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 2003-01-04
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
193 -ffixed-line-length-none -x f77-cpp-input gives: Warning: unknown
194 register name line-length-none
197 Richard Henderson (@email{rth@@redhat.com}) analyzed and improved the handling
198 of (no-)aliasing information for dummy arguments and improved the optimization
199 of induction variables in unrolled loops.
202 @heading In @code{GCC} 3.2 versus @code{GCC} 3.1:
205 Problem Reports fixed (in chronological order of submission):
208 gcc-3.x does not compile files with suffix .r (RATFOR) [Fixed in 3.2.1]
212 @heading In @code{GCC} 3.1 (formerly known as g77-0.5.27) versus @code{GCC} 3.0:
215 Problem Reports fixed (in chronological order of submission):
218 Data statement initialization with subscript of kind INTEGER*2
220 Reference to intrinsic `ISHFT' invalid
222 Function BESJN(integer,double) problems
224 g77 -pipe -xf77-cpp-input sends output to stdout
226 g77 -h" gives bogus output
228 ICE on valid input using CALL EXIT(%VAL(...))
230 g77 -v -c -xf77-version /dev/null -xnone causes ice
232 BACKSPACE example that doesn't work as of gcc/g77-3.0.x
234 g77 rejects accepted use of INTEGER*2 as type of DATA statement loop index
236 ICE on compiling source with 540 000 000 REAL array
238 ICE on BESJN(integer*8,real)
240 bug in loop unrolling
242 sparc-sun-solaris2.7 gcc-3.1 extra g77 testsuite failures w/-m64
244 Incorrect acces of integer*1 variables on PA
246 Failure of LAPACK test dtest on irix6.5 with -mabi=64 -O2
250 @command{g77} now has its man page generated from the texinfo documentation,
251 to guarantee that it remains up to date.
254 @command{g77} used to reject the following program on 32-bit targets:
257 DIMENSION A(140 000 000)
262 prog.f: In program `prog':
264 DIMENSION A(140 000 000)
266 Array `a' at (^) is too large to handle
268 because 140 000 000 REALs is larger than the largest bit-extent that can be
269 expressed in 32 bits. However, bit-sizes never play a role after offsets
270 have been converted to byte addresses. Therefore this check has been removed,
271 and the limit is now 2 Gbyte of memory (around 530 000 000 REALs).
272 Note: On GNU/Linux systems one has to compile and link programs that occupy
273 more than 1 Gbyte statically, i.e.@: @code{g77 -static ...}.
276 Based on work done by Juergen Pfeifer (@email{juergen.pfeifer@@gmx.net})
277 libf2c is now a shared library. One can still link in all objects with
278 the program by specifying the @option{-static} option.
281 Robert Anderson (@email{rwa@@alumni.princeton.edu}) thought up a two
282 line change that enables g77 to compile such code as:
286 DIMENSION A(N(1),N(2))
290 Note the use of array elements in the bounds of the adjustable array A.
293 George Helffrich (@email{george@@geo.titech.ac.jp}) implemented a change
294 in substring index checking (when specifying @option{-fbounds-check})
295 that permits the use of zero length substrings of the form
299 Based on code developed by Pedro Vazquez (@email{vazquez@@penelope.iqm.unicamp.br}),
300 the @code{libf2c} library is now able to read and write files larger than
301 2 Gbyte on 32-bit target machines, if the operating system supports this.
304 @heading In 0.5.26, @code{GCC} 3.0 versus @code{GCC} 2.95:
307 When a REWIND was issued after a WRITE statement on an unformatted
308 file, the implicit truncation was performed by copying the truncated
309 file to /tmp and copying the result back. This has been fixed by using
310 the @code{ftruncate} OS function. Thanks go to the GAMESS developers
311 for bringing this to our attention.
314 Using options @option{-g}, @option{-ggdb} or @option{-gdwarf[-2]} (where
315 appropriate for your target) now also enables debugging information
316 for COMMON BLOCK and EQUIVALENCE items to be emitted.
317 Thanks go to Andrew Vaught (@email{andy@@xena.eas.asu.edu}) and
318 George Helffrich (@email{george@@geology.bristol.ac.uk}) for
319 fixing this longstanding problem.
322 It is not necessary anymore to use the option @option{-femulate-complex}
323 to compile Fortran code using COMPLEX arithmetic, even on 64-bit machines
324 (like the Alpha). This will improve code generation.
327 INTRINSIC arithmetic functions are now treated as routines that do not
328 depend on anything but their argument(s). This enables further instruction
329 scheduling, because it is known that they cannot read or modify arbitrary
334 Upgrade to @code{libf2c} as of 2000-12-05.
336 This fixes a bug where a namelist containing initialization of LOGICAL
337 items and a variable starting with T or F would be read incorrectly.
340 The @code{TtyNam} intrinsics now set @var{Name} to all spaces (at run time)
341 if the system has no @code{ttyname} implementation available.
344 Upgrade to @code{libf2c} as of 1999-06-28.
346 This fixes a bug whereby
347 input to a @code{NAMELIST} read involving a repeat count,
348 such as @samp{K(5)=10*3},
349 was not properly handled by @code{libf2c}.
350 The first item was written to @samp{K(5)},
351 but the remaining nine were written elsewhere (still within the array),
352 not necessarily starting at @samp{K(6)}.
356 @heading In 0.5.25, @code{GCC} 2.95 (@code{EGCS} 1.2) versus @code{EGCS} 1.1.2:
360 @command{g77} no longer generates bad code for assignments,
361 or other conversions,
362 of @code{REAL} or @code{COMPLEX} constant expressions
363 to type @code{INTEGER(KIND=2)}
364 (often referred to as @code{INTEGER*8}).
366 For example, @samp{INTEGER*8 J; J = 4E10} now works as documented.
371 @command{g77} no longer truncates @code{INTEGER(KIND=2)}
372 (usually @code{INTEGER*8})
373 subscript expressions when evaluating array references
374 on systems with pointers widers than @code{INTEGER(KIND=1)}
380 @command{g77} no longer generates bad code
381 for an assignment to a @code{COMPLEX} variable or array
382 that partially overlaps one or more of the sources
383 of the same assignment
384 (a very rare construction).
385 It now assigns through a temporary,
386 in cases where such partial overlap is deemed possible.
391 @code{libg2c} (@code{libf2c}) no longer loses track
392 of the file being worked on
393 during a @code{BACKSPACE} operation.
398 @code{libg2c} (@code{libf2c}) fixes a bug whereby
399 input to a @code{NAMELIST} read involving a repeat count,
400 such as @samp{K(5)=10*3},
401 was not properly handled by @code{libf2c}.
402 The first item was written to @samp{K(5)},
403 but the remaining nine were written elsewhere (still within the array),
404 not necessarily starting at @samp{K(6)}.
409 @c Tim Prince reported this, regarding the TEST_FPU benchmark.
410 Automatic arrays now seem to be working on HP-UX systems.
415 The @code{Date} intrinsic now returns the correct result
416 on big-endian systems.
421 Fix @command{g77} so it no longer crashes when compiling
422 I/O statements using keywords that define @code{INTEGER} values,
423 such as @samp{IOSTAT=@var{j}},
424 where @var{j} is other than default @code{INTEGER}
425 (such as @code{INTEGER*2}).
426 Instead, it issues a diagnostic.
431 Fix @command{g77} so it properly handles @samp{DATA A/@var{rpt}*@var{val}/},
432 where @var{rpt} is not default @code{INTEGER}, such as @code{INTEGER*2},
433 instead of producing a spurious diagnostic.
434 Also fix @samp{DATA (A(I),I=1,N)},
435 where @samp{N} is not default @code{INTEGER}
436 to work instead of crashing @command{g77}.
441 The @option{-ax} option is now obeyed when compiling Fortran programs.
442 (It is passed to the @file{f771} driver.)
446 The new @option{-fbounds-check} option
447 causes @command{g77} to compile run-time bounds checks
448 of array subscripts, as well as of substring start and end points.
451 @code{libg2c} now supports building as multilibbed library,
452 which provides better support for systems
453 that require options such as @option{-mieee}
457 Source file names with the suffixes @samp{.FOR} and @samp{.FPP}
458 now are recognized by @command{g77}
459 as if they ended in @samp{.for} and @samp{.fpp}, respectively.
462 The order of arguments to the @emph{subroutine} forms of the
463 @code{CTime}, @code{DTime}, @code{ETime}, and @code{TtyNam}
464 intrinsics has been swapped.
465 The argument serving as the returned value
466 for the corresponding function forms
467 now is the @emph{second} argument,
468 making these consistent with the other subroutine forms
469 of @code{libU77} intrinsics.
472 @command{g77} now warns about a reference to an intrinsic
473 that has an interface that is not Year 2000 (Y2K) compliant.
474 Also, @code{libg2c} has been changed to increase the likelihood
475 of catching references to the implementations of these intrinsics
476 using the @code{EXTERNAL} mechanism
477 (which would avoid the new warnings).
480 @xref{Year 2000 (Y2K) Problems}, for more information.
485 @command{g77} now warns about a reference to a function
486 when the corresponding @emph{subsequent} function program unit
487 disagrees with the reference concerning the type of the function.
491 @option{-fno-emulate-complex} is now the default option.
492 This should result in improved performance
493 of code that uses the @code{COMPLEX} data type.
496 @cindex double-precision performance
497 @cindex -malign-double
499 The @option{-malign-double} option
500 now reliably aligns @emph{all} double-precision variables and arrays
501 on Intel x86 targets.
505 Even without the @option{-malign-double} option,
506 @command{g77} reliably aligns local double-precision variables
507 that are not in @code{EQUIVALENCE} areas
508 and not @code{SAVE}'d.
513 @command{g77} now open-codes (``inlines'') division of @code{COMPLEX} operands
514 instead of generating a run-time call to
515 the @code{libf2c} routines @code{c_div} or @code{z_div},
516 unless the @option{-Os} option is specified.
520 @command{g77} no longer generates code to maintain @code{errno},
521 a C-language concept,
522 when performing operations such as the @code{SqRt} intrinsic.
526 @command{g77} developers can temporarily use
527 the @option{-fflatten-arrays} option
528 to compare how the compiler handles code generation
529 using C-like constructs as compared to the
530 Fortran-like method constructs normally used.
535 A substantial portion of the @command{g77} front end's code-generation component
537 It now generates code using facilities more robustly supported
538 by the @command{gcc} back end.
539 One effect of this rewrite is that some codes no longer produce
540 a spurious ``label @var{lab} used before containing binding contour''
545 Support for the @option{-fugly} option has been removed.
549 Improve documentation and indexing,
550 including information on Year 2000 (Y2K) compliance,
551 and providing more information on internals of the front end.
556 Upgrade to @code{libf2c} as of 1999-05-10.
560 @heading In 0.5.24 versus 0.5.23:
562 There is no @command{g77} version 0.5.24 at this time,
564 0.5.24 is the version number designated for bug fixes and,
565 perhaps, some new features added,
567 Version 0.5.23 requires @command{gcc} 2.8.1,
568 as 0.5.24 was planned to require.
570 Due to @code{EGCS} becoming @code{GCC}
571 (which is now an acronym for ``GNU Compiler Collection''),
572 and @code{EGCS} 1.2 becoming officially designated @code{GCC} 2.95,
573 there seems to be no need for an actual 0.5.24 release.
575 To reduce the confusion already resulting from use of 0.5.24
576 to designate @command{g77} versions within @code{EGCS} versions 1.0 and 1.1,
577 as well as in versions of @command{g77} documentation and notices
579 ``mainline'' @command{g77} version numbering resumes
580 at 0.5.25 with @code{GCC} 2.95 (@code{EGCS} 1.2),
581 skipping over 0.5.24 as a placeholder version number.
583 To repeat, there is no @command{g77} 0.5.24, but there is now a 0.5.25.
584 Please remain calm and return to your keypunch units.
586 @c 1999-03-15: EGCS 1.1.2 released.
587 @heading In @code{EGCS} 1.1.2 versus @code{EGCS} 1.1.1:
591 Fix the @code{IDate} intrinsic (VXT) (in @code{libg2c})
592 so the returned year is in the documented, non-Y2K-compliant range
594 instead of being returned as 100 in the year 2000.
597 @xref{IDate Intrinsic (VXT)},
598 for more information.
602 Fix the @code{Date_and_Time} intrinsic (in @code{libg2c})
603 to return the milliseconds value properly
607 Fix the @code{LStat} intrinsic (in @code{libg2c})
608 to return device-ID information properly
612 Improve documentation.
616 @c 1998-12-04: EGCS 1.1.1 released.
617 @heading In @code{EGCS} 1.1.1 versus @code{EGCS} 1.1:
621 Fix @code{libg2c} so it performs an implicit @code{ENDFILE} operation
623 whenever a @code{REWIND} is done.
625 (This bug was introduced in 0.5.23 and @code{egcs} 1.1 in
626 @command{g77}'s version of @code{libf2c}.)
629 Fix @code{libg2c} so it no longer crashes with a spurious diagnostic
630 upon doing any I/O following a direct formatted write.
632 (This bug was introduced in 0.5.23 and @code{egcs} 1.1 in
633 @command{g77}'s version of @code{libf2c}.)
636 Fix @command{g77} so it no longer crashes compiling references
637 to the @code{Rand} intrinsic on some systems.
640 Fix @command{g77} portion of installation process so it works
641 better on some systems
642 (those with shells requiring @samp{else true} clauses
643 on @code{if} constructs
644 for the completion code to be set properly).
648 @c 1998-09-03: EGCS 1.1 released.
649 @heading In @code{EGCS} 1.1 versus @code{EGCS} 1.0.3:
653 Fix bugs in the @code{libU77} intrinsic @code{HostNm}
654 that wrote one byte beyond the end of its @code{CHARACTER}
656 and in the @code{libU77} intrinsics
657 @code{GMTime} and @code{LTime}
658 that overwrote their arguments.
663 Assumed arrays with negative bounds
664 (such as @samp{REAL A(-1:*)})
665 no longer elicit spurious diagnostics from @command{g77},
666 even on systems with pointers having
667 different sizes than integers.
669 This bug is not known to have existed in any
670 recent version of @command{gcc}.
671 It was introduced in an early release of @code{egcs}.
676 Valid combinations of @code{EXTERNAL},
677 passing that external as a dummy argument
678 without explicitly giving it a type,
679 and, in a subsequent program unit,
680 referencing that external as
681 an external function with a different type
682 no longer crash @command{g77}.
687 @code{CASE DEFAULT} no longer crashes @command{g77}.
692 The @option{-Wunused} option no longer issues a spurious
693 warning about the ``master'' procedure generated by
694 @command{g77} for procedures containing @code{ENTRY} statements.
698 Support @samp{FORMAT(I<@var{expr}>)} when @var{expr} is a
699 compile-time constant @code{INTEGER} expression.
702 Fix @command{g77} @option{-g} option so procedures that
703 use @code{ENTRY} can be stepped through, line by line,
707 Allow any @code{REAL} argument to intrinsics
708 @code{Second} and @code{CPU_Time}.
711 Use @code{tempnam}, if available, to open scratch files
712 (as in @samp{OPEN(STATUS='SCRATCH')})
713 so that the @code{TMPDIR} environment variable,
717 @command{g77}'s version of @code{libf2c} separates out
718 the setting of global state
719 (such as command-line arguments and signal handling)
720 from @file{main.o} into distinct, new library
723 This should make it easier to write portable applications
724 that have their own (non-Fortran) @code{main()} routine
725 properly set up the @code{libf2c} environment, even
726 when @code{libf2c} (now @code{libg2c}) is a shared library.
730 @command{g77} no longer installs the @file{f77} command
731 and @file{f77.1} man page
732 in the @file{/usr} or @file{/usr/local} hierarchy,
733 even if the @file{f77-install-ok} file exists
734 in the source or build directory.
735 See the installation documentation for more information.
740 @command{g77} no longer installs the @file{libf2c.a} library
741 and @file{f2c.h} include file
742 in the @file{/usr} or @file{/usr/local} hierarchy,
743 even if the @file{f2c-install-ok} or @file{f2c-exists-ok} files exist
744 in the source or build directory.
745 See the installation documentation for more information.
750 The @file{libf2c.a} library produced by @command{g77} has been
751 renamed to @file{libg2c.a}.
752 It is installed only in the @command{gcc} ``private''
753 directory hierarchy, @file{gcc-lib}.
754 This allows system administrators and users to choose which
755 version of the @code{libf2c} library from @code{netlib} they
756 wish to use on a case-by-case basis.
757 See the installation documentation for more information.
762 The @file{f2c.h} include (header) file produced by @command{g77}
763 has been renamed to @file{g2c.h}.
764 It is installed only in the @command{gcc} ``private''
765 directory hierarchy, @file{gcc-lib}.
766 This allows system administrators and users to choose which
767 version of the include file from @code{netlib} they
768 wish to use on a case-by-case basis.
769 See the installation documentation for more information.
773 The @command{g77} command now expects the run-time library
774 to be named @code{libg2c.a} instead of @code{libf2c.a},
775 to ensure that a version other than the one built and
776 installed as part of the same @command{g77} version is picked up.
780 During the configuration and build process,
781 @command{g77} creates subdirectories it needs only as it
783 Other cleaning up of the configuration and build process
784 has been performed as well.
789 @code{install-info} now used to update the directory of
790 Info documentation to contain an entry for @command{g77}
791 (during installation).
795 Some diagnostics have been changed from warnings to errors,
796 to prevent inadvertent use of the resulting, probably buggy,
798 These mostly include diagnostics about use of unsupported features
799 in the @code{OPEN}, @code{INQUIRE}, @code{READ}, and
800 @code{WRITE} statements,
801 and about truncations of various sorts of constants.
805 Improve compilation of @code{FORMAT} expressions so that
806 a null byte is appended to the last operand if it
808 This provides a cleaner run-time diagnostic as provided
809 by @code{libf2c} for statements like @samp{PRINT '(I1', 42}.
814 Improve documentation and indexing.
819 The upgrade to @code{libf2c} as of 1998-06-18
820 should fix a variety of problems, including
821 those involving some uses of the @code{T} format
822 specifier, and perhaps some build (porting) problems
827 @c 1998-09-03: EGCS 1.1 released.
828 @heading In @code{EGCS} 1.1 versus @command{g77} 0.5.23:
832 @cindex stack, 387 coprocessor
836 Fix a code-generation bug that afflicted
837 Intel x86 targets when @option{-O2} was specified
838 compiling, for example, an old version of
839 the @code{DNRM2} routine.
841 The x87 coprocessor stack was being
842 mismanaged in cases involving assigned @code{GOTO}
848 @command{g77} no longer produces incorrect code
850 for @code{EQUIVALENCE} and @code{COMMON}
851 aggregates that, due to ``unnatural'' ordering of members
852 vis-a-vis their types, require initial padding.
857 Fix @command{g77} crash compiling code
858 containing the construct @samp{CMPLX(0.)} or similar.
863 @command{g77} no longer crashes when compiling code
864 containing specification statements such as
865 @samp{INTEGER(KIND=7) PTR}.
870 @command{g77} no longer crashes when compiling code
871 such as @samp{J = SIGNAL(1, 2)}.
875 @command{g77} now treats @samp{%LOC(@var{expr})} and
876 @samp{LOC(@var{expr})} as ``ordinary'' expressions
877 when they are used as arguments in procedure calls.
878 This change applies only to global (filewide) analysis,
879 making it consistent with
880 how @command{g77} actually generates code
883 Previously, @command{g77} treated these expressions
884 as denoting special ``pointer'' arguments
885 for the purposes of filewide analysis.
889 Fix @command{g77} crash
890 (or apparently infinite run-time)
891 when compiling certain complicated expressions
892 involving @code{COMPLEX} arithmetic
893 (especially multiplication).
897 @cindex double-precision performance
898 @cindex -malign-double
900 Align static double-precision variables and arrays
902 regardless of whether @option{-malign-double} is specified.
904 Generally, this affects only local variables and arrays
905 having the @code{SAVE} attribute
906 or given initial values via @code{DATA}.
909 The @command{g77} driver now ensures that @option{-lg2c}
910 is specified in the link phase prior to any
911 occurrence of @option{-lm}.
912 This prevents accidentally linking to a routine
913 in the SunOS4 @option{-lm} library
914 when the generated code wants to link to the one
915 in @code{libf2c} (@code{libg2c}).
918 @command{g77} emits more debugging information when
921 This new information allows, for example,
922 @kbd{which __g77_length_a} to be used in @command{gdb}
923 to determine the type of the phantom length argument
924 supplied with @code{CHARACTER} variables.
926 This information pertains to internally-generated
927 type, variable, and other information,
928 not to the longstanding deficiencies vis-a-vis
929 @code{COMMON} and @code{EQUIVALENCE}.
932 The F90 @code{Date_and_Time} intrinsic now is
936 The F90 @code{System_Clock} intrinsic allows
937 the optional arguments (except for the @code{Count}
938 argument) to be omitted.
942 Upgrade to @code{libf2c} as of 1998-06-18.
947 Improve documentation and indexing.
952 @heading In previous versions:
954 Information on previous versions is not provided
955 in this @file{@value{path-g77}/NEWS} file,
957 See @file{@value{path-g77}/news.texi},
958 or any of its other derivations
959 (Info, HTML, dvi forms)
960 for such information.
964 @c 1998-05-20: 0.5.23 released.
965 @heading In 0.5.23 versus 0.5.22:
968 This release contains several regressions against
969 version 0.5.22 of @command{g77}, due to using the
970 ``vanilla'' @command{gcc} back end instead of patching
971 it to fix a few bugs and improve performance in a
974 Features that have been dropped from this version
975 of @command{g77} due to their being implemented
976 via @command{g77}-specific patches to the @command{gcc}
977 back end in previous releases include:
981 Support for @code{__restrict__} keyword,
982 the options @option{-fargument-alias}, @option{-fargument-noalias},
983 and @option{-fargument-noalias-global},
984 and the corresponding alias-analysis code.
986 (@code{egcs} has the alias-analysis
987 code, but not the @code{__restrict__} keyword.
988 @code{egcs} @command{g77} users benefit from the alias-analysis
989 code despite the lack of the @code{__restrict__} keyword,
990 which is a C-language construct.)
993 Support for the GNU compiler options
994 @option{-fmove-all-movables},
995 @option{-freduce-all-givs},
996 and @option{-frerun-loop-opt}.
998 (@code{egcs} supports these options.
999 @command{g77} users of @code{egcs} benefit from them even if
1000 they are not explicitly specified,
1001 because the defaults are optimized for @command{g77} users.)
1004 Support for the @option{-W} option warning about
1005 integer division by zero.
1008 The Intel x86-specific option @option{-malign-double}
1009 applying to stack-allocated data
1010 as well as statically-allocate data.
1013 @ifclear USERVISONLY
1014 Note that the @file{gcc/f/gbe/} subdirectory has been removed
1015 from this distribution as a result of @command{g77} no longer
1016 including patches for the @command{gcc} back end.
1019 @ifclear USERVISONLY
1021 Fix bugs in the @code{libU77} intrinsic @code{HostNm}
1022 that wrote one byte beyond the end of its @code{CHARACTER}
1024 and in the @code{libU77} intrinsics
1025 @code{GMTime} and @code{LTime}
1026 that overwrote their arguments.
1030 Support @command{gcc} version 2.8,
1031 and remove support for prior versions of @command{gcc}.
1033 @cindex -@w{}-driver option
1034 @cindex @command{g77} options, -@w{}-driver
1035 @cindex options, -@w{}-driver
1037 Remove support for the @option{--driver} option,
1038 as @command{g77} now does all the driving,
1039 just like @command{gcc}.
1041 @ifclear USERVISONLY
1043 @code{CASE DEFAULT} no longer crashes @command{g77}.
1046 @ifclear USERVISONLY
1048 Valid combinations of @code{EXTERNAL},
1049 passing that external as a dummy argument
1050 without explicitly giving it a type,
1051 and, in a subsequent program unit,
1052 referencing that external as
1053 an external function with a different type
1054 no longer crash @command{g77}.
1057 @ifclear USERVISONLY
1059 @command{g77} no longer installs the @file{f77} command
1060 and @file{f77.1} man page
1061 in the @file{/usr} or @file{/usr/local} hierarchy,
1062 even if the @file{f77-install-ok} file exists
1063 in the source or build directory.
1064 See the installation documentation for more information.
1067 @ifclear USERVISONLY
1069 @command{g77} no longer installs the @file{libf2c.a} library
1070 and @file{f2c.h} include file
1071 in the @file{/usr} or @file{/usr/local} hierarchy,
1072 even if the @file{f2c-install-ok} or @file{f2c-exists-ok} files exist
1073 in the source or build directory.
1074 See the installation documentation for more information.
1077 @ifclear USERVISONLY
1079 The @file{libf2c.a} library produced by @command{g77} has been
1080 renamed to @file{libg2c.a}.
1081 It is installed only in the @command{gcc} ``private''
1082 directory hierarchy, @file{gcc-lib}.
1083 This allows system administrators and users to choose which
1084 version of the @code{libf2c} library from @code{netlib} they
1085 wish to use on a case-by-case basis.
1086 See the installation documentation for more information.
1089 @ifclear USERVISONLY
1091 The @file{f2c.h} include (header) file produced by @command{g77}
1092 has been renamed to @file{g2c.h}.
1093 It is installed only in the @command{gcc} ``private''
1094 directory hierarchy, @file{gcc-lib}.
1095 This allows system administrators and users to choose which
1096 version of the include file from @code{netlib} they
1097 wish to use on a case-by-case basis.
1098 See the installation documentation for more information.
1102 The @command{g77} command now expects the run-time library
1103 to be named @code{libg2c.a} instead of @code{libf2c.a},
1104 to ensure that a version other than the one built and
1105 installed as part of the same @command{g77} version is picked up.
1107 @ifclear USERVISONLY
1109 The @option{-Wunused} option no longer issues a spurious
1110 warning about the ``master'' procedure generated by
1111 @command{g77} for procedures containing @code{ENTRY} statements.
1115 @command{g77}'s version of @code{libf2c} separates out
1116 the setting of global state
1117 (such as command-line arguments and signal handling)
1118 from @file{main.o} into distinct, new library
1121 This should make it easier to write portable applications
1122 that have their own (non-Fortran) @code{main()} routine
1123 properly set up the @code{libf2c} environment, even
1124 when @code{libf2c} (now @code{libg2c}) is a shared library.
1126 @ifclear USERVISONLY
1128 During the configuration and build process,
1129 @command{g77} creates subdirectories it needs only as it
1130 needs them, thus avoiding unnecessary creation of, for example,
1131 @file{stage1/f/runtime} when doing a non-bootstrap build.
1132 Other cleaning up of the configuration and build process
1133 has been performed as well.
1136 @ifclear USERVISONLY
1138 @code{install-info} now used to update the directory of
1139 Info documentation to contain an entry for @command{g77}
1140 (during installation).
1144 Some diagnostics have been changed from warnings to errors,
1145 to prevent inadvertent use of the resulting, probably buggy,
1147 These mostly include diagnostics about use of unsupported features
1148 in the @code{OPEN}, @code{INQUIRE}, @code{READ}, and
1149 @code{WRITE} statements,
1150 and about truncations of various sorts of constants.
1152 @ifclear USERVISONLY
1154 Improve documentation and indexing.
1157 @ifclear USERVISONLY
1159 Upgrade to @code{libf2c} as of 1998-04-20.
1161 This should fix a variety of problems, including
1162 those involving some uses of the @code{T} format
1163 specifier, and perhaps some build (porting) problems
1168 @c 1998-03-16: 0.5.22 released.
1169 @heading In 0.5.22 versus 0.5.21:
1171 @ifclear USERVISONLY
1173 Fix code generation for iterative @code{DO} loops that
1174 have one or more references to the iteration variable,
1175 or to aliases of it, in their control expressions.
1176 For example, @samp{DO 10 J=2,J} now is compiled correctly.
1179 @ifclear USERVISONLY
1181 @cindex stack, 387 coprocessor
1185 Fix a code-generation bug that afflicted
1186 Intel x86 targets when @option{-O2} was specified
1187 compiling, for example, an old version of
1188 the @code{DNRM2} routine.
1190 The x87 coprocessor stack was being
1191 mismanaged in cases involving assigned @code{GOTO}
1195 @ifclear USERVISONLY
1197 Fix @code{DTime} intrinsic so as not to truncate
1198 results to integer values (on some systems).
1202 Fix @code{Signal} intrinsic so it offers portable
1203 support for 64-bit systems (such as Digital Alphas
1206 @ifclear USERVISONLY
1208 Fix run-time crash involving @code{NAMELIST} on 64-bit
1209 machines such as Alphas.
1212 @ifclear USERVISONLY
1214 Fix @command{g77} version of @code{libf2c} so it no longer
1215 produces a spurious @samp{I/O recursion} diagnostic at run time
1216 when an I/O operation (such as @samp{READ *,I}) is interrupted
1217 in a manner that causes the program to be terminated
1218 via the @code{f_exit} routine (such as via @kbd{C-c}).
1221 @ifclear USERVISONLY
1223 Fix @command{g77} crash triggered by @code{CASE} statement with
1224 an omitted lower or upper bound.
1227 @ifclear USERVISONLY
1229 Fix @command{g77} crash compiling references to @code{CPU_Time}
1233 @ifclear USERVISONLY
1235 Fix @command{g77} crash
1236 (or apparently infinite run-time)
1237 when compiling certain complicated expressions
1238 involving @code{COMPLEX} arithmetic
1239 (especially multiplication).
1242 @ifclear USERVISONLY
1244 Fix @command{g77} crash on statements such as
1245 @samp{PRINT *, (REAL(Z(I)),I=1,2)}, where
1246 @samp{Z} is @code{DOUBLE COMPLEX}.
1249 @ifclear USERVISONLY
1251 Fix a @command{g++} crash.
1255 Support @samp{FORMAT(I<@var{expr}>)} when @var{expr} is a
1256 compile-time constant @code{INTEGER} expression.
1259 Fix @command{g77} @option{-g} option so procedures that
1260 use @code{ENTRY} can be stepped through, line by line,
1263 @ifclear USERVISONLY
1265 Fix a profiling-related bug in @command{gcc} back end for
1266 Intel x86 architecture.
1270 Allow any @code{REAL} argument to intrinsics
1271 @code{Second} and @code{CPU_Time}.
1274 Allow any numeric argument to intrinsics
1275 @code{Int2} and @code{Int8}.
1278 Use @code{tempnam}, if available, to open scratch files
1279 (as in @samp{OPEN(STATUS='SCRATCH')})
1280 so that the @code{TMPDIR} environment variable,
1281 if present, is used.
1284 Rename the @command{gcc} keyword @code{restrict} to
1285 @code{__restrict__}, to avoid rejecting valid, existing,
1287 Support for @code{restrict} is now more like support
1290 @ifclear USERVISONLY
1292 Fix @option{-fpedantic} to not reject procedure invocations
1293 such as @samp{I=J()} and @samp{CALL FOO()}.
1297 Fix @option{-fugly-comma} to affect invocations of
1298 only external procedures.
1299 Restore rejection of gratuitous trailing omitted
1300 arguments to intrinsics, as in @samp{I=MAX(3,4,,)}.
1303 Fix compiler so it accepts @option{-fgnu-intrinsics-*} and
1304 @option{-fbadu77-intrinsics-*} options.
1306 @ifclear USERVISONLY
1308 Improve diagnostic messages from @code{libf2c}
1309 so it is more likely that the printing of the
1310 active format string is limited to the string,
1311 with no trailing garbage being printed.
1313 (Unlike @command{f2c}, @command{g77} did not append
1314 a null byte to its compiled form of every
1315 format string specified via a @code{FORMAT} statement.
1316 However, @command{f2c} would exhibit the problem
1317 anyway for a statement like @samp{PRINT '(I)garbage', 1}
1318 by printing @samp{(I)garbage} as the format string.)
1321 @ifclear USERVISONLY
1323 Improve compilation of @code{FORMAT} expressions so that
1324 a null byte is appended to the last operand if it
1326 This provides a cleaner run-time diagnostic as provided
1327 by @code{libf2c} for statements like @samp{PRINT '(I1', 42}.
1330 @ifclear USERVISONLY
1332 Fix various crashes involving code with diagnosed errors.
1335 @ifclear USERVISONLY
1337 Fix cross-compilation bug when configuring @code{libf2c}.
1340 @ifclear USERVISONLY
1342 Improve diagnostics.
1345 @ifclear USERVISONLY
1347 Improve documentation and indexing.
1350 @ifclear USERVISONLY
1352 Upgrade to @code{libf2c} as of 1997-09-23.
1353 This fixes a formatted-I/O bug that afflicted
1354 64-bit systems with 32-bit integers
1355 (such as Digital Alpha running GNU/Linux).
1359 @c 1998-03-18: EGCS 1.0.2 released.
1360 @heading In @code{EGCS} 1.0.2 versus @code{EGCS} 1.0.1:
1362 @ifclear USERVISONLY
1364 Fix @command{g77} crash triggered by @code{CASE} statement with
1365 an omitted lower or upper bound.
1368 @ifclear USERVISONLY
1370 Fix @command{g77} crash on statements such as
1371 @samp{PRINT *, (REAL(Z(I)),I=1,2)}, where
1372 @samp{Z} is @code{DOUBLE COMPLEX}.
1375 @ifclear USERVISONLY
1377 @cindex support, ELF
1378 @cindex -fPIC option
1379 @cindex options, -fPIC
1381 Fix @option{-fPIC} (such as compiling for ELF targets)
1382 on the Intel x86 architecture target
1383 so invalid assembler code is no longer produced.
1386 @ifclear USERVISONLY
1388 Fix @option{-fpedantic} to not reject procedure invocations
1389 such as @samp{I=J()} and @samp{CALL FOO()}.
1392 @ifclear USERVISONLY
1394 Fix @option{-fugly-comma} to affect invocations of
1395 only external procedures.
1396 Restore rejection of gratuitous trailing omitted
1397 arguments to intrinsics, as in @samp{I=MAX(3,4,,)}.
1401 Fix compiler so it accepts @option{-fgnu-intrinsics-*} and
1402 @option{-fbadu77-intrinsics-*} options.
1405 @c 1998-01-06: EGCS 1.0.1 released.
1406 @heading In @code{EGCS} 1.0.1 versus @code{EGCS} 1.0:
1407 @ifclear USERVISONLY
1410 Fix run-time crash involving @code{NAMELIST} on 64-bit
1411 machines such as Alphas.
1415 @c 1997-12-03: EGCS 1.0 released.
1416 @heading In @code{EGCS} 1.0 versus @command{g77} 0.5.21:
1419 Version 1.0 of @code{egcs}
1420 contains several regressions against
1421 version 0.5.21 of @command{g77},
1423 ``vanilla'' @command{gcc} back end instead of patching
1424 it to fix a few bugs and improve performance in a
1427 Features that have been dropped from this version
1428 of @command{g77} due to their being implemented
1429 via @command{g77}-specific patches to the @command{gcc}
1430 back end in previous releases include:
1434 Support for the C-language @code{restrict} keyword.
1437 Support for the @option{-W} option warning about
1438 integer division by zero.
1441 The Intel x86-specific option @option{-malign-double}
1442 applying to stack-allocated data
1443 as well as statically-allocate data.
1446 @ifclear USERVISONLY
1447 Note that the @file{gcc/f/gbe/} subdirectory has been removed
1448 from this distribution as a result of @command{g77}
1449 being fully integrated with
1450 the @code{egcs} variant of the @command{gcc} back end.
1453 @ifclear USERVISONLY
1455 Fix code generation for iterative @code{DO} loops that
1456 have one or more references to the iteration variable,
1457 or to aliases of it, in their control expressions.
1458 For example, @samp{DO 10 J=2,J} now is compiled correctly.
1461 @ifclear USERVISONLY
1463 Fix @code{DTime} intrinsic so as not to truncate
1464 results to integer values (on some systems).
1467 @ifclear USERVISONLY
1469 @c Toon Moene discovered these.
1470 Some Fortran code, miscompiled
1471 by @command{g77} built on @command{gcc} version 2.8.1
1472 on m68k-next-nextstep3 configurations
1473 when using the @option{-O2} option,
1474 is now compiled correctly.
1475 It is believed that a C function known to miscompile
1476 on that configuration
1477 when using the @samp{-O2 -funroll-loops} options
1478 also is now compiled correctly.
1481 @ifclear USERVISONLY
1483 Remove support for non-@code{egcs} versions of @command{gcc}.
1486 @cindex -@w{}-driver option
1487 @cindex @command{g77} options, -@w{}-driver
1488 @cindex options, -@w{}-driver
1490 Remove support for the @option{--driver} option,
1491 as @command{g77} now does all the driving,
1492 just like @command{gcc}.
1495 Allow any numeric argument to intrinsics
1496 @code{Int2} and @code{Int8}.
1498 @ifclear USERVISONLY
1500 Improve diagnostic messages from @code{libf2c}
1501 so it is more likely that the printing of the
1502 active format string is limited to the string,
1503 with no trailing garbage being printed.
1505 (Unlike @command{f2c}, @command{g77} did not append
1506 a null byte to its compiled form of every
1507 format string specified via a @code{FORMAT} statement.
1508 However, @code{f2c} would exhibit the problem
1509 anyway for a statement like @samp{PRINT '(I)garbage', 1}
1510 by printing @samp{(I)garbage} as the format string.)
1513 @ifclear USERVISONLY
1515 Upgrade to @code{libf2c} as of 1997-09-23.
1516 This fixes a formatted-I/O bug that afflicted
1517 64-bit systems with 32-bit integers
1518 (such as Digital Alpha running GNU/Linux).
1522 @c 1997-09-09: 0.5.21 released.
1525 @ifclear USERVISONLY
1527 Fix a code-generation bug introduced by 0.5.20
1528 caused by loop unrolling (by specifying
1529 @option{-funroll-loops} or similar).
1530 This bug afflicted all code compiled by
1531 version 2.7.2.2.f.2 of @command{gcc} (C, C++,
1532 Fortran, and so on).
1535 @ifclear USERVISONLY
1537 Fix a code-generation bug manifested when
1538 combining local @code{EQUIVALENCE} with a
1539 @code{DATA} statement that follows
1540 the first executable statement (or is
1541 treated as an executable-context statement
1542 as a result of using the @option{-fpedantic}
1546 @ifclear USERVISONLY
1548 Fix a compiler crash that occurred when an
1549 integer division by a constant zero is detected.
1550 Instead, when the @option{-W} option is specified,
1551 the @command{gcc} back end issues a warning about such a case.
1552 This bug afflicted all code compiled by
1553 version 2.7.2.2.f.2 of @command{gcc} (C, C++,
1554 Fortran, and so on).
1558 When the @option{-W} option is specified, @command{gcc}, @command{g77},
1559 and other GNU compilers that incorporate the @command{gcc}
1560 back end as modified by @command{g77}, issue
1561 a warning about integer division by constant zero.
1564 @ifclear USERVISONLY
1566 Fix a compiler crash that occurred in some cases
1567 of procedure inlining.
1568 (Such cases became more frequent in 0.5.20.)
1571 @ifclear USERVISONLY
1573 Fix a compiler crash resulting from using @code{DATA}
1574 or similar to initialize a @code{COMPLEX} variable or
1578 @ifclear USERVISONLY
1580 Fix compiler crashes involving use of @code{AND}, @code{OR},
1581 or @code{XOR} intrinsics.
1584 @ifclear USERVISONLY
1586 Fix compiler bug triggered when using a @code{COMMON}
1587 or @code{EQUIVALENCE} variable
1588 as the target of an @code{ASSIGN}
1589 or assigned-@code{GOTO} statement.
1592 @ifclear USERVISONLY
1594 Fix compiler crashes due to using the name of a some
1595 non-standard intrinsics (such as @code{FTell} or
1596 @code{FPutC}) as such and as the name of a procedure
1598 Such dual use of a name in a program is allowed by
1602 @c @command{g77}'s version of @code{libf2c} has been modified
1603 @c so that the external names of library's procedures do not
1604 @c conflict with names used for Fortran procedures compiled
1605 @c by @command{g77}.
1606 @c An additional layer of jacket procedures has been added
1607 @c to @code{libf2c} to map the old names to the new names,
1608 @c for automatic use by programs that interface to the
1609 @c library procedures via the external-procedure mechanism.
1611 @c For example, the intrinsic @code{FPUTC} previously was
1612 @c implemented by @command{g77} as a call to the @code{libf2c}
1613 @c routine @code{fputc_}.
1614 @c This would conflict with a Fortran procedure named @code{FPUTC}
1615 @c (using default compiler options), and this conflict
1616 @c would cause a crash under certain circumstances.
1618 @c Now, the intrinsic @code{FPUTC} calls @code{G77_fputc_0},
1619 @c which does not conflict with the @code{fputc_} external
1620 @c that implements a Fortran procedure named @code{FPUTC}.
1622 @c Programs that refer to @code{FPUTC} as an external procedure
1623 @c without supplying their own implementation will link to
1624 @c the new @code{libf2c} routine @code{fputc_}, which is
1625 @c simply a jacket routine that calls @code{G77_fputc_0}.
1627 @ifclear USERVISONLY
1629 Place automatic arrays on the stack, even if
1630 @code{SAVE} or the @option{-fno-automatic} option
1632 This avoids a compiler crash in some cases.
1635 @ifclear USERVISONLY
1637 The @option{-malign-double} option now reliably aligns
1638 @code{DOUBLE PRECISION} optimally on Pentium and
1639 Pentium Pro architectures (586 and 686 in @command{gcc}).
1643 New option @option{-Wno-globals} disables warnings
1644 about ``suspicious'' use of a name both as a global
1645 name and as the implicit name of an intrinsic, and
1646 warnings about disagreements over the number or natures of
1647 arguments passed to global procedures, or the
1648 natures of the procedures themselves.
1650 The default is to issue such warnings, which are
1651 new as of this version of @command{g77}.
1654 New option @option{-fno-globals} disables diagnostics
1655 about potentially fatal disagreements
1656 analysis problems, such as disagreements over the
1657 number or natures of arguments passed to global
1658 procedures, or the natures of those procedures themselves.
1660 The default is to issue such diagnostics and flag
1661 the compilation as unsuccessful.
1662 With this option, the diagnostics are issued as
1663 warnings, or, if @option{-Wno-globals} is specified,
1664 are not issued at all.
1666 This option also disables inlining of global procedures,
1667 to avoid compiler crashes resulting from coding errors
1668 that these diagnostics normally would identify.
1670 @ifclear USERVISONLY
1672 Diagnose cases where a reference to a procedure
1673 disagrees with the type of that procedure, or
1674 where disagreements about the number or nature
1676 This avoids a compiler crash.
1679 @ifclear USERVISONLY
1681 Fix parsing bug whereby @command{g77} rejected a
1682 second initialization specification immediately
1683 following the first's closing @samp{/} without
1684 an intervening comma in a @code{DATA} statement,
1685 and the second specification was an implied-DO list.
1688 @ifclear USERVISONLY
1690 Improve performance of the @command{gcc} back end so
1691 certain complicated expressions involving @code{COMPLEX}
1692 arithmetic (especially multiplication) don't appear to
1693 take forever to compile.
1696 @ifclear USERVISONLY
1698 Fix a couple of profiling-related bugs in @command{gcc}
1702 @ifclear USERVISONLY
1704 Integrate GNU Ada's (GNAT's) changes to the back end,
1705 which consist almost entirely of bug fixes.
1706 These fixes are circa version 3.10p of GNAT.
1709 @ifclear USERVISONLY
1711 Include some other @command{gcc} fixes that seem useful in
1712 @command{g77}'s version of @command{gcc}.
1713 (See @file{gcc/ChangeLog} for details---compare it
1714 to that file in the vanilla @code{gcc-2.7.2.3.tar.gz}
1719 Fix @code{libU77} routines that accept file and other names
1720 to strip trailing blanks from them, for consistency
1721 with other implementations.
1722 Blanks may be forcibly appended to such names by
1723 appending a single null character (@samp{CHAR(0)})
1724 to the significant trailing blanks.
1727 Fix @code{CHMOD} intrinsic to work with file names
1728 that have embedded blanks, commas, and so on.
1731 Fix @code{SIGNAL} intrinsic so it accepts an
1732 optional third @code{Status} argument.
1734 @ifclear USERVISONLY
1736 Fix @code{IDATE()} intrinsic subroutine (VXT form)
1737 so it accepts arguments in the correct order.
1738 Documentation fixed accordingly, and for
1739 @code{GMTIME()} and @code{LTIME()} as well.
1743 Make many changes to @code{libU77} intrinsics to
1744 support existing code more directly.
1746 Such changes include allowing both subroutine and
1747 function forms of many routines, changing @code{MCLOCK()}
1748 and @code{TIME()} to return @code{INTEGER(KIND=1)} values,
1749 introducing @code{MCLOCK8()} and @code{TIME8()} to
1750 return @code{INTEGER(KIND=2)} values,
1751 and placing functions that are intended to perform
1752 side effects in a new intrinsic group, @code{badu77}.
1754 @ifclear USERVISONLY
1756 Improve @code{libU77} so it is more portable.
1760 Add options @option{-fbadu77-intrinsics-delete},
1761 @option{-fbadu77-intrinsics-hide}, and so on.
1763 @ifclear USERVISONLY
1765 Fix crashes involving diagnosed or invalid code.
1768 @ifclear USERVISONLY
1770 @command{g77} and @command{gcc} now do a somewhat better
1771 job detecting and diagnosing arrays that are too
1772 large to handle before these cause diagnostics
1773 during the assembler or linker phase, a compiler
1774 crash, or generation of incorrect code.
1777 @ifclear USERVISONLY
1779 Make some fixes to alias analysis code.
1782 @ifclear USERVISONLY
1784 Add support for @code{restrict} keyword in @command{gcc}
1788 @ifclear USERVISONLY
1790 Support @command{gcc} version 2.7.2.3
1791 (modified by @command{g77} into version 2.7.2.3.f.1),
1793 support for prior versions of @command{gcc}.
1796 @ifclear USERVISONLY
1798 Incorporate GNAT's patches to the @command{gcc} back
1799 end into @command{g77}'s, so GNAT users do not need
1800 to apply GNAT's patches to build both GNAT and @command{g77}
1801 from the same source tree.
1804 @ifclear USERVISONLY
1806 Modify @command{make} rules and related code so that
1807 generation of Info documentation doesn't require
1808 compilation using @command{gcc}.
1809 Now, any ANSI C compiler should be adequate to
1810 produce the @command{g77} documentation (in particular,
1811 the tables of intrinsics) from scratch.
1815 Add @code{INT2} and @code{INT8} intrinsics.
1818 Add @code{CPU_TIME} intrinsic.
1821 Add @code{ALARM} intrinsic.
1824 @code{CTIME} intrinsic now accepts any @code{INTEGER}
1825 argument, not just @code{INTEGER(KIND=2)}.
1827 @ifclear USERVISONLY
1829 Warn when explicit type declaration disagrees with
1830 the type of an intrinsic invocation.
1833 @ifclear USERVISONLY
1835 Support @samp{*f771} entry in @command{gcc} @file{specs} file.
1838 @ifclear USERVISONLY
1840 Fix typo in @command{make} rule @command{g77-cross}, used only for
1844 @ifclear USERVISONLY
1846 Fix @code{libf2c} build procedure to re-archive library
1847 if previous attempt to archive was interrupted.
1850 @ifclear USERVISONLY
1852 Change @command{gcc} to unroll loops only during the last
1853 invocation (of as many as two invocations) of loop
1857 @ifclear USERVISONLY
1859 Improve handling of @option{-fno-f2c} so that code that
1860 attempts to pass an intrinsic as an actual argument,
1861 such as @samp{CALL FOO(ABS)}, is rejected due to the fact
1862 that the run-time-library routine is, effectively,
1863 compiled with @option{-ff2c} in effect.
1866 @ifclear USERVISONLY
1868 Fix @command{g77} driver to recognize @option{-fsyntax-only}
1869 as an option that inhibits linking, just like @option{-c} or
1870 @option{-S}, and to recognize and properly handle the
1871 @option{-nostdlib}, @option{-M}, @option{-MM}, @option{-nodefaultlibs},
1872 and @option{-Xlinker} options.
1875 @ifclear USERVISONLY
1877 Upgrade to @code{libf2c} as of 1997-08-16.
1880 @ifclear USERVISONLY
1882 Modify @code{libf2c} to consistently and clearly diagnose
1883 recursive I/O (at run time).
1887 @command{g77} driver now prints version information (such as produced
1888 by @kbd{g77 -v}) to @code{stderr} instead of @code{stdout}.
1891 The @samp{.r} suffix now designates a Ratfor source file,
1892 to be preprocessed via the @command{ratfor} command, available
1895 @ifclear USERVISONLY
1897 Fix some aspects of how @command{gcc} determines what kind of
1898 system is being configured and what kinds are supported.
1899 For example, GNU Linux/Alpha ELF systems now are directly
1903 @ifclear USERVISONLY
1905 Improve diagnostics.
1908 @ifclear USERVISONLY
1910 Improve documentation and indexing.
1913 @ifclear USERVISONLY
1915 Include all pertinent files for @code{libf2c} that come
1916 from @code{netlib.bell-labs.com}; give any such files
1917 that aren't quite accurate in @command{g77}'s version of
1918 @code{libf2c} the suffix @samp{.netlib}.
1921 @ifclear USERVISONLY
1923 Reserve @code{INTEGER(KIND=0)} for future use.
1927 @c 1997-02-28: 0.5.20 released.
1931 The @option{-fno-typeless-boz} option is now the default.
1933 This option specifies that non-decimal-radix
1934 constants using the prefixed-radix form (such as @samp{Z'1234'})
1935 are to be interpreted as @code{INTEGER(KIND=1)} constants.
1936 Specify @option{-ftypeless-boz} to cause such
1937 constants to be interpreted as typeless.
1939 (Version 0.5.19 introduced @option{-fno-typeless-boz} and
1943 @xref{Fortran Dialect Options,,Options Controlling Fortran Dialect},
1944 for information on the @option{-ftypeless-boz} option.
1948 Options @option{-ff90-intrinsics-enable} and
1949 @option{-fvxt-intrinsics-enable} now are the
1952 Some programs might use names that clash with
1953 intrinsic names defined (and now enabled) by these
1954 options or by the new @code{libU77} intrinsics.
1955 Users of such programs might need to compile them
1956 differently (using, for example, @option{-ff90-intrinsics-disable})
1957 or, better yet, insert appropriate @code{EXTERNAL}
1958 statements specifying that these names are not intended
1959 to be names of intrinsics.
1962 The @code{ALWAYS_FLUSH} macro is no longer defined when
1963 building @code{libf2c}, which should result in improved
1964 I/O performance, especially over NFS.
1966 @emph{Note:} If you have code that depends on the behavior
1967 of @code{libf2c} when built with @code{ALWAYS_FLUSH} defined,
1968 you will have to modify @code{libf2c} accordingly before
1969 building it from this and future versions of @command{g77}.
1972 @xref{Output Assumed To Flush}, for more information.
1976 Dave Love's implementation of @code{libU77} has been
1977 added to the version of @code{libf2c} distributed with
1978 and built as part of @command{g77}.
1979 @command{g77} now knows about the routines in this library
1983 New option @option{-fvxt} specifies that the
1984 source file is written in VXT Fortran, instead of GNU Fortran.
1987 @xref{VXT Fortran}, for more information on the constructs
1988 recognized when the @option{-fvxt} option is specified.
1992 The @option{-fvxt-not-f90} option has been deleted,
1993 along with its inverse, @option{-ff90-not-vxt}.
1995 If you used one of these deleted options, you should
1996 re-read the pertinent documentation to determine which
1997 options, if any, are appropriate for compiling your
1998 code with this version of @command{g77}.
2001 @xref{Other Dialects}, for more information.
2005 The @option{-fugly} option now issues a warning, as it
2006 likely will be removed in a future version.
2008 (Enabling all the @option{-fugly-*} options is unlikely
2009 to be feasible, or sensible, in the future,
2010 so users should learn to specify only those
2011 @option{-fugly-*} options they really need for a
2012 particular source file.)
2015 The @option{-fugly-assumed} option, introduced in
2016 version 0.5.19, has been changed to
2017 better accommodate old and new code.
2020 @xref{Ugly Assumed-Size Arrays}, for more information.
2023 @ifclear USERVISONLY
2025 Make a number of fixes to the @command{g77} front end and
2026 the @command{gcc} back end to better support Alpha (AXP)
2028 This includes providing at least one bug-fix to the
2029 @command{gcc} back end for Alphas.
2033 Related to supporting Alpha (AXP) machines, the @code{LOC()}
2034 intrinsic and @code{%LOC()} construct now return
2035 values of @code{INTEGER(KIND=0)} type,
2036 as defined by the GNU Fortran language.
2038 This type is wide enough
2039 (holds the same number of bits)
2040 as the character-pointer type on the machine.
2042 On most machines, this won't make a difference,
2043 whereas, on Alphas and other systems with 64-bit pointers,
2044 the @code{INTEGER(KIND=0)} type is equivalent to @code{INTEGER(KIND=2)}
2045 (often referred to as @code{INTEGER*8})
2046 instead of the more common @code{INTEGER(KIND=1)}
2047 (often referred to as @code{INTEGER*4}).
2050 Emulate @code{COMPLEX} arithmetic in the @command{g77} front
2051 end, to avoid bugs in @code{complex} support in the
2052 @command{gcc} back end.
2053 New option @option{-fno-emulate-complex}
2054 causes @command{g77} to revert the 0.5.19 behavior.
2056 @ifclear USERVISONLY
2058 Fix bug whereby @samp{REAL A(1)}, for example, caused
2059 a compiler crash if @option{-fugly-assumed} was in effect
2060 and @var{A} was a local (automatic) array.
2061 That case is no longer affected by the new
2062 handling of @option{-fugly-assumed}.
2065 @ifclear USERVISONLY
2067 Fix @command{g77} command driver so that @samp{g77 -o foo.f}
2068 no longer deletes @file{foo.f} before issuing other
2069 diagnostics, and so the @option{-x} option is properly
2073 @ifclear USERVISONLY
2075 Enable inlining of subroutines and functions by the @command{gcc}
2077 This works as it does for @command{gcc} itself---program units
2078 may be inlined for invocations that follow them in the same
2079 program unit, as long as the appropriate compile-time
2080 options are specified.
2084 Dummy arguments are no longer assumed to potentially alias
2086 other dummy arguments or @code{COMMON} areas when any of
2087 these are defined (assigned to) by Fortran code.
2089 This can result in faster and/or smaller programs when
2090 compiling with optimization enabled, though on some
2091 systems this effect is observed only when @option{-fforce-addr}
2094 New options @option{-falias-check}, @option{-fargument-alias},
2095 @option{-fargument-noalias},
2096 and @option{-fno-argument-noalias-global} control the
2097 way @command{g77} handles potential aliasing.
2100 @xref{Aliasing Assumed To Work}, for detailed information on why the
2101 new defaults might result in some programs no longer working the way they
2102 did when compiled by previous versions of @command{g77}.
2105 @ifclear USERVISONLY
2107 The @code{CONJG()} and @code{DCONJG()} intrinsics now
2108 are compiled in-line.
2111 @ifclear USERVISONLY
2113 The bug-fix for 0.5.19.1 has been re-done.
2114 The @command{g77} compiler has been changed back to
2115 assume @code{libf2c} has no aliasing problems in
2116 its implementations of the @code{COMPLEX} (and
2117 @code{DOUBLE COMPLEX}) intrinsics.
2118 The @code{libf2c} has been changed to have no such
2121 As a result, 0.5.20 is expected to offer improved performance
2122 over 0.5.19.1, perhaps as good as 0.5.19 in most
2123 or all cases, due to this change alone.
2125 @emph{Note:} This change requires version 0.5.20 of
2126 @code{libf2c}, at least, when linking code produced
2127 by any versions of @command{g77} other than 0.5.19.1.
2128 Use @samp{g77 -v} to determine the version numbers
2129 of the @code{libF77}, @code{libI77}, and @code{libU77}
2130 components of the @code{libf2c} library.
2131 (If these version numbers are not printed---in
2132 particular, if the linker complains about unresolved
2133 references to names like @samp{g77__fvers__}---that
2134 strongly suggests your installation has an obsolete
2135 version of @code{libf2c}.)
2139 New option @option{-fugly-assign} specifies that the
2140 same memory locations are to be used to hold the
2141 values assigned by both statements @samp{I = 3} and
2142 @samp{ASSIGN 10 TO I}, for example.
2143 (Normally, @command{g77} uses a separate memory location
2144 to hold assigned statement labels.)
2147 @xref{Ugly Assigned Labels}, for more information.
2151 @code{FORMAT} and @code{ENTRY} statements now are allowed to
2152 precede @code{IMPLICIT NONE} statements.
2154 @ifclear USERVISONLY
2156 Produce diagnostic for unsupported @code{SELECT CASE} on
2157 @code{CHARACTER} type, instead of crashing, at compile time.
2160 @ifclear USERVISONLY
2162 Fix crashes involving diagnosed or invalid code.
2165 @ifclear USERVISONLY
2167 Change approach to building @code{libf2c} archive
2168 (@file{libf2c.a}) so that members are added to it
2169 only when truly necessary, so the user that installs
2170 an already-built @command{g77} doesn't need to have write
2171 access to the build tree (whereas the user doing the
2172 build might not have access to install new software
2176 @ifclear USERVISONLY
2178 Support @command{gcc} version 2.7.2.2
2179 (modified by @command{g77} into version 2.7.2.2.f.2),
2181 support for prior versions of @command{gcc}.
2184 @ifclear USERVISONLY
2186 Upgrade to @code{libf2c} as of 1997-02-08, and
2187 fix up some of the build procedures.
2190 @ifclear USERVISONLY
2192 Improve general build procedures for @command{g77},
2193 fixing minor bugs (such as deletion of any file
2194 named @file{f771} in the parent directory of @code{gcc/}).
2198 Enable full support of @code{INTEGER(KIND=2)}
2199 (often referred to as @code{INTEGER*8})
2201 @code{libf2c} and @file{f2c.h} so that @command{f2c} users
2202 may make full use of its features via the @command{g77}
2203 version of @file{f2c.h} and the @code{INTEGER(KIND=2)}
2204 support routines in the @command{g77} version of @code{libf2c}.
2207 Improve @command{g77} driver and @code{libf2c} so that @samp{g77 -v}
2208 yields version information on the library.
2211 The @code{SNGL} and @code{FLOAT} intrinsics now are
2212 specific intrinsics, instead of synonyms for the
2213 generic intrinsic @code{REAL}.
2216 New intrinsics have been added.
2217 These are @code{REALPART}, @code{IMAGPART},
2219 @code{LONG}, and @code{SHORT}.
2222 A new group of intrinsics, @code{gnu}, has been added
2223 to contain the new @code{REALPART}, @code{IMAGPART},
2224 and @code{COMPLEX} intrinsics.
2225 An old group, @code{dcp}, has been removed.
2228 Complain about industry-wide ambiguous references
2229 @samp{REAL(@var{expr})} and @samp{AIMAG(@var{expr})},
2230 where @var{expr} is @code{DOUBLE COMPLEX} (or any
2231 complex type other than @code{COMPLEX}), unless
2232 @option{-ff90} option specifies Fortran 90 interpretation
2233 or new @option{-fugly-complex} option, in conjunction with
2234 @option{-fnot-f90}, specifies @command{f2c} interpretation.
2236 @ifclear USERVISONLY
2238 Make improvements to diagnostics.
2241 @ifclear USERVISONLY
2243 Speed up compiler a bit.
2246 @ifclear USERVISONLY
2248 Improvements to documentation and indexing, including
2249 a new chapter containing information on one, later
2250 more, diagnostics that users are directed to pull
2251 up automatically via a message in the diagnostic itself.
2253 (Hence the menu item @code{M} for the node
2254 @code{Diagnostics} in the top-level menu of
2255 the Info documentation.)
2259 @ifclear DOC-OLDNEWS
2260 @heading In previous versions:
2262 Information on previous versions is archived
2263 in @file{@value{path-g77}/news.texi}
2264 following the test of the @code{DOC-OLDNEWS} macro.
2268 @c 1997-02-01: 0.5.19.1 released.
2269 @heading In 0.5.19.1:
2272 Code-generation bugs afflicting operations on complex
2273 data have been fixed.
2275 These bugs occurred when assigning the result of an
2276 operation to a complex variable (or array element)
2277 that also served as an input to that operation.
2279 The operations affected by this bug were: @code{CONJG()},
2280 @code{DCONJG()}, @code{CCOS()}, @code{CDCOS()},
2281 @code{CLOG()}, @code{CDLOG()}, @code{CSIN()}, @code{CDSIN()},
2282 @code{CSQRT()}, @code{CDSQRT()}, complex division, and
2283 raising a @code{DOUBLE COMPLEX} operand to an @code{INTEGER}
2285 (The related generic and @samp{Z}-prefixed intrinsics,
2286 such as @code{ZSIN()}, also were affected.)
2288 For example, @samp{C = CSQRT(C)}, @samp{Z = Z/C}, and @samp{Z = Z**I}
2289 (where @samp{C} is @code{COMPLEX} and @samp{Z} is
2290 @code{DOUBLE COMPLEX}) have been fixed.
2293 @c 1996-12-07: 0.5.19 released.
2297 Fix @code{FORMAT} statement parsing so negative values for
2298 specifiers such as @code{P} (e.g. @samp{FORMAT(-1PF8.1)})
2299 are correctly processed as negative.
2302 Fix @code{SIGNAL} intrinsic so it once again accepts a
2303 procedure as its second argument.
2306 A temporary kludge option provides bare-bones information on
2307 @code{COMMON} and @code{EQUIVALENCE} members at debug time.
2310 New @option{-fonetrip} option specifies FORTRAN-66-style
2311 one-trip @code{DO} loops.
2314 New @option{-fno-silent} option causes names of program units
2315 to be printed as they are compiled, in a fashion similar to
2316 UNIX @command{f77} and @command{f2c}.
2319 New @option{-fugly-assumed} option specifies that arrays
2320 dimensioned via @samp{DIMENSION X(1)}, for example, are to be
2321 treated as assumed-size.
2324 New @option{-fno-typeless-boz} option specifies that non-decimal-radix
2325 constants using the prefixed-radix form (such as @samp{Z'1234'})
2326 are to be interpreted as @code{INTEGER} constants.
2329 New @option{-ff66} option is a ``shorthand'' option that specifies
2330 behaviors considered appropriate for FORTRAN 66 programs.
2333 New @option{-ff77} option is a ``shorthand'' option that specifies
2334 behaviors considered appropriate for UNIX @command{f77} programs.
2337 New @option{-fugly-comma} and @option{-fugly-logint} options provided
2338 to perform some of what @option{-fugly} used to do.
2339 @option{-fugly} and @option{-fno-ugly} are now ``shorthand'' options,
2340 in that they do nothing more than enable (or disable) other
2341 @option{-fugly-*} options.
2344 Fix parsing of assignment statements involving targets that
2345 are substrings of elements of @code{CHARACTER} arrays having
2346 names such as @samp{READ}, @samp{WRITE}, @samp{GOTO}, and
2347 @samp{REALFUNCTIONFOO}.
2350 Fix crashes involving diagnosed code.
2353 Fix handling of local @code{EQUIVALENCE} areas so certain cases
2354 of valid Fortran programs are not misdiagnosed as improperly
2355 extending the area backwards.
2358 Support @command{gcc} version 2.7.2.1.
2361 Upgrade to @code{libf2c} as of 1996-09-26, and
2362 fix up some of the build procedures.
2365 Change code generation for list-directed I/O so it allows
2366 for new versions of @code{libf2c} that might return nonzero
2367 status codes for some operations previously assumed to always
2370 This change not only affects how @code{IOSTAT=} variables
2371 are set by list-directed I/O, it also affects whether
2372 @code{END=} and @code{ERR=} labels are reached by these
2376 Add intrinsic support for new @code{FTELL} and @code{FSEEK}
2377 procedures in @code{libf2c}.
2380 Modify @code{fseek_()} in @code{libf2c} to be more portable
2381 (though, in practice, there might be no systems where this
2382 matters) and to catch invalid @code{whence} arguments.
2385 Some useless warnings from the @option{-Wunused} option have
2389 Fix a problem building the @file{f771} executable
2390 on AIX systems by linking with the @option{-bbigtoc} option.
2393 Abort configuration if @command{gcc} has not been patched
2394 using the patch file provided in the @file{gcc/f/gbe/}
2398 Add options @option{--help} and @option{--version} to the
2399 @command{g77} command, to conform to GNU coding guidelines.
2400 Also add printing of @command{g77} version number when
2401 the @option{--verbose} (@option{-v}) option is used.
2404 Change internally generated name for local @code{EQUIVALENCE}
2405 areas to one based on the alphabetically sorted first name
2406 in the list of names for entities placed at the beginning
2410 Improvements to documentation and indexing.
2413 @c 1996-04-01: 0.5.18 released.
2417 Add some rudimentary support for @code{INTEGER*1},
2418 @code{INTEGER*2}, @code{INTEGER*8},
2419 and their @code{LOGICAL} equivalents.
2420 (This support works on most, maybe all, @command{gcc} targets.)
2422 Thanks to Scott Snyder (@email{snyder@@d0sgif.fnal.gov})
2423 for providing the patch for this!
2425 Among the missing elements from the support for these
2426 features are full intrinsic support and constants.
2429 Add some rudimentary support for the @code{BYTE} and
2430 @code{WORD} type-declaration statements.
2431 @code{BYTE} corresponds to @code{INTEGER*1},
2432 while @code{WORD} corresponds to @code{INTEGER*2}.
2434 Thanks to Scott Snyder (@email{snyder@@d0sgif.fnal.gov})
2435 for providing the patch for this!
2438 The compiler code handling intrinsics has been largely
2439 rewritten to accommodate the new types.
2440 No new intrinsics or arguments for existing
2441 intrinsics have been added, so there is, at this
2442 point, no intrinsic to convert to @code{INTEGER*8},
2446 Support automatic arrays in procedures.
2449 Reduce space/time requirements for handling large
2450 @emph{sparsely} initialized aggregate arrays.
2451 This improvement applies to only a subset of
2452 the general problem to be addressed in 0.6.
2455 Treat initial values of zero as if they weren't
2456 specified (in DATA and type-declaration statements).
2457 The initial values will be set to zero anyway, but the amount
2458 of compile time processing them will be reduced,
2459 in some cases significantly (though, again, this
2460 is only a subset of the general problem to be
2463 A new option, @option{-fzeros}, is introduced to
2464 enable the traditional treatment of zeros as any
2468 With @option{-ff90} in force, @command{g77} incorrectly
2469 interpreted @samp{REAL(Z)} as returning a @code{REAL}
2470 result, instead of as a @code{DOUBLE PRECISION}
2472 (Here, @samp{Z} is @code{DOUBLE COMPLEX}.)
2474 With @option{-fno-f90} in force, the interpretation remains
2475 unchanged, since this appears to be how at least some
2476 F77 code using the @code{DOUBLE COMPLEX} extension expected
2479 Essentially, @samp{REAL(Z)} in F90 is the same as
2480 @samp{DBLE(Z)}, while in extended F77, it appears to
2481 be the same as @samp{REAL(REAL(Z))}.
2484 An expression involving exponentiation, where both operands
2485 were type @code{INTEGER} and the right-hand operand
2486 was negative, was erroneously evaluated.
2489 Fix bugs involving @code{DATA} implied-@code{DO} constructs
2490 (these involved an errant diagnostic and a crash, both on good
2491 code, one involving subsequent statement-function definition).
2494 Close @code{INCLUDE} files after processing them, so compiling source
2495 files with lots of @code{INCLUDE} statements does not result in
2496 being unable to open @code{INCLUDE} files after all the available
2497 file descriptors are used up.
2500 Speed up compiling, especially of larger programs, and perhaps
2501 slightly reduce memory utilization while compiling (this is
2502 @emph{not} the improvement planned for 0.6 involving large aggregate
2503 areas)---these improvements result from simply turning
2504 off some low-level code to do self-checking that hasn't been
2505 triggered in a long time.
2508 Introduce three new options that
2509 implement optimizations in the @command{gcc} back end (GBE).
2510 These options are @option{-fmove-all-movables}, @option{-freduce-all-givs},
2511 and @option{-frerun-loop-opt}, which are enabled, by default,
2512 for Fortran compilations.
2513 These optimizations are intended to help toon Fortran programs.
2516 Patch the GBE to do a better job optimizing certain
2517 kinds of references to array elements.
2520 Due to patches to the GBE, the version number of @command{gcc}
2521 also is patched to make it easier to manage installations,
2522 especially useful if it turns out a @command{g77} change to the
2525 The @command{g77}-modified version number is the @command{gcc}
2526 version number with the string @samp{.f.@var{n}} appended,
2527 where @samp{f} identifies the version as enhanced for
2528 Fortran, and @var{n} is @samp{1} for the first Fortran
2529 patch for that version of @command{gcc}, @samp{2} for the
2532 So, this introduces version 2.7.2.f.1 of @command{gcc}.
2535 Make several improvements and fixes to diagnostics, including
2536 the removal of two that were inappropriate or inadequate.
2539 Warning about two successive arithmetic operators, produced
2540 by @option{-Wsurprising}, now produced @emph{only} when both
2541 operators are, indeed, arithmetic (not relational/boolean).
2544 @option{-Wsurprising} now warns about the remaining cases
2545 of using non-integral variables for implied-@code{DO}
2546 loops, instead of these being rejected unless @option{-fpedantic}
2547 or @option{-fugly} specified.
2550 Allow @code{SAVE} of a local variable or array, even after
2551 it has been given an initial value via @code{DATA}, for example.
2554 Introduce an Info version of @command{g77} documentation, which
2555 supersedes @file{gcc/f/CREDITS}, @file{gcc/f/DOC}, and
2556 @file{gcc/f/PROJECTS}.
2557 These files will be removed in a future release.
2558 The files @file{gcc/f/BUGS}, @file{gcc/f/INSTALL}, and
2559 @file{gcc/f/NEWS} now are automatically built from
2560 the texinfo source when distributions are made.
2562 This effort was inspired by a first pass at translating
2563 @file{g77-0.5.16/f/DOC} that was contributed to Craig by
2564 David Ronis (@email{ronis@@onsager.chem.mcgill.ca}).
2567 New @option{-fno-second-underscore} option to specify
2568 that, when @option{-funderscoring} is in effect, a second
2569 underscore is not to be appended to Fortran names already
2570 containing an underscore.
2573 Change the way iterative @code{DO} loops work to follow
2575 In particular, calculation of the iteration count is
2576 still done by converting the start, end, and increment
2577 parameters to the type of the @code{DO} variable, but
2578 the result of the calculation is always converted to
2579 the default @code{INTEGER} type.
2581 (This should have no effect on existing code compiled
2582 by @command{g77}, but code written to assume that use
2583 of a @emph{wider} type for the @code{DO} variable
2584 will result in an iteration count being fully calculated
2585 using that wider type (wider
2586 than default @code{INTEGER}) must be rewritten.)
2589 Support @command{gcc} version 2.7.2.
2592 Upgrade to @code{libf2c} as of 1996-03-23, and
2593 fix up some of the build procedures.
2595 Note that the email addresses related to @command{f2c}
2596 have changed---the distribution site now is
2597 named @code{netlib.bell-labs.com}, and the
2598 maintainer's new address is @email{dmg@@bell-labs.com}.
2601 @c 1995-11-18: 0.5.17 released.
2605 @strong{Fix serious bug} in @samp{g77 -v} command that can cause removal of a
2606 system's @file{/dev/null} special file if run by user @code{root}.
2608 @strong{All users} of version 0.5.16 should ensure that
2609 they have not removed @file{/dev/null} or replaced it with an ordinary
2610 file (e.g. by comparing the output of @samp{ls -l /dev/null} with
2611 @samp{ls -l /dev/zero}.
2612 If the output isn't basically the
2613 same, contact your system
2614 administrator about restoring @file{/dev/null} to its proper status).
2616 This bug is particularly insidious because removing @file{/dev/null} as
2617 a special file can go undetected for quite a while, aside from
2618 various applications and programs exhibiting sudden, strange
2621 I sincerely apologize for not realizing the
2622 implications of the fact that when @samp{g77 -v} runs the @command{ld} command
2623 with @samp{-o /dev/null} that @command{ld} tries to @emph{remove} the executable
2624 it is supposed to build (especially if it reports unresolved
2625 references, which it should in this case)!
2628 Fix crash on @samp{CHARACTER*(*) FOO} in a main or block data program unit.
2631 Fix crash that can occur when diagnostics given outside of any
2632 program unit (such as when input file contains @samp{@@foo}).
2635 Fix crashes, infinite loops (hangs), and such involving diagnosed code.
2638 Fix @code{ASSIGN}'ed variables so they can be @code{SAVE}'d or dummy arguments,
2639 and issue clearer error message in cases where target of @code{ASSIGN}
2640 or @code{ASSIGN}ed @code{GOTO}/@code{FORMAT} is too small (which should
2644 Make @code{libf2c} build procedures work on more systems again by
2645 eliminating unnecessary invocations of @samp{ld -r -x} and @command{mv}.
2648 Fix omission of @option{-funix-intrinsics-@dots{}} options in list of permitted
2649 options to compiler.
2652 Fix failure to always diagnose missing type declaration for
2653 @code{IMPLICIT NONE}.
2656 Fix compile-time performance problem (which could sometimes
2657 crash the compiler, cause a hang, or whatever, due to a bug
2658 in the back end) involving exponentiation with a large @code{INTEGER}
2659 constant for the right-hand operator (e.g. @samp{I**32767}).
2662 Fix build procedures so cross-compiling @command{g77} (the @command{fini}
2663 utility in particular) is properly built using the host compiler.
2666 Add new @option{-Wsurprising} option to warn about constructs that are
2667 interpreted by the Fortran standard (and @command{g77}) in ways that
2668 are surprising to many programmers.
2671 Add @code{ERF()} and @code{ERFC()} as generic intrinsics mapping to existing
2672 @code{ERF}/@code{DERF} and @code{ERFC}/@code{DERFC} specific intrinsics.
2674 @emph{Note:} You should
2675 specify @samp{INTRINSIC ERF,ERFC} in any code where you might use
2676 these as generic intrinsics, to improve likelihood of diagnostics
2677 (instead of subtle run-time bugs) when using a compiler that
2678 doesn't support these as intrinsics (e.g. @command{f2c}).
2681 Remove from @option{-fno-pedantic} the diagnostic about @code{DO}
2682 with non-@code{INTEGER} index variable; issue that under
2683 @option{-Wsurprising} instead.
2686 Clarify some diagnostics that say things like ``ignored'' when that's
2690 Clarify diagnostic on use of @code{.EQ.}/@code{.NE.} on @code{LOGICAL}
2694 Minor improvements to code generation for various operations on
2695 @code{LOGICAL} operands.
2698 Minor improvement to code generation for some @code{DO} loops on some
2702 Support @command{gcc} version 2.7.1.
2705 Upgrade to @code{libf2c} as of 1995-11-15.
2708 @c 1995-08-30: 0.5.16 released.
2712 Fix a code-generation bug involving complicated @code{EQUIVALENCE} statements
2713 not involving @code{COMMON}.
2716 Fix code-generation bugs involving invoking ``gratis'' library procedures
2717 in @code{libf2c} from code compiled with @option{-fno-f2c} by making these
2718 procedures known to @command{g77} as intrinsics (not affected by -fno-f2c).
2719 This is known to fix code invoking @code{ERF()}, @code{ERFC()},
2720 @code{DERF()}, and @code{DERFC()}.
2723 Update @code{libf2c} to include netlib patches through 1995-08-16, and
2724 @code{#define} @code{WANT_LEAD_0} to 1 to make @command{g77}-compiled code more
2725 consistent with other Fortran implementations by outputting
2726 leading zeros in formatted and list-directed output.
2729 Fix a code-generation bug involving adjustable dummy arrays with high
2730 bounds whose primaries are changed during procedure execution, and
2731 which might well improve code-generation performance for such arrays
2732 compared to @command{f2c} plus @command{gcc} (but apparently only when using
2733 @file{gcc-2.7.0} or later).
2736 Fix a code-generation bug involving invocation of @code{COMPLEX} and
2737 @code{DOUBLE COMPLEX} @code{FUNCTION}s and doing @code{COMPLEX} and
2738 @code{DOUBLE COMPLEX} divides, when the result
2739 of the invocation or divide is assigned directly to a variable
2740 that overlaps one or more of the arguments to the invocation or divide.
2743 Fix crash by not generating new optimal code for @samp{X**I} if @samp{I} is
2744 nonconstant and the expression is used to dimension a dummy
2745 array, since the @command{gcc} back end does not support the necessary
2746 mechanics (and the @command{gcc} front end rejects the equivalent
2747 construct, as it turns out).
2750 Fix crash on expressions like @samp{COMPLEX**INTEGER}.
2753 Fix crash on expressions like @samp{(1D0,2D0)**2}, i.e. raising a
2754 @code{DOUBLE COMPLEX} constant to an @code{INTEGER} constant power.
2757 Fix crashes and such involving diagnosed code.
2760 Diagnose, instead of crashing on, statement function definitions
2761 having duplicate dummy argument names.
2764 Fix bug causing rejection of good code involving statement function
2768 Fix bug resulting in debugger not knowing size of local equivalence
2769 area when any member of area has initial value (via @code{DATA},
2773 Fix installation bug that prevented installation of @command{g77} driver.
2774 Provide for easy selection of whether to install copy of @command{g77}
2775 as @command{f77} to replace the broken code.
2778 Fix @command{gcc} driver (affects @command{g77} thereby) to not
2779 gratuitously invoke the
2780 @code{f771} program (e.g. when @option{-E} is specified).
2783 Fix diagnostic to point to correct source line when it immediately
2784 follows an @code{INCLUDE} statement.
2787 Support more compiler options in @command{gcc}/@command{g77} when
2788 compiling Fortran files.
2789 These options include @option{-p}, @option{-pg}, @option{-aux-info}, @option{-P},
2790 correct setting of version-number macros for preprocessing, full
2791 recognition of @option{-O0}, and
2792 automatic insertion of configuration-specific linker specs.
2795 Add new intrinsics that interface to existing routines in @code{libf2c}:
2796 @code{ABORT}, @code{DERF}, @code{DERFC}, @code{ERF}, @code{ERFC}, @code{EXIT},
2797 @code{FLUSH}, @code{GETARG}, @code{GETENV}, @code{IARGC},
2798 @code{SIGNAL}, and @code{SYSTEM}.
2799 Note that @code{ABORT}, @code{EXIT}, @code{FLUSH}, @code{SIGNAL}, and
2800 @code{SYSTEM} are intrinsic subroutines, not functions (since they
2801 have side effects), so to get the return values from @code{SIGNAL}
2802 and @code{SYSTEM}, append a final argument specifying an @code{INTEGER}
2803 variable or array element (e.g. @samp{CALL SYSTEM('rm foo',ISTAT)}).
2806 Add new intrinsic group named @code{unix} to contain the new intrinsics,
2807 and by default enable this new group.
2810 Move @code{LOC()} intrinsic out of the @code{vxt} group to the new
2814 Improve @command{g77} so that @samp{g77 -v} by itself (or with
2815 certain other options, including @option{-B}, @option{-b}, @option{-i},
2816 @option{-nostdlib}, and @option{-V}) reports lots more useful
2817 version info, and so that long-form options @command{gcc} accepts are
2818 understood by @command{g77} as well (even in truncated, unambiguous forms).
2821 Add new @command{g77} option @option{--driver=name} to specify driver when
2822 default, @command{gcc}, isn't appropriate.
2825 Add support for @samp{#} directives (as output by the preprocessor) in the
2826 compiler, and enable generation of those directives by the
2827 preprocessor (when compiling @samp{.F} files) so diagnostics and debugging
2828 info are more useful to users of the preprocessor.
2831 Produce better diagnostics, more like @command{gcc}, with info such as
2832 @samp{In function `foo':} and @samp{In file included from...:}.
2835 Support @command{gcc}'s @option{-fident} and @option{-fno-ident} options.
2838 When @option{-Wunused} in effect, don't warn about local variables used as
2839 statement-function dummy arguments or @code{DATA} implied-@code{DO} iteration
2840 variables, even though, strictly speaking, these are not uses
2841 of the variables themselves.
2844 When @samp{-W -Wunused} in effect, don't warn about unused dummy arguments
2845 at all, since there's no way to turn this off for individual
2846 cases (@command{g77} might someday start warning about these)---applies
2847 to @command{gcc} versions 2.7.0 and later, since earlier versions didn't
2848 warn about unused dummy arguments.
2851 New option @option{-fno-underscoring} that inhibits transformation of names
2852 (by appending one or two underscores) so users may experiment
2853 with implications of such an environment.
2856 Minor improvement to @file{gcc/f/info} module to make it easier to build
2857 @command{g77} using the native (non-@command{gcc}) compiler on certain machines
2858 (but definitely not all machines nor all non-@command{gcc} compilers).
2860 do not report bugs showing problems compilers have with
2861 macros defined in @file{gcc/f/target.h} and used in places like
2862 @file{gcc/f/expr.c}.
2865 Add warning to be printed for each invocation of the compiler
2866 if the target machine @code{INTEGER}, @code{REAL}, or @code{LOGICAL} size
2868 since @command{g77} is known to not work well for such cases.
2871 Lots of new documentation (though work is still needed to put it into
2872 canonical GNU format).
2875 Build @code{libf2c} with @option{-g0}, not @option{-g2}, in effect
2876 (by default), to produce
2877 smaller library without lots of debugging clutter.
2880 @c 1995-05-19: 0.5.15 released.
2884 Fix bad code generation involving @samp{X**I} and temporary, internal variables
2885 generated by @command{g77} and the back end (such as for @code{DO} loops).
2888 Fix crash given @samp{CHARACTER A;DATA A/.TRUE./}.
2891 Replace crash with diagnostic given @samp{CHARACTER A;DATA A/1.0/}.
2894 Fix crash or other erratic behavior when null character constant
2895 (@samp{''}) is encountered.
2898 Fix crash or other erratic behavior involving diagnosed code.
2901 Fix code generation for external functions returning type @code{REAL} when
2902 the @option{-ff2c} option is in force (which it is by default) so that
2903 @command{f2c} compatibility is indeed provided.
2906 Disallow @samp{COMMON I(10)} if @samp{I} has previously been specified
2907 with an array declarator.
2910 New @option{-ffixed-line-length-@var{n}} option, where @var{n} is the
2912 of a typical fixed-form line, defaulting to 72 columns, such
2913 that characters beyond column @var{n} are ignored, or @var{n} is @samp{none},
2914 meaning no characters are ignored.
2915 does not affect lines
2916 with @samp{&} in column 1, which are always processed as if
2917 @option{-ffixed-line-length-none} was in effect.
2920 No longer generate better code for some kinds of array references,
2921 as @command{gcc} back end is to be fixed to do this even better, and it
2922 turned out to slow down some code in some cases after all.
2925 In @code{COMMON} and @code{EQUIVALENCE} areas with any members given initial
2926 values (e.g. via @code{DATA}), uninitialized members now always
2927 initialized to binary zeros (though this is not required by
2928 the standard, and might not be done in future versions
2930 Previously, in some @code{COMMON}/@code{EQUIVALENCE} areas
2931 (essentially those with members of more than one type), the
2932 uninitialized members were initialized to spaces, to
2933 cater to @code{CHARACTER} types, but it seems no existing code expects
2934 that, while much existing code expects binary zeros.
2940 Don't emit bad code when low bound of adjustable array is nonconstant
2941 and thus might vary as an expression at run time.
2944 Emit correct code for calculation of number of trips in @code{DO} loops
2946 where the loop should not execute at all.
2947 (This bug affected cases
2948 where the difference between the begin and end values was less
2949 than the step count, though probably not for floating-point cases.)
2952 Fix crash when extra parentheses surround item in
2953 @code{DATA} implied-@code{DO} list.
2956 Fix crash over minor internal inconsistencies in handling diagnostics,
2957 just substitute dummy strings where necessary.
2960 Fix crash on some systems when compiling call to @code{MVBITS()} intrinsic.
2963 Fix crash on array assignment @samp{TYPE@var{ddd}(@dots{})=@dots{}}, where @var{ddd}
2964 is a string of one or more digits.
2967 Fix crash on @code{DCMPLX()} with a single @code{INTEGER} argument.
2970 Fix various crashes involving code with diagnosed errors.
2973 Support @option{-I} option for @code{INCLUDE} statement, plus @command{gcc}'s
2974 @file{header.gcc} facility for handling systems like MS-DOS.
2977 Allow @code{INCLUDE} statement to be continued across multiple lines,
2978 even allow it to coexist with other statements on the same line.
2981 Incorporate Bellcore fixes to @code{libf2c} through 1995-03-15---this
2982 fixes a bug involving infinite loops reading EOF with empty list-directed
2986 Remove all the @command{g77}-specific auto-configuration scripts, code,
2988 except for temporary substitutes for bsearch() and strtoul(), as
2989 too many configure/build problems were reported in these areas.
2990 People will have to fix their systems' problems themselves, or at
2991 least somewhere other than @command{g77}, which expects a working ANSI C
2992 environment (and, for now, a GNU C compiler to compile @command{g77} itself).
2995 Complain if initialized common redeclared as larger in subsequent program
2999 Warn if blank common initialized, since its size can vary and hence
3000 related warnings that might be helpful won't be seen.
3003 New @option{-fbackslash} option, on by default, that causes @samp{\}
3004 within @code{CHARACTER}
3005 and Hollerith constants to be interpreted a la GNU C.
3007 this behavior is somewhat different from @command{f2c}'s, which supports only
3008 a limited subset of backslash (escape) sequences.
3011 Make @option{-fugly-args} the default.
3014 New @option{-fugly-init} option, on by default, that allows typeless/Hollerith
3015 to be specified as initial values for variables or named constants
3016 (@code{PARAMETER}), and also allows character<->numeric conversion in
3017 those contexts---turn off via @option{-fno-ugly-init}.
3020 New @option{-finit-local-zero} option to initialize
3021 local variables to binary zeros.
3022 This does not affect whether they are @code{SAVE}d, i.e. made
3023 automatic or static.
3026 New @option{-Wimplicit} option to warn about implicitly typed variables, arrays,
3028 (Basically causes all program units to default to @code{IMPLICIT NONE}.)
3031 @option{-Wall} now implies @option{-Wuninitialized} as with @command{gcc}
3032 (i.e. unless @option{-O} not specified, since @option{-Wuninitialized}
3033 requires @option{-O}), and implies @option{-Wunused} as well.
3036 @option{-Wunused} no longer gives spurious messages for unused
3037 @code{EXTERNAL} names (since they are assumed to refer to block data
3038 program units, to make use of libraries more reliable).
3041 Support @code{%LOC()} and @code{LOC()} of character arguments.
3044 Support null (zero-length) character constants and expressions.
3047 Support @command{f2c}'s @code{IMAG()} generic intrinsic.
3050 Support @code{ICHAR()}, @code{IACHAR()}, and @code{LEN()} of
3051 character expressions that are valid in assignments but
3052 not normally as actual arguments.
3055 Support @command{f2c}-style @samp{&} in column 1 to mean continuation line.
3058 Allow @code{NAMELIST}, @code{EXTERNAL}, @code{INTRINSIC}, and @code{VOLATILE}
3059 in @code{BLOCK DATA}, even though these are not allowed by the standard.
3062 Allow @code{RETURN} in main program unit.
3065 Changes to Hollerith-constant support to obey Appendix C of the
3070 Now padded on the right with zeros, not spaces.
3073 Hollerith ``format specifications'' in the form of arrays of
3074 non-character allowed.
3077 Warnings issued when non-space truncation occurs when converting
3081 When specified as actual argument, now passed
3082 by reference to @code{INTEGER} (padded on right with spaces if constant
3083 too small, otherwise fully intact if constant wider the @code{INTEGER}
3084 type) instead of by value.
3087 @strong{Warning:} @command{f2c} differs on the
3088 interpretation of @samp{CALL FOO(1HX)}, which it treats exactly the
3089 same as @samp{CALL FOO('X')}, but which the standard and @command{g77} treat
3090 as @samp{CALL FOO(%REF('X '))} (padded with as many spaces as necessary
3091 to widen to @code{INTEGER}), essentially.
3094 Changes and fixes to typeless-constant support:
3098 Now treated as a typeless double-length @code{INTEGER} value.
3101 Warnings issued when overflow occurs.
3104 Padded on the left with zeros when converting
3108 Should be properly aligned and ordered on
3109 the target machine for whatever type it is turned into.
3112 When specified as actual argument, now passed as reference to
3113 a default @code{INTEGER} constant.
3117 @code{%DESCR()} of a non-@code{CHARACTER} expression now passes a pointer to
3118 the expression plus a length for the expression just as if
3119 it were a @code{CHARACTER} expression.
3120 For example, @samp{CALL FOO(%DESCR(D))}, where
3121 @samp{D} is @code{REAL*8}, is the same as @samp{CALL FOO(D,%VAL(8)))}.
3124 Name of multi-entrypoint master function changed to incorporate
3125 the name of the primary entry point instead of a decimal
3126 value, so the name of the master function for @samp{SUBROUTINE X}
3127 with alternate entry points is now @samp{__g77_masterfun_x}.
3130 Remove redundant message about zero-step-count @code{DO} loops.
3133 Clean up diagnostic messages, shortening many of them.
3136 Fix typo in @command{g77} man page.
3139 Clarify implications of constant-handling bugs in @file{f/BUGS}.
3142 Generate better code for @samp{**} operator with a right-hand operand of
3143 type @code{INTEGER}.
3146 Generate better code for @code{SQRT()} and @code{DSQRT()},
3147 also when @option{-ffast-math}
3148 specified, enable better code generation for @code{SIN()} and @code{COS()}.
3151 Generate better code for some kinds of array references.
3154 Speed up lexing somewhat (this makes the compilation phase noticeably