c} is used, to avoid bugs when 1832they are actually called by code expecting the GNU calling 1833conventions to work. 1834 1835For example, @samp{INTRINSIC ABS;CALL FOO(ABS)} is 1836rejected when @option{-fno-f2c} is in force. 1837(Future versions of the @command{g77} run-time library might 1838offer routines that provide GNU-callable versions of the 1839routines that implement the @command{f2c} intrinsics 1840that may be passed as actual arguments, so that 1841valid programs need not be rejected when @option{-fno-f2c} 1842is used.) 1843 1844@strong{Caution:} If @option{-fno-f2c} is used when compiling any 1845source file used in a program, it must be used when compiling 1846@emph{all} Fortran source files used in that program. 1847 1848@c seems kinda dumb to tell people about an option they can't use -- jcb 1849@c then again, we want users building future-compatible libraries with it. 1850@cindex -ff2c-library option 1851@cindex options, -ff2c-library 1852@item -ff2c-library 1853Specify that use of @code{libg2c} (or the original @code{libf2c}) 1854is required. 1855This is the default for the current version of @command{g77} 1856 1857Currently it is not 1858valid to specify @option{-fno-f2c-library}. 1859This option is provided so users can specify it in shell 1860scripts that build programs and libraries that require the 1861@code{libf2c} library, even when being compiled by future 1862versions of @command{g77} that might otherwise default to 1863generating code for an incompatible library. 1864 1865@cindex -fno-underscoring option 1866@cindex options, -fno-underscoring 1867@item -fno-underscoring 1868@cindex underscore 1869@cindex symbol names, underscores 1870@cindex transforming symbol names 1871@cindex symbol names, transforming 1872Do not transform names of entities specified in the Fortran 1873source file by appending underscores to them. 1874 1875With @option{-funderscoring} in effect, @command{g77} appends two underscores 1876to names with underscores and one underscore to external names with 1877no underscores. (@command{g77} also appends two underscores to internal 1878names with underscores to avoid naming collisions with external names. 1879The @option{-fno-second-underscore} option disables appending of the 1880second underscore in all cases.) 1881 1882This is done to ensure compatibility with code produced by many 1883UNIX Fortran compilers, including @command{f2c} which perform the 1884same transformations. 1885 1886Use of @option{-fno-underscoring} is not recommended unless you are 1887experimenting with issues such as integration of (GNU) Fortran into 1888existing system environments (vis-a-vis existing libraries, tools, and 1889so on). 1890 1891For example, with @option{-funderscoring}, and assuming other defaults like 1892@option{-fcase-lower} and that @samp{j()} and @samp{max_count()} are 1893external functions while @samp{my_var} and @samp{lvar} are local variables, 1894a statement like 1895 1896@smallexample 1897I = J() + MAX_COUNT (MY_VAR, LVAR) 1898@end smallexample 1899 1900@noindent 1901is implemented as something akin to: 1902 1903@smallexample 1904i = j_() + max_count__(&my_var__, &lvar); 1905@end smallexample 1906 1907With @option{-fno-underscoring}, the same statement is implemented as: 1908 1909@smallexample 1910i = j() + max_count(&my_var, &lvar); 1911@end smallexample 1912 1913Use of @option{-fno-underscoring} allows direct specification of 1914user-defined names while debugging and when interfacing @command{g77} 1915code with other languages. 1916 1917Note that just because the names match does @emph{not} mean that the 1918interface implemented by @command{g77} for an external name matches the 1919interface implemented by some other language for that same name. 1920That is, getting code produced by @command{g77} to link to code produced 1921by some other compiler using this or any other method can be only a 1922small part of the overall solution---getting the code generated by 1923both compilers to agree on issues other than naming can require 1924significant effort, and, unlike naming disagreements, linkers normally 1925cannot detect disagreements in these other areas. 1926 1927Also, note that with @option{-fno-underscoring}, the lack of appended 1928underscores introduces the very real possibility that a user-defined 1929external name will conflict with a name in a system library, which 1930could make finding unresolved-reference bugs quite difficult in some 1931cases---they might occur at program run time, and show up only as 1932buggy behavior at run time. 1933 1934In future versions of @command{g77} we hope to improve naming and linking 1935issues so that debugging always involves using the names as they appear 1936in the source, even if the names as seen by the linker are mangled to 1937prevent accidental linking between procedures with incompatible 1938interfaces. 1939 1940@cindex -fno-second-underscore option 1941@cindex options, -fno-second-underscore 1942@item -fno-second-underscore 1943@cindex underscore 1944@cindex symbol names, underscores 1945@cindex transforming symbol names 1946@cindex symbol names, transforming 1947Do not append a second underscore to names of entities specified 1948in the Fortran source file. 1949 1950This option has no effect if @option{-fno-underscoring} is 1951in effect. 1952 1953Otherwise, with this option, an external name such as @samp{MAX_COUNT} 1954is implemented as a reference to the link-time external symbol 1955@samp{max_count_}, instead of @samp{max_count__}. 1956 1957@cindex -fno-ident option 1958@cindex options, -fno-ident 1959@item -fno-ident 1960Ignore the @samp{#ident} directive. 1961 1962@cindex -fzeros option 1963@cindex options, -fzeros 1964@item -fzeros 1965Treat initial values of zero as if they were any other value. 1966 1967As of version 0.5.18, @command{g77} normally treats @code{DATA} and 1968other statements that are used to specify initial values of zero 1969for variables and arrays as if no values were actually specified, 1970in the sense that no diagnostics regarding multiple initializations 1971are produced. 1972 1973This is done to speed up compiling of programs that initialize 1974large arrays to zeros. 1975 1976Use @option{-fzeros} to revert to the simpler, slower behavior 1977that can catch multiple initializations by keeping track of 1978all initializations, zero or otherwise. 1979 1980@emph{Caution:} Future versions of @command{g77} might disregard this option 1981(and its negative form, the default) or interpret it somewhat 1982differently. 1983The interpretation changes will affect only non-standard 1984programs; standard-conforming programs should not be affected. 1985 1986@cindex -femulate-complex option 1987@cindex options, -femulate-complex 1988@item -femulate-complex 1989Implement @code{COMPLEX} arithmetic via emulation, 1990instead of using the facilities of 1991the @command{gcc} back end that provide direct support of 1992@code{complex} arithmetic. 1993 1994(@command{gcc} had some bugs in its back-end support 1995for @code{complex} arithmetic, due primarily to the support not being 1996completed as of version 2.8.1 and @code{egcs} 1.1.2.) 1997 1998Use @option{-femulate-complex} if you suspect code-generation bugs, 1999or experience compiler crashes, 2000that might result from @command{g77} using the @code{COMPLEX} support 2001in the @command{gcc} back end. 2002If using that option fixes the bugs or crashes you are seeing, 2003that indicates a likely @command{g77} bugs 2004(though, all compiler crashes are considered bugs), 2005so, please report it. 2006(Note that the known bugs, now believed fixed, produced compiler crashes 2007rather than causing the generation of incorrect code.) 2008 2009Use of this option should not affect how Fortran code compiled 2010by @command{g77} works in terms of its interfaces to other code, 2011e.g. that compiled by @command{f2c} 2012 2013As of GCC version 3.0, this option is not necessary anymore. 2014 2015@emph{Caution:} Future versions of @command{g77} might ignore both forms 2016of this option. 2017 2018@cindex -falias-check option 2019@cindex options, -falias-check 2020@cindex -fargument-alias option 2021@cindex options, -fargument-alias 2022@cindex -fargument-noalias option 2023@cindex options, -fargument-noalias 2024@cindex -fno-argument-noalias-global option 2025@cindex options, -fno-argument-noalias-global 2026@item -falias-check 2027@item -fargument-alias 2028@item -fargument-noalias 2029@item -fno-argument-noalias-global 2030@emph{Version info:} 2031These options are not supported by 2032versions of @command{g77} based on @command{gcc} version 2.8. 2033 2034These options specify to what degree aliasing 2035(overlap) 2036is permitted between 2037arguments (passed as pointers) and @code{COMMON} (external, or 2038public) storage. 2039 2040The default for Fortran code, as mandated by the FORTRAN 77 and 2041Fortran 90 standards, is @option{-fargument-noalias-global}. 2042The default for code written in the C language family is 2043@option{-fargument-alias}. 2044 2045Note that, on some systems, compiling with @option{-fforce-addr} in 2046effect can produce more optimal code when the default aliasing 2047options are in effect (and when optimization is enabled). 2048 2049@xref{Aliasing Assumed To Work}, for detailed information on the implications 2050of compiling Fortran code that depends on the ability to alias dummy 2051arguments. 2052 2053@cindex -fno-globals option 2054@cindex options, -fno-globals 2055@item -fno-globals 2056@cindex global names, warning 2057@cindex warnings, global names 2058@cindex in-line code 2059@cindex compilation, in-line 2060Disable diagnostics about inter-procedural 2061analysis problems, such as disagreements about the 2062type of a function or a procedure's argument, 2063that might cause a compiler crash when attempting 2064to inline a reference to a procedure within a 2065program unit. 2066(The diagnostics themselves are still produced, but 2067as warnings, unless @option{-Wno-globals} is specified, 2068in which case no relevant diagnostics are produced.) 2069 2070Further, this option disables such inlining, to 2071avoid compiler crashes resulting from incorrect 2072code that would otherwise be diagnosed. 2073 2074As such, this option might be quite useful when 2075compiling existing, ``working'' code that happens 2076to have a few bugs that do not generally show themselves, 2077but which @command{g77} diagnoses. 2078 2079Use of this option therefore has the effect of 2080instructing @command{g77} to behave more like it did 2081up through version 0.5.19.1, when it paid little or 2082no attention to disagreements between program units 2083about a procedure's type and argument information, 2084and when it performed no inlining of procedures 2085(except statement functions). 2086 2087Without this option, @command{g77} defaults to performing 2088the potentially inlining procedures as it started doing 2089in version 0.5.20, but as of version 0.5.21, it also 2090diagnoses disagreements that might cause such inlining 2091to crash the compiler as (fatal) errors, 2092and warns about similar disagreements 2093that are currently believed to not 2094likely to result in the compiler later crashing 2095or producing incorrect code. 2096 2097@cindex -fflatten-arrays option 2098@item -fflatten-arrays 2099@cindex array performance 2100@cindex arrays, flattening 2101Use back end's C-like constructs 2102(pointer plus offset) 2103instead of its @code{ARRAY_REF} construct 2104to handle all array references. 2105 2106@emph{Note:} This option is not supported. 2107It is intended for use only by @command{g77} developers, 2108to evaluate code-generation issues. 2109It might be removed at any time. 2110 2111@cindex -fbounds-check option 2112@cindex -ffortran-bounds-check option 2113@item -fbounds-check 2114@itemx -ffortran-bounds-check 2115@cindex bounds checking 2116@cindex range checking 2117@cindex array bounds checking 2118@cindex subscript checking 2119@cindex substring checking 2120@cindex checking subscripts 2121@cindex checking substrings 2122Enable generation of run-time checks for array subscripts 2123and substring start and end points 2124against the (locally) declared minimum and maximum values. 2125 2126The current implementation uses the @code{libf2c} 2127library routine @code{s_rnge} to print the diagnostic. 2128 2129However, whereas @command{f2c} generates a single check per 2130reference for a multi-dimensional array, of the computed 2131offset against the valid offset range (0 through the size of the array), 2132@command{g77} generates a single check per @emph{subscript} expression. 2133This catches some cases of potential bugs that @command{f2c} does not, 2134such as references to below the beginning of an assumed-size array. 2135 2136@command{g77} also generates checks for @code{CHARACTER} substring references, 2137something @command{f2c} currently does not do. 2138 2139Use the new @option{-ffortran-bounds-check} option 2140to specify bounds-checking for only the Fortran code you are compiling, 2141not necessarily for code written in other languages. 2142 2143@emph{Note:} To provide more detailed information on the offending subscript, 2144@command{g77} provides the @code{libg2c} run-time library routine @code{s_rnge} 2145with somewhat differently-formatted information. 2146Here's a sample diagnostic: 2147 2148@smallexample 2149Subscript out of range on file line 4, procedure rnge.f/bf. 2150Attempt to access the -6-th element of variable b[subscript-2-of-2]. 2151Aborted 2152@end smallexample 2153 2154The above message indicates that the offending source line is 2155line 4 of the file @file{rnge.f}, 2156within the program unit (or statement function) named @samp{bf}. 2157The offended array is named @samp{b}. 2158The offended array dimension is the second for a two-dimensional array, 2159and the offending, computed subscript expression was @samp{-6}. 2160 2161For a @code{CHARACTER} substring reference, the second line has 2162this appearance: 2163 2164@smallexample 2165Attempt to access the 11-th element of variable a[start-substring]. 2166@end smallexample 2167 2168This indicates that the offended @code{CHARACTER} variable or array 2169is named @samp{a}, 2170the offended substring position is the starting (leftmost) position, 2171and the offending substring expression is @samp{11}. 2172 2173(Though the verbage of @code{s_rnge} is not ideal 2174for the purpose of the @command{g77} compiler, 2175the above information should provide adequate diagnostic abilities 2176to it users.) 2177@end table 2178 2179@xref{Code Gen Options,,Options for Code Generation Conventions, 2180gcc,Using the GNU Compiler Collection (GCC)}, for information on more options 2181offered by the GBE 2182shared by @command{g77} @command{gcc} and other GNU compilers. 2183 2184Some of these do @emph{not} work when compiling programs written in Fortran: 2185 2186@table @gcctabopt 2187@cindex -fpcc-struct-return option 2188@cindex options, -fpcc-struct-return 2189@item -fpcc-struct-return 2190@cindex -freg-struct-return option 2191@cindex options, -freg-struct-return 2192@item -freg-struct-return 2193You should not use these except strictly the same way as you 2194used them to build the version of @code{libg2c} with which 2195you will be linking all code compiled by @command{g77} with the 2196same option. 2197 2198@cindex -fshort-double option 2199@cindex options, -fshort-double 2200@item -fshort-double 2201This probably either has no effect on Fortran programs, or 2202makes them act loopy. 2203 2204@cindex -fno-common option 2205@cindex options, -fno-common 2206@item -fno-common 2207Do not use this when compiling Fortran programs, 2208or there will be Trouble. 2209 2210@cindex -fpack-struct option 2211@cindex options, -fpack-struct 2212@item -fpack-struct 2213This probably will break any calls to the @code{libg2c} library, 2214at the very least, even if it is built with the same option. 2215@end table 2216 2217@c man end 2218 2219@node Environment Variables 2220@section Environment Variables Affecting GNU Fortran 2221@cindex environment variables 2222 2223@c man begin ENVIRONMENT 2224 2225GNU Fortran currently does not make use of any environment 2226variables to control its operation above and beyond those 2227that affect the operation of @command{gcc}. 2228 2229@xref{Environment Variables,,Environment Variables Affecting GCC, 2230gcc,Using the GNU Compiler Collection (GCC)}, for information on environment 2231variables. 2232 2233@c man end 2234