1 @c Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002
3 @c Free Software Foundation, Inc.
4 @c This is part of the G77 manual.
5 @c For copying conditions, see the file g77.texi.
7 @c The text of this file appears in the file NEWS
8 @c in the G77 distribution, as well as in the G77 manual.
10 @c Keep this the same as the dates above, since it's used
11 @c in the standalone derivations of this file (e.g. NEWS).
12 @set copyrights-news 1995,1996,1997,1998,1999,2000,2001,2002,2003
14 @set last-update-news 2003-01-31
18 @c The immediately following lines apply to the NEWS file
19 @c which is derived from this file.
20 @emph{Note:} This file is automatically generated from the files
21 @file{news0.texi} and @file{news.texi}.
22 @file{NEWS} is @emph{not} a source file,
23 although it is normally included within source distributions.
25 This file lists news about the @value{which-g77} version
26 (and some other versions) of the GNU Fortran compiler.
27 Copyright (C) @value{copyrights-news} Free Software Foundation, Inc.
28 You may copy, distribute, and modify it freely as long as you preserve
29 this copyright notice and permission notice.
32 @chapter News About GNU Fortran
38 @chapter User-visible Changes
39 @cindex versions, recent
40 @cindex recent versions
41 @cindex changes, user-visible
42 @cindex user-visible changes
44 This chapter describes changes to @command{g77} that are visible
45 to the programmers who actually write and maintain Fortran
46 code they compile with @command{g77}.
47 Information on changes to installation procedures,
48 changes to the documentation, and bug fixes is
49 not provided here, unless it is likely to affect how
50 users use @command{g77}.
51 @xref{News,,News About GNU Fortran}, for information on
52 such changes to @command{g77}.
57 @chapter News About GNU Fortran
58 @cindex versions, recent
59 @cindex recent versions
64 Changes made to recent versions of GNU Fortran are listed
65 below, with the most recent version first.
67 The changes are generally listed in order:
71 Code-generation and run-time-library bug-fixes
74 Compiler and run-time-library crashes involving valid code
81 Fixes and enhancements to existing features
93 This order is not strict---for example, some items
94 involve a combination of these elements.
97 Note that two variants of @command{g77} are tracked below.
98 The @code{egcs} variant is described vis-a-vis
99 previous versions of @code{egcs} and/or
100 an official FSF version, as appropriate.
101 Note that all such variants are obsolete @emph{as of July 1999} -
102 the information is retained here only for its historical value.
104 Therefore, @code{egcs} versions sometimes have multiple listings
105 to help clarify how they differ from other versions,
106 though this can make getting a complete picture
107 of what a particular @code{egcs} version contains
108 somewhat more difficult.
111 For information on bugs in the @value{which-g77} version of @command{g77},
112 see @ref{Known Bugs,,Known Bugs In GNU Fortran}.
116 For information on bugs in the @value{which-g77} version of @command{g77},
117 see @file{@value{path-g77}/BUGS}.
121 @emph{Warning:} The information below is still under development,
122 and might not accurately reflect the @command{g77} code base
123 of which it is a part.
124 Efforts are made to keep it somewhat up-to-date,
125 but they are particularly concentrated
126 on any version of this information
127 that is distributed as part of a @emph{released} @command{g77}.
129 In particular, while this information is intended to apply to
130 the @value{which-g77} version of @command{g77},
131 only an official @emph{release} of that version
132 is expected to contain documentation that is
133 most consistent with the @command{g77} product in that version.
135 Nevertheless, information on @emph{previous} releases of @command{g77}, below,
136 is likely to be more up-to-date and accurate
137 than the equivalent information that accompanied
139 assuming the last-updated date of the information below
140 is later than the dates of those releases.
142 That's due to attempts to keep this development version
143 of news about previous @command{g77} versions up-to-date.
147 An online, ``live'' version of this document
148 (derived directly from the mainline, development version
149 of @command{g77} within @command{gcc})
151 @uref{http://gcc.gnu.org/onlinedocs/g77/News.html}.
154 The following information was last updated on @value{last-update-news}:
156 @heading In @code{GCC} 3.4 versus @code{GCC} 3.3:
162 @heading In @code{GCC} 3.3 versus @code{GCC} 3.2:
165 Problem Reports fixed (in chronological order of submission):
168 g77 generates code which is rejected by GAS if COFF debugging info is
171 Broken links on web pages
173 (libf2c) multiple repeat counts confuse namelist read into array
175 Logical operations error on logicals when using -fugly-logint
177 Generation of C++ Prototype for FORTRAN and extern "C"
179 Failure of g77.f-torture/execute/f90-intrinsic-bit.f -Os on irix6.5
181 (libf2c) OPEN(...,RECL=nnn,...) without ACCESS='DIRECT' should assume a direct
184 g77 "bug"; the executable misbehave (use of options -O2 -fno-automatic
187 (libf2c) DATE_AND_TIME milliseconds field inactive on Windows
189 Incorrect output with 0-based array of characters
191 Double complex zero ** double precision number -> NaN instead of zero
194 Richard Henderson (@email{rth@@redhat.com}) analyzed and improved the handling
195 of (no-)aliasing information for dummy arguments and improved the optimization
196 of induction variables in unrolled loops.
199 @heading In @code{GCC} 3.2 versus @code{GCC} 3.1:
202 Problem Reports fixed (in chronological order of submission):
205 ICE in compensate_edge, at reg-stack.c:2591
207 gcc-3.x does not compile files with suffix .r (RATFOR) [Fixed in 3.2.1]
209 [3.2/3.3/3.4 regression] ICE in compensate_edge, at reg-stack.c:2589
213 @heading In @code{GCC} 3.1 (formerly known as g77-0.5.27) versus @code{GCC} 3.0:
216 Problem Reports fixed (in chronological order of submission):
219 Data statement initialization with subscript of kind INTEGER*2
221 Reference to intrinsic `ISHFT' invalid
223 Function BESJN(integer,double) problems
225 g77 -pipe -xf77-cpp-input sends output to stdout
227 g77 -h" gives bogus output
229 ICE on valid input using CALL EXIT(%VAL(...))
231 g77 -v -c -xf77-version /dev/null -xnone causes ice
233 BACKSPACE example that doesn't work as of gcc/g77-3.0.x
235 g77 rejects accepted use of INTEGER*2 as type of DATA statement loop index
237 ICE on compiling source with 540 000 000 REAL array
239 ICE on BESJN(integer*8,real)
241 bug in loop unrolling
243 sparc-sun-solaris2.7 gcc-3.1 extra g77 testsuite failures w/-m64
245 Incorrect acces of integer*1 variables on PA
247 Failure of LAPACK test dtest on irix6.5 with -mabi=64 -O2
251 @command{g77} now has its man page generated from the texinfo documentation,
252 to guarantee that it remains up to date.
255 @command{g77} used to reject the following program on 32-bit targets:
258 DIMENSION A(140 000 000)
263 prog.f: In program `prog':
265 DIMENSION A(140 000 000)
267 Array `a' at (^) is too large to handle
269 because 140 000 000 REALs is larger than the largest bit-extent that can be
270 expressed in 32 bits. However, bit-sizes never play a role after offsets
271 have been converted to byte addresses. Therefore this check has been removed,
272 and the limit is now 2 Gbyte of memory (around 530 000 000 REALs).
273 Note: On GNU/Linux systems one has to compile and link programs that occupy
274 more than 1 Gbyte statically, i.e.@: @code{g77 -static ...}.
277 Based on work done by Juergen Pfeifer (@email{juergen.pfeifer@@gmx.net})
278 libf2c is now a shared library. One can still link in all objects with
279 the program by specifying the @option{-static} option.
282 Robert Anderson (@email{rwa@@alumni.princeton.edu}) thought up a two
283 line change that enables g77 to compile such code as:
287 DIMENSION A(N(1),N(2))
291 Note the use of array elements in the bounds of the adjustable array A.
294 George Helffrich (@email{george@@geo.titech.ac.jp}) implemented a change
295 in substring index checking (when specifying @option{-fbounds-check})
296 that permits the use of zero length substrings of the form
300 Based on code developed by Pedro Vazquez (@email{vazquez@@penelope.iqm.unicamp.br}),
301 the @code{libf2c} library is now able to read and write files larger than
302 2 Gbyte on 32-bit target machines, if the operating system supports this.
305 @heading In 0.5.26, @code{GCC} 3.0 versus @code{GCC} 2.95:
308 When a REWIND was issued after a WRITE statement on an unformatted
309 file, the implicit truncation was performed by copying the truncated
310 file to /tmp and copying the result back. This has been fixed by using
311 the @code{ftruncate} OS function. Thanks go to the GAMESS developers
312 for bringing this to our attention.
315 Using options @option{-g}, @option{-ggdb} or @option{-gdwarf[-2]} (where
316 appropriate for your target) now also enables debugging information
317 for COMMON BLOCK and EQUIVALENCE items to be emitted.
318 Thanks go to Andrew Vaught (@email{andy@@xena.eas.asu.edu}) and
319 George Helffrich (@email{george@@geology.bristol.ac.uk}) for
320 fixing this longstanding problem.
323 It is not necessary anymore to use the option @option{-femulate-complex}
324 to compile Fortran code using COMPLEX arithmetic, even on 64-bit machines
325 (like the Alpha). This will improve code generation.
328 INTRINSIC arithmetic functions are now treated as routines that do not
329 depend on anything but their argument(s). This enables further instruction
330 scheduling, because it is known that they cannot read or modify arbitrary
335 Upgrade to @code{libf2c} as of 2000-12-05.
337 This fixes a bug where a namelist containing initialization of LOGICAL
338 items and a variable starting with T or F would be read incorrectly.
341 The @code{TtyNam} intrinsics now set @var{Name} to all spaces (at run time)
342 if the system has no @code{ttyname} implementation available.
345 Upgrade to @code{libf2c} as of 1999-06-28.
347 This fixes a bug whereby
348 input to a @code{NAMELIST} read involving a repeat count,
349 such as @samp{K(5)=10*3},
350 was not properly handled by @code{libf2c}.
351 The first item was written to @samp{K(5)},
352 but the remaining nine were written elsewhere (still within the array),
353 not necessarily starting at @samp{K(6)}.
357 @heading In 0.5.25, @code{GCC} 2.95 (@code{EGCS} 1.2) versus @code{EGCS} 1.1.2:
361 @command{g77} no longer generates bad code for assignments,
362 or other conversions,
363 of @code{REAL} or @code{COMPLEX} constant expressions
364 to type @code{INTEGER(KIND=2)}
365 (often referred to as @code{INTEGER*8}).
367 For example, @samp{INTEGER*8 J; J = 4E10} now works as documented.
372 @command{g77} no longer truncates @code{INTEGER(KIND=2)}
373 (usually @code{INTEGER*8})
374 subscript expressions when evaluating array references
375 on systems with pointers widers than @code{INTEGER(KIND=1)}
381 @command{g77} no longer generates bad code
382 for an assignment to a @code{COMPLEX} variable or array
383 that partially overlaps one or more of the sources
384 of the same assignment
385 (a very rare construction).
386 It now assigns through a temporary,
387 in cases where such partial overlap is deemed possible.
392 @code{libg2c} (@code{libf2c}) no longer loses track
393 of the file being worked on
394 during a @code{BACKSPACE} operation.
399 @code{libg2c} (@code{libf2c}) fixes a bug whereby
400 input to a @code{NAMELIST} read involving a repeat count,
401 such as @samp{K(5)=10*3},
402 was not properly handled by @code{libf2c}.
403 The first item was written to @samp{K(5)},
404 but the remaining nine were written elsewhere (still within the array),
405 not necessarily starting at @samp{K(6)}.
410 @c Tim Prince reported this, regarding the TEST_FPU benchmark.
411 Automatic arrays now seem to be working on HP-UX systems.
416 The @code{Date} intrinsic now returns the correct result
417 on big-endian systems.
422 Fix @command{g77} so it no longer crashes when compiling
423 I/O statements using keywords that define @code{INTEGER} values,
424 such as @samp{IOSTAT=@var{j}},
425 where @var{j} is other than default @code{INTEGER}
426 (such as @code{INTEGER*2}).
427 Instead, it issues a diagnostic.
432 Fix @command{g77} so it properly handles @samp{DATA A/@var{rpt}*@var{val}/},
433 where @var{rpt} is not default @code{INTEGER}, such as @code{INTEGER*2},
434 instead of producing a spurious diagnostic.
435 Also fix @samp{DATA (A(I),I=1,N)},
436 where @samp{N} is not default @code{INTEGER}
437 to work instead of crashing @command{g77}.
442 The @option{-ax} option is now obeyed when compiling Fortran programs.
443 (It is passed to the @file{f771} driver.)
447 The new @option{-fbounds-check} option
448 causes @command{g77} to compile run-time bounds checks
449 of array subscripts, as well as of substring start and end points.
452 @code{libg2c} now supports building as multilibbed library,
453 which provides better support for systems
454 that require options such as @option{-mieee}
458 Source file names with the suffixes @samp{.FOR} and @samp{.FPP}
459 now are recognized by @command{g77}
460 as if they ended in @samp{.for} and @samp{.fpp}, respectively.
463 The order of arguments to the @emph{subroutine} forms of the
464 @code{CTime}, @code{DTime}, @code{ETime}, and @code{TtyNam}
465 intrinsics has been swapped.
466 The argument serving as the returned value
467 for the corresponding function forms
468 now is the @emph{second} argument,
469 making these consistent with the other subroutine forms
470 of @code{libU77} intrinsics.
473 @command{g77} now warns about a reference to an intrinsic
474 that has an interface that is not Year 2000 (Y2K) compliant.
475 Also, @code{libg2c} has been changed to increase the likelihood
476 of catching references to the implementations of these intrinsics
477 using the @code{EXTERNAL} mechanism
478 (which would avoid the new warnings).
481 @xref{Year 2000 (Y2K) Problems}, for more information.
486 @command{g77} now warns about a reference to a function
487 when the corresponding @emph{subsequent} function program unit
488 disagrees with the reference concerning the type of the function.
492 @option{-fno-emulate-complex} is now the default option.
493 This should result in improved performance
494 of code that uses the @code{COMPLEX} data type.
497 @cindex double-precision performance
498 @cindex -malign-double
500 The @option{-malign-double} option
501 now reliably aligns @emph{all} double-precision variables and arrays
502 on Intel x86 targets.
506 Even without the @option{-malign-double} option,
507 @command{g77} reliably aligns local double-precision variables
508 that are not in @code{EQUIVALENCE} areas
509 and not @code{SAVE}'d.
514 @command{g77} now open-codes (``inlines'') division of @code{COMPLEX} operands
515 instead of generating a run-time call to
516 the @code{libf2c} routines @code{c_div} or @code{z_div},
517 unless the @option{-Os} option is specified.
521 @command{g77} no longer generates code to maintain @code{errno},
522 a C-language concept,
523 when performing operations such as the @code{SqRt} intrinsic.
527 @command{g77} developers can temporarily use
528 the @option{-fflatten-arrays} option
529 to compare how the compiler handles code generation
530 using C-like constructs as compared to the
531 Fortran-like method constructs normally used.
536 A substantial portion of the @command{g77} front end's code-generation component
538 It now generates code using facilities more robustly supported
539 by the @command{gcc} back end.
540 One effect of this rewrite is that some codes no longer produce
541 a spurious ``label @var{lab} used before containing binding contour''
546 Support for the @option{-fugly} option has been removed.
550 Improve documentation and indexing,
551 including information on Year 2000 (Y2K) compliance,
552 and providing more information on internals of the front end.
557 Upgrade to @code{libf2c} as of 1999-05-10.
561 @heading In 0.5.24 versus 0.5.23:
563 There is no @command{g77} version 0.5.24 at this time,
565 0.5.24 is the version number designated for bug fixes and,
566 perhaps, some new features added,
568 Version 0.5.23 requires @command{gcc} 2.8.1,
569 as 0.5.24 was planned to require.
571 Due to @code{EGCS} becoming @code{GCC}
572 (which is now an acronym for ``GNU Compiler Collection''),
573 and @code{EGCS} 1.2 becoming officially designated @code{GCC} 2.95,
574 there seems to be no need for an actual 0.5.24 release.
576 To reduce the confusion already resulting from use of 0.5.24
577 to designate @command{g77} versions within @code{EGCS} versions 1.0 and 1.1,
578 as well as in versions of @command{g77} documentation and notices
580 ``mainline'' @command{g77} version numbering resumes
581 at 0.5.25 with @code{GCC} 2.95 (@code{EGCS} 1.2),
582 skipping over 0.5.24 as a placeholder version number.
584 To repeat, there is no @command{g77} 0.5.24, but there is now a 0.5.25.
585 Please remain calm and return to your keypunch units.
587 @c 1999-03-15: EGCS 1.1.2 released.
588 @heading In @code{EGCS} 1.1.2 versus @code{EGCS} 1.1.1:
592 Fix the @code{IDate} intrinsic (VXT) (in @code{libg2c})
593 so the returned year is in the documented, non-Y2K-compliant range
595 instead of being returned as 100 in the year 2000.
598 @xref{IDate Intrinsic (VXT)},
599 for more information.
603 Fix the @code{Date_and_Time} intrinsic (in @code{libg2c})
604 to return the milliseconds value properly
608 Fix the @code{LStat} intrinsic (in @code{libg2c})
609 to return device-ID information properly
613 Improve documentation.
617 @c 1998-12-04: EGCS 1.1.1 released.
618 @heading In @code{EGCS} 1.1.1 versus @code{EGCS} 1.1:
622 Fix @code{libg2c} so it performs an implicit @code{ENDFILE} operation
624 whenever a @code{REWIND} is done.
626 (This bug was introduced in 0.5.23 and @code{egcs} 1.1 in
627 @command{g77}'s version of @code{libf2c}.)
630 Fix @code{libg2c} so it no longer crashes with a spurious diagnostic
631 upon doing any I/O following a direct formatted write.
633 (This bug was introduced in 0.5.23 and @code{egcs} 1.1 in
634 @command{g77}'s version of @code{libf2c}.)
637 Fix @command{g77} so it no longer crashes compiling references
638 to the @code{Rand} intrinsic on some systems.
641 Fix @command{g77} portion of installation process so it works
642 better on some systems
643 (those with shells requiring @samp{else true} clauses
644 on @code{if} constructs
645 for the completion code to be set properly).
649 @c 1998-09-03: EGCS 1.1 released.
650 @heading In @code{EGCS} 1.1 versus @code{EGCS} 1.0.3:
654 Fix bugs in the @code{libU77} intrinsic @code{HostNm}
655 that wrote one byte beyond the end of its @code{CHARACTER}
657 and in the @code{libU77} intrinsics
658 @code{GMTime} and @code{LTime}
659 that overwrote their arguments.
664 Assumed arrays with negative bounds
665 (such as @samp{REAL A(-1:*)})
666 no longer elicit spurious diagnostics from @command{g77},
667 even on systems with pointers having
668 different sizes than integers.
670 This bug is not known to have existed in any
671 recent version of @command{gcc}.
672 It was introduced in an early release of @code{egcs}.
677 Valid combinations of @code{EXTERNAL},
678 passing that external as a dummy argument
679 without explicitly giving it a type,
680 and, in a subsequent program unit,
681 referencing that external as
682 an external function with a different type
683 no longer crash @command{g77}.
688 @code{CASE DEFAULT} no longer crashes @command{g77}.
693 The @option{-Wunused} option no longer issues a spurious
694 warning about the ``master'' procedure generated by
695 @command{g77} for procedures containing @code{ENTRY} statements.
699 Support @samp{FORMAT(I<@var{expr}>)} when @var{expr} is a
700 compile-time constant @code{INTEGER} expression.
703 Fix @command{g77} @option{-g} option so procedures that
704 use @code{ENTRY} can be stepped through, line by line,
708 Allow any @code{REAL} argument to intrinsics
709 @code{Second} and @code{CPU_Time}.
712 Use @code{tempnam}, if available, to open scratch files
713 (as in @samp{OPEN(STATUS='SCRATCH')})
714 so that the @code{TMPDIR} environment variable,
718 @command{g77}'s version of @code{libf2c} separates out
719 the setting of global state
720 (such as command-line arguments and signal handling)
721 from @file{main.o} into distinct, new library
724 This should make it easier to write portable applications
725 that have their own (non-Fortran) @code{main()} routine
726 properly set up the @code{libf2c} environment, even
727 when @code{libf2c} (now @code{libg2c}) is a shared library.
731 @command{g77} no longer installs the @file{f77} command
732 and @file{f77.1} man page
733 in the @file{/usr} or @file{/usr/local} hierarchy,
734 even if the @file{f77-install-ok} file exists
735 in the source or build directory.
736 See the installation documentation for more information.
741 @command{g77} no longer installs the @file{libf2c.a} library
742 and @file{f2c.h} include file
743 in the @file{/usr} or @file{/usr/local} hierarchy,
744 even if the @file{f2c-install-ok} or @file{f2c-exists-ok} files exist
745 in the source or build directory.
746 See the installation documentation for more information.
751 The @file{libf2c.a} library produced by @command{g77} has been
752 renamed to @file{libg2c.a}.
753 It is installed only in the @command{gcc} ``private''
754 directory hierarchy, @file{gcc-lib}.
755 This allows system administrators and users to choose which
756 version of the @code{libf2c} library from @code{netlib} they
757 wish to use on a case-by-case basis.
758 See the installation documentation for more information.
763 The @file{f2c.h} include (header) file produced by @command{g77}
764 has been renamed to @file{g2c.h}.
765 It is installed only in the @command{gcc} ``private''
766 directory hierarchy, @file{gcc-lib}.
767 This allows system administrators and users to choose which
768 version of the include file from @code{netlib} they
769 wish to use on a case-by-case basis.
770 See the installation documentation for more information.
774 The @command{g77} command now expects the run-time library
775 to be named @code{libg2c.a} instead of @code{libf2c.a},
776 to ensure that a version other than the one built and
777 installed as part of the same @command{g77} version is picked up.
781 During the configuration and build process,
782 @command{g77} creates subdirectories it needs only as it
784 Other cleaning up of the configuration and build process
785 has been performed as well.
790 @code{install-info} now used to update the directory of
791 Info documentation to contain an entry for @command{g77}
792 (during installation).
796 Some diagnostics have been changed from warnings to errors,
797 to prevent inadvertent use of the resulting, probably buggy,
799 These mostly include diagnostics about use of unsupported features
800 in the @code{OPEN}, @code{INQUIRE}, @code{READ}, and
801 @code{WRITE} statements,
802 and about truncations of various sorts of constants.
806 Improve compilation of @code{FORMAT} expressions so that
807 a null byte is appended to the last operand if it
809 This provides a cleaner run-time diagnostic as provided
810 by @code{libf2c} for statements like @samp{PRINT '(I1', 42}.
815 Improve documentation and indexing.
820 The upgrade to @code{libf2c} as of 1998-06-18
821 should fix a variety of problems, including
822 those involving some uses of the @code{T} format
823 specifier, and perhaps some build (porting) problems
828 @c 1998-09-03: EGCS 1.1 released.
829 @heading In @code{EGCS} 1.1 versus @command{g77} 0.5.23:
833 @cindex stack, 387 coprocessor
837 Fix a code-generation bug that afflicted
838 Intel x86 targets when @option{-O2} was specified
839 compiling, for example, an old version of
840 the @code{DNRM2} routine.
842 The x87 coprocessor stack was being
843 mismanaged in cases involving assigned @code{GOTO}
849 @command{g77} no longer produces incorrect code
851 for @code{EQUIVALENCE} and @code{COMMON}
852 aggregates that, due to ``unnatural'' ordering of members
853 vis-a-vis their types, require initial padding.
858 Fix @command{g77} crash compiling code
859 containing the construct @samp{CMPLX(0.)} or similar.
864 @command{g77} no longer crashes when compiling code
865 containing specification statements such as
866 @samp{INTEGER(KIND=7) PTR}.
871 @command{g77} no longer crashes when compiling code
872 such as @samp{J = SIGNAL(1, 2)}.
876 @command{g77} now treats @samp{%LOC(@var{expr})} and
877 @samp{LOC(@var{expr})} as ``ordinary'' expressions
878 when they are used as arguments in procedure calls.
879 This change applies only to global (filewide) analysis,
880 making it consistent with
881 how @command{g77} actually generates code
884 Previously, @command{g77} treated these expressions
885 as denoting special ``pointer'' arguments
886 for the purposes of filewide analysis.
890 Fix @command{g77} crash
891 (or apparently infinite run-time)
892 when compiling certain complicated expressions
893 involving @code{COMPLEX} arithmetic
894 (especially multiplication).
898 @cindex double-precision performance
899 @cindex -malign-double
901 Align static double-precision variables and arrays
903 regardless of whether @option{-malign-double} is specified.
905 Generally, this affects only local variables and arrays
906 having the @code{SAVE} attribute
907 or given initial values via @code{DATA}.
910 The @command{g77} driver now ensures that @option{-lg2c}
911 is specified in the link phase prior to any
912 occurrence of @option{-lm}.
913 This prevents accidentally linking to a routine
914 in the SunOS4 @option{-lm} library
915 when the generated code wants to link to the one
916 in @code{libf2c} (@code{libg2c}).
919 @command{g77} emits more debugging information when
922 This new information allows, for example,
923 @kbd{which __g77_length_a} to be used in @command{gdb}
924 to determine the type of the phantom length argument
925 supplied with @code{CHARACTER} variables.
927 This information pertains to internally-generated
928 type, variable, and other information,
929 not to the longstanding deficiencies vis-a-vis
930 @code{COMMON} and @code{EQUIVALENCE}.
933 The F90 @code{Date_and_Time} intrinsic now is
937 The F90 @code{System_Clock} intrinsic allows
938 the optional arguments (except for the @code{Count}
939 argument) to be omitted.
943 Upgrade to @code{libf2c} as of 1998-06-18.
948 Improve documentation and indexing.
953 @heading In previous versions:
955 Information on previous versions is not provided
956 in this @file{@value{path-g77}/NEWS} file,
958 See @file{@value{path-g77}/news.texi},
959 or any of its other derivations
960 (Info, HTML, dvi forms)
961 for such information.
965 @c 1998-05-20: 0.5.23 released.
966 @heading In 0.5.23 versus 0.5.22:
969 This release contains several regressions against
970 version 0.5.22 of @command{g77}, due to using the
971 ``vanilla'' @command{gcc} back end instead of patching
972 it to fix a few bugs and improve performance in a
975 Features that have been dropped from this version
976 of @command{g77} due to their being implemented
977 via @command{g77}-specific patches to the @command{gcc}
978 back end in previous releases include:
982 Support for @code{__restrict__} keyword,
983 the options @option{-fargument-alias}, @option{-fargument-noalias},
984 and @option{-fargument-noalias-global},
985 and the corresponding alias-analysis code.
987 (@code{egcs} has the alias-analysis
988 code, but not the @code{__restrict__} keyword.
989 @code{egcs} @command{g77} users benefit from the alias-analysis
990 code despite the lack of the @code{__restrict__} keyword,
991 which is a C-language construct.)
994 Support for the GNU compiler options
995 @option{-fmove-all-movables},
996 @option{-freduce-all-givs},
997 and @option{-frerun-loop-opt}.
999 (@code{egcs} supports these options.
1000 @command{g77} users of @code{egcs} benefit from them even if
1001 they are not explicitly specified,
1002 because the defaults are optimized for @command{g77} users.)
1005 Support for the @option{-W} option warning about
1006 integer division by zero.
1009 The Intel x86-specific option @option{-malign-double}
1010 applying to stack-allocated data
1011 as well as statically-allocate data.
1014 @ifclear USERVISONLY
1015 Note that the @file{gcc/f/gbe/} subdirectory has been removed
1016 from this distribution as a result of @command{g77} no longer
1017 including patches for the @command{gcc} back end.
1020 @ifclear USERVISONLY
1022 Fix bugs in the @code{libU77} intrinsic @code{HostNm}
1023 that wrote one byte beyond the end of its @code{CHARACTER}
1025 and in the @code{libU77} intrinsics
1026 @code{GMTime} and @code{LTime}
1027 that overwrote their arguments.
1031 Support @command{gcc} version 2.8,
1032 and remove support for prior versions of @command{gcc}.
1034 @cindex -@w{}-driver option
1035 @cindex @command{g77} options, -@w{}-driver
1036 @cindex options, -@w{}-driver
1038 Remove support for the @option{--driver} option,
1039 as @command{g77} now does all the driving,
1040 just like @command{gcc}.
1042 @ifclear USERVISONLY
1044 @code{CASE DEFAULT} no longer crashes @command{g77}.
1047 @ifclear USERVISONLY
1049 Valid combinations of @code{EXTERNAL},
1050 passing that external as a dummy argument
1051 without explicitly giving it a type,
1052 and, in a subsequent program unit,
1053 referencing that external as
1054 an external function with a different type
1055 no longer crash @command{g77}.
1058 @ifclear USERVISONLY
1060 @command{g77} no longer installs the @file{f77} command
1061 and @file{f77.1} man page
1062 in the @file{/usr} or @file{/usr/local} hierarchy,
1063 even if the @file{f77-install-ok} file exists
1064 in the source or build directory.
1065 See the installation documentation for more information.
1068 @ifclear USERVISONLY
1070 @command{g77} no longer installs the @file{libf2c.a} library
1071 and @file{f2c.h} include file
1072 in the @file{/usr} or @file{/usr/local} hierarchy,
1073 even if the @file{f2c-install-ok} or @file{f2c-exists-ok} files exist
1074 in the source or build directory.
1075 See the installation documentation for more information.
1078 @ifclear USERVISONLY
1080 The @file{libf2c.a} library produced by @command{g77} has been
1081 renamed to @file{libg2c.a}.
1082 It is installed only in the @command{gcc} ``private''
1083 directory hierarchy, @file{gcc-lib}.
1084 This allows system administrators and users to choose which
1085 version of the @code{libf2c} library from @code{netlib} they
1086 wish to use on a case-by-case basis.
1087 See the installation documentation for more information.
1090 @ifclear USERVISONLY
1092 The @file{f2c.h} include (header) file produced by @command{g77}
1093 has been renamed to @file{g2c.h}.
1094 It is installed only in the @command{gcc} ``private''
1095 directory hierarchy, @file{gcc-lib}.
1096 This allows system administrators and users to choose which
1097 version of the include file from @code{netlib} they
1098 wish to use on a case-by-case basis.
1099 See the installation documentation for more information.
1103 The @command{g77} command now expects the run-time library
1104 to be named @code{libg2c.a} instead of @code{libf2c.a},
1105 to ensure that a version other than the one built and
1106 installed as part of the same @command{g77} version is picked up.
1108 @ifclear USERVISONLY
1110 The @option{-Wunused} option no longer issues a spurious
1111 warning about the ``master'' procedure generated by
1112 @command{g77} for procedures containing @code{ENTRY} statements.
1116 @command{g77}'s version of @code{libf2c} separates out
1117 the setting of global state
1118 (such as command-line arguments and signal handling)
1119 from @file{main.o} into distinct, new library
1122 This should make it easier to write portable applications
1123 that have their own (non-Fortran) @code{main()} routine
1124 properly set up the @code{libf2c} environment, even
1125 when @code{libf2c} (now @code{libg2c}) is a shared library.
1127 @ifclear USERVISONLY
1129 During the configuration and build process,
1130 @command{g77} creates subdirectories it needs only as it
1131 needs them, thus avoiding unnecessary creation of, for example,
1132 @file{stage1/f/runtime} when doing a non-bootstrap build.
1133 Other cleaning up of the configuration and build process
1134 has been performed as well.
1137 @ifclear USERVISONLY
1139 @code{install-info} now used to update the directory of
1140 Info documentation to contain an entry for @command{g77}
1141 (during installation).
1145 Some diagnostics have been changed from warnings to errors,
1146 to prevent inadvertent use of the resulting, probably buggy,
1148 These mostly include diagnostics about use of unsupported features
1149 in the @code{OPEN}, @code{INQUIRE}, @code{READ}, and
1150 @code{WRITE} statements,
1151 and about truncations of various sorts of constants.
1153 @ifclear USERVISONLY
1155 Improve documentation and indexing.
1158 @ifclear USERVISONLY
1160 Upgrade to @code{libf2c} as of 1998-04-20.
1162 This should fix a variety of problems, including
1163 those involving some uses of the @code{T} format
1164 specifier, and perhaps some build (porting) problems
1169 @c 1998-03-16: 0.5.22 released.
1170 @heading In 0.5.22 versus 0.5.21:
1172 @ifclear USERVISONLY
1174 Fix code generation for iterative @code{DO} loops that
1175 have one or more references to the iteration variable,
1176 or to aliases of it, in their control expressions.
1177 For example, @samp{DO 10 J=2,J} now is compiled correctly.
1180 @ifclear USERVISONLY
1182 @cindex stack, 387 coprocessor
1186 Fix a code-generation bug that afflicted
1187 Intel x86 targets when @option{-O2} was specified
1188 compiling, for example, an old version of
1189 the @code{DNRM2} routine.
1191 The x87 coprocessor stack was being
1192 mismanaged in cases involving assigned @code{GOTO}
1196 @ifclear USERVISONLY
1198 Fix @code{DTime} intrinsic so as not to truncate
1199 results to integer values (on some systems).
1203 Fix @code{Signal} intrinsic so it offers portable
1204 support for 64-bit systems (such as Digital Alphas
1207 @ifclear USERVISONLY
1209 Fix run-time crash involving @code{NAMELIST} on 64-bit
1210 machines such as Alphas.
1213 @ifclear USERVISONLY
1215 Fix @command{g77} version of @code{libf2c} so it no longer
1216 produces a spurious @samp{I/O recursion} diagnostic at run time
1217 when an I/O operation (such as @samp{READ *,I}) is interrupted
1218 in a manner that causes the program to be terminated
1219 via the @code{f_exit} routine (such as via @kbd{C-c}).
1222 @ifclear USERVISONLY
1224 Fix @command{g77} crash triggered by @code{CASE} statement with
1225 an omitted lower or upper bound.
1228 @ifclear USERVISONLY
1230 Fix @command{g77} crash compiling references to @code{CPU_Time}
1234 @ifclear USERVISONLY
1236 Fix @command{g77} crash
1237 (or apparently infinite run-time)
1238 when compiling certain complicated expressions
1239 involving @code{COMPLEX} arithmetic
1240 (especially multiplication).
1243 @ifclear USERVISONLY
1245 Fix @command{g77} crash on statements such as
1246 @samp{PRINT *, (REAL(Z(I)),I=1,2)}, where
1247 @samp{Z} is @code{DOUBLE COMPLEX}.
1250 @ifclear USERVISONLY
1252 Fix a @command{g++} crash.
1256 Support @samp{FORMAT(I<@var{expr}>)} when @var{expr} is a
1257 compile-time constant @code{INTEGER} expression.
1260 Fix @command{g77} @option{-g} option so procedures that
1261 use @code{ENTRY} can be stepped through, line by line,
1264 @ifclear USERVISONLY
1266 Fix a profiling-related bug in @command{gcc} back end for
1267 Intel x86 architecture.
1271 Allow any @code{REAL} argument to intrinsics
1272 @code{Second} and @code{CPU_Time}.
1275 Allow any numeric argument to intrinsics
1276 @code{Int2} and @code{Int8}.
1279 Use @code{tempnam}, if available, to open scratch files
1280 (as in @samp{OPEN(STATUS='SCRATCH')})
1281 so that the @code{TMPDIR} environment variable,
1282 if present, is used.
1285 Rename the @command{gcc} keyword @code{restrict} to
1286 @code{__restrict__}, to avoid rejecting valid, existing,
1288 Support for @code{restrict} is now more like support
1291 @ifclear USERVISONLY
1293 Fix @option{-fpedantic} to not reject procedure invocations
1294 such as @samp{I=J()} and @samp{CALL FOO()}.
1298 Fix @option{-fugly-comma} to affect invocations of
1299 only external procedures.
1300 Restore rejection of gratuitous trailing omitted
1301 arguments to intrinsics, as in @samp{I=MAX(3,4,,)}.
1304 Fix compiler so it accepts @option{-fgnu-intrinsics-*} and
1305 @option{-fbadu77-intrinsics-*} options.
1307 @ifclear USERVISONLY
1309 Improve diagnostic messages from @code{libf2c}
1310 so it is more likely that the printing of the
1311 active format string is limited to the string,
1312 with no trailing garbage being printed.
1314 (Unlike @command{f2c}, @command{g77} did not append
1315 a null byte to its compiled form of every
1316 format string specified via a @code{FORMAT} statement.
1317 However, @command{f2c} would exhibit the problem
1318 anyway for a statement like @samp{PRINT '(I)garbage', 1}
1319 by printing @samp{(I)garbage} as the format string.)
1322 @ifclear USERVISONLY
1324 Improve compilation of @code{FORMAT} expressions so that
1325 a null byte is appended to the last operand if it
1327 This provides a cleaner run-time diagnostic as provided
1328 by @code{libf2c} for statements like @samp{PRINT '(I1', 42}.
1331 @ifclear USERVISONLY
1333 Fix various crashes involving code with diagnosed errors.
1336 @ifclear USERVISONLY
1338 Fix cross-compilation bug when configuring @code{libf2c}.
1341 @ifclear USERVISONLY
1343 Improve diagnostics.
1346 @ifclear USERVISONLY
1348 Improve documentation and indexing.
1351 @ifclear USERVISONLY
1353 Upgrade to @code{libf2c} as of 1997-09-23.
1354 This fixes a formatted-I/O bug that afflicted
1355 64-bit systems with 32-bit integers
1356 (such as Digital Alpha running GNU/Linux).
1360 @c 1998-03-18: EGCS 1.0.2 released.
1361 @heading In @code{EGCS} 1.0.2 versus @code{EGCS} 1.0.1:
1363 @ifclear USERVISONLY
1365 Fix @command{g77} crash triggered by @code{CASE} statement with
1366 an omitted lower or upper bound.
1369 @ifclear USERVISONLY
1371 Fix @command{g77} crash on statements such as
1372 @samp{PRINT *, (REAL(Z(I)),I=1,2)}, where
1373 @samp{Z} is @code{DOUBLE COMPLEX}.
1376 @ifclear USERVISONLY
1378 @cindex support, ELF
1379 @cindex -fPIC option
1380 @cindex options, -fPIC
1382 Fix @option{-fPIC} (such as compiling for ELF targets)
1383 on the Intel x86 architecture target
1384 so invalid assembler code is no longer produced.
1387 @ifclear USERVISONLY
1389 Fix @option{-fpedantic} to not reject procedure invocations
1390 such as @samp{I=J()} and @samp{CALL FOO()}.
1393 @ifclear USERVISONLY
1395 Fix @option{-fugly-comma} to affect invocations of
1396 only external procedures.
1397 Restore rejection of gratuitous trailing omitted
1398 arguments to intrinsics, as in @samp{I=MAX(3,4,,)}.
1402 Fix compiler so it accepts @option{-fgnu-intrinsics-*} and
1403 @option{-fbadu77-intrinsics-*} options.
1406 @c 1998-01-06: EGCS 1.0.1 released.
1407 @heading In @code{EGCS} 1.0.1 versus @code{EGCS} 1.0:
1408 @ifclear USERVISONLY
1411 Fix run-time crash involving @code{NAMELIST} on 64-bit
1412 machines such as Alphas.
1416 @c 1997-12-03: EGCS 1.0 released.
1417 @heading In @code{EGCS} 1.0 versus @command{g77} 0.5.21:
1420 Version 1.0 of @code{egcs}
1421 contains several regressions against
1422 version 0.5.21 of @command{g77},
1424 ``vanilla'' @command{gcc} back end instead of patching
1425 it to fix a few bugs and improve performance in a
1428 Features that have been dropped from this version
1429 of @command{g77} due to their being implemented
1430 via @command{g77}-specific patches to the @command{gcc}
1431 back end in previous releases include:
1435 Support for the C-language @code{restrict} keyword.
1438 Support for the @option{-W} option warning about
1439 integer division by zero.
1442 The Intel x86-specific option @option{-malign-double}
1443 applying to stack-allocated data
1444 as well as statically-allocate data.
1447 @ifclear USERVISONLY
1448 Note that the @file{gcc/f/gbe/} subdirectory has been removed
1449 from this distribution as a result of @command{g77}
1450 being fully integrated with
1451 the @code{egcs} variant of the @command{gcc} back end.
1454 @ifclear USERVISONLY
1456 Fix code generation for iterative @code{DO} loops that
1457 have one or more references to the iteration variable,
1458 or to aliases of it, in their control expressions.
1459 For example, @samp{DO 10 J=2,J} now is compiled correctly.
1462 @ifclear USERVISONLY
1464 Fix @code{DTime} intrinsic so as not to truncate
1465 results to integer values (on some systems).
1468 @ifclear USERVISONLY
1470 @c Toon Moene discovered these.
1471 Some Fortran code, miscompiled
1472 by @command{g77} built on @command{gcc} version 2.8.1
1473 on m68k-next-nextstep3 configurations
1474 when using the @option{-O2} option,
1475 is now compiled correctly.
1476 It is believed that a C function known to miscompile
1477 on that configuration
1478 when using the @samp{-O2 -funroll-loops} options
1479 also is now compiled correctly.
1482 @ifclear USERVISONLY
1484 Remove support for non-@code{egcs} versions of @command{gcc}.
1487 @cindex -@w{}-driver option
1488 @cindex @command{g77} options, -@w{}-driver
1489 @cindex options, -@w{}-driver
1491 Remove support for the @option{--driver} option,
1492 as @command{g77} now does all the driving,
1493 just like @command{gcc}.
1496 Allow any numeric argument to intrinsics
1497 @code{Int2} and @code{Int8}.
1499 @ifclear USERVISONLY
1501 Improve diagnostic messages from @code{libf2c}
1502 so it is more likely that the printing of the
1503 active format string is limited to the string,
1504 with no trailing garbage being printed.
1506 (Unlike @command{f2c}, @command{g77} did not append
1507 a null byte to its compiled form of every
1508 format string specified via a @code{FORMAT} statement.
1509 However, @code{f2c} would exhibit the problem
1510 anyway for a statement like @samp{PRINT '(I)garbage', 1}
1511 by printing @samp{(I)garbage} as the format string.)
1514 @ifclear USERVISONLY
1516 Upgrade to @code{libf2c} as of 1997-09-23.
1517 This fixes a formatted-I/O bug that afflicted
1518 64-bit systems with 32-bit integers
1519 (such as Digital Alpha running GNU/Linux).
1523 @c 1997-09-09: 0.5.21 released.
1526 @ifclear USERVISONLY
1528 Fix a code-generation bug introduced by 0.5.20
1529 caused by loop unrolling (by specifying
1530 @option{-funroll-loops} or similar).
1531 This bug afflicted all code compiled by
1532 version 2.7.2.2.f.2 of @command{gcc} (C, C++,
1533 Fortran, and so on).
1536 @ifclear USERVISONLY
1538 Fix a code-generation bug manifested when
1539 combining local @code{EQUIVALENCE} with a
1540 @code{DATA} statement that follows
1541 the first executable statement (or is
1542 treated as an executable-context statement
1543 as a result of using the @option{-fpedantic}
1547 @ifclear USERVISONLY
1549 Fix a compiler crash that occurred when an
1550 integer division by a constant zero is detected.
1551 Instead, when the @option{-W} option is specified,
1552 the @command{gcc} back end issues a warning about such a case.
1553 This bug afflicted all code compiled by
1554 version 2.7.2.2.f.2 of @command{gcc} (C, C++,
1555 Fortran, and so on).
1559 When the @option{-W} option is specified, @command{gcc}, @command{g77},
1560 and other GNU compilers that incorporate the @command{gcc}
1561 back end as modified by @command{g77}, issue
1562 a warning about integer division by constant zero.
1565 @ifclear USERVISONLY
1567 Fix a compiler crash that occurred in some cases
1568 of procedure inlining.
1569 (Such cases became more frequent in 0.5.20.)
1572 @ifclear USERVISONLY
1574 Fix a compiler crash resulting from using @code{DATA}
1575 or similar to initialize a @code{COMPLEX} variable or
1579 @ifclear USERVISONLY
1581 Fix compiler crashes involving use of @code{AND}, @code{OR},
1582 or @code{XOR} intrinsics.
1585 @ifclear USERVISONLY
1587 Fix compiler bug triggered when using a @code{COMMON}
1588 or @code{EQUIVALENCE} variable
1589 as the target of an @code{ASSIGN}
1590 or assigned-@code{GOTO} statement.
1593 @ifclear USERVISONLY
1595 Fix compiler crashes due to using the name of a some
1596 non-standard intrinsics (such as @code{FTell} or
1597 @code{FPutC}) as such and as the name of a procedure
1599 Such dual use of a name in a program is allowed by
1603 @c @command{g77}'s version of @code{libf2c} has been modified
1604 @c so that the external names of library's procedures do not
1605 @c conflict with names used for Fortran procedures compiled
1606 @c by @command{g77}.
1607 @c An additional layer of jacket procedures has been added
1608 @c to @code{libf2c} to map the old names to the new names,
1609 @c for automatic use by programs that interface to the
1610 @c library procedures via the external-procedure mechanism.
1612 @c For example, the intrinsic @code{FPUTC} previously was
1613 @c implemented by @command{g77} as a call to the @code{libf2c}
1614 @c routine @code{fputc_}.
1615 @c This would conflict with a Fortran procedure named @code{FPUTC}
1616 @c (using default compiler options), and this conflict
1617 @c would cause a crash under certain circumstances.
1619 @c Now, the intrinsic @code{FPUTC} calls @code{G77_fputc_0},
1620 @c which does not conflict with the @code{fputc_} external
1621 @c that implements a Fortran procedure named @code{FPUTC}.
1623 @c Programs that refer to @code{FPUTC} as an external procedure
1624 @c without supplying their own implementation will link to
1625 @c the new @code{libf2c} routine @code{fputc_}, which is
1626 @c simply a jacket routine that calls @code{G77_fputc_0}.
1628 @ifclear USERVISONLY
1630 Place automatic arrays on the stack, even if
1631 @code{SAVE} or the @option{-fno-automatic} option
1633 This avoids a compiler crash in some cases.
1636 @ifclear USERVISONLY
1638 The @option{-malign-double} option now reliably aligns
1639 @code{DOUBLE PRECISION} optimally on Pentium and
1640 Pentium Pro architectures (586 and 686 in @command{gcc}).
1644 New option @option{-Wno-globals} disables warnings
1645 about ``suspicious'' use of a name both as a global
1646 name and as the implicit name of an intrinsic, and
1647 warnings about disagreements over the number or natures of
1648 arguments passed to global procedures, or the
1649 natures of the procedures themselves.
1651 The default is to issue such warnings, which are
1652 new as of this version of @command{g77}.
1655 New option @option{-fno-globals} disables diagnostics
1656 about potentially fatal disagreements
1657 analysis problems, such as disagreements over the
1658 number or natures of arguments passed to global
1659 procedures, or the natures of those procedures themselves.
1661 The default is to issue such diagnostics and flag
1662 the compilation as unsuccessful.
1663 With this option, the diagnostics are issued as
1664 warnings, or, if @option{-Wno-globals} is specified,
1665 are not issued at all.
1667 This option also disables inlining of global procedures,
1668 to avoid compiler crashes resulting from coding errors
1669 that these diagnostics normally would identify.
1671 @ifclear USERVISONLY
1673 Diagnose cases where a reference to a procedure
1674 disagrees with the type of that procedure, or
1675 where disagreements about the number or nature
1677 This avoids a compiler crash.
1680 @ifclear USERVISONLY
1682 Fix parsing bug whereby @command{g77} rejected a
1683 second initialization specification immediately
1684 following the first's closing @samp{/} without
1685 an intervening comma in a @code{DATA} statement,
1686 and the second specification was an implied-DO list.
1689 @ifclear USERVISONLY
1691 Improve performance of the @command{gcc} back end so
1692 certain complicated expressions involving @code{COMPLEX}
1693 arithmetic (especially multiplication) don't appear to
1694 take forever to compile.
1697 @ifclear USERVISONLY
1699 Fix a couple of profiling-related bugs in @command{gcc}
1703 @ifclear USERVISONLY
1705 Integrate GNU Ada's (GNAT's) changes to the back end,
1706 which consist almost entirely of bug fixes.
1707 These fixes are circa version 3.10p of GNAT.
1710 @ifclear USERVISONLY
1712 Include some other @command{gcc} fixes that seem useful in
1713 @command{g77}'s version of @command{gcc}.
1714 (See @file{gcc/ChangeLog} for details---compare it
1715 to that file in the vanilla @code{gcc-2.7.2.3.tar.gz}
1720 Fix @code{libU77} routines that accept file and other names
1721 to strip trailing blanks from them, for consistency
1722 with other implementations.
1723 Blanks may be forcibly appended to such names by
1724 appending a single null character (@samp{CHAR(0)})
1725 to the significant trailing blanks.
1728 Fix @code{CHMOD} intrinsic to work with file names
1729 that have embedded blanks, commas, and so on.
1732 Fix @code{SIGNAL} intrinsic so it accepts an
1733 optional third @code{Status} argument.
1735 @ifclear USERVISONLY
1737 Fix @code{IDATE()} intrinsic subroutine (VXT form)
1738 so it accepts arguments in the correct order.
1739 Documentation fixed accordingly, and for
1740 @code{GMTIME()} and @code{LTIME()} as well.
1744 Make many changes to @code{libU77} intrinsics to
1745 support existing code more directly.
1747 Such changes include allowing both subroutine and
1748 function forms of many routines, changing @code{MCLOCK()}
1749 and @code{TIME()} to return @code{INTEGER(KIND=1)} values,
1750 introducing @code{MCLOCK8()} and @code{TIME8()} to
1751 return @code{INTEGER(KIND=2)} values,
1752 and placing functions that are intended to perform
1753 side effects in a new intrinsic group, @code{badu77}.
1755 @ifclear USERVISONLY
1757 Improve @code{libU77} so it is more portable.
1761 Add options @option{-fbadu77-intrinsics-delete},
1762 @option{-fbadu77-intrinsics-hide}, and so on.
1764 @ifclear USERVISONLY
1766 Fix crashes involving diagnosed or invalid code.
1769 @ifclear USERVISONLY
1771 @command{g77} and @command{gcc} now do a somewhat better
1772 job detecting and diagnosing arrays that are too
1773 large to handle before these cause diagnostics
1774 during the assembler or linker phase, a compiler
1775 crash, or generation of incorrect code.
1778 @ifclear USERVISONLY
1780 Make some fixes to alias analysis code.
1783 @ifclear USERVISONLY
1785 Add support for @code{restrict} keyword in @command{gcc}
1789 @ifclear USERVISONLY
1791 Support @command{gcc} version 2.7.2.3
1792 (modified by @command{g77} into version 2.7.2.3.f.1),
1794 support for prior versions of @command{gcc}.
1797 @ifclear USERVISONLY
1799 Incorporate GNAT's patches to the @command{gcc} back
1800 end into @command{g77}'s, so GNAT users do not need
1801 to apply GNAT's patches to build both GNAT and @command{g77}
1802 from the same source tree.
1805 @ifclear USERVISONLY
1807 Modify @command{make} rules and related code so that
1808 generation of Info documentation doesn't require
1809 compilation using @command{gcc}.
1810 Now, any ANSI C compiler should be adequate to
1811 produce the @command{g77} documentation (in particular,
1812 the tables of intrinsics) from scratch.
1816 Add @code{INT2} and @code{INT8} intrinsics.
1819 Add @code{CPU_TIME} intrinsic.
1822 Add @code{ALARM} intrinsic.
1825 @code{CTIME} intrinsic now accepts any @code{INTEGER}
1826 argument, not just @code{INTEGER(KIND=2)}.
1828 @ifclear USERVISONLY
1830 Warn when explicit type declaration disagrees with
1831 the type of an intrinsic invocation.
1834 @ifclear USERVISONLY
1836 Support @samp{*f771} entry in @command{gcc} @file{specs} file.
1839 @ifclear USERVISONLY
1841 Fix typo in @command{make} rule @command{g77-cross}, used only for
1845 @ifclear USERVISONLY
1847 Fix @code{libf2c} build procedure to re-archive library
1848 if previous attempt to archive was interrupted.
1851 @ifclear USERVISONLY
1853 Change @command{gcc} to unroll loops only during the last
1854 invocation (of as many as two invocations) of loop
1858 @ifclear USERVISONLY
1860 Improve handling of @option{-fno-f2c} so that code that
1861 attempts to pass an intrinsic as an actual argument,
1862 such as @samp{CALL FOO(ABS)}, is rejected due to the fact
1863 that the run-time-library routine is, effectively,
1864 compiled with @option{-ff2c} in effect.
1867 @ifclear USERVISONLY
1869 Fix @command{g77} driver to recognize @option{-fsyntax-only}
1870 as an option that inhibits linking, just like @option{-c} or
1871 @option{-S}, and to recognize and properly handle the
1872 @option{-nostdlib}, @option{-M}, @option{-MM}, @option{-nodefaultlibs},
1873 and @option{-Xlinker} options.
1876 @ifclear USERVISONLY
1878 Upgrade to @code{libf2c} as of 1997-08-16.
1881 @ifclear USERVISONLY
1883 Modify @code{libf2c} to consistently and clearly diagnose
1884 recursive I/O (at run time).
1888 @command{g77} driver now prints version information (such as produced
1889 by @kbd{g77 -v}) to @code{stderr} instead of @code{stdout}.
1892 The @samp{.r} suffix now designates a Ratfor source file,
1893 to be preprocessed via the @command{ratfor} command, available
1896 @ifclear USERVISONLY
1898 Fix some aspects of how @command{gcc} determines what kind of
1899 system is being configured and what kinds are supported.
1900 For example, GNU Linux/Alpha ELF systems now are directly
1904 @ifclear USERVISONLY
1906 Improve diagnostics.
1909 @ifclear USERVISONLY
1911 Improve documentation and indexing.
1914 @ifclear USERVISONLY
1916 Include all pertinent files for @code{libf2c} that come
1917 from @code{netlib.bell-labs.com}; give any such files
1918 that aren't quite accurate in @command{g77}'s version of
1919 @code{libf2c} the suffix @samp{.netlib}.
1922 @ifclear USERVISONLY
1924 Reserve @code{INTEGER(KIND=0)} for future use.
1928 @c 1997-02-28: 0.5.20 released.
1932 The @option{-fno-typeless-boz} option is now the default.
1934 This option specifies that non-decimal-radix
1935 constants using the prefixed-radix form (such as @samp{Z'1234'})
1936 are to be interpreted as @code{INTEGER(KIND=1)} constants.
1937 Specify @option{-ftypeless-boz} to cause such
1938 constants to be interpreted as typeless.
1940 (Version 0.5.19 introduced @option{-fno-typeless-boz} and
1944 @xref{Fortran Dialect Options,,Options Controlling Fortran Dialect},
1945 for information on the @option{-ftypeless-boz} option.
1949 Options @option{-ff90-intrinsics-enable} and
1950 @option{-fvxt-intrinsics-enable} now are the
1953 Some programs might use names that clash with
1954 intrinsic names defined (and now enabled) by these
1955 options or by the new @code{libU77} intrinsics.
1956 Users of such programs might need to compile them
1957 differently (using, for example, @option{-ff90-intrinsics-disable})
1958 or, better yet, insert appropriate @code{EXTERNAL}
1959 statements specifying that these names are not intended
1960 to be names of intrinsics.
1963 The @code{ALWAYS_FLUSH} macro is no longer defined when
1964 building @code{libf2c}, which should result in improved
1965 I/O performance, especially over NFS.
1967 @emph{Note:} If you have code that depends on the behavior
1968 of @code{libf2c} when built with @code{ALWAYS_FLUSH} defined,
1969 you will have to modify @code{libf2c} accordingly before
1970 building it from this and future versions of @command{g77}.
1973 @xref{Output Assumed To Flush}, for more information.
1977 Dave Love's implementation of @code{libU77} has been
1978 added to the version of @code{libf2c} distributed with
1979 and built as part of @command{g77}.
1980 @command{g77} now knows about the routines in this library
1984 New option @option{-fvxt} specifies that the
1985 source file is written in VXT Fortran, instead of GNU Fortran.
1988 @xref{VXT Fortran}, for more information on the constructs
1989 recognized when the @option{-fvxt} option is specified.
1993 The @option{-fvxt-not-f90} option has been deleted,
1994 along with its inverse, @option{-ff90-not-vxt}.
1996 If you used one of these deleted options, you should
1997 re-read the pertinent documentation to determine which
1998 options, if any, are appropriate for compiling your
1999 code with this version of @command{g77}.
2002 @xref{Other Dialects}, for more information.
2006 The @option{-fugly} option now issues a warning, as it
2007 likely will be removed in a future version.
2009 (Enabling all the @option{-fugly-*} options is unlikely
2010 to be feasible, or sensible, in the future,
2011 so users should learn to specify only those
2012 @option{-fugly-*} options they really need for a
2013 particular source file.)
2016 The @option{-fugly-assumed} option, introduced in
2017 version 0.5.19, has been changed to
2018 better accommodate old and new code.
2021 @xref{Ugly Assumed-Size Arrays}, for more information.
2024 @ifclear USERVISONLY
2026 Make a number of fixes to the @command{g77} front end and
2027 the @command{gcc} back end to better support Alpha (AXP)
2029 This includes providing at least one bug-fix to the
2030 @command{gcc} back end for Alphas.
2034 Related to supporting Alpha (AXP) machines, the @code{LOC()}
2035 intrinsic and @code{%LOC()} construct now return
2036 values of @code{INTEGER(KIND=0)} type,
2037 as defined by the GNU Fortran language.
2039 This type is wide enough
2040 (holds the same number of bits)
2041 as the character-pointer type on the machine.
2043 On most machines, this won't make a difference,
2044 whereas, on Alphas and other systems with 64-bit pointers,
2045 the @code{INTEGER(KIND=0)} type is equivalent to @code{INTEGER(KIND=2)}
2046 (often referred to as @code{INTEGER*8})
2047 instead of the more common @code{INTEGER(KIND=1)}
2048 (often referred to as @code{INTEGER*4}).
2051 Emulate @code{COMPLEX} arithmetic in the @command{g77} front
2052 end, to avoid bugs in @code{complex} support in the
2053 @command{gcc} back end.
2054 New option @option{-fno-emulate-complex}
2055 causes @command{g77} to revert the 0.5.19 behavior.
2057 @ifclear USERVISONLY
2059 Fix bug whereby @samp{REAL A(1)}, for example, caused
2060 a compiler crash if @option{-fugly-assumed} was in effect
2061 and @var{A} was a local (automatic) array.
2062 That case is no longer affected by the new
2063 handling of @option{-fugly-assumed}.
2066 @ifclear USERVISONLY
2068 Fix @command{g77} command driver so that @samp{g77 -o foo.f}
2069 no longer deletes @file{foo.f} before issuing other
2070 diagnostics, and so the @option{-x} option is properly
2074 @ifclear USERVISONLY
2076 Enable inlining of subroutines and functions by the @command{gcc}
2078 This works as it does for @command{gcc} itself---program units
2079 may be inlined for invocations that follow them in the same
2080 program unit, as long as the appropriate compile-time
2081 options are specified.
2085 Dummy arguments are no longer assumed to potentially alias
2087 other dummy arguments or @code{COMMON} areas when any of
2088 these are defined (assigned to) by Fortran code.
2090 This can result in faster and/or smaller programs when
2091 compiling with optimization enabled, though on some
2092 systems this effect is observed only when @option{-fforce-addr}
2095 New options @option{-falias-check}, @option{-fargument-alias},
2096 @option{-fargument-noalias},
2097 and @option{-fno-argument-noalias-global} control the
2098 way @command{g77} handles potential aliasing.
2101 @xref{Aliasing Assumed To Work}, for detailed information on why the
2102 new defaults might result in some programs no longer working the way they
2103 did when compiled by previous versions of @command{g77}.
2106 @ifclear USERVISONLY
2108 The @code{CONJG()} and @code{DCONJG()} intrinsics now
2109 are compiled in-line.
2112 @ifclear USERVISONLY
2114 The bug-fix for 0.5.19.1 has been re-done.
2115 The @command{g77} compiler has been changed back to
2116 assume @code{libf2c} has no aliasing problems in
2117 its implementations of the @code{COMPLEX} (and
2118 @code{DOUBLE COMPLEX}) intrinsics.
2119 The @code{libf2c} has been changed to have no such
2122 As a result, 0.5.20 is expected to offer improved performance
2123 over 0.5.19.1, perhaps as good as 0.5.19 in most
2124 or all cases, due to this change alone.
2126 @emph{Note:} This change requires version 0.5.20 of
2127 @code{libf2c}, at least, when linking code produced
2128 by any versions of @command{g77} other than 0.5.19.1.
2129 Use @samp{g77 -v} to determine the version numbers
2130 of the @code{libF77}, @code{libI77}, and @code{libU77}
2131 components of the @code{libf2c} library.
2132 (If these version numbers are not printed---in
2133 particular, if the linker complains about unresolved
2134 references to names like @samp{g77__fvers__}---that
2135 strongly suggests your installation has an obsolete
2136 version of @code{libf2c}.)
2140 New option @option{-fugly-assign} specifies that the
2141 same memory locations are to be used to hold the
2142 values assigned by both statements @samp{I = 3} and
2143 @samp{ASSIGN 10 TO I}, for example.
2144 (Normally, @command{g77} uses a separate memory location
2145 to hold assigned statement labels.)
2148 @xref{Ugly Assigned Labels}, for more information.
2152 @code{FORMAT} and @code{ENTRY} statements now are allowed to
2153 precede @code{IMPLICIT NONE} statements.
2155 @ifclear USERVISONLY
2157 Produce diagnostic for unsupported @code{SELECT CASE} on
2158 @code{CHARACTER} type, instead of crashing, at compile time.
2161 @ifclear USERVISONLY
2163 Fix crashes involving diagnosed or invalid code.
2166 @ifclear USERVISONLY
2168 Change approach to building @code{libf2c} archive
2169 (@file{libf2c.a}) so that members are added to it
2170 only when truly necessary, so the user that installs
2171 an already-built @command{g77} doesn't need to have write
2172 access to the build tree (whereas the user doing the
2173 build might not have access to install new software
2177 @ifclear USERVISONLY
2179 Support @command{gcc} version 2.7.2.2
2180 (modified by @command{g77} into version 2.7.2.2.f.2),
2182 support for prior versions of @command{gcc}.
2185 @ifclear USERVISONLY
2187 Upgrade to @code{libf2c} as of 1997-02-08, and
2188 fix up some of the build procedures.
2191 @ifclear USERVISONLY
2193 Improve general build procedures for @command{g77},
2194 fixing minor bugs (such as deletion of any file
2195 named @file{f771} in the parent directory of @code{gcc/}).
2199 Enable full support of @code{INTEGER(KIND=2)}
2200 (often referred to as @code{INTEGER*8})
2202 @code{libf2c} and @file{f2c.h} so that @command{f2c} users
2203 may make full use of its features via the @command{g77}
2204 version of @file{f2c.h} and the @code{INTEGER(KIND=2)}
2205 support routines in the @command{g77} version of @code{libf2c}.
2208 Improve @command{g77} driver and @code{libf2c} so that @samp{g77 -v}
2209 yields version information on the library.
2212 The @code{SNGL} and @code{FLOAT} intrinsics now are
2213 specific intrinsics, instead of synonyms for the
2214 generic intrinsic @code{REAL}.
2217 New intrinsics have been added.
2218 These are @code{REALPART}, @code{IMAGPART},
2220 @code{LONG}, and @code{SHORT}.
2223 A new group of intrinsics, @code{gnu}, has been added
2224 to contain the new @code{REALPART}, @code{IMAGPART},
2225 and @code{COMPLEX} intrinsics.
2226 An old group, @code{dcp}, has been removed.
2229 Complain about industry-wide ambiguous references
2230 @samp{REAL(@var{expr})} and @samp{AIMAG(@var{expr})},
2231 where @var{expr} is @code{DOUBLE COMPLEX} (or any
2232 complex type other than @code{COMPLEX}), unless
2233 @option{-ff90} option specifies Fortran 90 interpretation
2234 or new @option{-fugly-complex} option, in conjunction with
2235 @option{-fnot-f90}, specifies @command{f2c} interpretation.
2237 @ifclear USERVISONLY
2239 Make improvements to diagnostics.
2242 @ifclear USERVISONLY
2244 Speed up compiler a bit.
2247 @ifclear USERVISONLY
2249 Improvements to documentation and indexing, including
2250 a new chapter containing information on one, later
2251 more, diagnostics that users are directed to pull
2252 up automatically via a message in the diagnostic itself.
2254 (Hence the menu item @code{M} for the node
2255 @code{Diagnostics} in the top-level menu of
2256 the Info documentation.)
2260 @ifclear DOC-OLDNEWS
2261 @heading In previous versions:
2263 Information on previous versions is archived
2264 in @file{@value{path-g77}/news.texi}
2265 following the test of the @code{DOC-OLDNEWS} macro.
2269 @c 1997-02-01: 0.5.19.1 released.
2270 @heading In 0.5.19.1:
2273 Code-generation bugs afflicting operations on complex
2274 data have been fixed.
2276 These bugs occurred when assigning the result of an
2277 operation to a complex variable (or array element)
2278 that also served as an input to that operation.
2280 The operations affected by this bug were: @code{CONJG()},
2281 @code{DCONJG()}, @code{CCOS()}, @code{CDCOS()},
2282 @code{CLOG()}, @code{CDLOG()}, @code{CSIN()}, @code{CDSIN()},
2283 @code{CSQRT()}, @code{CDSQRT()}, complex division, and
2284 raising a @code{DOUBLE COMPLEX} operand to an @code{INTEGER}
2286 (The related generic and @samp{Z}-prefixed intrinsics,
2287 such as @code{ZSIN()}, also were affected.)
2289 For example, @samp{C = CSQRT(C)}, @samp{Z = Z/C}, and @samp{Z = Z**I}
2290 (where @samp{C} is @code{COMPLEX} and @samp{Z} is
2291 @code{DOUBLE COMPLEX}) have been fixed.
2294 @c 1996-12-07: 0.5.19 released.
2298 Fix @code{FORMAT} statement parsing so negative values for
2299 specifiers such as @code{P} (e.g. @samp{FORMAT(-1PF8.1)})
2300 are correctly processed as negative.
2303 Fix @code{SIGNAL} intrinsic so it once again accepts a
2304 procedure as its second argument.
2307 A temporary kludge option provides bare-bones information on
2308 @code{COMMON} and @code{EQUIVALENCE} members at debug time.
2311 New @option{-fonetrip} option specifies FORTRAN-66-style
2312 one-trip @code{DO} loops.
2315 New @option{-fno-silent} option causes names of program units
2316 to be printed as they are compiled, in a fashion similar to
2317 UNIX @command{f77} and @command{f2c}.
2320 New @option{-fugly-assumed} option specifies that arrays
2321 dimensioned via @samp{DIMENSION X(1)}, for example, are to be
2322 treated as assumed-size.
2325 New @option{-fno-typeless-boz} option specifies that non-decimal-radix
2326 constants using the prefixed-radix form (such as @samp{Z'1234'})
2327 are to be interpreted as @code{INTEGER} constants.
2330 New @option{-ff66} option is a ``shorthand'' option that specifies
2331 behaviors considered appropriate for FORTRAN 66 programs.
2334 New @option{-ff77} option is a ``shorthand'' option that specifies
2335 behaviors considered appropriate for UNIX @command{f77} programs.
2338 New @option{-fugly-comma} and @option{-fugly-logint} options provided
2339 to perform some of what @option{-fugly} used to do.
2340 @option{-fugly} and @option{-fno-ugly} are now ``shorthand'' options,
2341 in that they do nothing more than enable (or disable) other
2342 @option{-fugly-*} options.
2345 Fix parsing of assignment statements involving targets that
2346 are substrings of elements of @code{CHARACTER} arrays having
2347 names such as @samp{READ}, @samp{WRITE}, @samp{GOTO}, and
2348 @samp{REALFUNCTIONFOO}.
2351 Fix crashes involving diagnosed code.
2354 Fix handling of local @code{EQUIVALENCE} areas so certain cases
2355 of valid Fortran programs are not misdiagnosed as improperly
2356 extending the area backwards.
2359 Support @command{gcc} version 2.7.2.1.
2362 Upgrade to @code{libf2c} as of 1996-09-26, and
2363 fix up some of the build procedures.
2366 Change code generation for list-directed I/O so it allows
2367 for new versions of @code{libf2c} that might return nonzero
2368 status codes for some operations previously assumed to always
2371 This change not only affects how @code{IOSTAT=} variables
2372 are set by list-directed I/O, it also affects whether
2373 @code{END=} and @code{ERR=} labels are reached by these
2377 Add intrinsic support for new @code{FTELL} and @code{FSEEK}
2378 procedures in @code{libf2c}.
2381 Modify @code{fseek_()} in @code{libf2c} to be more portable
2382 (though, in practice, there might be no systems where this
2383 matters) and to catch invalid @code{whence} arguments.
2386 Some useless warnings from the @option{-Wunused} option have
2390 Fix a problem building the @file{f771} executable
2391 on AIX systems by linking with the @option{-bbigtoc} option.
2394 Abort configuration if @command{gcc} has not been patched
2395 using the patch file provided in the @file{gcc/f/gbe/}
2399 Add options @option{--help} and @option{--version} to the
2400 @command{g77} command, to conform to GNU coding guidelines.
2401 Also add printing of @command{g77} version number when
2402 the @option{--verbose} (@option{-v}) option is used.
2405 Change internally generated name for local @code{EQUIVALENCE}
2406 areas to one based on the alphabetically sorted first name
2407 in the list of names for entities placed at the beginning
2411 Improvements to documentation and indexing.
2414 @c 1996-04-01: 0.5.18 released.
2418 Add some rudimentary support for @code{INTEGER*1},
2419 @code{INTEGER*2}, @code{INTEGER*8},
2420 and their @code{LOGICAL} equivalents.
2421 (This support works on most, maybe all, @command{gcc} targets.)
2423 Thanks to Scott Snyder (@email{snyder@@d0sgif.fnal.gov})
2424 for providing the patch for this!
2426 Among the missing elements from the support for these
2427 features are full intrinsic support and constants.
2430 Add some rudimentary support for the @code{BYTE} and
2431 @code{WORD} type-declaration statements.
2432 @code{BYTE} corresponds to @code{INTEGER*1},
2433 while @code{WORD} corresponds to @code{INTEGER*2}.
2435 Thanks to Scott Snyder (@email{snyder@@d0sgif.fnal.gov})
2436 for providing the patch for this!
2439 The compiler code handling intrinsics has been largely
2440 rewritten to accommodate the new types.
2441 No new intrinsics or arguments for existing
2442 intrinsics have been added, so there is, at this
2443 point, no intrinsic to convert to @code{INTEGER*8},
2447 Support automatic arrays in procedures.
2450 Reduce space/time requirements for handling large
2451 @emph{sparsely} initialized aggregate arrays.
2452 This improvement applies to only a subset of
2453 the general problem to be addressed in 0.6.
2456 Treat initial values of zero as if they weren't
2457 specified (in DATA and type-declaration statements).
2458 The initial values will be set to zero anyway, but the amount
2459 of compile time processing them will be reduced,
2460 in some cases significantly (though, again, this
2461 is only a subset of the general problem to be
2464 A new option, @option{-fzeros}, is introduced to
2465 enable the traditional treatment of zeros as any
2469 With @option{-ff90} in force, @command{g77} incorrectly
2470 interpreted @samp{REAL(Z)} as returning a @code{REAL}
2471 result, instead of as a @code{DOUBLE PRECISION}
2473 (Here, @samp{Z} is @code{DOUBLE COMPLEX}.)
2475 With @option{-fno-f90} in force, the interpretation remains
2476 unchanged, since this appears to be how at least some
2477 F77 code using the @code{DOUBLE COMPLEX} extension expected
2480 Essentially, @samp{REAL(Z)} in F90 is the same as
2481 @samp{DBLE(Z)}, while in extended F77, it appears to
2482 be the same as @samp{REAL(REAL(Z))}.
2485 An expression involving exponentiation, where both operands
2486 were type @code{INTEGER} and the right-hand operand
2487 was negative, was erroneously evaluated.
2490 Fix bugs involving @code{DATA} implied-@code{DO} constructs
2491 (these involved an errant diagnostic and a crash, both on good
2492 code, one involving subsequent statement-function definition).
2495 Close @code{INCLUDE} files after processing them, so compiling source
2496 files with lots of @code{INCLUDE} statements does not result in
2497 being unable to open @code{INCLUDE} files after all the available
2498 file descriptors are used up.
2501 Speed up compiling, especially of larger programs, and perhaps
2502 slightly reduce memory utilization while compiling (this is
2503 @emph{not} the improvement planned for 0.6 involving large aggregate
2504 areas)---these improvements result from simply turning
2505 off some low-level code to do self-checking that hasn't been
2506 triggered in a long time.
2509 Introduce three new options that
2510 implement optimizations in the @command{gcc} back end (GBE).
2511 These options are @option{-fmove-all-movables}, @option{-freduce-all-givs},
2512 and @option{-frerun-loop-opt}, which are enabled, by default,
2513 for Fortran compilations.
2514 These optimizations are intended to help toon Fortran programs.
2517 Patch the GBE to do a better job optimizing certain
2518 kinds of references to array elements.
2521 Due to patches to the GBE, the version number of @command{gcc}
2522 also is patched to make it easier to manage installations,
2523 especially useful if it turns out a @command{g77} change to the
2526 The @command{g77}-modified version number is the @command{gcc}
2527 version number with the string @samp{.f.@var{n}} appended,
2528 where @samp{f} identifies the version as enhanced for
2529 Fortran, and @var{n} is @samp{1} for the first Fortran
2530 patch for that version of @command{gcc}, @samp{2} for the
2533 So, this introduces version 2.7.2.f.1 of @command{gcc}.
2536 Make several improvements and fixes to diagnostics, including
2537 the removal of two that were inappropriate or inadequate.
2540 Warning about two successive arithmetic operators, produced
2541 by @option{-Wsurprising}, now produced @emph{only} when both
2542 operators are, indeed, arithmetic (not relational/boolean).
2545 @option{-Wsurprising} now warns about the remaining cases
2546 of using non-integral variables for implied-@code{DO}
2547 loops, instead of these being rejected unless @option{-fpedantic}
2548 or @option{-fugly} specified.
2551 Allow @code{SAVE} of a local variable or array, even after
2552 it has been given an initial value via @code{DATA}, for example.
2555 Introduce an Info version of @command{g77} documentation, which
2556 supersedes @file{gcc/f/CREDITS}, @file{gcc/f/DOC}, and
2557 @file{gcc/f/PROJECTS}.
2558 These files will be removed in a future release.
2559 The files @file{gcc/f/BUGS}, @file{gcc/f/INSTALL}, and
2560 @file{gcc/f/NEWS} now are automatically built from
2561 the texinfo source when distributions are made.
2563 This effort was inspired by a first pass at translating
2564 @file{g77-0.5.16/f/DOC} that was contributed to Craig by
2565 David Ronis (@email{ronis@@onsager.chem.mcgill.ca}).
2568 New @option{-fno-second-underscore} option to specify
2569 that, when @option{-funderscoring} is in effect, a second
2570 underscore is not to be appended to Fortran names already
2571 containing an underscore.
2574 Change the way iterative @code{DO} loops work to follow
2576 In particular, calculation of the iteration count is
2577 still done by converting the start, end, and increment
2578 parameters to the type of the @code{DO} variable, but
2579 the result of the calculation is always converted to
2580 the default @code{INTEGER} type.
2582 (This should have no effect on existing code compiled
2583 by @command{g77}, but code written to assume that use
2584 of a @emph{wider} type for the @code{DO} variable
2585 will result in an iteration count being fully calculated
2586 using that wider type (wider
2587 than default @code{INTEGER}) must be rewritten.)
2590 Support @command{gcc} version 2.7.2.
2593 Upgrade to @code{libf2c} as of 1996-03-23, and
2594 fix up some of the build procedures.
2596 Note that the email addresses related to @command{f2c}
2597 have changed---the distribution site now is
2598 named @code{netlib.bell-labs.com}, and the
2599 maintainer's new address is @email{dmg@@bell-labs.com}.
2602 @c 1995-11-18: 0.5.17 released.
2606 @strong{Fix serious bug} in @samp{g77 -v} command that can cause removal of a
2607 system's @file{/dev/null} special file if run by user @code{root}.
2609 @strong{All users} of version 0.5.16 should ensure that
2610 they have not removed @file{/dev/null} or replaced it with an ordinary
2611 file (e.g. by comparing the output of @samp{ls -l /dev/null} with
2612 @samp{ls -l /dev/zero}.
2613 If the output isn't basically the
2614 same, contact your system
2615 administrator about restoring @file{/dev/null} to its proper status).
2617 This bug is particularly insidious because removing @file{/dev/null} as
2618 a special file can go undetected for quite a while, aside from
2619 various applications and programs exhibiting sudden, strange
2622 I sincerely apologize for not realizing the
2623 implications of the fact that when @samp{g77 -v} runs the @command{ld} command
2624 with @samp{-o /dev/null} that @command{ld} tries to @emph{remove} the executable
2625 it is supposed to build (especially if it reports unresolved
2626 references, which it should in this case)!
2629 Fix crash on @samp{CHARACTER*(*) FOO} in a main or block data program unit.
2632 Fix crash that can occur when diagnostics given outside of any
2633 program unit (such as when input file contains @samp{@@foo}).
2636 Fix crashes, infinite loops (hangs), and such involving diagnosed code.
2639 Fix @code{ASSIGN}'ed variables so they can be @code{SAVE}'d or dummy arguments,
2640 and issue clearer error message in cases where target of @code{ASSIGN}
2641 or @code{ASSIGN}ed @code{GOTO}/@code{FORMAT} is too small (which should
2645 Make @code{libf2c} build procedures work on more systems again by
2646 eliminating unnecessary invocations of @samp{ld -r -x} and @command{mv}.
2649 Fix omission of @option{-funix-intrinsics-@dots{}} options in list of permitted
2650 options to compiler.
2653 Fix failure to always diagnose missing type declaration for
2654 @code{IMPLICIT NONE}.
2657 Fix compile-time performance problem (which could sometimes
2658 crash the compiler, cause a hang, or whatever, due to a bug
2659 in the back end) involving exponentiation with a large @code{INTEGER}
2660 constant for the right-hand operator (e.g. @samp{I**32767}).
2663 Fix build procedures so cross-compiling @command{g77} (the @command{fini}
2664 utility in particular) is properly built using the host compiler.
2667 Add new @option{-Wsurprising} option to warn about constructs that are
2668 interpreted by the Fortran standard (and @command{g77}) in ways that
2669 are surprising to many programmers.
2672 Add @code{ERF()} and @code{ERFC()} as generic intrinsics mapping to existing
2673 @code{ERF}/@code{DERF} and @code{ERFC}/@code{DERFC} specific intrinsics.
2675 @emph{Note:} You should
2676 specify @samp{INTRINSIC ERF,ERFC} in any code where you might use
2677 these as generic intrinsics, to improve likelihood of diagnostics
2678 (instead of subtle run-time bugs) when using a compiler that
2679 doesn't support these as intrinsics (e.g. @command{f2c}).
2682 Remove from @option{-fno-pedantic} the diagnostic about @code{DO}
2683 with non-@code{INTEGER} index variable; issue that under
2684 @option{-Wsurprising} instead.
2687 Clarify some diagnostics that say things like ``ignored'' when that's
2691 Clarify diagnostic on use of @code{.EQ.}/@code{.NE.} on @code{LOGICAL}
2695 Minor improvements to code generation for various operations on
2696 @code{LOGICAL} operands.
2699 Minor improvement to code generation for some @code{DO} loops on some
2703 Support @command{gcc} version 2.7.1.
2706 Upgrade to @code{libf2c} as of 1995-11-15.
2709 @c 1995-08-30: 0.5.16 released.
2713 Fix a code-generation bug involving complicated @code{EQUIVALENCE} statements
2714 not involving @code{COMMON}.
2717 Fix code-generation bugs involving invoking ``gratis'' library procedures
2718 in @code{libf2c} from code compiled with @option{-fno-f2c} by making these
2719 procedures known to @command{g77} as intrinsics (not affected by -fno-f2c).
2720 This is known to fix code invoking @code{ERF()}, @code{ERFC()},
2721 @code{DERF()}, and @code{DERFC()}.
2724 Update @code{libf2c} to include netlib patches through 1995-08-16, and
2725 @code{#define} @code{WANT_LEAD_0} to 1 to make @command{g77}-compiled code more
2726 consistent with other Fortran implementations by outputting
2727 leading zeros in formatted and list-directed output.
2730 Fix a code-generation bug involving adjustable dummy arrays with high
2731 bounds whose primaries are changed during procedure execution, and
2732 which might well improve code-generation performance for such arrays
2733 compared to @command{f2c} plus @command{gcc} (but apparently only when using
2734 @file{gcc-2.7.0} or later).
2737 Fix a code-generation bug involving invocation of @code{COMPLEX} and
2738 @code{DOUBLE COMPLEX} @code{FUNCTION}s and doing @code{COMPLEX} and
2739 @code{DOUBLE COMPLEX} divides, when the result
2740 of the invocation or divide is assigned directly to a variable
2741 that overlaps one or more of the arguments to the invocation or divide.
2744 Fix crash by not generating new optimal code for @samp{X**I} if @samp{I} is
2745 nonconstant and the expression is used to dimension a dummy
2746 array, since the @command{gcc} back end does not support the necessary
2747 mechanics (and the @command{gcc} front end rejects the equivalent
2748 construct, as it turns out).
2751 Fix crash on expressions like @samp{COMPLEX**INTEGER}.
2754 Fix crash on expressions like @samp{(1D0,2D0)**2}, i.e. raising a
2755 @code{DOUBLE COMPLEX} constant to an @code{INTEGER} constant power.
2758 Fix crashes and such involving diagnosed code.
2761 Diagnose, instead of crashing on, statement function definitions
2762 having duplicate dummy argument names.
2765 Fix bug causing rejection of good code involving statement function
2769 Fix bug resulting in debugger not knowing size of local equivalence
2770 area when any member of area has initial value (via @code{DATA},
2774 Fix installation bug that prevented installation of @command{g77} driver.
2775 Provide for easy selection of whether to install copy of @command{g77}
2776 as @command{f77} to replace the broken code.
2779 Fix @command{gcc} driver (affects @command{g77} thereby) to not
2780 gratuitously invoke the
2781 @code{f771} program (e.g. when @option{-E} is specified).
2784 Fix diagnostic to point to correct source line when it immediately
2785 follows an @code{INCLUDE} statement.
2788 Support more compiler options in @command{gcc}/@command{g77} when
2789 compiling Fortran files.
2790 These options include @option{-p}, @option{-pg}, @option{-aux-info}, @option{-P},
2791 correct setting of version-number macros for preprocessing, full
2792 recognition of @option{-O0}, and
2793 automatic insertion of configuration-specific linker specs.
2796 Add new intrinsics that interface to existing routines in @code{libf2c}:
2797 @code{ABORT}, @code{DERF}, @code{DERFC}, @code{ERF}, @code{ERFC}, @code{EXIT},
2798 @code{FLUSH}, @code{GETARG}, @code{GETENV}, @code{IARGC},
2799 @code{SIGNAL}, and @code{SYSTEM}.
2800 Note that @code{ABORT}, @code{EXIT}, @code{FLUSH}, @code{SIGNAL}, and
2801 @code{SYSTEM} are intrinsic subroutines, not functions (since they
2802 have side effects), so to get the return values from @code{SIGNAL}
2803 and @code{SYSTEM}, append a final argument specifying an @code{INTEGER}
2804 variable or array element (e.g. @samp{CALL SYSTEM('rm foo',ISTAT)}).
2807 Add new intrinsic group named @code{unix} to contain the new intrinsics,
2808 and by default enable this new group.
2811 Move @code{LOC()} intrinsic out of the @code{vxt} group to the new
2815 Improve @command{g77} so that @samp{g77 -v} by itself (or with
2816 certain other options, including @option{-B}, @option{-b}, @option{-i},
2817 @option{-nostdlib}, and @option{-V}) reports lots more useful
2818 version info, and so that long-form options @command{gcc} accepts are
2819 understood by @command{g77} as well (even in truncated, unambiguous forms).
2822 Add new @command{g77} option @option{--driver=name} to specify driver when
2823 default, @command{gcc}, isn't appropriate.
2826 Add support for @samp{#} directives (as output by the preprocessor) in the
2827 compiler, and enable generation of those directives by the
2828 preprocessor (when compiling @samp{.F} files) so diagnostics and debugging
2829 info are more useful to users of the preprocessor.
2832 Produce better diagnostics, more like @command{gcc}, with info such as
2833 @samp{In function `foo':} and @samp{In file included from...:}.
2836 Support @command{gcc}'s @option{-fident} and @option{-fno-ident} options.
2839 When @option{-Wunused} in effect, don't warn about local variables used as
2840 statement-function dummy arguments or @code{DATA} implied-@code{DO} iteration
2841 variables, even though, strictly speaking, these are not uses
2842 of the variables themselves.
2845 When @samp{-W -Wunused} in effect, don't warn about unused dummy arguments
2846 at all, since there's no way to turn this off for individual
2847 cases (@command{g77} might someday start warning about these)---applies
2848 to @command{gcc} versions 2.7.0 and later, since earlier versions didn't
2849 warn about unused dummy arguments.
2852 New option @option{-fno-underscoring} that inhibits transformation of names
2853 (by appending one or two underscores) so users may experiment
2854 with implications of such an environment.
2857 Minor improvement to @file{gcc/f/info} module to make it easier to build
2858 @command{g77} using the native (non-@command{gcc}) compiler on certain machines
2859 (but definitely not all machines nor all non-@command{gcc} compilers).
2861 do not report bugs showing problems compilers have with
2862 macros defined in @file{gcc/f/target.h} and used in places like
2863 @file{gcc/f/expr.c}.
2866 Add warning to be printed for each invocation of the compiler
2867 if the target machine @code{INTEGER}, @code{REAL}, or @code{LOGICAL} size
2869 since @command{g77} is known to not work well for such cases.
2872 Lots of new documentation (though work is still needed to put it into
2873 canonical GNU format).
2876 Build @code{libf2c} with @option{-g0}, not @option{-g2}, in effect
2877 (by default), to produce
2878 smaller library without lots of debugging clutter.
2881 @c 1995-05-19: 0.5.15 released.
2885 Fix bad code generation involving @samp{X**I} and temporary, internal variables
2886 generated by @command{g77} and the back end (such as for @code{DO} loops).
2889 Fix crash given @samp{CHARACTER A;DATA A/.TRUE./}.
2892 Replace crash with diagnostic given @samp{CHARACTER A;DATA A/1.0/}.
2895 Fix crash or other erratic behavior when null character constant
2896 (@samp{''}) is encountered.
2899 Fix crash or other erratic behavior involving diagnosed code.
2902 Fix code generation for external functions returning type @code{REAL} when
2903 the @option{-ff2c} option is in force (which it is by default) so that
2904 @command{f2c} compatibility is indeed provided.
2907 Disallow @samp{COMMON I(10)} if @samp{I} has previously been specified
2908 with an array declarator.
2911 New @option{-ffixed-line-length-@var{n}} option, where @var{n} is the
2913 of a typical fixed-form line, defaulting to 72 columns, such
2914 that characters beyond column @var{n} are ignored, or @var{n} is @samp{none},
2915 meaning no characters are ignored.
2916 does not affect lines
2917 with @samp{&} in column 1, which are always processed as if
2918 @option{-ffixed-line-length-none} was in effect.
2921 No longer generate better code for some kinds of array references,
2922 as @command{gcc} back end is to be fixed to do this even better, and it
2923 turned out to slow down some code in some cases after all.
2926 In @code{COMMON} and @code{EQUIVALENCE} areas with any members given initial
2927 values (e.g. via @code{DATA}), uninitialized members now always
2928 initialized to binary zeros (though this is not required by
2929 the standard, and might not be done in future versions
2931 Previously, in some @code{COMMON}/@code{EQUIVALENCE} areas
2932 (essentially those with members of more than one type), the
2933 uninitialized members were initialized to spaces, to
2934 cater to @code{CHARACTER} types, but it seems no existing code expects
2935 that, while much existing code expects binary zeros.
2941 Don't emit bad code when low bound of adjustable array is nonconstant
2942 and thus might vary as an expression at run time.
2945 Emit correct code for calculation of number of trips in @code{DO} loops
2947 where the loop should not execute at all.
2948 (This bug affected cases
2949 where the difference between the begin and end values was less
2950 than the step count, though probably not for floating-point cases.)
2953 Fix crash when extra parentheses surround item in
2954 @code{DATA} implied-@code{DO} list.
2957 Fix crash over minor internal inconsistencies in handling diagnostics,
2958 just substitute dummy strings where necessary.
2961 Fix crash on some systems when compiling call to @code{MVBITS()} intrinsic.
2964 Fix crash on array assignment @samp{TYPE@var{ddd}(@dots{})=@dots{}}, where @var{ddd}
2965 is a string of one or more digits.
2968 Fix crash on @code{DCMPLX()} with a single @code{INTEGER} argument.
2971 Fix various crashes involving code with diagnosed errors.
2974 Support @option{-I} option for @code{INCLUDE} statement, plus @command{gcc}'s
2975 @file{header.gcc} facility for handling systems like MS-DOS.
2978 Allow @code{INCLUDE} statement to be continued across multiple lines,
2979 even allow it to coexist with other statements on the same line.
2982 Incorporate Bellcore fixes to @code{libf2c} through 1995-03-15---this
2983 fixes a bug involving infinite loops reading EOF with empty list-directed
2987 Remove all the @command{g77}-specific auto-configuration scripts, code,
2989 except for temporary substitutes for bsearch() and strtoul(), as
2990 too many configure/build problems were reported in these areas.
2991 People will have to fix their systems' problems themselves, or at
2992 least somewhere other than @command{g77}, which expects a working ANSI C
2993 environment (and, for now, a GNU C compiler to compile @command{g77} itself).
2996 Complain if initialized common redeclared as larger in subsequent program
3000 Warn if blank common initialized, since its size can vary and hence
3001 related warnings that might be helpful won't be seen.
3004 New @option{-fbackslash} option, on by default, that causes @samp{\}
3005 within @code{CHARACTER}
3006 and Hollerith constants to be interpreted a la GNU C.
3008 this behavior is somewhat different from @command{f2c}'s, which supports only
3009 a limited subset of backslash (escape) sequences.
3012 Make @option{-fugly-args} the default.
3015 New @option{-fugly-init} option, on by default, that allows typeless/Hollerith
3016 to be specified as initial values for variables or named constants
3017 (@code{PARAMETER}), and also allows character<->numeric conversion in
3018 those contexts---turn off via @option{-fno-ugly-init}.
3021 New @option{-finit-local-zero} option to initialize
3022 local variables to binary zeros.
3023 This does not affect whether they are @code{SAVE}d, i.e. made
3024 automatic or static.
3027 New @option{-Wimplicit} option to warn about implicitly typed variables, arrays,
3029 (Basically causes all program units to default to @code{IMPLICIT NONE}.)
3032 @option{-Wall} now implies @option{-Wuninitialized} as with @command{gcc}
3033 (i.e. unless @option{-O} not specified, since @option{-Wuninitialized}
3034 requires @option{-O}), and implies @option{-Wunused} as well.
3037 @option{-Wunused} no longer gives spurious messages for unused
3038 @code{EXTERNAL} names (since they are assumed to refer to block data
3039 program units, to make use of libraries more reliable).
3042 Support @code{%LOC()} and @code{LOC()} of character arguments.
3045 Support null (zero-length) character constants and expressions.
3048 Support @command{f2c}'s @code{IMAG()} generic intrinsic.
3051 Support @code{ICHAR()}, @code{IACHAR()}, and @code{LEN()} of
3052 character expressions that are valid in assignments but
3053 not normally as actual arguments.
3056 Support @command{f2c}-style @samp{&} in column 1 to mean continuation line.
3059 Allow @code{NAMELIST}, @code{EXTERNAL}, @code{INTRINSIC}, and @code{VOLATILE}
3060 in @code{BLOCK DATA}, even though these are not allowed by the standard.
3063 Allow @code{RETURN} in main program unit.
3066 Changes to Hollerith-constant support to obey Appendix C of the
3071 Now padded on the right with zeros, not spaces.
3074 Hollerith ``format specifications'' in the form of arrays of
3075 non-character allowed.
3078 Warnings issued when non-space truncation occurs when converting
3082 When specified as actual argument, now passed
3083 by reference to @code{INTEGER} (padded on right with spaces if constant
3084 too small, otherwise fully intact if constant wider the @code{INTEGER}
3085 type) instead of by value.
3088 @strong{Warning:} @command{f2c} differs on the
3089 interpretation of @samp{CALL FOO(1HX)}, which it treats exactly the
3090 same as @samp{CALL FOO('X')}, but which the standard and @command{g77} treat
3091 as @samp{CALL FOO(%REF('X '))} (padded with as many spaces as necessary
3092 to widen to @code{INTEGER}), essentially.
3095 Changes and fixes to typeless-constant support:
3099 Now treated as a typeless double-length @code{INTEGER} value.
3102 Warnings issued when overflow occurs.
3105 Padded on the left with zeros when converting
3109 Should be properly aligned and ordered on
3110 the target machine for whatever type it is turned into.
3113 When specified as actual argument, now passed as reference to
3114 a default @code{INTEGER} constant.
3118 @code{%DESCR()} of a non-@code{CHARACTER} expression now passes a pointer to
3119 the expression plus a length for the expression just as if
3120 it were a @code{CHARACTER} expression.
3121 For example, @samp{CALL FOO(%DESCR(D))}, where
3122 @samp{D} is @code{REAL*8}, is the same as @samp{CALL FOO(D,%VAL(8)))}.
3125 Name of multi-entrypoint master function changed to incorporate
3126 the name of the primary entry point instead of a decimal
3127 value, so the name of the master function for @samp{SUBROUTINE X}
3128 with alternate entry points is now @samp{__g77_masterfun_x}.
3131 Remove redundant message about zero-step-count @code{DO} loops.
3134 Clean up diagnostic messages, shortening many of them.
3137 Fix typo in @command{g77} man page.
3140 Clarify implications of constant-handling bugs in @file{f/BUGS}.
3143 Generate better code for @samp{**} operator with a right-hand operand of
3144 type @code{INTEGER}.
3147 Generate better code for @code{SQRT()} and @code{DSQRT()},
3148 also when @option{-ffast-math}
3149 specified, enable better code generation for @code{SIN()} and @code{COS()}.
3152 Generate better code for some kinds of array references.
3155 Speed up lexing somewhat (this makes the compilation phase noticeably