1 \input texinfo @c -*-texinfo-*-
3 @setfilename gfortran.info
4 @set copyrights-gfortran 1999-2006
6 @include gcc-common.texi
8 @settitle The GNU Fortran Compiler
10 @c Create a separate index for command line options
12 @c Merge the standard indexes into a single one.
19 @c TODO: The following "Part" definitions are included here temporarily
20 @c until they are incorporated into the official Texinfo distribution.
21 @c They borrow heavily from Texinfo's \unnchapentry definitions.
28 \vglue\titlepagetopglue
30 \leftline{Part #1:@* #2}
31 \vskip4pt \hrule height 4pt width \hsize \vskip4pt
33 \writetocentry{part}{#2}{#1}
36 \writetocentry{blankpart}{}{}
38 % Part TOC-entry definition for summary contents.
39 \gdef\dosmallpartentry#1#2#3#4{%
40 \vskip .5\baselineskip plus.2\baselineskip
43 \tocentry{Part #2: #1}{\doshortpageno\bgroup#4\egroup}
46 \gdef\dosmallblankpartentry#1#2#3#4{%
47 \vskip .5\baselineskip plus.2\baselineskip
49 % Part TOC-entry definition for regular contents. This has to be
50 % equated to an existing entry to not cause problems when the PDF
52 \gdef\dopartentry#1#2#3#4{%
53 \unnchapentry{Part #2: #1}{}{#3}{#4}
55 \gdef\doblankpartentry#1#2#3#4{}
60 @c Use with @@smallbook.
62 @c %** start of document
64 @c Cause even numbered pages to be printed on the left hand side of
65 @c the page and odd numbered pages to be printed on the right hand
66 @c side of the page. Using this, you can print on both sides of a
67 @c sheet of paper and have the text on the same part of the sheet.
69 @c The text on right hand pages is pushed towards the right hand
70 @c margin and the text on left hand pages is pushed toward the left
72 @c (To provide the reverse effect, set bindingoffset to -0.75in.)
75 @c \global\bindingoffset=0.75in
76 @c \global\normaloffset =0.75in
80 Copyright @copyright{} @value{copyrights-gfortran} Free Software Foundation, Inc.
82 Permission is granted to copy, distribute and/or modify this document
83 under the terms of the GNU Free Documentation License, Version 1.1 or
84 any later version published by the Free Software Foundation; with the
85 Invariant Sections being ``GNU General Public License'' and ``Funding
86 Free Software'', the Front-Cover
87 texts being (a) (see below), and with the Back-Cover Texts being (b)
88 (see below). A copy of the license is included in the section entitled
89 ``GNU Free Documentation License''.
91 (a) The FSF's Front-Cover Text is:
95 (b) The FSF's Back-Cover Text is:
97 You have freedom to copy and modify this GNU Manual, like GNU
98 software. Copies published by the Free Software Foundation raise
99 funds for GNU development.
103 @dircategory Software development
105 * gfortran: (gfortran). The GNU Fortran Compiler.
107 This file documents the use and the internals of
108 the GNU Fortran compiler, (@command{gfortran}).
110 Published by the Free Software Foundation
111 51 Franklin Street, Fifth Floor
112 Boston, MA 02110-1301 USA
118 @setchapternewpage odd
120 @title Using GNU Fortran
122 @center The gfortran team
124 @vskip 0pt plus 1filll
125 For the @value{version-GCC} Version
127 Published by the Free Software Foundation@*
128 51 Franklin Street, Fifth Floor@*
129 Boston, MA 02110-1301, USA@*
130 @c Last printed ??ber, 19??.@*
131 @c Printed copies are available for $? each.@*
137 @c TODO: The following "Part" definitions are included here temporarily
138 @c until they are incorporated into the official Texinfo distribution.
141 \global\let\partentry=\dosmallpartentry
142 \global\let\blankpartentry=\dosmallblankpartentry
147 \global\let\partentry=\dopartentry
148 \global\let\blankpartentry=\doblankpartentry
154 @c ---------------------------------------------------------------------
155 @c TexInfo table of contents.
156 @c ---------------------------------------------------------------------
163 This manual documents the use of @command{gfortran},
164 the GNU Fortran compiler. You can find in this manual how to invoke
165 @command{gfortran}, as well as its features and incompatibilities.
168 @emph{Warning:} This document, and the compiler it describes, are still
169 under development. While efforts are made to keep it up-to-date, it might
170 not accurately reflect the status of the most recent GNU Fortran compiler.
174 @comment When you add a new menu item, please keep the right hand
175 @comment aligned to the same column. Do not use tabs. This provides
176 @comment better formatting.
181 Part I: Invoking GNU Fortran
182 * Invoking GNU Fortran:: Command options supported by @command{gfortran}.
183 * Runtime:: Influencing runtime behavior with environment variables.
185 Part II: Language Reference
186 * Fortran 2003 status:: Fortran 2003 features supported by GNU Fortran.
187 * Extensions:: Language extensions implemented by GNU Fortran.
188 * Intrinsic Procedures:: Intrinsic procedures supported by GNU Fortran.
190 * Contributing:: How you can help.
191 * Copying:: GNU General Public License says
192 how you can copy and share GNU Fortran.
193 * GNU Free Documentation License::
194 How you can copy and share this manual.
195 * Funding:: How to help assure continued work for free software.
196 * Index:: Index of this documentation.
200 @c ---------------------------------------------------------------------
202 @c ---------------------------------------------------------------------
205 @chapter Introduction
207 @c The following duplicates the text on the TexInfo table of contents.
209 This manual documents the use of @command{gfortran}, the GNU Fortran
210 compiler. You can find in this manual how to invoke @command{gfortran},
211 as well as its features and incompatibilities.
214 @emph{Warning:} This document, and the compiler it describes, are still
215 under development. While efforts are made to keep it up-to-date, it
216 might not accurately reflect the status of the most recent GNU Fortran
221 The GNU Fortran compiler front end was
222 designed initially as a free replacement for,
223 or alternative to, the unix @command{f95} command;
224 @command{gfortran} is the command you'll use to invoke the compiler.
227 * About GNU Fortran:: What you should know about the GNU Fortran compiler.
228 * GNU Fortran and GCC:: You can compile Fortran, C, or other programs.
229 * GNU Fortran and G77:: Why we chose to start from scratch.
230 * Project Status:: Status of GNU Fortran, roadmap, proposed extensions.
231 * Standards:: Standards supported by GNU Fortran.
235 @c ---------------------------------------------------------------------
237 @c ---------------------------------------------------------------------
239 @node About GNU Fortran
240 @section About GNU Fortran
242 The GNU Fortran compiler is still in an early state of development.
243 It can generate code for most constructs and expressions,
244 but much work remains to be done.
246 When the GNU Fortran compiler is finished,
247 it will do everything you expect from any decent compiler:
251 Read a user's program,
252 stored in a file and containing instructions written
253 in Fortran 77, Fortran 90, Fortran 95 or Fortran 2003.
254 This file contains @dfn{source code}.
257 Translate the user's program into instructions a computer
258 can carry out more quickly than it takes to translate the
259 instructions in the first
260 place. The result after compilation of a program is
262 code designed to be efficiently translated and processed
263 by a machine such as your computer.
264 Humans usually aren't as good writing machine code
265 as they are at writing Fortran (or C++, Ada, or Java),
266 because is easy to make tiny mistakes writing machine code.
269 Provide the user with information about the reasons why
270 the compiler is unable to create a binary from the source code.
271 Usually this will be the case if the source code is flawed.
272 When writing Fortran, it is easy to make big mistakes.
273 The Fortran 90 requires that the compiler can point out
274 mistakes to the user.
275 An incorrect usage of the language causes an @dfn{error message}.
277 The compiler will also attempt to diagnose cases where the
278 user's program contains a correct usage of the language,
279 but instructs the computer to do something questionable.
280 This kind of diagnostics message is called a @dfn{warning message}.
283 Provide optional information about the translation passes
284 from the source code to machine code.
285 This can help a user of the compiler to find the cause of
286 certain bugs which may not be obvious in the source code,
287 but may be more easily found at a lower level compiler output.
288 It also helps developers to find bugs in the compiler itself.
291 Provide information in the generated machine code that can
292 make it easier to find bugs in the program (using a debugging tool,
293 called a @dfn{debugger}, such as the GNU Debugger @command{gdb}).
296 Locate and gather machine code already generated to
297 perform actions requested by statements in the user's program.
298 This machine code is organized into @dfn{modules} and is located
299 and @dfn{linked} to the user program.
302 The GNU Fortran compiler consists of several components:
306 A version of the @command{gcc} command
307 (which also might be installed as the system's @command{cc} command)
308 that also understands and accepts Fortran source code.
309 The @command{gcc} command is the @dfn{driver} program for
310 all the languages in the GNU Compiler Collection (GCC);
312 you can compile the source code of any language for
313 which a front end is available in GCC.
316 The @command{gfortran} command itself,
317 which also might be installed as the
318 system's @command{f95} command.
319 @command{gfortran} is just another driver program,
320 but specifically for the Fortran compiler only.
321 The difference with @command{gcc} is that @command{gfortran}
322 will automatically link the correct libraries to your program.
325 A collection of run-time libraries.
326 These libraries contain the machine code needed to support
327 capabilities of the Fortran language that are not directly
328 provided by the machine code generated by the
329 @command{gfortran} compilation phase,
330 such as intrinsic functions and subroutines,
331 and routines for interaction with files and the operating system.
332 @c and mechanisms to spawn,
333 @c unleash and pause threads in parallelized code.
336 The Fortran compiler itself, (@command{f951}).
337 This is the GNU Fortran parser and code generator,
338 linked to and interfaced with the GCC backend library.
339 @command{f951} ``translates'' the source code to
340 assembler code. You would typically not use this
342 instead, the @command{gcc} or @command{gfortran} driver
343 programs will call it for you.
347 @c ---------------------------------------------------------------------
348 @c GNU Fortran and GCC
349 @c ---------------------------------------------------------------------
351 @node GNU Fortran and GCC
352 @section GNU Fortran and GCC
353 @cindex GNU Compiler Collection
355 GCC used to be the GNU ``C'' Compiler,
356 but is now known as the @dfn{GNU Compiler Collection}.
357 GCC provides the GNU system with a very versatile
358 compiler middle end (shared optimization passes),
359 and back ends (code generators) for many different
360 computer architectures and operating systems.
361 The code of the middle end and back end are shared by all
362 compiler front ends that are in the GNU Compiler Collection.
364 A GCC front end is essentially a source code parser
365 and an intermediate code generator. The code generator translates the
366 semantics of the source code into a language independent form called
369 The parser takes a source file written in a
370 particular computer language, reads and parses it,
371 and tries to make sure that the source code conforms to
373 Once the correctness of a program has been established,
374 the compiler will build a data structure known as the
375 @dfn{Abstract Syntax tree},
376 or just @dfn{AST} or ``tree'' for short.
377 This data structure represents the whole program
378 or a subroutine or a function.
379 The ``tree'' is passed to the GCC middle end,
380 which will perform optimization passes on it. The optimized AST is then
381 handed off too the back end which assembles the program unit.
383 Different phases in this translation process can be,
384 and in fact @emph{are} merged in many compiler front ends.
385 GNU Fortran has a strict separation between the
386 parser and code generator.
388 The goal of the GNU Fortran project is to build a new front end for GCC.
389 Specifically, a Fortran 95 front end.
390 In a non-@command{gfortran} installation,
391 @command{gcc} will not be able to compile Fortran source code
392 (only the ``C'' front end has to be compiled if you want to build GCC,
393 all other languages are optional).
394 If you build GCC with @command{gfortran}, @command{gcc} will recognize
395 @file{.f/.f90/.f95} source files and accepts Fortran specific
396 command line options.
399 @c ---------------------------------------------------------------------
400 @c GNU Fortran and G77
401 @c ---------------------------------------------------------------------
403 @node GNU Fortran and G77
404 @section GNU Fortran and G77
408 Why do we write a compiler front end from scratch?
409 There's a fine Fortran 77 compiler in the
410 GNU Compiler Collection that accepts some features
411 of the Fortran 90 standard as extensions.
412 Why not start from there and revamp it?
414 One of the reasons is that Craig Burley, the author of G77,
415 has decided to stop working on the G77 front end.
416 On @uref{http://world.std.com/~burley/g77-why.html,
417 Craig explains the reasons for his decision to stop working on G77}
418 in one of the pages in his homepage.
419 Among the reasons is a lack of interest in improvements to
421 Users appear to be quite satisfied with @command{g77} as it is.
422 While @command{g77} is still being maintained (by Toon Moene),
423 it is unlikely that sufficient people will be willing
424 to completely rewrite the existing code.
426 But there are other reasons to start from scratch.
427 Many people, including Craig Burley,
428 no longer agreed with certain design decisions in the G77 front end.
429 Also, the interface of @command{g77} to the back end is written in
430 a style which is confusing and not up to date on recommended practice.
431 In fact, a full rewrite had already been planned for GCC 3.0.
433 When Craig decided to stop,
434 it just seemed to be a better idea to start a new project from scratch,
435 because it was expected to be easier to maintain code we
436 develop ourselves than to do a major overhaul of @command{g77} first,
437 and then build a Fortran 95 compiler out of it.
440 @c ---------------------------------------------------------------------
442 @c ---------------------------------------------------------------------
445 @section Project Status
448 As soon as @command{gfortran} can parse all of the statements correctly,
449 it will be in the ``larva'' state.
450 When we generate code, the ``puppa'' state.
451 When @command{gfortran} is done,
452 we'll see if it will be a beautiful butterfly,
453 or just a big bug....
455 --Andy Vaught, April 2000
458 The start of the GNU Fortran 95 project was announced on
459 the GCC homepage in March 18, 2000
460 (even though Andy had already been working on it for a while,
463 The GNU Fortran compiler is able to compile nearly all
464 standard-compliant Fortran 95, Fortran 90, and Fortran 77 programs,
465 including a number of standard and non-standard extensions, and can be
466 used on real-world programs. In particular, the supported extensions
467 include OpenMP, Cray-style pointers, and several Fortran 2003 features
468 such as enumeration, stream I/O, and some of the enhancements to
469 allocatable array support from TR 15581. However, it is still under
470 development and has a few remaining rough edges.
472 At present, the GNU Fortran compiler passes the
473 @uref{http://www.fortran-2000.com/ArnaudRecipes/fcvs21_f95.html,
474 NIST Fortran 77 Test Suite}, and produces acceptable results on the
475 @uref{http://www.netlib.org/lapack/faq.html#1.21, LAPACK Test Suite}.
476 It also provides respectable performance on
477 the @uref{http://www.polyhedron.com/pb05.html, Polyhedron Fortran
478 compiler benchmarks} and the
479 @uref{http://www.llnl.gov/asci_benchmarks/asci/limited/lfk/README.html,
480 Livermore Fortran Kernels test}. It has been used to compile a number of
481 large real-world programs, including
482 @uref{http://mysite.verizon.net/serveall/moene.pdf, the HIRLAM
483 weather-forecasting code} and
484 @uref{http://www.theochem.uwa.edu.au/tonto/, the Tonto quantum
485 chemistry package}; see @url{http://gcc.gnu.org/wiki/GfortranApps} for an
488 Among other things, the GNU Fortran compiler is intended as a replacement
489 for G77. At this point, nearly all programs that could be compiled with
490 G77 can be compiled with GNU Fortran, although there are a few minor known
493 The primary work remaining to be done on GNU Fortran falls into three
494 categories: bug fixing (primarily regarding the treatment of invalid code
495 and providing useful error messages), improving the compiler optimizations
496 and the performance of compiled code, and extending the compiler to support
497 future standards---in particular, Fortran 2003.
500 @c ---------------------------------------------------------------------
502 @c ---------------------------------------------------------------------
508 The GNU Fortran compiler implements
509 ISO/IEC 1539:1997 (Fortran 95). As such, it can also compile essentially all
510 standard-compliant Fortran 90 and Fortran 77 programs. It also supports
511 the ISO/IEC TR-15581 enhancements to allocatable arrays, and
512 the @uref{http://www.openmp.org/drupal/mp-documents/spec25.pdf,
513 OpenMP Application Program Interface v2.5} specification.
515 In the future, the GNU Fortran compiler may also support other standard
516 variants of and extensions to the Fortran language. These include
517 ISO/IEC 1539-1:2004 (Fortran 2003).
520 @c =====================================================================
521 @c PART I: INVOCATION REFERENCE
522 @c =====================================================================
525 \part{I}{Invoking GNU Fortran}
528 @c ---------------------------------------------------------------------
530 @c ---------------------------------------------------------------------
535 @c ---------------------------------------------------------------------
537 @c ---------------------------------------------------------------------
540 @chapter Runtime: Influencing runtime behavior with environment variables
543 The behavior of the @command{gfortran} can be influenced by
544 environment variables.
546 Malformed environment variables are silently ignored.
549 * GFORTRAN_STDIN_UNIT:: Unit number for standard input
550 * GFORTRAN_STDOUT_UNIT:: Unit number for standard output
551 * GFORTRAN_STDERR_UNIT:: Unit number for standard error
552 * GFORTRAN_USE_STDERR:: Send library output to standard error
553 * GFORTRAN_TMPDIR:: Directory for scratch files
554 * GFORTRAN_UNBUFFERED_ALL:: Don't buffer output
555 * GFORTRAN_SHOW_LOCUS:: Show location for runtime errors
556 * GFORTRAN_OPTIONAL_PLUS:: Print leading + where permitted
557 * GFORTRAN_DEFAULT_RECL:: Default record length for new files
558 * GFORTRAN_LIST_SEPARATOR:: Separator for list output
559 * GFORTRAN_CONVERT_UNIT:: Set endianness for unformatted I/O
562 @node GFORTRAN_STDIN_UNIT
563 @section @env{GFORTRAN_STDIN_UNIT}---Unit number for standard input
565 This environment variable can be used to select the unit number
566 preconnected to standard input. This must be a positive integer.
567 The default value is 5.
569 @node GFORTRAN_STDOUT_UNIT
570 @section @env{GFORTRAN_STDOUT_UNIT}---Unit number for standard output
572 This environment variable can be used to select the unit number
573 preconnected to standard output. This must be a positive integer.
574 The default value is 6.
576 @node GFORTRAN_STDERR_UNIT
577 @section @env{GFORTRAN_STDERR_UNIT}---Unit number for standard error
579 This environment variable can be used to select the unit number
580 preconnected to standard error. This must be a positive integer.
581 The default value is 0.
583 @node GFORTRAN_USE_STDERR
584 @section @env{GFORTRAN_USE_STDERR}---Send library output to standard error
586 This environment variable controls where library output is sent.
587 If the first letter is @samp{y}, @samp{Y} or @samp{1}, standard
588 error is used. If the first letter is @samp{n}, @samp{N} or
589 @samp{0}, standard output is used.
591 @node GFORTRAN_TMPDIR
592 @section @env{GFORTRAN_TMPDIR}---Directory for scratch files
594 This environment variable controls where scratch files are
595 created. If this environment variable is missing,
596 GNU Fortran searches for the environment variable @env{TMP}. If
597 this is also missing, the default is @file{/tmp}.
599 @node GFORTRAN_UNBUFFERED_ALL
600 @section @env{GFORTRAN_UNBUFFERED_ALL}---Don't buffer output
602 This environment variable controls whether all output is unbuffered.
603 If the first letter is @samp{y}, @samp{Y} or @samp{1}, all output is
604 unbuffered. This will slow down large writes. If the first letter is
605 @samp{n}, @samp{N} or @samp{0}, output is buffered. This is the
608 @node GFORTRAN_SHOW_LOCUS
609 @section @env{GFORTRAN_SHOW_LOCUS}---Show location for runtime errors
611 If the first letter is @samp{y}, @samp{Y} or @samp{1}, filename and
612 line numbers for runtime errors are printed. If the first letter is
613 @samp{n}, @samp{N} or @samp{0}, don't print filename and line numbers
614 for runtime errors. The default is to print the location.
616 @node GFORTRAN_OPTIONAL_PLUS
617 @section @env{GFORTRAN_OPTIONAL_PLUS}---Print leading + where permitted
619 If the first letter is @samp{y}, @samp{Y} or @samp{1},
620 a plus sign is printed
621 where permitted by the Fortran standard. If the first letter
622 is @samp{n}, @samp{N} or @samp{0}, a plus sign is not printed
623 in most cases. Default is not to print plus signs.
625 @node GFORTRAN_DEFAULT_RECL
626 @section @env{GFORTRAN_DEFAULT_RECL}---Default record length for new files
628 This environment variable specifies the default record length, in
629 bytes, for files which are opened without a @code{RECL} tag in the
630 @code{OPEN} statement. This must be a positive integer. The
631 default value is 1073741824 bytes (1 GB).
633 @node GFORTRAN_LIST_SEPARATOR
634 @section @env{GFORTRAN_LIST_SEPARATOR}---Separator for list output
636 This environment variable specifies the separator when writing
637 list-directed output. It may contain any number of spaces and
638 at most one comma. If you specify this on the command line,
639 be sure to quote spaces, as in
641 $ GFORTRAN_LIST_SEPARATOR=' , ' ./a.out
643 when @command{a.out} is the compiled Fortran program that you want to run.
644 Default is a single space.
646 @node GFORTRAN_CONVERT_UNIT
647 @section @env{GFORTRAN_CONVERT_UNIT}---Set endianness for unformatted I/O
649 By setting the @env{GFORTRAN_CONVERT_UNIT} variable, it is possible
650 to change the representation of data for unformatted files.
651 The syntax for the @env{GFORTRAN_CONVERT_UNIT} variable is:
653 GFORTRAN_CONVERT_UNIT: mode | mode ';' exception ;
654 mode: 'native' | 'swap' | 'big_endian' | 'little_endian' ;
655 exception: mode ':' unit_list | unit_list ;
656 unit_list: unit_spec | unit_list unit_spec ;
657 unit_spec: INTEGER | INTEGER '-' INTEGER ;
659 The variable consists of an optional default mode, followed by
660 a list of optional exceptions, which are separated by semicolons
661 from the preceding default and each other. Each exception consists
662 of a format and a comma-separated list of units. Valid values for
663 the modes are the same as for the @code{CONVERT} specifier:
666 @item @code{NATIVE} Use the native format. This is the default.
667 @item @code{SWAP} Swap between little- and big-endian.
668 @item @code{LITTLE_ENDIAN} Use the little-endian format
669 for unformatted files.
670 @item @code{BIG_ENDIAN} Use the big-endian format for unformatted files.
672 A missing mode for an exception is taken to mean @code{BIG_ENDIAN}.
673 Examples of values for @env{GFORTRAN_CONVERT_UNIT} are:
675 @item @code{'big_endian'} Do all unformatted I/O in big_endian mode.
676 @item @code{'little_endian;native:10-20,25'} Do all unformatted I/O
677 in little_endian mode, except for units 10 to 20 and 25, which are in
679 @item @code{'10-20'} Units 10 to 20 are big-endian, the rest is native.
682 Setting the environment variables should be done on the command
683 line or via the @command{export}
684 command for @command{sh}-compatible shells and via @command{setenv}
685 for @command{csh}-compatible shells.
687 Example for @command{sh}:
690 $ GFORTRAN_CONVERT_UNIT='big_endian;native:10-20' ./a.out
693 Example code for @command{csh}:
696 % setenv GFORTRAN_CONVERT_UNIT 'big_endian;native:10-20'
700 Using anything but the native representation for unformatted data
701 carries a significant speed overhead. If speed in this area matters
702 to you, it is best if you use this only for data that needs to be
705 @xref{CONVERT specifier}, for an alternative way to specify the
706 data representation for unformatted files. @xref{Runtime Options}, for
707 setting a default data representation for the whole program. The
708 @code{CONVERT} specifier overrides the @option{-fconvert} compile options.
711 @c =====================================================================
712 @c PART II: LANGUAGE REFERENCE
713 @c =====================================================================
716 \part{II}{Language Reference}
719 @c ---------------------------------------------------------------------
720 @c Fortran 2003 Status
721 @c ---------------------------------------------------------------------
723 @node Fortran 2003 status
724 @chapter Fortran 2003 Status
726 Although GNU Fortran focuses on implementing the Fortran 95
727 standard for the time being, a few Fortran 2003 features are currently
732 Intrinsics @code{command_argument_count}, @code{get_command},
733 @code{get_command_argument}, @code{get_environment_variable}, and
737 @cindex Array constructors
739 Array constructors using square brackets. That is, @code{[...]} rather
743 @cindex @code{FLUSH} statement
744 @code{FLUSH} statement.
747 @cindex @code{IOMSG=} specifier
748 @code{IOMSG=} specifier for I/O statements.
751 @cindex @code{ENUM} statement
752 @cindex @code{ENUMERATOR} statement
753 @cindex @code{-fshort-enums} option
754 Support for the declaration of enumeration constants via the
755 @code{ENUM} and @code{ENUMERATOR} statements. Interoperability with
756 @command{gcc} is guaranteed also for the case where the
757 @command{-fshort-enums} command line option is given.
764 @cindex @code{ALLOCATABLE} dummy arguments
765 @code{ALLOCATABLE} dummy arguments.
767 @cindex @code{ALLOCATABLE} function results
768 @code{ALLOCATABLE} function results
770 @cindex @code{ALLOCATABLE} components of derived types
771 @code{ALLOCATABLE} components of derived types
775 @cindex @code{STREAM} I/O
776 @cindex @code{ACCESS='STREAM'} I/O
777 The @code{OPEN} statement supports the @code{ACCESS='STREAM'} specifier,
778 allowing I/O without any record structure.
781 Namelist input/output for internal files.
784 @cindex @code{PROTECTED}
785 The @code{PROTECTED} statement and attribute.
789 The @code{VALUE} statement and attribute.
792 @cindex @code{VOLATILE}
793 The @code{VOLATILE} statement and attribute.
796 @cindex @code{IMPORT}
797 The @code{IMPORT} statement, allowing to import
798 host-associated derived types.
801 @cindex @code{USE, INTRINSIC}
802 @cindex @code{ISO_FORTRAN_ENV}
803 @code{USE} statement with @code{INTRINSIC} and @code{NON_INTRINSIC}
804 attribute; supported intrinsic modules: @code{ISO_FORTRAN_ENV},
805 @code{OMP_LIB} and @code{OMP_LIB_KINDS}.
810 @c ---------------------------------------------------------------------
812 @c ---------------------------------------------------------------------
814 @c Maybe this chapter should be merged with the 'Standards' section,
815 @c whenever that is written :-)
821 GNU Fortran implements a number of extensions over standard
822 Fortran. This chapter contains information on their syntax and
823 meaning. There are currently two categories of GNU Fortran
824 extensions, those that provide functionality beyond that provided
825 by any standard, and those that are supported by GNU Fortran
826 purely for backward compatibility with legacy compilers. By default,
827 @option{-std=gnu} allows the compiler to accept both types of
828 extensions, but to warn about the use of the latter. Specifying
829 either @option{-std=f95} or @option{-std=f2003} disables both types
830 of extensions, and @option{-std=legacy} allows both without warning.
833 * Old-style kind specifications::
834 * Old-style variable initialization::
835 * Extensions to namelist::
836 * X format descriptor without count field::
837 * Commas in FORMAT specifications::
838 * Missing period in FORMAT specifications::
840 * BOZ literal constants::
841 * Real array indices::
843 * Implicitly convert LOGICAL and INTEGER values::
844 * Hollerith constants support::
846 * CONVERT specifier::
850 @node Old-style kind specifications
851 @section Old-style kind specifications
852 @cindex Kind specifications
854 GNU Fortran allows old-style kind specifications in
855 declarations. These look like:
859 where @code{TYPESPEC} is a basic type (@code{INTEGER}, @code{REAL},
860 etc.), and where @code{k} is a valid kind number for that type. The
861 statement then declares @code{x}, @code{y} and @code{z} to be of
862 type @code{TYPESPEC} with kind @code{k}. This is equivalent to the
863 standard conforming declaration
868 @node Old-style variable initialization
869 @section Old-style variable initialization
870 @cindex Initialization
872 GNU Fortran allows old-style initialization of variables of the
876 REAL x(2,2) /3*0.,1./
878 The syntax for the initializers is as for the @code{DATA} statement, but
879 unlike in a @code{DATA} statement, an initializer only applies to the
880 variable immediately preceding the initialization. In other words,
881 something like @code{INTEGER I,J/2,3/} is not valid. This style of
882 initialization is only allowed in declarations without double colons
883 (@code{::}); the double colons were introduced in Fortran 90, which also
884 introduced a standard syntax for initializating variables in type
887 Examples of standard-conforming code equivalent to the above example
891 INTEGER :: i = 1, j = 2
892 REAL :: x(2,2) = RESHAPE((/0.,0.,0.,1./),SHAPE(x))
896 DATA i/1/, j/2/, x/3*0.,1./
899 Note that variables which are explicitly initialized in declarations
900 or in @code{DATA} statements automatically acquire the @code{SAVE}
903 @node Extensions to namelist
904 @section Extensions to namelist
907 GNU Fortran fully supports the Fortran 95 standard for namelist I/O
908 including array qualifiers, substrings and fully qualified derived types.
909 The output from a namelist write is compatible with namelist read. The
910 output has all names in upper case and indentation to column 1 after the
911 namelist name. Two extensions are permitted:
913 Old-style use of @samp{$} instead of @samp{&}
916 X(:)%Y(2) = 1.0 2.0 3.0
921 It should be noted that the default terminator is @samp{/} rather than
924 Querying of the namelist when inputting from stdin. After at least
925 one space, entering @samp{?} sends to stdout the namelist name and the names of
926 the variables in the namelist:
937 Entering @samp{=?} outputs the namelist to stdout, as if
938 @code{WRITE(*,NML = mynml)} had been called:
943 X(1)%Y= 0.000000 , 1.000000 , 0.000000 ,
944 X(2)%Y= 0.000000 , 2.000000 , 0.000000 ,
945 X(3)%Y= 0.000000 , 3.000000 , 0.000000 ,
949 To aid this dialog, when input is from stdin, errors send their
950 messages to stderr and execution continues, even if @code{IOSTAT} is set.
952 @code{PRINT} namelist is permitted. This causes an error if
953 @option{-std=f95} is used.
956 REAL, dimension (4) :: x = (/1.0, 2.0, 3.0, 4.0/)
959 END PROGRAM test_print
962 Expanded namelist reads are permitted. This causes an error if
963 @option{-std=f95} is used. In the following example, the first element
964 of the array will be given the value 0.00 and the two succeeding
965 elements will be given the values 1.00 and 2.00.
968 X(1,1) = 0.00 , 1.00 , 2.00
972 @node X format descriptor without count field
973 @section @code{X} format descriptor without count field
974 @cindex @code{X} format descriptor without count field
976 To support legacy codes, GNU Fortran permits the count field of the
977 @code{X} edit descriptor in @code{FORMAT} statements to be omitted.
978 When omitted, the count is implicitly assumed to be one.
982 10 FORMAT (I1, X, I1)
985 @node Commas in FORMAT specifications
986 @section Commas in @code{FORMAT} specifications
987 @cindex Commas in @code{FORMAT} specifications
989 To support legacy codes, GNU Fortran allows the comma separator
990 to be omitted immediately before and after character string edit
991 descriptors in @code{FORMAT} statements.
995 10 FORMAT ('FOO='I1' BAR='I2)
999 @node Missing period in FORMAT specifications
1000 @section Missing period in @code{FORMAT} specifications
1001 @cindex Missing period in @code{FORMAT} specifications
1003 To support legacy codes, GNU Fortran allows missing periods in format
1004 specifications if and only if @option{-std=legacy} is given on the
1005 command line. This is considered non-conforming code and is
1014 @node I/O item lists
1015 @section I/O item lists
1016 @cindex I/O item lists
1018 To support legacy codes, GNU Fortran allows the input item list
1019 of the @code{READ} statement, and the output item lists of the
1020 @code{WRITE} and @code{PRINT} statements, to start with a comma.
1022 @node BOZ literal constants
1023 @section BOZ literal constants
1024 @cindex BOZ literal constants
1026 As an extension, GNU Fortran allows hexadecimal BOZ literal constants to
1027 be specified using the X prefix, in addition to the standard Z prefix.
1028 BOZ literal constants can also be specified by adding a suffix to the
1029 string. For example, @code{Z'ABC'} and @code{'ABC'Z} are equivalent.
1031 The Fortran standard restricts the appearance of a BOZ literal constant
1032 to the @code{DATA} statement, and it is expected to be assigned to an
1033 @code{INTEGER} variable. GNU Fortran permits a BOZ literal to appear in
1034 any initialization expression as well as assignment statements.
1036 Attempts to use a BOZ literal constant to do a bitwise initialization of
1037 a variable can lead to confusion. A BOZ literal constant is converted
1038 to an @code{INTEGER} value with the kind type with the largest decimal
1039 representation, and this value is then converted numerically to the type
1040 and kind of the variable in question. Thus, one should not expect a
1041 bitwise copy of the BOZ literal constant to be assigned to a @code{REAL}
1044 Similarly, initializing an @code{INTEGER} variable with a statement such
1045 as @code{DATA i/Z'FFFFFFFF'/} will produce an integer overflow rather
1046 than the desired result of @math{-1} when @code{i} is a 32-bit integer
1047 on a system that supports 64-bit integers. The @samp{-fno-range-check}
1048 option can be used as a workaround for legacy code that initializes
1049 integers in this manner.
1051 @node Real array indices
1052 @section Real array indices
1053 @cindex Real array indices
1055 As an extension, GNU Fortran allows the use of @code{REAL} expressions
1056 or variables as array indices.
1058 @node Unary operators
1059 @section Unary operators
1060 @cindex Unary operators
1062 As an extension, GNU Fortran allows unary plus and unary minus operators
1063 to appear as the second operand of binary arithmetic operators without
1064 the need for parenthesis.
1070 @node Implicitly convert LOGICAL and INTEGER values
1071 @section Implicitly convert @code{LOGICAL} and @code{INTEGER} values
1072 @cindex Implicitly convert @code{LOGICAL} and @code{INTEGER} values
1074 As an extension for backwards compatibility with other compilers, GNU
1075 Fortran allows the implicit conversion of @code{LOGICAL} values to
1076 @code{INTEGER} values and vice versa. When converting from a
1077 @code{LOGICAL} to an @code{INTEGER}, @code{.FALSE.} is interpreted as
1078 zero, and @code{.TRUE.} is interpreted as one. When converting from
1079 @code{INTEGER} to @code{LOGICAL}, the value zero is interpreted as
1080 @code{.FALSE.} and any nonzero value is interpreted as @code{.TRUE.}.
1084 IF (i) PRINT *, 'True'
1087 @node Hollerith constants support
1088 @section Hollerith constants support
1089 @cindex Hollerith constants
1091 GNU Fortran supports Hollerith constants in assignments, function
1092 arguments, and @code{DATA} and @code{ASSIGN} statements. A Hollerith
1093 constant is written as a string of characters preceeded by an integer
1094 constant indicating the character count, and the letter @code{H} or
1095 @code{h}, and stored in bytewise fashion in a numeric (@code{INTEGER},
1096 @code{REAL}, or @code{complex}) or @code{LOGICAL} variable. The
1097 constant will be padded or truncated to fit the size of the variable in
1100 Examples of valid uses of Hollerith constants:
1103 data x /16Habcdefghijklmnop, 16Hqrstuvwxyz012345/
1104 x(1) = 16HABCDEFGHIJKLMNOP
1108 Invalid Hollerith constants examples:
1111 a = 8H12345678 ! Valid, but the Hollerith constant will be truncated.
1112 a = 0H ! At least one character is needed.
1115 In general, Hollerith constants were used to provide a rudimentary
1116 facility for handling character strings in early Fortran compilers,
1117 prior to the introduction of @code{CHARACTER} variables in Fortran 77;
1118 in those cases, the standard-compliant equivalent is to convert the
1119 program to use proper character strings. On occasion, there may be a
1120 case where the intent is specifically to initialize a numeric variable
1121 with a given byte sequence. In these cases, the same result can be
1122 obtained by using the @code{TRANSFER} statement, as in this example.
1124 INTEGER(KIND=4) :: a
1125 a = TRANSFER ("abcd", a) ! equivalent to: a = 4Habcd
1130 @section Cray pointers
1131 @cindex Cray pointers
1133 Cray pointers are part of a non-standard extension that provides a
1134 C-like pointer in Fortran. This is accomplished through a pair of
1135 variables: an integer "pointer" that holds a memory address, and a
1136 "pointee" that is used to dereference the pointer.
1138 Pointer/pointee pairs are declared in statements of the form:
1140 pointer ( <pointer> , <pointee> )
1144 pointer ( <pointer1> , <pointee1> ), ( <pointer2> , <pointee2> ), ...
1146 The pointer is an integer that is intended to hold a memory address.
1147 The pointee may be an array or scalar. A pointee can be an assumed
1148 size array---that is, the last dimension may be left unspecified by
1149 using a @code{*} in place of a value---but a pointee cannot be an
1150 assumed shape array. No space is allocated for the pointee.
1152 The pointee may have its type declared before or after the pointer
1153 statement, and its array specification (if any) may be declared
1154 before, during, or after the pointer statement. The pointer may be
1155 declared as an integer prior to the pointer statement. However, some
1156 machines have default integer sizes that are different than the size
1157 of a pointer, and so the following code is not portable:
1162 If a pointer is declared with a kind that is too small, the compiler
1163 will issue a warning; the resulting binary will probably not work
1164 correctly, because the memory addresses stored in the pointers may be
1165 truncated. It is safer to omit the first line of the above example;
1166 if explicit declaration of ipt's type is omitted, then the compiler
1167 will ensure that ipt is an integer variable large enough to hold a
1170 Pointer arithmetic is valid with Cray pointers, but it is not the same
1171 as C pointer arithmetic. Cray pointers are just ordinary integers, so
1172 the user is responsible for determining how many bytes to add to a
1173 pointer in order to increment it. Consider the following example:
1177 pointer (ipt, pointee)
1181 The last statement does not set @code{ipt} to the address of
1182 @code{target(1)}, as it would in C pointer arithmetic. Adding @code{1}
1183 to @code{ipt} just adds one byte to the address stored in @code{ipt}.
1185 Any expression involving the pointee will be translated to use the
1186 value stored in the pointer as the base address.
1188 To get the address of elements, this extension provides an intrinsic
1189 function @code{LOC()}. The @code{LOC()} function is equivalent to the
1190 @code{&} operator in C, except the address is cast to an integer type:
1193 pointer(ipt, arpte(10))
1195 ipt = loc(ar) ! Makes arpte is an alias for ar
1196 arpte(1) = 1.0 ! Sets ar(1) to 1.0
1198 The pointer can also be set by a call to the @code{MALLOC} intrinsic
1201 Cray pointees often are used to alias an existing variable. For
1209 As long as @code{ipt} remains unchanged, @code{iarr} is now an alias for
1210 @code{target}. The optimizer, however, will not detect this aliasing, so
1211 it is unsafe to use @code{iarr} and @code{target} simultaneously. Using
1212 a pointee in any way that violates the Fortran aliasing rules or
1213 assumptions is illegal. It is the user's responsibility to avoid doing
1214 this; the compiler works under the assumption that no such aliasing
1217 Cray pointers will work correctly when there is no aliasing (i.e., when
1218 they are used to access a dynamically allocated block of memory), and
1219 also in any routine where a pointee is used, but any variable with which
1220 it shares storage is not used. Code that violates these rules may not
1221 run as the user intends. This is not a bug in the optimizer; any code
1222 that violates the aliasing rules is illegal. (Note that this is not
1223 unique to GNU Fortran; any Fortran compiler that supports Cray pointers
1224 will ``incorrectly'' optimize code with illegal aliasing.)
1226 There are a number of restrictions on the attributes that can be applied
1227 to Cray pointers and pointees. Pointees may not have the
1228 @code{ALLOCATABLE}, @code{INTENT}, @code{OPTIONAL}, @code{DUMMY},
1229 @code{TARGET}, @code{INTRINSIC}, or @code{POINTER} attributes. Pointers
1230 may not have the @code{DIMENSION}, @code{POINTER}, @code{TARGET},
1231 @code{ALLOCATABLE}, @code{EXTERNAL}, or @code{INTRINSIC} attributes.
1232 Pointees may not occur in more than one pointer statement. A pointee
1233 cannot be a pointer. Pointees cannot occur in equivalence, common, or
1236 A Cray pointer may also point to a function or a subroutine. For
1237 example, the following excerpt is valid:
1241 pointer (subptr,subpte)
1251 A pointer may be modified during the course of a program, and this
1252 will change the location to which the pointee refers. However, when
1253 pointees are passed as arguments, they are treated as ordinary
1254 variables in the invoked function. Subsequent changes to the pointer
1255 will not change the base address of the array that was passed.
1257 @node CONVERT specifier
1258 @section CONVERT specifier
1259 @cindex CONVERT specifier
1261 GNU Fortran allows the conversion of unformatted data between little-
1262 and big-endian representation to facilitate moving of data
1263 between different systems. The conversion can be indicated with
1264 the @code{CONVERT} specifier on the @code{OPEN} statement.
1265 @xref{GFORTRAN_CONVERT_UNIT}, for an alternative way of specifying
1266 the data format via an environment variable.
1268 Valid values for @code{CONVERT} are:
1270 @item @code{CONVERT='NATIVE'} Use the native format. This is the default.
1271 @item @code{CONVERT='SWAP'} Swap between little- and big-endian.
1272 @item @code{CONVERT='LITTLE_ENDIAN'} Use the little-endian representation
1273 for unformatted files.
1274 @item @code{CONVERT='BIG_ENDIAN'} Use the big-endian representation for
1278 Using the option could look like this:
1280 open(file='big.dat',form='unformatted',access='sequential', &
1281 convert='big_endian')
1284 The value of the conversion can be queried by using
1285 @code{INQUIRE(CONVERT=ch)}. The values returned are
1286 @code{'BIG_ENDIAN'} and @code{'LITTLE_ENDIAN'}.
1288 @code{CONVERT} works between big- and little-endian for
1289 @code{INTEGER} values of all supported kinds and for @code{REAL}
1290 on IEEE systems of kinds 4 and 8. Conversion between different
1291 ``extended double'' types on different architectures such as
1292 m68k and x86_64, which GNU Fortran
1293 supports as @code{REAL(KIND=10)} and @code{REAL(KIND=16)}, will
1296 @emph{Note that the values specified via the GFORTRAN_CONVERT_UNIT
1297 environment variable will override the CONVERT specifier in the
1298 open statement}. This is to give control over data formats to
1299 users who do not have the source code of their program available.
1301 Using anything but the native representation for unformatted data
1302 carries a significant speed overhead. If speed in this area matters
1303 to you, it is best if you use this only for data that needs to be
1310 GNU Fortran attempts to be OpenMP Application Program Interface v2.5
1311 compatible when invoked with the @option{-fopenmp} option. GNU Fortran
1312 then generates parallelized code according to the OpenMP directives
1313 used in the source. The OpenMP Fortran runtime library
1314 routines are provided both in a form of a Fortran 90 module named
1315 @code{omp_lib} and in a form of a Fortran @code{include} file named
1318 For details refer to the actual
1319 @uref{http://www.openmp.org/drupal/mp-documents/spec25.pdf,
1320 OpenMP Application Program Interface v2.5} specification.
1322 @c ---------------------------------------------------------------------
1323 @c Intrinsic Procedures
1324 @c ---------------------------------------------------------------------
1326 @include intrinsic.texi
1333 @c ---------------------------------------------------------------------
1335 @c ---------------------------------------------------------------------
1338 @unnumbered Contributing
1339 @cindex Contributing
1341 Free software is only possible if people contribute to efforts
1343 We're always in need of more people helping out with ideas
1344 and comments, writing documentation and contributing code.
1346 If you want to contribute to GNU Fortran,
1347 have a look at the long lists of projects you can take on.
1348 Some of these projects are small,
1349 some of them are large;
1350 some are completely orthogonal to the rest of what is
1351 happening on GNU Fortran,
1352 but others are ``mainstream'' projects in need of enthusiastic hackers.
1353 All of these projects are important!
1354 We'll eventually get around to the things here,
1355 but they are also things doable by someone who is willing and able.
1360 * Proposed Extensions::
1365 @section Contributors to GNU Fortran
1366 @cindex Contributors
1370 Most of the parser was hand-crafted by @emph{Andy Vaught}, who is
1371 also the initiator of the whole project. Thanks Andy!
1372 Most of the interface with GCC was written by @emph{Paul Brook}.
1374 The following individuals have contributed code and/or
1375 ideas and significant help to the GNU Fortran project
1376 (in no particular order):
1380 @item Katherine Holcomb
1381 @item Tobias Schl@"uter
1382 @item Steven Bosscher
1385 @item Niels Kristian Bech Jensen
1386 @item Steven Johnson
1391 @item Fran@,{c}ois-Xavier Coudert
1392 @item Steven G. Kargl
1394 @item Janne Blomqvist
1401 @item Richard Henderson
1402 @item Richard Sandiford
1403 @item Richard Guenther
1404 @item Bernhard Fischer
1407 The following people have contributed bug reports,
1408 smaller or larger patches,
1409 and much needed feedback and encouragement for the
1410 GNU Fortran project:
1413 @item Erik Schnetter
1418 Many other individuals have helped debug,
1419 test and improve the GNU Fortran compiler over the past few years,
1420 and we welcome you to do the same!
1421 If you already have done so,
1422 and you would like to see your name listed in the
1423 list above, please contact us.
1431 @item Help build the test suite
1432 Solicit more code for donation to the test suite.
1433 We can keep code private on request.
1435 @item Bug hunting/squishing
1436 Find bugs and write more test cases!
1437 Test cases are especially very welcome,
1438 because it allows us to concentrate on fixing bugs
1439 instead of isolating them.
1441 @item Smaller projects (``bug'' fixes):
1443 @item Allow init exprs to be numbers raised to integer powers.
1444 @item Implement correct rounding.
1445 @item Implement F restrictions on Fortran 95 syntax.
1446 @item See about making Emacs-parsable error messages.
1450 If you wish to work on the runtime libraries,
1451 please contact a project maintainer.
1455 @node Proposed Extensions
1456 @section Proposed Extensions
1458 Here's a list of proposed extensions for the GNU Fortran compiler, in no particular
1459 order. Most of these are necessary to be fully compatible with
1460 existing Fortran compilers, but they are not part of the official
1461 J3 Fortran 95 standard.
1463 @subsection Compiler extensions:
1466 User-specified alignment rules for structures.
1469 Flag to generate @code{Makefile} info.
1472 Automatically extend single precision constants to double.
1475 Compile code that conserves memory by dynamically allocating common and
1476 module storage either on stack or heap.
1479 Compile flag to generate code for array conformance checking (suggest -CC).
1482 User control of symbol names (underscores, etc).
1485 Compile setting for maximum size of stack frame size before spilling
1486 parts to static or heap.
1489 Flag to force local variables into static space.
1492 Flag to force local variables onto stack.
1495 Flag for maximum errors before ending compile.
1498 Option to initialize otherwise uninitialized integer and floating
1503 @subsection Environment Options
1506 Pluggable library modules for random numbers, linear algebra.
1507 LA should use BLAS calling conventions.
1510 Environment variables controlling actions on arithmetic exceptions like
1511 overflow, underflow, precision loss---Generate NaN, abort, default.
1515 Set precision for fp units that support it (i387).
1518 Variable for setting fp rounding mode.
1521 Variable to fill uninitialized variables with a user-defined bit
1525 Environment variable controlling filename that is opened for that unit
1529 Environment variable to clear/trash memory being freed.
1532 Environment variable to control tracing of allocations and frees.
1535 Environment variable to display allocated memory at normal program end.
1538 Environment variable for filename for * IO-unit.
1541 Environment variable for temporary file directory.
1544 Environment variable forcing standard output to be line buffered (unix).
1549 @c ---------------------------------------------------------------------
1550 @c GNU General Public License
1551 @c ---------------------------------------------------------------------
1557 @c ---------------------------------------------------------------------
1558 @c GNU Free Documentation License
1559 @c ---------------------------------------------------------------------
1565 @c ---------------------------------------------------------------------
1566 @c Funding Free Software
1567 @c ---------------------------------------------------------------------
1569 @include funding.texi
1571 @c ---------------------------------------------------------------------
1573 @c ---------------------------------------------------------------------