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
158 This manual documents the use of @command{gfortran},
159 the GNU Fortran compiler. You can find in this manual how to invoke
160 @command{gfortran}, as well as its features and incompatibilities.
163 @emph{Warning:} This document, and the compiler it describes, are still
164 under development. While efforts are made to keep it up-to-date, it might
165 not accurately reflect the status of the most recent GNU Fortran compiler.
169 @comment When you add a new menu item, please keep the right hand
170 @comment aligned to the same column. Do not use tabs. This provides
171 @comment better formatting.
174 Part I: About GNU Fortran
175 * Getting Started:: What you should know about GNU Fortran.
176 * GNU Fortran and GCC:: You can compile Fortran, C, or other programs.
177 * GNU Fortran and G77:: Why we chose to start from scratch.
178 * Project Status:: Status of GNU Fortran, roadmap, proposed extensions.
179 * Standards:: Standards supported by GNU Fortran.
181 Part II: Invoking GNU Fortran
182 * Invoking GNU Fortran:: Command options supported by @command{gfortran}.
183 * Runtime:: Influencing runtime behavior with environment variables.
185 Part III: 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 =====================================================================
201 @c PART I: ABOUT GNU FORTRAN
202 @c =====================================================================
205 \part{I}{About GNU Fortran}
208 @c ---------------------------------------------------------------------
210 @c ---------------------------------------------------------------------
212 @node Getting Started
213 @chapter Getting Started
215 The GNU Fortran compiler front end was
216 designed initially as a free replacement for,
217 or alternative to, the unix @command{f95} command;
218 @command{gfortran} is the command you'll use to invoke the compiler.
220 The GNU Fortran compiler is still in an early state of development.
221 It can generate code for most constructs and expressions,
222 but much work remains to be done.
224 When the GNU Fortran compiler is finished,
225 it will do everything you expect from any decent compiler:
229 Read a user's program,
230 stored in a file and containing instructions written
231 in Fortran 77, Fortran 90, Fortran 95 or Fortran 2003.
232 This file contains @dfn{source code}.
235 Translate the user's program into instructions a computer
236 can carry out more quickly than it takes to translate the
237 instructions in the first
238 place. The result after compilation of a program is
240 code designed to be efficiently translated and processed
241 by a machine such as your computer.
242 Humans usually aren't as good writing machine code
243 as they are at writing Fortran (or C++, Ada, or Java),
244 because is easy to make tiny mistakes writing machine code.
247 Provide the user with information about the reasons why
248 the compiler is unable to create a binary from the source code.
249 Usually this will be the case if the source code is flawed.
250 When writing Fortran, it is easy to make big mistakes.
251 The Fortran 90 requires that the compiler can point out
252 mistakes to the user.
253 An incorrect usage of the language causes an @dfn{error message}.
255 The compiler will also attempt to diagnose cases where the
256 user's program contains a correct usage of the language,
257 but instructs the computer to do something questionable.
258 This kind of diagnostics message is called a @dfn{warning message}.
261 Provide optional information about the translation passes
262 from the source code to machine code.
263 This can help a user of the compiler to find the cause of
264 certain bugs which may not be obvious in the source code,
265 but may be more easily found at a lower level compiler output.
266 It also helps developers to find bugs in the compiler itself.
269 Provide information in the generated machine code that can
270 make it easier to find bugs in the program (using a debugging tool,
271 called a @dfn{debugger}, such as the GNU Debugger @command{gdb}).
274 Locate and gather machine code already generated to
275 perform actions requested by statements in the user's program.
276 This machine code is organized into @dfn{modules} and is located
277 and @dfn{linked} to the user program.
280 The GNU Fortran compiler consists of several components:
284 A version of the @command{gcc} command
285 (which also might be installed as the system's @command{cc} command)
286 that also understands and accepts Fortran source code.
287 The @command{gcc} command is the @dfn{driver} program for
288 all the languages in the GNU Compiler Collection (GCC);
290 you can compile the source code of any language for
291 which a front end is available in GCC.
294 The @command{gfortran} command itself,
295 which also might be installed as the
296 system's @command{f95} command.
297 @command{gfortran} is just another driver program,
298 but specifically for the Fortran compiler only.
299 The difference with @command{gcc} is that @command{gfortran}
300 will automatically link the correct libraries to your program.
303 A collection of run-time libraries.
304 These libraries contain the machine code needed to support
305 capabilities of the Fortran language that are not directly
306 provided by the machine code generated by the
307 @command{gfortran} compilation phase,
308 such as intrinsic functions and subroutines,
309 and routines for interaction with files and the operating system.
310 @c and mechanisms to spawn,
311 @c unleash and pause threads in parallelized code.
314 The Fortran compiler itself, (@command{f951}).
315 This is the GNU Fortran parser and code generator,
316 linked to and interfaced with the GCC backend library.
317 @command{f951} ``translates'' the source code to
318 assembler code. You would typically not use this
320 instead, the @command{gcc} or @command{gfortran} driver
321 programs will call it for you.
325 @c ---------------------------------------------------------------------
326 @c GNU Fortran and GCC
327 @c ---------------------------------------------------------------------
329 @node GNU Fortran and GCC
330 @chapter GNU Fortran and GCC
331 @cindex GNU Compiler Collection
333 GCC used to be the GNU ``C'' Compiler,
334 but is now known as the @dfn{GNU Compiler Collection}.
335 GCC provides the GNU system with a very versatile
336 compiler middle end (shared optimization passes),
337 and back ends (code generators) for many different
338 computer architectures and operating systems.
339 The code of the middle end and back end are shared by all
340 compiler front ends that are in the GNU Compiler Collection.
342 A GCC front end is essentially a source code parser
343 and an intermediate code generator. The code generator translates the
344 semantics of the source code into a language independent form called
347 The parser takes a source file written in a
348 particular computer language, reads and parses it,
349 and tries to make sure that the source code conforms to
351 Once the correctness of a program has been established,
352 the compiler will build a data structure known as the
353 @dfn{Abstract Syntax tree},
354 or just @dfn{AST} or ``tree'' for short.
355 This data structure represents the whole program
356 or a subroutine or a function.
357 The ``tree'' is passed to the GCC middle end,
358 which will perform optimization passes on it. The optimized AST is then
359 handed off too the back end which assembles the program unit.
361 Different phases in this translation process can be,
362 and in fact @emph{are} merged in many compiler front ends.
363 GNU Fortran has a strict separation between the
364 parser and code generator.
366 The goal of the GNU Fortran project is to build a new front end for GCC.
367 Specifically, a Fortran 95 front end.
368 In a non-@command{gfortran} installation,
369 @command{gcc} will not be able to compile Fortran source code
370 (only the ``C'' front end has to be compiled if you want to build GCC,
371 all other languages are optional).
372 If you build GCC with @command{gfortran}, @command{gcc} will recognize
373 @file{.f/.f90/.f95} source files and accepts Fortran specific
374 command line options.
377 @c ---------------------------------------------------------------------
378 @c GNU Fortran and G77
379 @c ---------------------------------------------------------------------
381 @node GNU Fortran and G77
382 @chapter GNU Fortran and G77
386 Why do we write a compiler front end from scratch?
387 There's a fine Fortran 77 compiler in the
388 GNU Compiler Collection that accepts some features
389 of the Fortran 90 standard as extensions.
390 Why not start from there and revamp it?
392 One of the reasons is that Craig Burley, the author of G77,
393 has decided to stop working on the G77 front end.
394 On @uref{http://world.std.com/~burley/g77-why.html,
395 Craig explains the reasons for his decision to stop working on G77}
396 in one of the pages in his homepage.
397 Among the reasons is a lack of interest in improvements to
399 Users appear to be quite satisfied with @command{g77} as it is.
400 While @command{g77} is still being maintained (by Toon Moene),
401 it is unlikely that sufficient people will be willing
402 to completely rewrite the existing code.
404 But there are other reasons to start from scratch.
405 Many people, including Craig Burley,
406 no longer agreed with certain design decisions in the G77 front end.
407 Also, the interface of @command{g77} to the back end is written in
408 a style which is confusing and not up to date on recommended practice.
409 In fact, a full rewrite had already been planned for GCC 3.0.
411 When Craig decided to stop,
412 it just seemed to be a better idea to start a new project from scratch,
413 because it was expected to be easier to maintain code we
414 develop ourselves than to do a major overhaul of @command{g77} first,
415 and then build a Fortran 95 compiler out of it.
418 @c ---------------------------------------------------------------------
420 @c ---------------------------------------------------------------------
423 @chapter Project Status
426 As soon as @command{gfortran} can parse all of the statements correctly,
427 it will be in the ``larva'' state.
428 When we generate code, the ``puppa'' state.
429 When @command{gfortran} is done,
430 we'll see if it will be a beautiful butterfly,
431 or just a big bug....
433 --Andy Vaught, April 2000
436 The start of the GNU Fortran 95 project was announced on
437 the GCC homepage in March 18, 2000
438 (even though Andy had already been working on it for a while,
441 The GNU Fortran compiler is able to compile nearly all
442 standard-compliant Fortran 95, Fortran 90, and Fortran 77 programs,
443 including a number of standard and non-standard extensions, and can be
444 used on real-world programs. In particular, the supported extensions
445 include OpenMP, Cray-style pointers, and several Fortran 2003 features
446 such as enumeration, stream I/O, and some of the enhancements to
447 allocatable array support from TR 15581. However, it is still under
448 development and has a few remaining rough edges.
450 At present, the GNU Fortran compiler passes the
451 @uref{http://www.fortran-2000.com/ArnaudRecipes/fcvs21_f95.html,
452 NIST Fortran 77 Test Suite}, and produces acceptable results on the
453 @uref{http://www.netlib.org/lapack/faq.html#1.21, LAPACK Test Suite}.
454 It also provides respectable performance on
455 the @uref{http://www.polyhedron.com/pb05.html, Polyhedron Fortran
456 compiler benchmarks} and the
457 @uref{http://www.llnl.gov/asci_benchmarks/asci/limited/lfk/README.html,
458 Livermore Fortran Kernels test}. It has been used to compile a number of
459 large real-world programs, including
460 @uref{http://mysite.verizon.net/serveall/moene.pdf, the HIRLAM
461 weather-forecasting code} and
462 @uref{http://www.theochem.uwa.edu.au/tonto/, the Tonto quantum
463 chemistry package}; see @url{http://gcc.gnu.org/wiki/GfortranApps} for an
466 Among other things, the GNU Fortran compiler is intended as a replacement
467 for G77. At this point, nearly all programs that could be compiled with
468 G77 can be compiled with GNU Fortran, although there are a few minor known
471 The primary work remaining to be done on GNU Fortran falls into three
472 categories: bug fixing (primarily regarding the treatment of invalid code
473 and providing useful error messages), improving the compiler optimizations
474 and the performance of compiled code, and extending the compiler to support
475 future standards---in particular, Fortran 2003.
478 @c ---------------------------------------------------------------------
480 @c ---------------------------------------------------------------------
486 The GNU Fortran compiler implements
487 ISO/IEC 1539:1997 (Fortran 95). As such, it can also compile essentially all
488 standard-compliant Fortran 90 and Fortran 77 programs. It also supports
489 the ISO/IEC TR-15581 enhancements to allocatable arrays, and
490 the @uref{http://www.openmp.org/drupal/mp-documents/spec25.pdf,
491 OpenMP Application Program Interface v2.5} specification.
493 In the future, the GNU Fortran compiler may also support other standard
494 variants of and extensions to the Fortran language. These include
495 ISO/IEC 1539-1:2004 (Fortran 2003).
498 @c =====================================================================
499 @c PART II: INVOCATION REFERENCE
500 @c =====================================================================
503 \part{II}{Invoking GNU Fortran}
506 @c ---------------------------------------------------------------------
508 @c ---------------------------------------------------------------------
513 @c ---------------------------------------------------------------------
515 @c ---------------------------------------------------------------------
518 @chapter Runtime: Influencing runtime behavior with environment variables
521 The behavior of the @command{gfortran} can be influenced by
522 environment variables.
524 Malformed environment variables are silently ignored.
527 * GFORTRAN_STDIN_UNIT:: Unit number for standard input
528 * GFORTRAN_STDOUT_UNIT:: Unit number for standard output
529 * GFORTRAN_STDERR_UNIT:: Unit number for standard error
530 * GFORTRAN_USE_STDERR:: Send library output to standard error
531 * GFORTRAN_TMPDIR:: Directory for scratch files
532 * GFORTRAN_UNBUFFERED_ALL:: Don't buffer output
533 * GFORTRAN_SHOW_LOCUS:: Show location for runtime errors
534 * GFORTRAN_OPTIONAL_PLUS:: Print leading + where permitted
535 * GFORTRAN_DEFAULT_RECL:: Default record length for new files
536 * GFORTRAN_LIST_SEPARATOR:: Separator for list output
537 * GFORTRAN_CONVERT_UNIT:: Set endianness for unformatted I/O
540 @node GFORTRAN_STDIN_UNIT
541 @section @env{GFORTRAN_STDIN_UNIT}---Unit number for standard input
543 This environment variable can be used to select the unit number
544 preconnected to standard input. This must be a positive integer.
545 The default value is 5.
547 @node GFORTRAN_STDOUT_UNIT
548 @section @env{GFORTRAN_STDOUT_UNIT}---Unit number for standard output
550 This environment variable can be used to select the unit number
551 preconnected to standard output. This must be a positive integer.
552 The default value is 6.
554 @node GFORTRAN_STDERR_UNIT
555 @section @env{GFORTRAN_STDERR_UNIT}---Unit number for standard error
557 This environment variable can be used to select the unit number
558 preconnected to standard error. This must be a positive integer.
559 The default value is 0.
561 @node GFORTRAN_USE_STDERR
562 @section @env{GFORTRAN_USE_STDERR}---Send library output to standard error
564 This environment variable controls where library output is sent.
565 If the first letter is @samp{y}, @samp{Y} or @samp{1}, standard
566 error is used. If the first letter is @samp{n}, @samp{N} or
567 @samp{0}, standard output is used.
569 @node GFORTRAN_TMPDIR
570 @section @env{GFORTRAN_TMPDIR}---Directory for scratch files
572 This environment variable controls where scratch files are
573 created. If this environment variable is missing,
574 GNU Fortran searches for the environment variable @env{TMP}. If
575 this is also missing, the default is @file{/tmp}.
577 @node GFORTRAN_UNBUFFERED_ALL
578 @section @env{GFORTRAN_UNBUFFERED_ALL}---Don't buffer output
580 This environment variable controls whether all output is unbuffered.
581 If the first letter is @samp{y}, @samp{Y} or @samp{1}, all output is
582 unbuffered. This will slow down large writes. If the first letter is
583 @samp{n}, @samp{N} or @samp{0}, output is buffered. This is the
586 @node GFORTRAN_SHOW_LOCUS
587 @section @env{GFORTRAN_SHOW_LOCUS}---Show location for runtime errors
589 If the first letter is @samp{y}, @samp{Y} or @samp{1}, filename and
590 line numbers for runtime errors are printed. If the first letter is
591 @samp{n}, @samp{N} or @samp{0}, don't print filename and line numbers
592 for runtime errors. The default is to print the location.
594 @node GFORTRAN_OPTIONAL_PLUS
595 @section @env{GFORTRAN_OPTIONAL_PLUS}---Print leading + where permitted
597 If the first letter is @samp{y}, @samp{Y} or @samp{1},
598 a plus sign is printed
599 where permitted by the Fortran standard. If the first letter
600 is @samp{n}, @samp{N} or @samp{0}, a plus sign is not printed
601 in most cases. Default is not to print plus signs.
603 @node GFORTRAN_DEFAULT_RECL
604 @section @env{GFORTRAN_DEFAULT_RECL}---Default record length for new files
606 This environment variable specifies the default record length, in
607 bytes, for files which are opened without a @code{RECL} tag in the
608 @code{OPEN} statement. This must be a positive integer. The
609 default value is 1073741824 bytes (1 GB).
611 @node GFORTRAN_LIST_SEPARATOR
612 @section @env{GFORTRAN_LIST_SEPARATOR}---Separator for list output
614 This environment variable specifies the separator when writing
615 list-directed output. It may contain any number of spaces and
616 at most one comma. If you specify this on the command line,
617 be sure to quote spaces, as in
619 $ GFORTRAN_LIST_SEPARATOR=' , ' ./a.out
621 when @command{a.out} is the compiled Fortran program that you want to run.
622 Default is a single space.
624 @node GFORTRAN_CONVERT_UNIT
625 @section @env{GFORTRAN_CONVERT_UNIT}---Set endianness for unformatted I/O
627 By setting the @env{GFORTRAN_CONVERT_UNIT} variable, it is possible
628 to change the representation of data for unformatted files.
629 The syntax for the @env{GFORTRAN_CONVERT_UNIT} variable is:
631 GFORTRAN_CONVERT_UNIT: mode | mode ';' exception ;
632 mode: 'native' | 'swap' | 'big_endian' | 'little_endian' ;
633 exception: mode ':' unit_list | unit_list ;
634 unit_list: unit_spec | unit_list unit_spec ;
635 unit_spec: INTEGER | INTEGER '-' INTEGER ;
637 The variable consists of an optional default mode, followed by
638 a list of optional exceptions, which are separated by semicolons
639 from the preceding default and each other. Each exception consists
640 of a format and a comma-separated list of units. Valid values for
641 the modes are the same as for the @code{CONVERT} specifier:
644 @item @code{NATIVE} Use the native format. This is the default.
645 @item @code{SWAP} Swap between little- and big-endian.
646 @item @code{LITTLE_ENDIAN} Use the little-endian format
647 for unformatted files.
648 @item @code{BIG_ENDIAN} Use the big-endian format for unformatted files.
650 A missing mode for an exception is taken to mean @code{BIG_ENDIAN}.
651 Examples of values for @env{GFORTRAN_CONVERT_UNIT} are:
653 @item @code{'big_endian'} Do all unformatted I/O in big_endian mode.
654 @item @code{'little_endian;native:10-20,25'} Do all unformatted I/O
655 in little_endian mode, except for units 10 to 20 and 25, which are in
657 @item @code{'10-20'} Units 10 to 20 are big-endian, the rest is native.
660 Setting the environment variables should be done on the command
661 line or via the @command{export}
662 command for @command{sh}-compatible shells and via @command{setenv}
663 for @command{csh}-compatible shells.
665 Example for @command{sh}:
668 $ GFORTRAN_CONVERT_UNIT='big_endian;native:10-20' ./a.out
671 Example code for @command{csh}:
674 % setenv GFORTRAN_CONVERT_UNIT 'big_endian;native:10-20'
678 Using anything but the native representation for unformatted data
679 carries a significant speed overhead. If speed in this area matters
680 to you, it is best if you use this only for data that needs to be
683 @xref{CONVERT specifier}, for an alternative way to specify the
684 data representation for unformatted files. @xref{Runtime Options}, for
685 setting a default data representation for the whole program. The
686 @code{CONVERT} specifier overrides the @option{-fconvert} compile options.
689 @c =====================================================================
690 @c PART III: LANGUAGE REFERENCE
691 @c =====================================================================
694 \part{III}{Language Reference}
697 @c ---------------------------------------------------------------------
698 @c Fortran 2003 Status
699 @c ---------------------------------------------------------------------
701 @node Fortran 2003 status
702 @chapter Fortran 2003 Status
704 Although GNU Fortran focuses on implementing the Fortran 95
705 standard for the time being, a few Fortran 2003 features are currently
710 Intrinsics @code{command_argument_count}, @code{get_command},
711 @code{get_command_argument}, @code{get_environment_variable}, and
715 @cindex Array constructors
717 Array constructors using square brackets. That is, @code{[...]} rather
721 @cindex @code{FLUSH} statement
722 @code{FLUSH} statement.
725 @cindex @code{IOMSG=} specifier
726 @code{IOMSG=} specifier for I/O statements.
729 @cindex @code{ENUM} statement
730 @cindex @code{ENUMERATOR} statement
731 @cindex @code{-fshort-enums} option
732 Support for the declaration of enumeration constants via the
733 @code{ENUM} and @code{ENUMERATOR} statements. Interoperability with
734 @command{gcc} is guaranteed also for the case where the
735 @command{-fshort-enums} command line option is given.
742 @cindex @code{ALLOCATABLE} dummy arguments
743 @code{ALLOCATABLE} dummy arguments.
745 @cindex @code{ALLOCATABLE} function results
746 @code{ALLOCATABLE} function results
748 @cindex @code{ALLOCATABLE} components of derived types
749 @code{ALLOCATABLE} components of derived types
753 @cindex @code{STREAM} I/O
754 @cindex @code{ACCESS='STREAM'} I/O
755 The @code{OPEN} statement supports the @code{ACCESS='STREAM'} specifier,
756 allowing I/O without any record structure.
759 Namelist input/output for internal files.
762 @cindex @code{PROTECTED}
763 The @code{PROTECTED} statement and attribute.
767 The @code{VALUE} statement and attribute.
770 @cindex @code{VOLATILE}
771 The @code{VOLATILE} statement and attribute.
774 @cindex @code{IMPORT}
775 The @code{IMPORT} statement, allowing to import
776 host-associated derived types.
779 @cindex @code{USE, INTRINSIC}
780 @cindex @code{ISO_FORTRAN_ENV}
781 @code{USE} statement with @code{INTRINSIC} and @code{NON_INTRINSIC}
782 attribute; supported intrinsic modules: @code{ISO_FORTRAN_ENV},
783 @code{OMP_LIB} and @code{OMP_LIB_KINDS}.
788 @c ---------------------------------------------------------------------
790 @c ---------------------------------------------------------------------
792 @c Maybe this chapter should be merged with the 'Standards' section,
793 @c whenever that is written :-)
799 GNU Fortran implements a number of extensions over standard
800 Fortran. This chapter contains information on their syntax and
801 meaning. There are currently two categories of GNU Fortran
802 extensions, those that provide functionality beyond that provided
803 by any standard, and those that are supported by GNU Fortran
804 purely for backward compatibility with legacy compilers. By default,
805 @option{-std=gnu} allows the compiler to accept both types of
806 extensions, but to warn about the use of the latter. Specifying
807 either @option{-std=f95} or @option{-std=f2003} disables both types
808 of extensions, and @option{-std=legacy} allows both without warning.
811 * Old-style kind specifications::
812 * Old-style variable initialization::
813 * Extensions to namelist::
814 * X format descriptor without count field::
815 * Commas in FORMAT specifications::
816 * Missing period in FORMAT specifications::
818 * BOZ literal constants::
819 * Real array indices::
821 * Implicitly convert LOGICAL and INTEGER values::
822 * Hollerith constants support::
824 * CONVERT specifier::
828 @node Old-style kind specifications
829 @section Old-style kind specifications
830 @cindex Kind specifications
832 GNU Fortran allows old-style kind specifications in
833 declarations. These look like:
837 where @code{TYPESPEC} is a basic type (@code{INTEGER}, @code{REAL},
838 etc.), and where @code{k} is a valid kind number for that type. The
839 statement then declares @code{x}, @code{y} and @code{z} to be of
840 type @code{TYPESPEC} with kind @code{k}. This is equivalent to the
841 standard conforming declaration
846 @node Old-style variable initialization
847 @section Old-style variable initialization
848 @cindex Initialization
850 GNU Fortran allows old-style initialization of variables of the
854 REAL x(2,2) /3*0.,1./
856 The syntax for the initializers is as for the @code{DATA} statement, but
857 unlike in a @code{DATA} statement, an initializer only applies to the
858 variable immediately preceding the initialization. In other words,
859 something like @code{INTEGER I,J/2,3/} is not valid. This style of
860 initialization is only allowed in declarations without double colons
861 (@code{::}); the double colons were introduced in Fortran 90, which also
862 introduced a standard syntax for initializating variables in type
865 Examples of standard-conforming code equivalent to the above example
869 INTEGER :: i = 1, j = 2
870 REAL :: x(2,2) = RESHAPE((/0.,0.,0.,1./),SHAPE(x))
874 DATA i/1/, j/2/, x/3*0.,1./
877 Note that variables which are explicitly initialized in declarations
878 or in @code{DATA} statements automatically acquire the @code{SAVE}
881 @node Extensions to namelist
882 @section Extensions to namelist
885 GNU Fortran fully supports the Fortran 95 standard for namelist I/O
886 including array qualifiers, substrings and fully qualified derived types.
887 The output from a namelist write is compatible with namelist read. The
888 output has all names in upper case and indentation to column 1 after the
889 namelist name. Two extensions are permitted:
891 Old-style use of @samp{$} instead of @samp{&}
894 X(:)%Y(2) = 1.0 2.0 3.0
899 It should be noted that the default terminator is @samp{/} rather than
902 Querying of the namelist when inputting from stdin. After at least
903 one space, entering @samp{?} sends to stdout the namelist name and the names of
904 the variables in the namelist:
915 Entering @samp{=?} outputs the namelist to stdout, as if
916 @code{WRITE(*,NML = mynml)} had been called:
921 X(1)%Y= 0.000000 , 1.000000 , 0.000000 ,
922 X(2)%Y= 0.000000 , 2.000000 , 0.000000 ,
923 X(3)%Y= 0.000000 , 3.000000 , 0.000000 ,
927 To aid this dialog, when input is from stdin, errors send their
928 messages to stderr and execution continues, even if @code{IOSTAT} is set.
930 @code{PRINT} namelist is permitted. This causes an error if
931 @option{-std=f95} is used.
934 REAL, dimension (4) :: x = (/1.0, 2.0, 3.0, 4.0/)
937 END PROGRAM test_print
940 Expanded namelist reads are permitted. This causes an error if
941 @option{-std=f95} is used. In the following example, the first element
942 of the array will be given the value 0.00 and the two succeeding
943 elements will be given the values 1.00 and 2.00.
946 X(1,1) = 0.00 , 1.00 , 2.00
950 @node X format descriptor without count field
951 @section @code{X} format descriptor without count field
952 @cindex @code{X} format descriptor without count field
954 To support legacy codes, GNU Fortran permits the count field of the
955 @code{X} edit descriptor in @code{FORMAT} statements to be omitted.
956 When omitted, the count is implicitly assumed to be one.
960 10 FORMAT (I1, X, I1)
963 @node Commas in FORMAT specifications
964 @section Commas in @code{FORMAT} specifications
965 @cindex Commas in @code{FORMAT} specifications
967 To support legacy codes, GNU Fortran allows the comma separator
968 to be omitted immediately before and after character string edit
969 descriptors in @code{FORMAT} statements.
973 10 FORMAT ('FOO='I1' BAR='I2)
977 @node Missing period in FORMAT specifications
978 @section Missing period in @code{FORMAT} specifications
979 @cindex Missing period in @code{FORMAT} specifications
981 To support legacy codes, GNU Fortran allows missing periods in format
982 specifications if and only if @option{-std=legacy} is given on the
983 command line. This is considered non-conforming code and is
993 @section I/O item lists
994 @cindex I/O item lists
996 To support legacy codes, GNU Fortran allows the input item list
997 of the @code{READ} statement, and the output item lists of the
998 @code{WRITE} and @code{PRINT} statements, to start with a comma.
1000 @node BOZ literal constants
1001 @section BOZ literal constants
1002 @cindex BOZ literal constants
1004 As an extension, GNU Fortran allows hexadecimal BOZ literal constants to
1005 be specified using the X prefix, in addition to the standard Z prefix.
1006 BOZ literal constants can also be specified by adding a suffix to the
1007 string. For example, @code{Z'ABC'} and @code{'ABC'Z} are equivalent.
1009 The Fortran standard restricts the appearance of a BOZ literal constant
1010 to the @code{DATA} statement, and it is expected to be assigned to an
1011 @code{INTEGER} variable. GNU Fortran permits a BOZ literal to appear in
1012 any initialization expression as well as assignment statements.
1014 Attempts to use a BOZ literal constant to do a bitwise initialization of
1015 a variable can lead to confusion. A BOZ literal constant is converted
1016 to an @code{INTEGER} value with the kind type with the largest decimal
1017 representation, and this value is then converted numerically to the type
1018 and kind of the variable in question. Thus, one should not expect a
1019 bitwise copy of the BOZ literal constant to be assigned to a @code{REAL}
1022 Similarly, initializing an @code{INTEGER} variable with a statement such
1023 as @code{DATA i/Z'FFFFFFFF'/} will produce an integer overflow rather
1024 than the desired result of @math{-1} when @code{i} is a 32-bit integer
1025 on a system that supports 64-bit integers. The @samp{-fno-range-check}
1026 option can be used as a workaround for legacy code that initializes
1027 integers in this manner.
1029 @node Real array indices
1030 @section Real array indices
1031 @cindex Real array indices
1033 As an extension, GNU Fortran allows the use of @code{REAL} expressions
1034 or variables as array indices.
1036 @node Unary operators
1037 @section Unary operators
1038 @cindex Unary operators
1040 As an extension, GNU Fortran allows unary plus and unary minus operators
1041 to appear as the second operand of binary arithmetic operators without
1042 the need for parenthesis.
1048 @node Implicitly convert LOGICAL and INTEGER values
1049 @section Implicitly convert @code{LOGICAL} and @code{INTEGER} values
1050 @cindex Implicitly convert @code{LOGICAL} and @code{INTEGER} values
1052 As an extension for backwards compatibility with other compilers, GNU
1053 Fortran allows the implicit conversion of @code{LOGICAL} values to
1054 @code{INTEGER} values and vice versa. When converting from a
1055 @code{LOGICAL} to an @code{INTEGER}, @code{.FALSE.} is interpreted as
1056 zero, and @code{.TRUE.} is interpreted as one. When converting from
1057 @code{INTEGER} to @code{LOGICAL}, the value zero is interpreted as
1058 @code{.FALSE.} and any nonzero value is interpreted as @code{.TRUE.}.
1062 IF (i) PRINT *, 'True'
1065 @node Hollerith constants support
1066 @section Hollerith constants support
1067 @cindex Hollerith constants
1069 GNU Fortran supports Hollerith constants in assignments, function
1070 arguments, and @code{DATA} and @code{ASSIGN} statements. A Hollerith
1071 constant is written as a string of characters preceeded by an integer
1072 constant indicating the character count, and the letter @code{H} or
1073 @code{h}, and stored in bytewise fashion in a numeric (@code{INTEGER},
1074 @code{REAL}, or @code{complex}) or @code{LOGICAL} variable. The
1075 constant will be padded or truncated to fit the size of the variable in
1078 Examples of valid uses of Hollerith constants:
1081 data x /16Habcdefghijklmnop, 16Hqrstuvwxyz012345/
1082 x(1) = 16HABCDEFGHIJKLMNOP
1086 Invalid Hollerith constants examples:
1089 a = 8H12345678 ! Valid, but the Hollerith constant will be truncated.
1090 a = 0H ! At least one character is needed.
1093 In general, Hollerith constants were used to provide a rudimentary
1094 facility for handling character strings in early Fortran compilers,
1095 prior to the introduction of @code{CHARACTER} variables in Fortran 77;
1096 in those cases, the standard-compliant equivalent is to convert the
1097 program to use proper character strings. On occasion, there may be a
1098 case where the intent is specifically to initialize a numeric variable
1099 with a given byte sequence. In these cases, the same result can be
1100 obtained by using the @code{TRANSFER} statement, as in this example.
1102 INTEGER(KIND=4) :: a
1103 a = TRANSFER ("abcd", a) ! equivalent to: a = 4Habcd
1108 @section Cray pointers
1109 @cindex Cray pointers
1111 Cray pointers are part of a non-standard extension that provides a
1112 C-like pointer in Fortran. This is accomplished through a pair of
1113 variables: an integer "pointer" that holds a memory address, and a
1114 "pointee" that is used to dereference the pointer.
1116 Pointer/pointee pairs are declared in statements of the form:
1118 pointer ( <pointer> , <pointee> )
1122 pointer ( <pointer1> , <pointee1> ), ( <pointer2> , <pointee2> ), ...
1124 The pointer is an integer that is intended to hold a memory address.
1125 The pointee may be an array or scalar. A pointee can be an assumed
1126 size array---that is, the last dimension may be left unspecified by
1127 using a @code{*} in place of a value---but a pointee cannot be an
1128 assumed shape array. No space is allocated for the pointee.
1130 The pointee may have its type declared before or after the pointer
1131 statement, and its array specification (if any) may be declared
1132 before, during, or after the pointer statement. The pointer may be
1133 declared as an integer prior to the pointer statement. However, some
1134 machines have default integer sizes that are different than the size
1135 of a pointer, and so the following code is not portable:
1140 If a pointer is declared with a kind that is too small, the compiler
1141 will issue a warning; the resulting binary will probably not work
1142 correctly, because the memory addresses stored in the pointers may be
1143 truncated. It is safer to omit the first line of the above example;
1144 if explicit declaration of ipt's type is omitted, then the compiler
1145 will ensure that ipt is an integer variable large enough to hold a
1148 Pointer arithmetic is valid with Cray pointers, but it is not the same
1149 as C pointer arithmetic. Cray pointers are just ordinary integers, so
1150 the user is responsible for determining how many bytes to add to a
1151 pointer in order to increment it. Consider the following example:
1155 pointer (ipt, pointee)
1159 The last statement does not set @code{ipt} to the address of
1160 @code{target(1)}, as it would in C pointer arithmetic. Adding @code{1}
1161 to @code{ipt} just adds one byte to the address stored in @code{ipt}.
1163 Any expression involving the pointee will be translated to use the
1164 value stored in the pointer as the base address.
1166 To get the address of elements, this extension provides an intrinsic
1167 function @code{LOC()}. The @code{LOC()} function is equivalent to the
1168 @code{&} operator in C, except the address is cast to an integer type:
1171 pointer(ipt, arpte(10))
1173 ipt = loc(ar) ! Makes arpte is an alias for ar
1174 arpte(1) = 1.0 ! Sets ar(1) to 1.0
1176 The pointer can also be set by a call to the @code{MALLOC} intrinsic
1179 Cray pointees often are used to alias an existing variable. For
1187 As long as @code{ipt} remains unchanged, @code{iarr} is now an alias for
1188 @code{target}. The optimizer, however, will not detect this aliasing, so
1189 it is unsafe to use @code{iarr} and @code{target} simultaneously. Using
1190 a pointee in any way that violates the Fortran aliasing rules or
1191 assumptions is illegal. It is the user's responsibility to avoid doing
1192 this; the compiler works under the assumption that no such aliasing
1195 Cray pointers will work correctly when there is no aliasing (i.e., when
1196 they are used to access a dynamically allocated block of memory), and
1197 also in any routine where a pointee is used, but any variable with which
1198 it shares storage is not used. Code that violates these rules may not
1199 run as the user intends. This is not a bug in the optimizer; any code
1200 that violates the aliasing rules is illegal. (Note that this is not
1201 unique to GNU Fortran; any Fortran compiler that supports Cray pointers
1202 will ``incorrectly'' optimize code with illegal aliasing.)
1204 There are a number of restrictions on the attributes that can be applied
1205 to Cray pointers and pointees. Pointees may not have the
1206 @code{ALLOCATABLE}, @code{INTENT}, @code{OPTIONAL}, @code{DUMMY},
1207 @code{TARGET}, @code{INTRINSIC}, or @code{POINTER} attributes. Pointers
1208 may not have the @code{DIMENSION}, @code{POINTER}, @code{TARGET},
1209 @code{ALLOCATABLE}, @code{EXTERNAL}, or @code{INTRINSIC} attributes.
1210 Pointees may not occur in more than one pointer statement. A pointee
1211 cannot be a pointer. Pointees cannot occur in equivalence, common, or
1214 A Cray pointer may also point to a function or a subroutine. For
1215 example, the following excerpt is valid:
1219 pointer (subptr,subpte)
1229 A pointer may be modified during the course of a program, and this
1230 will change the location to which the pointee refers. However, when
1231 pointees are passed as arguments, they are treated as ordinary
1232 variables in the invoked function. Subsequent changes to the pointer
1233 will not change the base address of the array that was passed.
1235 @node CONVERT specifier
1236 @section CONVERT specifier
1237 @cindex CONVERT specifier
1239 GNU Fortran allows the conversion of unformatted data between little-
1240 and big-endian representation to facilitate moving of data
1241 between different systems. The conversion can be indicated with
1242 the @code{CONVERT} specifier on the @code{OPEN} statement.
1243 @xref{GFORTRAN_CONVERT_UNIT}, for an alternative way of specifying
1244 the data format via an environment variable.
1246 Valid values for @code{CONVERT} are:
1248 @item @code{CONVERT='NATIVE'} Use the native format. This is the default.
1249 @item @code{CONVERT='SWAP'} Swap between little- and big-endian.
1250 @item @code{CONVERT='LITTLE_ENDIAN'} Use the little-endian representation
1251 for unformatted files.
1252 @item @code{CONVERT='BIG_ENDIAN'} Use the big-endian representation for
1256 Using the option could look like this:
1258 open(file='big.dat',form='unformatted',access='sequential', &
1259 convert='big_endian')
1262 The value of the conversion can be queried by using
1263 @code{INQUIRE(CONVERT=ch)}. The values returned are
1264 @code{'BIG_ENDIAN'} and @code{'LITTLE_ENDIAN'}.
1266 @code{CONVERT} works between big- and little-endian for
1267 @code{INTEGER} values of all supported kinds and for @code{REAL}
1268 on IEEE systems of kinds 4 and 8. Conversion between different
1269 ``extended double'' types on different architectures such as
1270 m68k and x86_64, which GNU Fortran
1271 supports as @code{REAL(KIND=10)} and @code{REAL(KIND=16)}, will
1274 @emph{Note that the values specified via the GFORTRAN_CONVERT_UNIT
1275 environment variable will override the CONVERT specifier in the
1276 open statement}. This is to give control over data formats to
1277 users who do not have the source code of their program available.
1279 Using anything but the native representation for unformatted data
1280 carries a significant speed overhead. If speed in this area matters
1281 to you, it is best if you use this only for data that needs to be
1288 GNU Fortran attempts to be OpenMP Application Program Interface v2.5
1289 compatible when invoked with the @option{-fopenmp} option. GNU Fortran
1290 then generates parallelized code according to the OpenMP directives
1291 used in the source. The OpenMP Fortran runtime library
1292 routines are provided both in a form of a Fortran 90 module named
1293 @code{omp_lib} and in a form of a Fortran @code{include} file named
1296 For details refer to the actual
1297 @uref{http://www.openmp.org/drupal/mp-documents/spec25.pdf,
1298 OpenMP Application Program Interface v2.5} specification.
1300 @c ---------------------------------------------------------------------
1301 @c Intrinsic Procedures
1302 @c ---------------------------------------------------------------------
1304 @include intrinsic.texi
1311 @c ---------------------------------------------------------------------
1313 @c ---------------------------------------------------------------------
1316 @unnumbered Contributing
1317 @cindex Contributing
1319 Free software is only possible if people contribute to efforts
1321 We're always in need of more people helping out with ideas
1322 and comments, writing documentation and contributing code.
1324 If you want to contribute to GNU Fortran,
1325 have a look at the long lists of projects you can take on.
1326 Some of these projects are small,
1327 some of them are large;
1328 some are completely orthogonal to the rest of what is
1329 happening on GNU Fortran,
1330 but others are ``mainstream'' projects in need of enthusiastic hackers.
1331 All of these projects are important!
1332 We'll eventually get around to the things here,
1333 but they are also things doable by someone who is willing and able.
1338 * Proposed Extensions::
1343 @section Contributors to GNU Fortran
1344 @cindex Contributors
1348 Most of the parser was hand-crafted by @emph{Andy Vaught}, who is
1349 also the initiator of the whole project. Thanks Andy!
1350 Most of the interface with GCC was written by @emph{Paul Brook}.
1352 The following individuals have contributed code and/or
1353 ideas and significant help to the GNU Fortran project
1354 (in no particular order):
1358 @item Katherine Holcomb
1359 @item Tobias Schl@"uter
1360 @item Steven Bosscher
1363 @item Niels Kristian Bech Jensen
1364 @item Steven Johnson
1369 @item Fran@,{c}ois-Xavier Coudert
1370 @item Steven G. Kargl
1372 @item Janne Blomqvist
1379 @item Richard Henderson
1380 @item Richard Sandiford
1381 @item Richard Guenther
1382 @item Bernhard Fischer
1385 The following people have contributed bug reports,
1386 smaller or larger patches,
1387 and much needed feedback and encouragement for the
1388 GNU Fortran project:
1391 @item Erik Schnetter
1396 Many other individuals have helped debug,
1397 test and improve the GNU Fortran compiler over the past few years,
1398 and we welcome you to do the same!
1399 If you already have done so,
1400 and you would like to see your name listed in the
1401 list above, please contact us.
1409 @item Help build the test suite
1410 Solicit more code for donation to the test suite.
1411 We can keep code private on request.
1413 @item Bug hunting/squishing
1414 Find bugs and write more test cases!
1415 Test cases are especially very welcome,
1416 because it allows us to concentrate on fixing bugs
1417 instead of isolating them.
1419 @item Smaller projects (``bug'' fixes):
1421 @item Allow init exprs to be numbers raised to integer powers.
1422 @item Implement correct rounding.
1423 @item Implement F restrictions on Fortran 95 syntax.
1424 @item See about making Emacs-parsable error messages.
1428 If you wish to work on the runtime libraries,
1429 please contact a project maintainer.
1433 @node Proposed Extensions
1434 @section Proposed Extensions
1436 Here's a list of proposed extensions for the GNU Fortran compiler, in no particular
1437 order. Most of these are necessary to be fully compatible with
1438 existing Fortran compilers, but they are not part of the official
1439 J3 Fortran 95 standard.
1441 @subsection Compiler extensions:
1444 User-specified alignment rules for structures.
1447 Flag to generate @code{Makefile} info.
1450 Automatically extend single precision constants to double.
1453 Compile code that conserves memory by dynamically allocating common and
1454 module storage either on stack or heap.
1457 Compile flag to generate code for array conformance checking (suggest -CC).
1460 User control of symbol names (underscores, etc).
1463 Compile setting for maximum size of stack frame size before spilling
1464 parts to static or heap.
1467 Flag to force local variables into static space.
1470 Flag to force local variables onto stack.
1473 Flag for maximum errors before ending compile.
1476 Option to initialize otherwise uninitialized integer and floating
1481 @subsection Environment Options
1484 Pluggable library modules for random numbers, linear algebra.
1485 LA should use BLAS calling conventions.
1488 Environment variables controlling actions on arithmetic exceptions like
1489 overflow, underflow, precision loss---Generate NaN, abort, default.
1493 Set precision for fp units that support it (i387).
1496 Variable for setting fp rounding mode.
1499 Variable to fill uninitialized variables with a user-defined bit
1503 Environment variable controlling filename that is opened for that unit
1507 Environment variable to clear/trash memory being freed.
1510 Environment variable to control tracing of allocations and frees.
1513 Environment variable to display allocated memory at normal program end.
1516 Environment variable for filename for * IO-unit.
1519 Environment variable for temporary file directory.
1522 Environment variable forcing standard output to be line buffered (unix).
1527 @c ---------------------------------------------------------------------
1528 @c GNU General Public License
1529 @c ---------------------------------------------------------------------
1535 @c ---------------------------------------------------------------------
1536 @c GNU Free Documentation License
1537 @c ---------------------------------------------------------------------
1543 @c ---------------------------------------------------------------------
1544 @c Funding Free Software
1545 @c ---------------------------------------------------------------------
1547 @include funding.texi
1549 @c ---------------------------------------------------------------------
1551 @c ---------------------------------------------------------------------