1 @c Copyright (C) 1995, 1996, 1997, 1998, 1999 Free Software Foundation, Inc.
2 @c This is part of the G77 manual.
3 @c For copying conditions, see the file g77.texi.
5 @c The text of this file appears in the file NEWS
6 @c in the G77 distribution, as well as in the G77 manual.
8 @c Keep this the same as the dates above, since it's used
9 @c in the standalone derivations of this file (e.g. NEWS).
10 @set copyrights-news 1995-1999
12 @set last-update-news 1999-07-08
17 @c The immediately following lines apply to the NEWS file
18 @c which is derived from this file.
19 @emph{Note:} This file is automatically generated from the files
20 @file{news0.texi} and @file{news.texi}.
21 @file{NEWS} is @emph{not} a source file,
22 although it is normally included within source distributions.
24 This file lists news about the @value{which-g77} version
25 (and some other versions) of the GNU Fortran compiler.
26 Copyright (C) @value{copyrights-news} Free Software Foundation, Inc.
27 You may copy, distribute, and modify it freely as long as you preserve
28 this copyright notice and permission notice.
31 @chapter News About GNU Fortran
37 @chapter User-visible Changes
38 @cindex versions, recent
39 @cindex recent versions
40 @cindex changes, user-visible
41 @cindex user-visible changes
43 This chapter describes changes to @code{g77} that are visible
44 to the programmers who actually write and maintain Fortran
45 code they compile with @code{g77}.
46 Information on changes to installation procedures,
47 changes to the documentation, and bug fixes is
48 not provided here, unless it is likely to affect how
50 @xref{News,,News About GNU Fortran}, for information on
51 such changes to @code{g77}.
56 @chapter News About GNU Fortran
57 @cindex versions, recent
58 @cindex recent versions
63 Changes made to recent versions of GNU Fortran are listed
64 below, with the most recent version first.
66 The changes are generally listed in order:
70 Code-generation and run-time-library bug-fixes
73 Compiler and run-time-library crashes involving valid code
80 Fixes and enhancements to existing features
92 This order is not strict---for example, some items
93 involve a combination of these elements.
96 Note that two variants of @code{g77} are tracked below.
97 The @code{egcs} variant is described vis-a-vis
98 previous versions of @code{egcs} and/or
99 an official FSF version,
102 Therefore, @code{egcs} versions sometimes have multiple listings
103 to help clarify how they differ from other versions,
104 though this can make getting a complete picture
105 of what a particular @code{egcs} version contains
106 somewhat more difficult.
109 For information on bugs in the @value{which-g77} version of @code{g77},
110 see @ref{Known Bugs,,Known Bugs In GNU Fortran}.
114 For information on bugs in the @value{which-g77} version of @code{g77},
115 see @file{@value{path-g77}/BUGS}.
119 @emph{Warning:} The information below is still under development,
120 and might not accurately reflect the @code{g77} code base
121 of which it is a part.
122 Efforts are made to keep it somewhat up-to-date,
123 but they are particularly concentrated
124 on any version of this information
125 that is distributed as part of a @emph{released} @code{g77}.
127 In particular, while this information is intended to apply to
128 the @value{which-g77} version of @code{g77},
129 only an official @emph{release} of that version
130 is expected to contain documentation that is
131 most consistent with the @code{g77} product in that version.
133 Nevertheless, information on @emph{previous} releases of @code{g77}, below,
134 is likely to be more up-to-date and accurate
135 than the equivalent information that accompanied
137 assuming the last-updated date of the information below
138 is later than the dates of those releases.
140 That's due to attempts to keep this development version
141 of news about previous @code{g77} versions up-to-date.
145 An online, ``live'' version of this document
146 (derived directly from the mainline, development version
147 of @code{g77} within @code{egcs})
149 @uref{http://www.gnu.org/software/gcc/onlinedocs/g77_news.html}.
152 The following information was last updated on @value{last-update-news}:
154 @heading In development:
158 Using options @code{-g}, @code{-ggdb} or @code{-gdwarf[-2]} (where
159 appropriate for your target) now also enables debugging information
160 for COMMON BLOCK and EQUIVALENCE items to be emitted.
165 The @code{TtyNam} intrinsics now set @var{Name} to all spaces (at run time)
166 if the system has no @code{ttyname} implementation available.
171 Upgrade to @code{libf2c} as of 1999-06-28.
173 This fixes a bug whereby
174 input to a @code{NAMELIST} read involving a repeat count,
175 such as @samp{K(5)=10*3},
176 was not properly handled by @code{libf2c}.
177 The first item was written to @samp{K(5)},
178 but the remaining nine were written elsewhere (still within the array),
179 not necessarily starting at @samp{K(6)}.
183 @heading In 0.5.25, @code{GCC} 2.95 (@code{EGCS} 1.2) versus @code{EGCS} 1.1.2:
187 @code{g77} no longer generates bad code for assignments,
188 or other conversions,
189 of @code{REAL} or @code{COMPLEX} constant expressions
190 to type @code{INTEGER(KIND=2)}
191 (often referred to as @code{INTEGER*8}).
193 For example, @samp{INTEGER*8 J; J = 4E10} now works as documented.
198 @code{g77} no longer truncates @code{INTEGER(KIND=2)}
199 (usually @code{INTEGER*8})
200 subscript expressions when evaluating array references
201 on systems with pointers widers than @code{INTEGER(KIND=1)}
207 @code{g77} no longer generates bad code
208 for an assignment to a @code{COMPLEX} variable or array
209 that partially overlaps one or more of the sources
210 of the same assignment
211 (a very rare construction).
212 It now assigns through a temporary,
213 in cases where such partial overlap is deemed possible.
218 @code{libg2c} (@code{libf2c}) no longer loses track
219 of the file being worked on
220 during a @code{BACKSPACE} operation.
225 @code{libg2c} (@code{libf2c}) fixes a bug whereby
226 input to a @code{NAMELIST} read involving a repeat count,
227 such as @samp{K(5)=10*3},
228 was not properly handled by @code{libf2c}.
229 The first item was written to @samp{K(5)},
230 but the remaining nine were written elsewhere (still within the array),
231 not necessarily starting at @samp{K(6)}.
236 @c Tim Prince reported this, regarding the TEST_FPU benchmark.
237 Automatic arrays now seem to be working on HP-UX systems.
242 The @code{Date} intrinsic now returns the correct result
243 on big-endian systems.
248 Fix @code{g77} so it no longer crashes when compiling
249 I/O statements using keywords that define @code{INTEGER} values,
250 such as @samp{IOSTAT=@var{j}},
251 where @var{j} is other than default @code{INTEGER}
252 (such as @code{INTEGER*2}).
253 Instead, it issues a diagnostic.
258 Fix @code{g77} so it properly handles @samp{DATA A/@var{rpt}*@var{val}/},
259 where @var{rpt} is not default @code{INTEGER}, such as @code{INTEGER*2},
260 instead of producing a spurious diagnostic.
261 Also fix @samp{DATA (A(I),I=1,N)},
262 where @samp{N} is not default @code{INTEGER}
263 to work instead of crashing @code{g77}.
268 The @samp{-ax} option is now obeyed when compiling Fortran programs.
269 (It is passed to the @file{f771} driver.)
273 The new @samp{-fbounds-check} option
274 causes @code{g77} to compile run-time bounds checks
275 of array subscripts, as well as of substring start and end points.
278 @code{libg2c} now supports building as multilibbed library,
279 which provides better support for systems
280 that require options such as @samp{-mieee}
284 Source file names with the suffixes @samp{.FOR} and @samp{.FPP}
285 now are recognized by @code{g77}
286 as if they ended in @samp{.for} and @samp{.fpp}, respectively.
289 The order of arguments to the @emph{subroutine} forms of the
290 @code{CTime}, @code{DTime}, @code{ETime}, and @code{TtyNam}
291 intrinsics has been swapped.
292 The argument serving as the returned value
293 for the corresponding function forms
294 now is the @emph{second} argument,
295 making these consistent with the other subroutine forms
296 of @code{libU77} intrinsics.
299 @code{g77} now warns about a reference to an intrinsic
300 that has an interface that is not Year 2000 (Y2K) compliant.
301 Also, @code{libg2c} has been changed to increase the likelihood
302 of catching references to the implementations of these intrinsics
303 using the @code{EXTERNAL} mechanism
304 (which would avoid the new warnings).
307 @xref{Year 2000 (Y2K) Problems}, for more information.
312 @code{g77} now warns about a reference to a function
313 when the corresponding @emph{subsequent} function program unit
314 disagrees with the reference concerning the type of the function.
318 @samp{-fno-emulate-complex} is now the default option.
319 This should result in improved performance
320 of code that uses the @code{COMPLEX} data type.
323 @cindex double-precision performance
324 @cindex -malign-double
326 The @samp{-malign-double} option
327 now reliably aligns @emph{all} double-precision variables and arrays
328 on Intel x86 targets.
332 Even without the @samp{-malign-double} option,
333 @code{g77} reliably aligns local double-precision variables
334 that are not in @code{EQUIVALENCE} areas
335 and not @code{SAVE}'d.
340 @code{g77} now open-codes (``inlines'') division of @code{COMPLEX} operands
341 instead of generating a run-time call to
342 the @code{libf2c} routines @code{c_div} or @code{z_div},
343 unless the @samp{-Os} option is specified.
347 @code{g77} no longer generates code to maintain @code{errno},
348 a C-language concept,
349 when performing operations such as the @code{SqRt} intrinsic.
353 @code{g77} developers can temporarily use
354 the @samp{-fflatten-arrays} option
355 to compare how the compiler handles code generation
356 using C-like constructs as compared to the
357 Fortran-like method constructs normally used.
362 A substantial portion of the @code{g77} front end's code-generation component
364 It now generates code using facilities more robustly supported
365 by the @code{gcc} back end.
366 One effect of this rewrite is that some codes no longer produce
367 a spurious ``label @var{lab} used before containing binding contour''
372 Support for the @samp{-fugly} option has been removed.
376 Improve documentation and indexing,
377 including information on Year 2000 (Y2K) compliance,
378 and providing more information on internals of the front end.
383 Upgrade to @code{libf2c} as of 1999-05-10.
387 @heading In 0.5.24 versus 0.5.23:
389 There is no @code{g77} version 0.5.24 at this time,
391 0.5.24 is the version number designated for bug fixes and,
392 perhaps, some new features added,
394 Version 0.5.23 requires @code{gcc} 2.8.1,
395 as 0.5.24 was planned to require.
397 Due to @code{EGCS} becoming @code{GCC}
398 (which is now an acronym for ``GNU Compiler Collection''),
399 and @code{EGCS} 1.2 becoming officially designated @code{GCC} 2.95,
400 there seems to be no need for an actual 0.5.24 release.
402 To reduce the confusion already resulting from use of 0.5.24
403 to designate @code{g77} versions within @code{EGCS} versions 1.0 and 1.1,
404 as well as in versions of @code{g77} documentation and notices
406 ``mainline'' @code{g77} version numbering resumes
407 at 0.5.25 with @code{GCC} 2.95 (@code{EGCS} 1.2),
408 skipping over 0.5.24 as a placeholder version number.
410 To repeat, there is no @code{g77} 0.5.24, but there is now a 0.5.25.
411 Please remain calm and return to your keypunch units.
413 @c 1999-03-15: EGCS 1.1.2 released.
414 @heading In @code{EGCS} 1.1.2 versus @code{EGCS} 1.1.1:
418 Fix the @code{IDate} intrinsic (VXT) (in @code{libg2c})
419 so the returned year is in the documented, non-Y2K-compliant range
421 instead of being returned as 100 in the year 2000.
424 @xref{IDate Intrinsic (VXT)},
425 for more information.
429 Fix the @code{Date_and_Time} intrinsic (in @code{libg2c})
430 to return the milliseconds value properly
434 Fix the @code{LStat} intrinsic (in @code{libg2c})
435 to return device-ID information properly
439 Improve documentation.
443 @c 1998-12-04: EGCS 1.1.1 released.
444 @heading In @code{EGCS} 1.1.1 versus @code{EGCS} 1.1:
448 Fix @code{libg2c} so it performs an implicit @code{ENDFILE} operation
450 whenever a @code{REWIND} is done.
452 (This bug was introduced in 0.5.23 and @code{egcs} 1.1 in
453 @code{g77}'s version of @code{libf2c}.)
456 Fix @code{libg2c} so it no longer crashes with a spurious diagnostic
457 upon doing any I/O following a direct formatted write.
459 (This bug was introduced in 0.5.23 and @code{egcs} 1.1 in
460 @code{g77}'s version of @code{libf2c}.)
463 Fix @code{g77} so it no longer crashes compiling references
464 to the @code{Rand} intrinsic on some systems.
467 Fix @code{g77} portion of installation process so it works
468 better on some systems
469 (those with shells requiring @samp{else true} clauses
470 on @code{if} constructs
471 for the completion code to be set properly).
475 @c 1998-09-03: EGCS 1.1 released.
476 @heading In @code{EGCS} 1.1 versus @code{EGCS} 1.0.3:
480 Fix bugs in the @code{libU77} intrinsic @code{HostNm}
481 that wrote one byte beyond the end of its @code{CHARACTER}
483 and in the @code{libU77} intrinsics
484 @code{GMTime} and @code{LTime}
485 that overwrote their arguments.
490 Assumed arrays with negative bounds
491 (such as @samp{REAL A(-1:*)})
492 no longer elicit spurious diagnostics from @code{g77},
493 even on systems with pointers having
494 different sizes than integers.
496 This bug is not known to have existed in any
497 recent version of @code{gcc}.
498 It was introduced in an early release of @code{egcs}.
503 Valid combinations of @code{EXTERNAL},
504 passing that external as a dummy argument
505 without explicitly giving it a type,
506 and, in a subsequent program unit,
507 referencing that external as
508 an external function with a different type
509 no longer crash @code{g77}.
514 @code{CASE DEFAULT} no longer crashes @code{g77}.
519 The @samp{-Wunused} option no longer issues a spurious
520 warning about the ``master'' procedure generated by
521 @code{g77} for procedures containing @code{ENTRY} statements.
525 Support @samp{FORMAT(I<@var{expr}>)} when @var{expr} is a
526 compile-time constant @code{INTEGER} expression.
529 Fix @code{g77} @samp{-g} option so procedures that
530 use @code{ENTRY} can be stepped through, line by line,
534 Allow any @code{REAL} argument to intrinsics
535 @code{Second} and @code{CPU_Time}.
538 Use @code{tempnam}, if available, to open scratch files
539 (as in @samp{OPEN(STATUS='SCRATCH')})
540 so that the @code{TMPDIR} environment variable,
544 @code{g77}'s version of @code{libf2c} separates out
545 the setting of global state
546 (such as command-line arguments and signal handling)
547 from @file{main.o} into distinct, new library
550 This should make it easier to write portable applications
551 that have their own (non-Fortran) @code{main()} routine
552 properly set up the @code{libf2c} environment, even
553 when @code{libf2c} (now @code{libg2c}) is a shared library.
557 @code{g77} no longer installs the @file{f77} command
558 and @file{f77.1} man page
559 in the @file{/usr} or @file{/usr/local} heirarchy,
560 even if the @file{f77-install-ok} file exists
561 in the source or build directory.
562 See the installation documentation for more information.
567 @code{g77} no longer installs the @file{libf2c.a} library
568 and @file{f2c.h} include file
569 in the @file{/usr} or @file{/usr/local} heirarchy,
570 even if the @file{f2c-install-ok} or @file{f2c-exists-ok} files exist
571 in the source or build directory.
572 See the installation documentation for more information.
577 The @file{libf2c.a} library produced by @code{g77} has been
578 renamed to @file{libg2c.a}.
579 It is installed only in the @code{gcc} ``private''
580 directory heirarchy, @file{gcc-lib}.
581 This allows system administrators and users to choose which
582 version of the @code{libf2c} library from @code{netlib} they
583 wish to use on a case-by-case basis.
584 See the installation documentation for more information.
589 The @file{f2c.h} include (header) file produced by @code{g77}
590 has been renamed to @file{g2c.h}.
591 It is installed only in the @code{gcc} ``private''
592 directory heirarchy, @file{gcc-lib}.
593 This allows system administrators and users to choose which
594 version of the include file from @code{netlib} they
595 wish to use on a case-by-case basis.
596 See the installation documentation for more information.
600 The @code{g77} command now expects the run-time library
601 to be named @code{libg2c.a} instead of @code{libf2c.a},
602 to ensure that a version other than the one built and
603 installed as part of the same @code{g77} version is picked up.
607 During the configuration and build process,
608 @code{g77} creates subdirectories it needs only as it
610 Other cleaning up of the configuration and build process
611 has been performed as well.
616 @code{install-info} now used to update the directory of
617 Info documentation to contain an entry for @code{g77}
618 (during installation).
622 Some diagnostics have been changed from warnings to errors,
623 to prevent inadvertent use of the resulting, probably buggy,
625 These mostly include diagnostics about use of unsupported features
626 in the @code{OPEN}, @code{INQUIRE}, @code{READ}, and
627 @code{WRITE} statements,
628 and about truncations of various sorts of constants.
632 Improve compilation of @code{FORMAT} expressions so that
633 a null byte is appended to the last operand if it
635 This provides a cleaner run-time diagnostic as provided
636 by @code{libf2c} for statements like @samp{PRINT '(I1', 42}.
641 Improve documentation and indexing.
646 The upgrade to @code{libf2c} as of 1998-06-18
647 should fix a variety of problems, including
648 those involving some uses of the @code{T} format
649 specifier, and perhaps some build (porting) problems
654 @c 1998-09-03: EGCS 1.1 released.
655 @heading In @code{EGCS} 1.1 versus @code{g77} 0.5.23:
659 @cindex stack, 387 coprocessor
663 Fix a code-generation bug that afflicted
664 Intel x86 targets when @samp{-O2} was specified
665 compiling, for example, an old version of
666 the @code{DNRM2} routine.
668 The x87 coprocessor stack was being
669 mismanaged in cases involving assigned @code{GOTO}
675 @code{g77} no longer produces incorrect code
677 for @code{EQUIVALENCE} and @code{COMMON}
678 aggregates that, due to ``unnatural'' ordering of members
679 vis-a-vis their types, require initial padding.
684 Fix @code{g77} crash compiling code
685 containing the construct @samp{CMPLX(0.)} or similar.
690 @code{g77} no longer crashes when compiling code
691 containing specification statements such as
692 @samp{INTEGER(KIND=7) PTR}.
697 @code{g77} no longer crashes when compiling code
698 such as @samp{J = SIGNAL(1, 2)}.
702 @code{g77} now treats @samp{%LOC(@var{expr})} and
703 @samp{LOC(@var{expr})} as ``ordinary'' expressions
704 when they are used as arguments in procedure calls.
705 This change applies only to global (filewide) analysis,
706 making it consistent with
707 how @code{g77} actually generates code
710 Previously, @code{g77} treated these expressions
711 as denoting special ``pointer'' arguments
712 for the purposes of filewide analysis.
717 (or apparently infinite run-time)
718 when compiling certain complicated expressions
719 involving @code{COMPLEX} arithmetic
720 (especially multiplication).
724 @cindex double-precision performance
725 @cindex -malign-double
727 Align static double-precision variables and arrays
729 regardless of whether @samp{-malign-double} is specified.
731 Generally, this affects only local variables and arrays
732 having the @code{SAVE} attribute
733 or given initial values via @code{DATA}.
736 The @code{g77} driver now ensures that @samp{-lg2c}
737 is specified in the link phase prior to any
738 occurrence of @samp{-lm}.
739 This prevents accidentally linking to a routine
740 in the SunOS4 @samp{-lm} library
741 when the generated code wants to link to the one
742 in @code{libf2c} (@code{libg2c}).
745 @code{g77} emits more debugging information when
748 This new information allows, for example,
749 @kbd{which __g77_length_a} to be used in @code{gdb}
750 to determine the type of the phantom length argument
751 supplied with @code{CHARACTER} variables.
753 This information pertains to internally-generated
754 type, variable, and other information,
755 not to the longstanding deficiencies vis-a-vis
756 @code{COMMON} and @code{EQUIVALENCE}.
759 The F90 @code{Date_and_Time} intrinsic now is
763 The F90 @code{System_Clock} intrinsic allows
764 the optional arguments (except for the @code{Count}
765 argument) to be omitted.
769 Upgrade to @code{libf2c} as of 1998-06-18.
774 Improve documentation and indexing.
779 @heading In previous versions:
781 Information on previous versions is not provided
782 in this @file{@value{path-g77}/NEWS} file,
784 See @file{@value{path-g77}/news.texi},
785 or any of its other derivations
786 (Info, HTML, dvi forms)
787 for such information.
791 @c 1998-05-20: 0.5.23 released.
792 @heading In 0.5.23 versus 0.5.22:
795 This release contains several regressions against
796 version 0.5.22 of @code{g77}, due to using the
797 ``vanilla'' @code{gcc} back end instead of patching
798 it to fix a few bugs and improve performance in a
801 Features that have been dropped from this version
802 of @code{g77} due to their being implemented
803 via @code{g77}-specific patches to the @code{gcc}
804 back end in previous releases include:
808 Support for @code{__restrict__} keyword,
809 the options @samp{-fargument-alias}, @samp{-fargument-noalias},
810 and @samp{-fargument-noalias-global},
811 and the corresponding alias-analysis code.
813 (@code{egcs} has the alias-analysis
814 code, but not the @code{__restrict__} keyword.
815 @code{egcs} @code{g77} users benefit from the alias-analysis
816 code despite the lack of the @code{__restrict__} keyword,
817 which is a C-language construct.)
820 Support for the GNU compiler options
821 @samp{-fmove-all-movables},
822 @samp{-freduce-all-givs},
823 and @samp{-frerun-loop-opt}.
825 (@code{egcs} supports these options.
826 @code{g77} users of @code{egcs} benefit from them even if
827 they are not explicitly specified,
828 because the defaults are optimized for @code{g77} users.)
831 Support for the @samp{-W} option warning about
832 integer division by zero.
835 The Intel x86-specific option @samp{-malign-double}
836 applying to stack-allocated data
837 as well as statically-allocate data.
841 Note that the @file{gcc/f/gbe/} subdirectory has been removed
842 from this distribution as a result of @code{g77} no longer
843 including patches for the @code{gcc} back end.
848 Fix bugs in the @code{libU77} intrinsic @code{HostNm}
849 that wrote one byte beyond the end of its @code{CHARACTER}
851 and in the @code{libU77} intrinsics
852 @code{GMTime} and @code{LTime}
853 that overwrote their arguments.
857 Support @code{gcc} version 2.8,
858 and remove support for prior versions of @code{gcc}.
860 @cindex -@w{}-driver option
861 @cindex @code{g77} options, -@w{}-driver
862 @cindex options, -@w{}-driver
864 Remove support for the @samp{--driver} option,
865 as @code{g77} now does all the driving,
866 just like @code{gcc}.
870 @code{CASE DEFAULT} no longer crashes @code{g77}.
875 Valid combinations of @code{EXTERNAL},
876 passing that external as a dummy argument
877 without explicitly giving it a type,
878 and, in a subsequent program unit,
879 referencing that external as
880 an external function with a different type
881 no longer crash @code{g77}.
886 @code{g77} no longer installs the @file{f77} command
887 and @file{f77.1} man page
888 in the @file{/usr} or @file{/usr/local} heirarchy,
889 even if the @file{f77-install-ok} file exists
890 in the source or build directory.
891 See the installation documentation for more information.
896 @code{g77} no longer installs the @file{libf2c.a} library
897 and @file{f2c.h} include file
898 in the @file{/usr} or @file{/usr/local} heirarchy,
899 even if the @file{f2c-install-ok} or @file{f2c-exists-ok} files exist
900 in the source or build directory.
901 See the installation documentation for more information.
906 The @file{libf2c.a} library produced by @code{g77} has been
907 renamed to @file{libg2c.a}.
908 It is installed only in the @code{gcc} ``private''
909 directory heirarchy, @file{gcc-lib}.
910 This allows system administrators and users to choose which
911 version of the @code{libf2c} library from @code{netlib} they
912 wish to use on a case-by-case basis.
913 See the installation documentation for more information.
918 The @file{f2c.h} include (header) file produced by @code{g77}
919 has been renamed to @file{g2c.h}.
920 It is installed only in the @code{gcc} ``private''
921 directory heirarchy, @file{gcc-lib}.
922 This allows system administrators and users to choose which
923 version of the include file from @code{netlib} they
924 wish to use on a case-by-case basis.
925 See the installation documentation for more information.
929 The @code{g77} command now expects the run-time library
930 to be named @code{libg2c.a} instead of @code{libf2c.a},
931 to ensure that a version other than the one built and
932 installed as part of the same @code{g77} version is picked up.
936 The @samp{-Wunused} option no longer issues a spurious
937 warning about the ``master'' procedure generated by
938 @code{g77} for procedures containing @code{ENTRY} statements.
942 @code{g77}'s version of @code{libf2c} separates out
943 the setting of global state
944 (such as command-line arguments and signal handling)
945 from @file{main.o} into distinct, new library
948 This should make it easier to write portable applications
949 that have their own (non-Fortran) @code{main()} routine
950 properly set up the @code{libf2c} environment, even
951 when @code{libf2c} (now @code{libg2c}) is a shared library.
955 During the configuration and build process,
956 @code{g77} creates subdirectories it needs only as it
957 needs them, thus avoiding unnecessary creation of, for example,
958 @file{stage1/f/runtime} when doing a non-bootstrap build.
959 Other cleaning up of the configuration and build process
960 has been performed as well.
965 @code{install-info} now used to update the directory of
966 Info documentation to contain an entry for @code{g77}
967 (during installation).
971 Some diagnostics have been changed from warnings to errors,
972 to prevent inadvertent use of the resulting, probably buggy,
974 These mostly include diagnostics about use of unsupported features
975 in the @code{OPEN}, @code{INQUIRE}, @code{READ}, and
976 @code{WRITE} statements,
977 and about truncations of various sorts of constants.
981 Improve documentation and indexing.
986 Upgrade to @code{libf2c} as of 1998-04-20.
988 This should fix a variety of problems, including
989 those involving some uses of the @code{T} format
990 specifier, and perhaps some build (porting) problems
995 @c 1998-03-16: 0.5.22 released.
996 @heading In 0.5.22 versus 0.5.21:
1000 Fix code generation for iterative @code{DO} loops that
1001 have one or more references to the iteration variable,
1002 or to aliases of it, in their control expressions.
1003 For example, @samp{DO 10 J=2,J} now is compiled correctly.
1006 @ifclear USERVISONLY
1008 @cindex stack, 387 coprocessor
1012 Fix a code-generation bug that afflicted
1013 Intel x86 targets when @samp{-O2} was specified
1014 compiling, for example, an old version of
1015 the @code{DNRM2} routine.
1017 The x87 coprocessor stack was being
1018 mismanaged in cases involving assigned @code{GOTO}
1022 @ifclear USERVISONLY
1024 Fix @code{DTime} intrinsic so as not to truncate
1025 results to integer values (on some systems).
1029 Fix @code{Signal} intrinsic so it offers portable
1030 support for 64-bit systems (such as Digital Alphas
1033 @ifclear USERVISONLY
1035 Fix run-time crash involving @code{NAMELIST} on 64-bit
1036 machines such as Alphas.
1039 @ifclear USERVISONLY
1041 Fix @code{g77} version of @code{libf2c} so it no longer
1042 produces a spurious @samp{I/O recursion} diagnostic at run time
1043 when an I/O operation (such as @samp{READ *,I}) is interrupted
1044 in a manner that causes the program to be terminated
1045 via the @code{f_exit} routine (such as via @kbd{C-c}).
1048 @ifclear USERVISONLY
1050 Fix @code{g77} crash triggered by @code{CASE} statement with
1051 an omitted lower or upper bound.
1054 @ifclear USERVISONLY
1056 Fix @code{g77} crash compiling references to @code{CPU_Time}
1060 @ifclear USERVISONLY
1062 Fix @code{g77} crash
1063 (or apparently infinite run-time)
1064 when compiling certain complicated expressions
1065 involving @code{COMPLEX} arithmetic
1066 (especially multiplication).
1069 @ifclear USERVISONLY
1071 Fix @code{g77} crash on statements such as
1072 @samp{PRINT *, (REAL(Z(I)),I=1,2)}, where
1073 @samp{Z} is @code{DOUBLE COMPLEX}.
1076 @ifclear USERVISONLY
1078 Fix a @code{g++} crash.
1082 Support @samp{FORMAT(I<@var{expr}>)} when @var{expr} is a
1083 compile-time constant @code{INTEGER} expression.
1086 Fix @code{g77} @samp{-g} option so procedures that
1087 use @code{ENTRY} can be stepped through, line by line,
1090 @ifclear USERVISONLY
1092 Fix a profiling-related bug in @code{gcc} back end for
1093 Intel x86 architecture.
1097 Allow any @code{REAL} argument to intrinsics
1098 @code{Second} and @code{CPU_Time}.
1101 Allow any numeric argument to intrinsics
1102 @code{Int2} and @code{Int8}.
1105 Use @code{tempnam}, if available, to open scratch files
1106 (as in @samp{OPEN(STATUS='SCRATCH')})
1107 so that the @code{TMPDIR} environment variable,
1108 if present, is used.
1111 Rename the @code{gcc} keyword @code{restrict} to
1112 @code{__restrict__}, to avoid rejecting valid, existing,
1114 Support for @code{restrict} is now more like support
1117 @ifclear USERVISONLY
1119 Fix @samp{-fpedantic} to not reject procedure invocations
1120 such as @samp{I=J()} and @samp{CALL FOO()}.
1124 Fix @samp{-fugly-comma} to affect invocations of
1125 only external procedures.
1126 Restore rejection of gratuitous trailing omitted
1127 arguments to intrinsics, as in @samp{I=MAX(3,4,,)}.
1130 Fix compiler so it accepts @samp{-fgnu-intrinsics-*} and
1131 @samp{-fbadu77-intrinsics-*} options.
1133 @ifclear USERVISONLY
1135 Improve diagnostic messages from @code{libf2c}
1136 so it is more likely that the printing of the
1137 active format string is limited to the string,
1138 with no trailing garbage being printed.
1140 (Unlike @code{f2c}, @code{g77} did not append
1141 a null byte to its compiled form of every
1142 format string specified via a @code{FORMAT} statement.
1143 However, @code{f2c} would exhibit the problem
1144 anyway for a statement like @samp{PRINT '(I)garbage', 1}
1145 by printing @samp{(I)garbage} as the format string.)
1148 @ifclear USERVISONLY
1150 Improve compilation of @code{FORMAT} expressions so that
1151 a null byte is appended to the last operand if it
1153 This provides a cleaner run-time diagnostic as provided
1154 by @code{libf2c} for statements like @samp{PRINT '(I1', 42}.
1157 @ifclear USERVISONLY
1159 Fix various crashes involving code with diagnosed errors.
1162 @ifclear USERVISONLY
1164 Fix cross-compilation bug when configuring @code{libf2c}.
1167 @ifclear USERVISONLY
1169 Improve diagnostics.
1172 @ifclear USERVISONLY
1174 Improve documentation and indexing.
1177 @ifclear USERVISONLY
1179 Upgrade to @code{libf2c} as of 1997-09-23.
1180 This fixes a formatted-I/O bug that afflicted
1181 64-bit systems with 32-bit integers
1182 (such as Digital Alpha running GNU/Linux).
1186 @c 1998-03-18: EGCS 1.0.2 released.
1187 @heading In @code{EGCS} 1.0.2 versus @code{EGCS} 1.0.1:
1189 @ifclear USERVISONLY
1191 Fix @code{g77} crash triggered by @code{CASE} statement with
1192 an omitted lower or upper bound.
1195 @ifclear USERVISONLY
1197 Fix @code{g77} crash on statements such as
1198 @samp{PRINT *, (REAL(Z(I)),I=1,2)}, where
1199 @samp{Z} is @code{DOUBLE COMPLEX}.
1202 @ifclear USERVISONLY
1204 @cindex support, ELF
1205 @cindex -fPIC option
1206 @cindex options, -fPIC
1208 Fix @samp{-fPIC} (such as compiling for ELF targets)
1209 on the Intel x86 architecture target
1210 so invalid assembler code is no longer produced.
1213 @ifclear USERVISONLY
1215 Fix @samp{-fpedantic} to not reject procedure invocations
1216 such as @samp{I=J()} and @samp{CALL FOO()}.
1219 @ifclear USERVISONLY
1221 Fix @samp{-fugly-comma} to affect invocations of
1222 only external procedures.
1223 Restore rejection of gratuitous trailing omitted
1224 arguments to intrinsics, as in @samp{I=MAX(3,4,,)}.
1228 Fix compiler so it accepts @samp{-fgnu-intrinsics-*} and
1229 @samp{-fbadu77-intrinsics-*} options.
1232 @c 1998-01-06: EGCS 1.0.1 released.
1233 @heading In @code{EGCS} 1.0.1 versus @code{EGCS} 1.0:
1234 @ifclear USERVISONLY
1237 Fix run-time crash involving @code{NAMELIST} on 64-bit
1238 machines such as Alphas.
1242 @c 1997-12-03: EGCS 1.0 released.
1243 @heading In @code{EGCS} 1.0 versus @code{g77} 0.5.21:
1246 Version 1.0 of @code{egcs}
1247 contains several regressions against
1248 version 0.5.21 of @code{g77},
1250 ``vanilla'' @code{gcc} back end instead of patching
1251 it to fix a few bugs and improve performance in a
1254 Features that have been dropped from this version
1255 of @code{g77} due to their being implemented
1256 via @code{g77}-specific patches to the @code{gcc}
1257 back end in previous releases include:
1261 Support for the C-language @code{restrict} keyword.
1264 Support for the @samp{-W} option warning about
1265 integer division by zero.
1268 The Intel x86-specific option @samp{-malign-double}
1269 applying to stack-allocated data
1270 as well as statically-allocate data.
1273 @ifclear USERVISONLY
1274 Note that the @file{gcc/f/gbe/} subdirectory has been removed
1275 from this distribution as a result of @code{g77}
1276 being fully integrated with
1277 the @code{egcs} variant of the @code{gcc} back end.
1280 @ifclear USERVISONLY
1282 Fix code generation for iterative @code{DO} loops that
1283 have one or more references to the iteration variable,
1284 or to aliases of it, in their control expressions.
1285 For example, @samp{DO 10 J=2,J} now is compiled correctly.
1288 @ifclear USERVISONLY
1290 Fix @code{DTime} intrinsic so as not to truncate
1291 results to integer values (on some systems).
1294 @ifclear USERVISONLY
1296 @c Toon Moene discovered these.
1297 Some Fortran code, miscompiled
1298 by @code{g77} built on @code{gcc} version 2.8.1
1299 on m68k-next-nextstep3 configurations
1300 when using the @samp{-O2} option,
1301 is now compiled correctly.
1302 It is believed that a C function known to miscompile
1303 on that configuration
1304 when using the @samp{-O2 -funroll-loops} options
1305 also is now compiled correctly.
1308 @ifclear USERVISONLY
1310 Remove support for non-@code{egcs} versions of @code{gcc}.
1313 @cindex -@w{}-driver option
1314 @cindex @code{g77} options, -@w{}-driver
1315 @cindex options, -@w{}-driver
1317 Remove support for the @samp{--driver} option,
1318 as @code{g77} now does all the driving,
1319 just like @code{gcc}.
1322 Allow any numeric argument to intrinsics
1323 @code{Int2} and @code{Int8}.
1325 @ifclear USERVISONLY
1327 Improve diagnostic messages from @code{libf2c}
1328 so it is more likely that the printing of the
1329 active format string is limited to the string,
1330 with no trailing garbage being printed.
1332 (Unlike @code{f2c}, @code{g77} did not append
1333 a null byte to its compiled form of every
1334 format string specified via a @code{FORMAT} statement.
1335 However, @code{f2c} would exhibit the problem
1336 anyway for a statement like @samp{PRINT '(I)garbage', 1}
1337 by printing @samp{(I)garbage} as the format string.)
1340 @ifclear USERVISONLY
1342 Upgrade to @code{libf2c} as of 1997-09-23.
1343 This fixes a formatted-I/O bug that afflicted
1344 64-bit systems with 32-bit integers
1345 (such as Digital Alpha running GNU/Linux).
1349 @c 1997-09-09: 0.5.21 released.
1352 @ifclear USERVISONLY
1354 Fix a code-generation bug introduced by 0.5.20
1355 caused by loop unrolling (by specifying
1356 @samp{-funroll-loops} or similar).
1357 This bug afflicted all code compiled by
1358 version 2.7.2.2.f.2 of @code{gcc} (C, C++,
1359 Fortran, and so on).
1362 @ifclear USERVISONLY
1364 Fix a code-generation bug manifested when
1365 combining local @code{EQUIVALENCE} with a
1366 @code{DATA} statement that follows
1367 the first executable statement (or is
1368 treated as an executable-context statement
1369 as a result of using the @samp{-fpedantic}
1373 @ifclear USERVISONLY
1375 Fix a compiler crash that occured when an
1376 integer division by a constant zero is detected.
1377 Instead, when the @samp{-W} option is specified,
1378 the @code{gcc} back end issues a warning about such a case.
1379 This bug afflicted all code compiled by
1380 version 2.7.2.2.f.2 of @code{gcc} (C, C++,
1381 Fortran, and so on).
1385 When the @samp{-W} option is specified, @code{gcc}, @code{g77},
1386 and other GNU compilers that incorporate the @code{gcc}
1387 back end as modified by @code{g77}, issue
1388 a warning about integer division by constant zero.
1391 @ifclear USERVISONLY
1393 Fix a compiler crash that occurred in some cases
1394 of procedure inlining.
1395 (Such cases became more frequent in 0.5.20.)
1398 @ifclear USERVISONLY
1400 Fix a compiler crash resulting from using @code{DATA}
1401 or similar to initialize a @code{COMPLEX} variable or
1405 @ifclear USERVISONLY
1407 Fix compiler crashes involving use of @code{AND}, @code{OR},
1408 or @code{XOR} intrinsics.
1411 @ifclear USERVISONLY
1413 Fix compiler bug triggered when using a @code{COMMON}
1414 or @code{EQUIVALENCE} variable
1415 as the target of an @code{ASSIGN}
1416 or assigned-@code{GOTO} statement.
1419 @ifclear USERVISONLY
1421 Fix compiler crashes due to using the name of a some
1422 non-standard intrinsics (such as @code{FTell} or
1423 @code{FPutC}) as such and as the name of a procedure
1425 Such dual use of a name in a program is allowed by
1429 @c @code{g77}'s version of @code{libf2c} has been modified
1430 @c so that the external names of library's procedures do not
1431 @c conflict with names used for Fortran procedures compiled
1433 @c An additional layer of jacket procedures has been added
1434 @c to @code{libf2c} to map the old names to the new names,
1435 @c for automatic use by programs that interface to the
1436 @c library procedures via the external-procedure mechanism.
1438 @c For example, the intrinsic @code{FPUTC} previously was
1439 @c implemented by @code{g77} as a call to the @code{libf2c}
1440 @c routine @code{fputc_}.
1441 @c This would conflict with a Fortran procedure named @code{FPUTC}
1442 @c (using default compiler options), and this conflict
1443 @c would cause a crash under certain circumstances.
1445 @c Now, the intrinsic @code{FPUTC} calls @code{G77_fputc_0},
1446 @c which does not conflict with the @code{fputc_} external
1447 @c that implements a Fortran procedure named @code{FPUTC}.
1449 @c Programs that refer to @code{FPUTC} as an external procedure
1450 @c without supplying their own implementation will link to
1451 @c the new @code{libf2c} routine @code{fputc_}, which is
1452 @c simply a jacket routine that calls @code{G77_fputc_0}.
1454 @ifclear USERVISONLY
1456 Place automatic arrays on the stack, even if
1457 @code{SAVE} or the @samp{-fno-automatic} option
1459 This avoids a compiler crash in some cases.
1462 @ifclear USERVISONLY
1464 The @samp{-malign-double} option now reliably aligns
1465 @code{DOUBLE PRECISION} optimally on Pentium and
1466 Pentium Pro architectures (586 and 686 in @code{gcc}).
1470 New option @samp{-Wno-globals} disables warnings
1471 about ``suspicious'' use of a name both as a global
1472 name and as the implicit name of an intrinsic, and
1473 warnings about disagreements over the number or natures of
1474 arguments passed to global procedures, or the
1475 natures of the procedures themselves.
1477 The default is to issue such warnings, which are
1478 new as of this version of @code{g77}.
1481 New option @samp{-fno-globals} disables diagnostics
1482 about potentially fatal disagreements
1483 analysis problems, such as disagreements over the
1484 number or natures of arguments passed to global
1485 procedures, or the natures of those procedures themselves.
1487 The default is to issue such diagnostics and flag
1488 the compilation as unsuccessful.
1489 With this option, the diagnostics are issued as
1490 warnings, or, if @samp{-Wno-globals} is specified,
1491 are not issued at all.
1493 This option also disables inlining of global procedures,
1494 to avoid compiler crashes resulting from coding errors
1495 that these diagnostics normally would identify.
1497 @ifclear USERVISONLY
1499 Diagnose cases where a reference to a procedure
1500 disagrees with the type of that procedure, or
1501 where disagreements about the number or nature
1503 This avoids a compiler crash.
1506 @ifclear USERVISONLY
1508 Fix parsing bug whereby @code{g77} rejected a
1509 second initialization specification immediately
1510 following the first's closing @samp{/} without
1511 an intervening comma in a @code{DATA} statement,
1512 and the second specification was an implied-DO list.
1515 @ifclear USERVISONLY
1517 Improve performance of the @code{gcc} back end so
1518 certain complicated expressions involving @code{COMPLEX}
1519 arithmetic (especially multiplication) don't appear to
1520 take forever to compile.
1523 @ifclear USERVISONLY
1525 Fix a couple of profiling-related bugs in @code{gcc}
1529 @ifclear USERVISONLY
1531 Integrate GNU Ada's (GNAT's) changes to the back end,
1532 which consist almost entirely of bug fixes.
1533 These fixes are circa version 3.10p of GNAT.
1536 @ifclear USERVISONLY
1538 Include some other @code{gcc} fixes that seem useful in
1539 @code{g77}'s version of @code{gcc}.
1540 (See @file{gcc/ChangeLog} for details---compare it
1541 to that file in the vanilla @code{gcc-2.7.2.3.tar.gz}
1546 Fix @code{libU77} routines that accept file and other names
1547 to strip trailing blanks from them, for consistency
1548 with other implementations.
1549 Blanks may be forcibly appended to such names by
1550 appending a single null character (@samp{CHAR(0)})
1551 to the significant trailing blanks.
1554 Fix @code{CHMOD} intrinsic to work with file names
1555 that have embedded blanks, commas, and so on.
1558 Fix @code{SIGNAL} intrinsic so it accepts an
1559 optional third @code{Status} argument.
1561 @ifclear USERVISONLY
1563 Fix @code{IDATE()} intrinsic subroutine (VXT form)
1564 so it accepts arguments in the correct order.
1565 Documentation fixed accordingly, and for
1566 @code{GMTIME()} and @code{LTIME()} as well.
1570 Make many changes to @code{libU77} intrinsics to
1571 support existing code more directly.
1573 Such changes include allowing both subroutine and
1574 function forms of many routines, changing @code{MCLOCK()}
1575 and @code{TIME()} to return @code{INTEGER(KIND=1)} values,
1576 introducing @code{MCLOCK8()} and @code{TIME8()} to
1577 return @code{INTEGER(KIND=2)} values,
1578 and placing functions that are intended to perform
1579 side effects in a new intrinsic group, @code{badu77}.
1581 @ifclear USERVISONLY
1583 Improve @code{libU77} so it is more portable.
1587 Add options @samp{-fbadu77-intrinsics-delete},
1588 @samp{-fbadu77-intrinsics-hide}, and so on.
1590 @ifclear USERVISONLY
1592 Fix crashes involving diagnosed or invalid code.
1595 @ifclear USERVISONLY
1597 @code{g77} and @code{gcc} now do a somewhat better
1598 job detecting and diagnosing arrays that are too
1599 large to handle before these cause diagnostics
1600 during the assembler or linker phase, a compiler
1601 crash, or generation of incorrect code.
1604 @ifclear USERVISONLY
1606 Make some fixes to alias analysis code.
1609 @ifclear USERVISONLY
1611 Add support for @code{restrict} keyword in @code{gcc}
1615 @ifclear USERVISONLY
1617 Support @code{gcc} version 2.7.2.3
1618 (modified by @code{g77} into version 2.7.2.3.f.1),
1620 support for prior versions of @code{gcc}.
1623 @ifclear USERVISONLY
1625 Incorporate GNAT's patches to the @code{gcc} back
1626 end into @code{g77}'s, so GNAT users do not need
1627 to apply GNAT's patches to build both GNAT and @code{g77}
1628 from the same source tree.
1631 @ifclear USERVISONLY
1633 Modify @code{make} rules and related code so that
1634 generation of Info documentation doesn't require
1635 compilation using @code{gcc}.
1636 Now, any ANSI C compiler should be adequate to
1637 produce the @code{g77} documentation (in particular,
1638 the tables of intrinsics) from scratch.
1642 Add @code{INT2} and @code{INT8} intrinsics.
1645 Add @code{CPU_TIME} intrinsic.
1648 Add @code{ALARM} intrinsic.
1651 @code{CTIME} intrinsic now accepts any @code{INTEGER}
1652 argument, not just @code{INTEGER(KIND=2)}.
1654 @ifclear USERVISONLY
1656 Warn when explicit type declaration disagrees with
1657 the type of an intrinsic invocation.
1660 @ifclear USERVISONLY
1662 Support @samp{*f771} entry in @code{gcc} @file{specs} file.
1665 @ifclear USERVISONLY
1667 Fix typo in @code{make} rule @code{g77-cross}, used only for
1671 @ifclear USERVISONLY
1673 Fix @code{libf2c} build procedure to re-archive library
1674 if previous attempt to archive was interrupted.
1677 @ifclear USERVISONLY
1679 Change @code{gcc} to unroll loops only during the last
1680 invocation (of as many as two invocations) of loop
1684 @ifclear USERVISONLY
1686 Improve handling of @samp{-fno-f2c} so that code that
1687 attempts to pass an intrinsic as an actual argument,
1688 such as @samp{CALL FOO(ABS)}, is rejected due to the fact
1689 that the run-time-library routine is, effectively,
1690 compiled with @samp{-ff2c} in effect.
1693 @ifclear USERVISONLY
1695 Fix @code{g77} driver to recognize @samp{-fsyntax-only}
1696 as an option that inhibits linking, just like @samp{-c} or
1697 @samp{-S}, and to recognize and properly handle the
1698 @samp{-nostdlib}, @samp{-M}, @samp{-MM}, @samp{-nodefaultlibs},
1699 and @samp{-Xlinker} options.
1702 @ifclear USERVISONLY
1704 Upgrade to @code{libf2c} as of 1997-08-16.
1707 @ifclear USERVISONLY
1709 Modify @code{libf2c} to consistently and clearly diagnose
1710 recursive I/O (at run time).
1714 @code{g77} driver now prints version information (such as produced
1715 by @kbd{g77 -v}) to @code{stderr} instead of @code{stdout}.
1718 The @samp{.r} suffix now designates a Ratfor source file,
1719 to be preprocessed via the @code{ratfor} command, available
1722 @ifclear USERVISONLY
1724 Fix some aspects of how @code{gcc} determines what kind of
1725 system is being configured and what kinds are supported.
1726 For example, GNU Linux/Alpha ELF systems now are directly
1730 @ifclear USERVISONLY
1732 Improve diagnostics.
1735 @ifclear USERVISONLY
1737 Improve documentation and indexing.
1740 @ifclear USERVISONLY
1742 Include all pertinent files for @code{libf2c} that come
1743 from @code{netlib.bell-labs.com}; give any such files
1744 that aren't quite accurate in @code{g77}'s version of
1745 @code{libf2c} the suffix @samp{.netlib}.
1748 @ifclear USERVISONLY
1750 Reserve @code{INTEGER(KIND=0)} for future use.
1754 @c 1997-02-28: 0.5.20 released.
1758 The @samp{-fno-typeless-boz} option is now the default.
1760 This option specifies that non-decimal-radix
1761 constants using the prefixed-radix form (such as @samp{Z'1234'})
1762 are to be interpreted as @code{INTEGER(KIND=1)} constants.
1763 Specify @samp{-ftypeless-boz} to cause such
1764 constants to be interpreted as typeless.
1766 (Version 0.5.19 introduced @samp{-fno-typeless-boz} and
1770 @xref{Fortran Dialect Options,,Options Controlling Fortran Dialect},
1771 for information on the @samp{-ftypeless-boz} option.
1775 Options @samp{-ff90-intrinsics-enable} and
1776 @samp{-fvxt-intrinsics-enable} now are the
1779 Some programs might use names that clash with
1780 intrinsic names defined (and now enabled) by these
1781 options or by the new @code{libU77} intrinsics.
1782 Users of such programs might need to compile them
1783 differently (using, for example, @samp{-ff90-intrinsics-disable})
1784 or, better yet, insert appropriate @code{EXTERNAL}
1785 statements specifying that these names are not intended
1786 to be names of intrinsics.
1789 The @code{ALWAYS_FLUSH} macro is no longer defined when
1790 building @code{libf2c}, which should result in improved
1791 I/O performance, especially over NFS.
1793 @emph{Note:} If you have code that depends on the behavior
1794 of @code{libf2c} when built with @code{ALWAYS_FLUSH} defined,
1795 you will have to modify @code{libf2c} accordingly before
1796 building it from this and future versions of @code{g77}.
1799 @xref{Output Assumed To Flush}, for more information.
1803 Dave Love's implementation of @code{libU77} has been
1804 added to the version of @code{libf2c} distributed with
1805 and built as part of @code{g77}.
1806 @code{g77} now knows about the routines in this library
1810 New option @samp{-fvxt} specifies that the
1811 source file is written in VXT Fortran, instead of GNU Fortran.
1814 @xref{VXT Fortran}, for more information on the constructs
1815 recognized when the @samp{-fvxt} option is specified.
1819 The @samp{-fvxt-not-f90} option has been deleted,
1820 along with its inverse, @samp{-ff90-not-vxt}.
1822 If you used one of these deleted options, you should
1823 re-read the pertinent documentation to determine which
1824 options, if any, are appropriate for compiling your
1825 code with this version of @code{g77}.
1828 @xref{Other Dialects}, for more information.
1832 The @samp{-fugly} option now issues a warning, as it
1833 likely will be removed in a future version.
1835 (Enabling all the @samp{-fugly-*} options is unlikely
1836 to be feasible, or sensible, in the future,
1837 so users should learn to specify only those
1838 @samp{-fugly-*} options they really need for a
1839 particular source file.)
1842 The @samp{-fugly-assumed} option, introduced in
1843 version 0.5.19, has been changed to
1844 better accommodate old and new code.
1847 @xref{Ugly Assumed-Size Arrays}, for more information.
1850 @ifclear USERVISONLY
1852 Make a number of fixes to the @code{g77} front end and
1853 the @code{gcc} back end to better support Alpha (AXP)
1855 This includes providing at least one bug-fix to the
1856 @code{gcc} back end for Alphas.
1860 Related to supporting Alpha (AXP) machines, the @code{LOC()}
1861 intrinsic and @code{%LOC()} construct now return
1862 values of @code{INTEGER(KIND=0)} type,
1863 as defined by the GNU Fortran language.
1865 This type is wide enough
1866 (holds the same number of bits)
1867 as the character-pointer type on the machine.
1869 On most machines, this won't make a difference,
1870 whereas, on Alphas and other systems with 64-bit pointers,
1871 the @code{INTEGER(KIND=0)} type is equivalent to @code{INTEGER(KIND=2)}
1872 (often referred to as @code{INTEGER*8})
1873 instead of the more common @code{INTEGER(KIND=1)}
1874 (often referred to as @code{INTEGER*4}).
1877 Emulate @code{COMPLEX} arithmetic in the @code{g77} front
1878 end, to avoid bugs in @code{complex} support in the
1879 @code{gcc} back end.
1880 New option @samp{-fno-emulate-complex}
1881 causes @code{g77} to revert the 0.5.19 behavior.
1883 @ifclear USERVISONLY
1885 Fix bug whereby @samp{REAL A(1)}, for example, caused
1886 a compiler crash if @samp{-fugly-assumed} was in effect
1887 and @var{A} was a local (automatic) array.
1888 That case is no longer affected by the new
1889 handling of @samp{-fugly-assumed}.
1892 @ifclear USERVISONLY
1894 Fix @code{g77} command driver so that @samp{g77 -o foo.f}
1895 no longer deletes @file{foo.f} before issuing other
1896 diagnostics, and so the @samp{-x} option is properly
1900 @ifclear USERVISONLY
1902 Enable inlining of subroutines and functions by the @code{gcc}
1904 This works as it does for @code{gcc} itself---program units
1905 may be inlined for invocations that follow them in the same
1906 program unit, as long as the appropriate compile-time
1907 options are specified.
1911 Dummy arguments are no longer assumed to potentially alias
1913 other dummy arguments or @code{COMMON} areas when any of
1914 these are defined (assigned to) by Fortran code.
1916 This can result in faster and/or smaller programs when
1917 compiling with optimization enabled, though on some
1918 systems this effect is observed only when @samp{-fforce-addr}
1921 New options @samp{-falias-check}, @samp{-fargument-alias},
1922 @samp{-fargument-noalias},
1923 and @samp{-fno-argument-noalias-global} control the
1924 way @code{g77} handles potential aliasing.
1927 @xref{Aliasing Assumed To Work}, for detailed information on why the
1928 new defaults might result in some programs no longer working the way they
1929 did when compiled by previous versions of @code{g77}.
1932 @ifclear USERVISONLY
1934 The @code{CONJG()} and @code{DCONJG()} intrinsics now
1935 are compiled in-line.
1938 @ifclear USERVISONLY
1940 The bug-fix for 0.5.19.1 has been re-done.
1941 The @code{g77} compiler has been changed back to
1942 assume @code{libf2c} has no aliasing problems in
1943 its implementations of the @code{COMPLEX} (and
1944 @code{DOUBLE COMPLEX}) intrinsics.
1945 The @code{libf2c} has been changed to have no such
1948 As a result, 0.5.20 is expected to offer improved performance
1949 over 0.5.19.1, perhaps as good as 0.5.19 in most
1950 or all cases, due to this change alone.
1952 @emph{Note:} This change requires version 0.5.20 of
1953 @code{libf2c}, at least, when linking code produced
1954 by any versions of @code{g77} other than 0.5.19.1.
1955 Use @samp{g77 -v} to determine the version numbers
1956 of the @code{libF77}, @code{libI77}, and @code{libU77}
1957 components of the @code{libf2c} library.
1958 (If these version numbers are not printed---in
1959 particular, if the linker complains about unresolved
1960 references to names like @samp{g77__fvers__}---that
1961 strongly suggests your installation has an obsolete
1962 version of @code{libf2c}.)
1966 New option @samp{-fugly-assign} specifies that the
1967 same memory locations are to be used to hold the
1968 values assigned by both statements @samp{I = 3} and
1969 @samp{ASSIGN 10 TO I}, for example.
1970 (Normally, @code{g77} uses a separate memory location
1971 to hold assigned statement labels.)
1974 @xref{Ugly Assigned Labels}, for more information.
1978 @code{FORMAT} and @code{ENTRY} statements now are allowed to
1979 precede @code{IMPLICIT NONE} statements.
1981 @ifclear USERVISONLY
1983 Produce diagnostic for unsupported @code{SELECT CASE} on
1984 @code{CHARACTER} type, instead of crashing, at compile time.
1987 @ifclear USERVISONLY
1989 Fix crashes involving diagnosed or invalid code.
1992 @ifclear USERVISONLY
1994 Change approach to building @code{libf2c} archive
1995 (@file{libf2c.a}) so that members are added to it
1996 only when truly necessary, so the user that installs
1997 an already-built @code{g77} doesn't need to have write
1998 access to the build tree (whereas the user doing the
1999 build might not have access to install new software
2003 @ifclear USERVISONLY
2005 Support @code{gcc} version 2.7.2.2
2006 (modified by @code{g77} into version 2.7.2.2.f.2),
2008 support for prior versions of @code{gcc}.
2011 @ifclear USERVISONLY
2013 Upgrade to @code{libf2c} as of 1997-02-08, and
2014 fix up some of the build procedures.
2017 @ifclear USERVISONLY
2019 Improve general build procedures for @code{g77},
2020 fixing minor bugs (such as deletion of any file
2021 named @file{f771} in the parent directory of @code{gcc/}).
2025 Enable full support of @code{INTEGER(KIND=2)}
2026 (often referred to as @code{INTEGER*8})
2028 @code{libf2c} and @file{f2c.h} so that @code{f2c} users
2029 may make full use of its features via the @code{g77}
2030 version of @file{f2c.h} and the @code{INTEGER(KIND=2)}
2031 support routines in the @code{g77} version of @code{libf2c}.
2034 Improve @code{g77} driver and @code{libf2c} so that @samp{g77 -v}
2035 yields version information on the library.
2038 The @code{SNGL} and @code{FLOAT} intrinsics now are
2039 specific intrinsics, instead of synonyms for the
2040 generic intrinsic @code{REAL}.
2043 New intrinsics have been added.
2044 These are @code{REALPART}, @code{IMAGPART},
2046 @code{LONG}, and @code{SHORT}.
2049 A new group of intrinsics, @code{gnu}, has been added
2050 to contain the new @code{REALPART}, @code{IMAGPART},
2051 and @code{COMPLEX} intrinsics.
2052 An old group, @code{dcp}, has been removed.
2055 Complain about industry-wide ambiguous references
2056 @samp{REAL(@var{expr})} and @samp{AIMAG(@var{expr})},
2057 where @var{expr} is @code{DOUBLE COMPLEX} (or any
2058 complex type other than @code{COMPLEX}), unless
2059 @samp{-ff90} option specifies Fortran 90 interpretation
2060 or new @samp{-fugly-complex} option, in conjunction with
2061 @samp{-fnot-f90}, specifies @code{f2c} interpretation.
2063 @ifclear USERVISONLY
2065 Make improvements to diagnostics.
2068 @ifclear USERVISONLY
2070 Speed up compiler a bit.
2073 @ifclear USERVISONLY
2075 Improvements to documentation and indexing, including
2076 a new chapter containing information on one, later
2077 more, diagnostics that users are directed to pull
2078 up automatically via a message in the diagnostic itself.
2080 (Hence the menu item @code{M} for the node
2081 @code{Diagnostics} in the top-level menu of
2082 the Info documentation.)
2086 @ifclear DOC-OLDNEWS
2087 @heading In previous versions:
2089 Information on previous versions is archived
2090 in @file{@value{path-g77}/news.texi}
2091 following the test of the @code{DOC-OLDNEWS} macro.
2095 @c 1997-02-01: 0.5.19.1 released.
2096 @heading In 0.5.19.1:
2099 Code-generation bugs afflicting operations on complex
2100 data have been fixed.
2102 These bugs occurred when assigning the result of an
2103 operation to a complex variable (or array element)
2104 that also served as an input to that operation.
2106 The operations affected by this bug were: @code{CONJG()},
2107 @code{DCONJG()}, @code{CCOS()}, @code{CDCOS()},
2108 @code{CLOG()}, @code{CDLOG()}, @code{CSIN()}, @code{CDSIN()},
2109 @code{CSQRT()}, @code{CDSQRT()}, complex division, and
2110 raising a @code{DOUBLE COMPLEX} operand to an @code{INTEGER}
2112 (The related generic and @samp{Z}-prefixed intrinsics,
2113 such as @code{ZSIN()}, also were affected.)
2115 For example, @samp{C = CSQRT(C)}, @samp{Z = Z/C}, and @samp{Z = Z**I}
2116 (where @samp{C} is @code{COMPLEX} and @samp{Z} is
2117 @code{DOUBLE COMPLEX}) have been fixed.
2120 @c 1996-12-07: 0.5.19 released.
2124 Fix @code{FORMAT} statement parsing so negative values for
2125 specifiers such as @code{P} (e.g. @samp{FORMAT(-1PF8.1)})
2126 are correctly processed as negative.
2129 Fix @code{SIGNAL} intrinsic so it once again accepts a
2130 procedure as its second argument.
2133 A temporary kludge option provides bare-bones information on
2134 @code{COMMON} and @code{EQUIVALENCE} members at debug time.
2137 New @samp{-fonetrip} option specifies FORTRAN-66-style
2138 one-trip @code{DO} loops.
2141 New @samp{-fno-silent} option causes names of program units
2142 to be printed as they are compiled, in a fashion similar to
2143 UNIX @code{f77} and @code{f2c}.
2146 New @samp{-fugly-assumed} option specifies that arrays
2147 dimensioned via @samp{DIMENSION X(1)}, for example, are to be
2148 treated as assumed-size.
2151 New @samp{-fno-typeless-boz} option specifies that non-decimal-radix
2152 constants using the prefixed-radix form (such as @samp{Z'1234'})
2153 are to be interpreted as @code{INTEGER} constants.
2156 New @samp{-ff66} option is a ``shorthand'' option that specifies
2157 behaviors considered appropriate for FORTRAN 66 programs.
2160 New @samp{-ff77} option is a ``shorthand'' option that specifies
2161 behaviors considered appropriate for UNIX @code{f77} programs.
2164 New @samp{-fugly-comma} and @samp{-fugly-logint} options provided
2165 to perform some of what @samp{-fugly} used to do.
2166 @samp{-fugly} and @samp{-fno-ugly} are now ``shorthand'' options,
2167 in that they do nothing more than enable (or disable) other
2168 @samp{-fugly-*} options.
2171 Fix parsing of assignment statements involving targets that
2172 are substrings of elements of @code{CHARACTER} arrays having
2173 names such as @samp{READ}, @samp{WRITE}, @samp{GOTO}, and
2174 @samp{REALFUNCTIONFOO}.
2177 Fix crashes involving diagnosed code.
2180 Fix handling of local @code{EQUIVALENCE} areas so certain cases
2181 of valid Fortran programs are not misdiagnosed as improperly
2182 extending the area backwards.
2185 Support @code{gcc} version 2.7.2.1.
2188 Upgrade to @code{libf2c} as of 1996-09-26, and
2189 fix up some of the build procedures.
2192 Change code generation for list-directed I/O so it allows
2193 for new versions of @code{libf2c} that might return non-zero
2194 status codes for some operations previously assumed to always
2197 This change not only affects how @code{IOSTAT=} variables
2198 are set by list-directed I/O, it also affects whether
2199 @code{END=} and @code{ERR=} labels are reached by these
2203 Add intrinsic support for new @code{FTELL} and @code{FSEEK}
2204 procedures in @code{libf2c}.
2207 Modify @code{fseek_()} in @code{libf2c} to be more portable
2208 (though, in practice, there might be no systems where this
2209 matters) and to catch invalid @code{whence} arguments.
2212 Some useless warnings from the @samp{-Wunused} option have
2216 Fix a problem building the @file{f771} executable
2217 on AIX systems by linking with the @samp{-bbigtoc} option.
2220 Abort configuration if @code{gcc} has not been patched
2221 using the patch file provided in the @file{gcc/f/gbe/}
2225 Add options @samp{--help} and @samp{--version} to the
2226 @code{g77} command, to conform to GNU coding guidelines.
2227 Also add printing of @code{g77} version number when
2228 the @samp{--verbose} (@samp{-v}) option is used.
2231 Change internally generated name for local @code{EQUIVALENCE}
2232 areas to one based on the alphabetically sorted first name
2233 in the list of names for entities placed at the beginning
2237 Improvements to documentation and indexing.
2240 @c 1996-04-01: 0.5.18 released.
2244 Add some rudimentary support for @code{INTEGER*1},
2245 @code{INTEGER*2}, @code{INTEGER*8},
2246 and their @code{LOGICAL} equivalents.
2247 (This support works on most, maybe all, @code{gcc} targets.)
2249 Thanks to Scott Snyder (@email{snyder@@d0sgif.fnal.gov})
2250 for providing the patch for this!
2252 Among the missing elements from the support for these
2253 features are full intrinsic support and constants.
2256 Add some rudimentary support for the @code{BYTE} and
2257 @code{WORD} type-declaration statements.
2258 @code{BYTE} corresponds to @code{INTEGER*1},
2259 while @code{WORD} corresponds to @code{INTEGER*2}.
2261 Thanks to Scott Snyder (@email{snyder@@d0sgif.fnal.gov})
2262 for providing the patch for this!
2265 The compiler code handling intrinsics has been largely
2266 rewritten to accommodate the new types.
2267 No new intrinsics or arguments for existing
2268 intrinsics have been added, so there is, at this
2269 point, no intrinsic to convert to @code{INTEGER*8},
2273 Support automatic arrays in procedures.
2276 Reduce space/time requirements for handling large
2277 @emph{sparsely} initialized aggregate arrays.
2278 This improvement applies to only a subset of
2279 the general problem to be addressed in 0.6.
2282 Treat initial values of zero as if they weren't
2283 specified (in DATA and type-declaration statements).
2284 The initial values will be set to zero anyway, but the amount
2285 of compile time processing them will be reduced,
2286 in some cases significantly (though, again, this
2287 is only a subset of the general problem to be
2290 A new option, @samp{-fzeros}, is introduced to
2291 enable the traditional treatment of zeros as any
2295 With @samp{-ff90} in force, @code{g77} incorrectly
2296 interpreted @samp{REAL(Z)} as returning a @code{REAL}
2297 result, instead of as a @code{DOUBLE PRECISION}
2299 (Here, @samp{Z} is @code{DOUBLE COMPLEX}.)
2301 With @samp{-fno-f90} in force, the interpretation remains
2302 unchanged, since this appears to be how at least some
2303 F77 code using the @code{DOUBLE COMPLEX} extension expected
2306 Essentially, @samp{REAL(Z)} in F90 is the same as
2307 @samp{DBLE(Z)}, while in extended F77, it appears to
2308 be the same as @samp{REAL(REAL(Z))}.
2311 An expression involving exponentiation, where both operands
2312 were type @code{INTEGER} and the right-hand operand
2313 was negative, was erroneously evaluated.
2316 Fix bugs involving @code{DATA} implied-@code{DO} constructs
2317 (these involved an errant diagnostic and a crash, both on good
2318 code, one involving subsequent statement-function definition).
2321 Close @code{INCLUDE} files after processing them, so compiling source
2322 files with lots of @code{INCLUDE} statements does not result in
2323 being unable to open @code{INCLUDE} files after all the available
2324 file descriptors are used up.
2327 Speed up compiling, especially of larger programs, and perhaps
2328 slightly reduce memory utilization while compiling (this is
2329 @emph{not} the improvement planned for 0.6 involving large aggregate
2330 areas)---these improvements result from simply turning
2331 off some low-level code to do self-checking that hasn't been
2332 triggered in a long time.
2335 Introduce three new options that
2336 implement optimizations in the @code{gcc} back end (GBE).
2337 These options are @samp{-fmove-all-movables}, @samp{-freduce-all-givs},
2338 and @samp{-frerun-loop-opt}, which are enabled, by default,
2339 for Fortran compilations.
2340 These optimizations are intended to help toon Fortran programs.
2343 Patch the GBE to do a better job optimizing certain
2344 kinds of references to array elements.
2347 Due to patches to the GBE, the version number of @code{gcc}
2348 also is patched to make it easier to manage installations,
2349 especially useful if it turns out a @code{g77} change to the
2352 The @code{g77}-modified version number is the @code{gcc}
2353 version number with the string @samp{.f.@var{n}} appended,
2354 where @samp{f} identifies the version as enhanced for
2355 Fortran, and @var{n} is @samp{1} for the first Fortran
2356 patch for that version of @code{gcc}, @samp{2} for the
2359 So, this introduces version 2.7.2.f.1 of @code{gcc}.
2362 Make several improvements and fixes to diagnostics, including
2363 the removal of two that were inappropriate or inadequate.
2366 Warning about two successive arithmetic operators, produced
2367 by @samp{-Wsurprising}, now produced @emph{only} when both
2368 operators are, indeed, arithmetic (not relational/boolean).
2371 @samp{-Wsurprising} now warns about the remaining cases
2372 of using non-integral variables for implied-@code{DO}
2373 loops, instead of these being rejected unless @samp{-fpedantic}
2374 or @samp{-fugly} specified.
2377 Allow @code{SAVE} of a local variable or array, even after
2378 it has been given an initial value via @code{DATA}, for example.
2381 Introduce an Info version of @code{g77} documentation, which
2382 supercedes @file{gcc/f/CREDITS}, @file{gcc/f/DOC}, and
2383 @file{gcc/f/PROJECTS}.
2384 These files will be removed in a future release.
2385 The files @file{gcc/f/BUGS}, @file{gcc/f/INSTALL}, and
2386 @file{gcc/f/NEWS} now are automatically built from
2387 the texinfo source when distributions are made.
2389 This effort was inspired by a first pass at translating
2390 @file{g77-0.5.16/f/DOC} that was contributed to Craig by
2391 David Ronis (@email{ronis@@onsager.chem.mcgill.ca}).
2394 New @samp{-fno-second-underscore} option to specify
2395 that, when @samp{-funderscoring} is in effect, a second
2396 underscore is not to be appended to Fortran names already
2397 containing an underscore.
2400 Change the way iterative @code{DO} loops work to follow
2402 In particular, calculation of the iteration count is
2403 still done by converting the start, end, and increment
2404 parameters to the type of the @code{DO} variable, but
2405 the result of the calculation is always converted to
2406 the default @code{INTEGER} type.
2408 (This should have no effect on existing code compiled
2409 by @code{g77}, but code written to assume that use
2410 of a @emph{wider} type for the @code{DO} variable
2411 will result in an iteration count being fully calculated
2412 using that wider type (wider
2413 than default @code{INTEGER}) must be rewritten.)
2416 Support @code{gcc} version 2.7.2.
2419 Upgrade to @code{libf2c} as of 1996-03-23, and
2420 fix up some of the build procedures.
2422 Note that the email addresses related to @code{f2c}
2423 have changed---the distribution site now is
2424 named @code{netlib.bell-labs.com}, and the
2425 maintainer's new address is @email{dmg@@bell-labs.com}.
2428 @c 1995-11-18: 0.5.17 released.
2432 @strong{Fix serious bug} in @samp{g77 -v} command that can cause removal of a
2433 system's @file{/dev/null} special file if run by user @code{root}.
2435 @strong{All users} of version 0.5.16 should ensure that
2436 they have not removed @file{/dev/null} or replaced it with an ordinary
2437 file (e.g. by comparing the output of @samp{ls -l /dev/null} with
2438 @samp{ls -l /dev/zero}.
2439 If the output isn't basically the
2440 same, contact your system
2441 administrator about restoring @file{/dev/null} to its proper status).
2443 This bug is particularly insidious because removing @file{/dev/null} as
2444 a special file can go undetected for quite a while, aside from
2445 various applications and programs exhibiting sudden, strange
2448 I sincerely apologize for not realizing the
2449 implications of the fact that when @samp{g77 -v} runs the @code{ld} command
2450 with @samp{-o /dev/null} that @code{ld} tries to @emph{remove} the executable
2451 it is supposed to build (especially if it reports unresolved
2452 references, which it should in this case)!
2455 Fix crash on @samp{CHARACTER*(*) FOO} in a main or block data program unit.
2458 Fix crash that can occur when diagnostics given outside of any
2459 program unit (such as when input file contains @samp{@@foo}).
2462 Fix crashes, infinite loops (hangs), and such involving diagnosed code.
2465 Fix @code{ASSIGN}'ed variables so they can be @code{SAVE}'d or dummy arguments,
2466 and issue clearer error message in cases where target of @code{ASSIGN}
2467 or @code{ASSIGN}ed @code{GOTO}/@code{FORMAT} is too small (which should
2471 Make @code{libf2c} build procedures work on more systems again by
2472 eliminating unnecessary invocations of @samp{ld -r -x} and @code{mv}.
2475 Fix omission of @samp{-funix-intrinsics-@dots{}} options in list of permitted
2476 options to compiler.
2479 Fix failure to always diagnose missing type declaration for
2480 @code{IMPLICIT NONE}.
2483 Fix compile-time performance problem (which could sometimes
2484 crash the compiler, cause a hang, or whatever, due to a bug
2485 in the back end) involving exponentiation with a large @code{INTEGER}
2486 constant for the right-hand operator (e.g. @samp{I**32767}).
2489 Fix build procedures so cross-compiling @code{g77} (the @code{fini}
2490 utility in particular) is properly built using the host compiler.
2493 Add new @samp{-Wsurprising} option to warn about constructs that are
2494 interpreted by the Fortran standard (and @code{g77}) in ways that
2495 are surprising to many programmers.
2498 Add @code{ERF()} and @code{ERFC()} as generic intrinsics mapping to existing
2499 @code{ERF}/@code{DERF} and @code{ERFC}/@code{DERFC} specific intrinsics.
2501 @emph{Note:} You should
2502 specify @samp{INTRINSIC ERF,ERFC} in any code where you might use
2503 these as generic intrinsics, to improve likelihood of diagnostics
2504 (instead of subtle run-time bugs) when using a compiler that
2505 doesn't support these as intrinsics (e.g. @code{f2c}).
2508 Remove from @samp{-fno-pedantic} the diagnostic about @code{DO}
2509 with non-@code{INTEGER} index variable; issue that under
2510 @samp{-Wsurprising} instead.
2513 Clarify some diagnostics that say things like ``ignored'' when that's
2517 Clarify diagnostic on use of @code{.EQ.}/@code{.NE.} on @code{LOGICAL}
2521 Minor improvements to code generation for various operations on
2522 @code{LOGICAL} operands.
2525 Minor improvement to code generation for some @code{DO} loops on some
2529 Support @code{gcc} version 2.7.1.
2532 Upgrade to @code{libf2c} as of 1995-11-15.
2535 @c 1995-08-30: 0.5.16 released.
2539 Fix a code-generation bug involving complicated @code{EQUIVALENCE} statements
2540 not involving @code{COMMON}.
2543 Fix code-generation bugs involving invoking ``gratis'' library procedures
2544 in @code{libf2c} from code compiled with @samp{-fno-f2c} by making these
2545 procedures known to @code{g77} as intrinsics (not affected by -fno-f2c).
2546 This is known to fix code invoking @code{ERF()}, @code{ERFC()},
2547 @code{DERF()}, and @code{DERFC()}.
2550 Update @code{libf2c} to include netlib patches through 1995-08-16, and
2551 @code{#define} @code{WANT_LEAD_0} to 1 to make @code{g77}-compiled code more
2552 consistent with other Fortran implementations by outputting
2553 leading zeros in formatted and list-directed output.
2556 Fix a code-generation bug involving adjustable dummy arrays with high
2557 bounds whose primaries are changed during procedure execution, and
2558 which might well improve code-generation performance for such arrays
2559 compared to @code{f2c} plus @code{gcc} (but apparently only when using
2560 @file{gcc-2.7.0} or later).
2563 Fix a code-generation bug involving invocation of @code{COMPLEX} and
2564 @code{DOUBLE COMPLEX} @code{FUNCTION}s and doing @code{COMPLEX} and
2565 @code{DOUBLE COMPLEX} divides, when the result
2566 of the invocation or divide is assigned directly to a variable
2567 that overlaps one or more of the arguments to the invocation or divide.
2570 Fix crash by not generating new optimal code for @samp{X**I} if @samp{I} is
2571 nonconstant and the expression is used to dimension a dummy
2572 array, since the @code{gcc} back end does not support the necessary
2573 mechanics (and the @code{gcc} front end rejects the equivalent
2574 construct, as it turns out).
2577 Fix crash on expressions like @samp{COMPLEX**INTEGER}.
2580 Fix crash on expressions like @samp{(1D0,2D0)**2}, i.e. raising a
2581 @code{DOUBLE COMPLEX} constant to an @code{INTEGER} constant power.
2584 Fix crashes and such involving diagnosed code.
2587 Diagnose, instead of crashing on, statement function definitions
2588 having duplicate dummy argument names.
2591 Fix bug causing rejection of good code involving statement function
2595 Fix bug resulting in debugger not knowing size of local equivalence
2596 area when any member of area has initial value (via @code{DATA},
2600 Fix installation bug that prevented installation of @code{g77} driver.
2601 Provide for easy selection of whether to install copy of @code{g77}
2602 as @code{f77} to replace the broken code.
2605 Fix @code{gcc} driver (affects @code{g77} thereby) to not
2606 gratuitously invoke the
2607 @code{f771} program (e.g. when @samp{-E} is specified).
2610 Fix diagnostic to point to correct source line when it immediately
2611 follows an @code{INCLUDE} statement.
2614 Support more compiler options in @code{gcc}/@code{g77} when
2615 compiling Fortran files.
2616 These options include @samp{-p}, @samp{-pg}, @samp{-aux-info}, @samp{-P},
2617 correct setting of version-number macros for preprocessing, full
2618 recognition of @samp{-O0}, and
2619 automatic insertion of configuration-specific linker specs.
2622 Add new intrinsics that interface to existing routines in @code{libf2c}:
2623 @code{ABORT}, @code{DERF}, @code{DERFC}, @code{ERF}, @code{ERFC}, @code{EXIT},
2624 @code{FLUSH}, @code{GETARG}, @code{GETENV}, @code{IARGC},
2625 @code{SIGNAL}, and @code{SYSTEM}.
2626 Note that @code{ABORT}, @code{EXIT}, @code{FLUSH}, @code{SIGNAL}, and
2627 @code{SYSTEM} are intrinsic subroutines, not functions (since they
2628 have side effects), so to get the return values from @code{SIGNAL}
2629 and @code{SYSTEM}, append a final argument specifying an @code{INTEGER}
2630 variable or array element (e.g. @samp{CALL SYSTEM('rm foo',ISTAT)}).
2633 Add new intrinsic group named @code{unix} to contain the new intrinsics,
2634 and by default enable this new group.
2637 Move @code{LOC()} intrinsic out of the @code{vxt} group to the new
2641 Improve @code{g77} so that @samp{g77 -v} by itself (or with
2642 certain other options, including @samp{-B}, @samp{-b}, @samp{-i},
2643 @samp{-nostdlib}, and @samp{-V}) reports lots more useful
2644 version info, and so that long-form options @code{gcc} accepts are
2645 understood by @code{g77} as well (even in truncated, unambiguous forms).
2648 Add new @code{g77} option @samp{--driver=name} to specify driver when
2649 default, @code{gcc}, isn't appropriate.
2652 Add support for @samp{#} directives (as output by the preprocessor) in the
2653 compiler, and enable generation of those directives by the
2654 preprocessor (when compiling @samp{.F} files) so diagnostics and debugging
2655 info are more useful to users of the preprocessor.
2658 Produce better diagnostics, more like @code{gcc}, with info such as
2659 @samp{In function `foo':} and @samp{In file included from...:}.
2662 Support @code{gcc}'s @samp{-fident} and @samp{-fno-ident} options.
2665 When @samp{-Wunused} in effect, don't warn about local variables used as
2666 statement-function dummy arguments or @code{DATA} implied-@code{DO} iteration
2667 variables, even though, strictly speaking, these are not uses
2668 of the variables themselves.
2671 When @samp{-W -Wunused} in effect, don't warn about unused dummy arguments
2672 at all, since there's no way to turn this off for individual
2673 cases (@code{g77} might someday start warning about these)---applies
2674 to @code{gcc} versions 2.7.0 and later, since earlier versions didn't
2675 warn about unused dummy arguments.
2678 New option @samp{-fno-underscoring} that inhibits transformation of names
2679 (by appending one or two underscores) so users may experiment
2680 with implications of such an environment.
2683 Minor improvement to @file{gcc/f/info} module to make it easier to build
2684 @code{g77} using the native (non-@code{gcc}) compiler on certain machines
2685 (but definitely not all machines nor all non-@code{gcc} compilers).
2687 do not report bugs showing problems compilers have with
2688 macros defined in @file{gcc/f/target.h} and used in places like
2689 @file{gcc/f/expr.c}.
2692 Add warning to be printed for each invocation of the compiler
2693 if the target machine @code{INTEGER}, @code{REAL}, or @code{LOGICAL} size
2695 since @code{g77} is known to not work well for such cases.
2698 Lots of new documentation (though work is still needed to put it into
2699 canonical GNU format).
2702 Build @code{libf2c} with @samp{-g0}, not @samp{-g2}, in effect
2703 (by default), to produce
2704 smaller library without lots of debugging clutter.
2707 @c 1995-05-19: 0.5.15 released.
2711 Fix bad code generation involving @samp{X**I} and temporary, internal variables
2712 generated by @code{g77} and the back end (such as for @code{DO} loops).
2715 Fix crash given @samp{CHARACTER A;DATA A/.TRUE./}.
2718 Replace crash with diagnostic given @samp{CHARACTER A;DATA A/1.0/}.
2721 Fix crash or other erratic behavior when null character constant
2722 (@samp{''}) is encountered.
2725 Fix crash or other erratic behavior involving diagnosed code.
2728 Fix code generation for external functions returning type @code{REAL} when
2729 the @samp{-ff2c} option is in force (which it is by default) so that
2730 @code{f2c} compatibility is indeed provided.
2733 Disallow @samp{COMMON I(10)} if @samp{I} has previously been specified
2734 with an array declarator.
2737 New @samp{-ffixed-line-length-@var{n}} option, where @var{n} is the
2739 of a typical fixed-form line, defaulting to 72 columns, such
2740 that characters beyond column @var{n} are ignored, or @var{n} is @samp{none},
2741 meaning no characters are ignored.
2742 does not affect lines
2743 with @samp{&} in column 1, which are always processed as if
2744 @samp{-ffixed-line-length-none} was in effect.
2747 No longer generate better code for some kinds of array references,
2748 as @code{gcc} back end is to be fixed to do this even better, and it
2749 turned out to slow down some code in some cases after all.
2752 In @code{COMMON} and @code{EQUIVALENCE} areas with any members given initial
2753 values (e.g. via @code{DATA}), uninitialized members now always
2754 initialized to binary zeros (though this is not required by
2755 the standard, and might not be done in future versions
2757 Previously, in some @code{COMMON}/@code{EQUIVALENCE} areas
2758 (essentially those with members of more than one type), the
2759 uninitialized members were initialized to spaces, to
2760 cater to @code{CHARACTER} types, but it seems no existing code expects
2761 that, while much existing code expects binary zeros.
2767 Don't emit bad code when low bound of adjustable array is nonconstant
2768 and thus might vary as an expression at run time.
2771 Emit correct code for calculation of number of trips in @code{DO} loops
2773 where the loop should not execute at all.
2774 (This bug affected cases
2775 where the difference between the begin and end values was less
2776 than the step count, though probably not for floating-point cases.)
2779 Fix crash when extra parentheses surround item in
2780 @code{DATA} implied-@code{DO} list.
2783 Fix crash over minor internal inconsistencies in handling diagnostics,
2784 just substitute dummy strings where necessary.
2787 Fix crash on some systems when compiling call to @code{MVBITS()} intrinsic.
2790 Fix crash on array assignment @samp{TYPE@var{ddd}(@dots{})=@dots{}}, where @var{ddd}
2791 is a string of one or more digits.
2794 Fix crash on @code{DCMPLX()} with a single @code{INTEGER} argument.
2797 Fix various crashes involving code with diagnosed errors.
2800 Support @samp{-I} option for @code{INCLUDE} statement, plus @code{gcc}'s
2801 @file{header.gcc} facility for handling systems like MS-DOS.
2804 Allow @code{INCLUDE} statement to be continued across multiple lines,
2805 even allow it to coexist with other statements on the same line.
2808 Incorporate Bellcore fixes to @code{libf2c} through 1995-03-15---this
2809 fixes a bug involving infinite loops reading EOF with empty list-directed
2813 Remove all the @code{g77}-specific auto-configuration scripts, code,
2815 except for temporary substitutes for bsearch() and strtoul(), as
2816 too many configure/build problems were reported in these areas.
2817 People will have to fix their systems' problems themselves, or at
2818 least somewhere other than @code{g77}, which expects a working ANSI C
2819 environment (and, for now, a GNU C compiler to compile @code{g77} itself).
2822 Complain if initialized common redeclared as larger in subsequent program
2826 Warn if blank common initialized, since its size can vary and hence
2827 related warnings that might be helpful won't be seen.
2830 New @samp{-fbackslash} option, on by default, that causes @samp{\}
2831 within @code{CHARACTER}
2832 and Hollerith constants to be interpreted a la GNU C.
2834 this behavior is somewhat different from @code{f2c}'s, which supports only
2835 a limited subset of backslash (escape) sequences.
2838 Make @samp{-fugly-args} the default.
2841 New @samp{-fugly-init} option, on by default, that allows typeless/Hollerith
2842 to be specified as initial values for variables or named constants
2843 (@code{PARAMETER}), and also allows character<->numeric conversion in
2844 those contexts---turn off via @samp{-fno-ugly-init}.
2847 New @samp{-finit-local-zero} option to initialize
2848 local variables to binary zeros.
2849 This does not affect whether they are @code{SAVE}d, i.e. made
2850 automatic or static.
2853 New @samp{-Wimplicit} option to warn about implicitly typed variables, arrays,
2855 (Basically causes all program units to default to @code{IMPLICIT NONE}.)
2858 @samp{-Wall} now implies @samp{-Wuninitialized} as with @code{gcc}
2859 (i.e. unless @samp{-O} not specified, since @samp{-Wuninitialized}
2860 requires @samp{-O}), and implies @samp{-Wunused} as well.
2863 @samp{-Wunused} no longer gives spurious messages for unused
2864 @code{EXTERNAL} names (since they are assumed to refer to block data
2865 program units, to make use of libraries more reliable).
2868 Support @code{%LOC()} and @code{LOC()} of character arguments.
2871 Support null (zero-length) character constants and expressions.
2874 Support @code{f2c}'s @code{IMAG()} generic intrinsic.
2877 Support @code{ICHAR()}, @code{IACHAR()}, and @code{LEN()} of
2878 character expressions that are valid in assignments but
2879 not normally as actual arguments.
2882 Support @code{f2c}-style @samp{&} in column 1 to mean continuation line.
2885 Allow @code{NAMELIST}, @code{EXTERNAL}, @code{INTRINSIC}, and @code{VOLATILE}
2886 in @code{BLOCK DATA}, even though these are not allowed by the standard.
2889 Allow @code{RETURN} in main program unit.
2892 Changes to Hollerith-constant support to obey Appendix C of the
2897 Now padded on the right with zeros, not spaces.
2900 Hollerith ``format specifications'' in the form of arrays of
2901 non-character allowed.
2904 Warnings issued when non-space truncation occurs when converting
2908 When specified as actual argument, now passed
2909 by reference to @code{INTEGER} (padded on right with spaces if constant
2910 too small, otherwise fully intact if constant wider the @code{INTEGER}
2911 type) instead of by value.
2914 @strong{Warning:} @code{f2c} differs on the
2915 interpretation of @samp{CALL FOO(1HX)}, which it treats exactly the
2916 same as @samp{CALL FOO('X')}, but which the standard and @code{g77} treat
2917 as @samp{CALL FOO(%REF('X '))} (padded with as many spaces as necessary
2918 to widen to @code{INTEGER}), essentially.
2921 Changes and fixes to typeless-constant support:
2925 Now treated as a typeless double-length @code{INTEGER} value.
2928 Warnings issued when overflow occurs.
2931 Padded on the left with zeros when converting
2935 Should be properly aligned and ordered on
2936 the target machine for whatever type it is turned into.
2939 When specified as actual argument, now passed as reference to
2940 a default @code{INTEGER} constant.
2944 @code{%DESCR()} of a non-@code{CHARACTER} expression now passes a pointer to
2945 the expression plus a length for the expression just as if
2946 it were a @code{CHARACTER} expression.
2947 For example, @samp{CALL FOO(%DESCR(D))}, where
2948 @samp{D} is @code{REAL*8}, is the same as @samp{CALL FOO(D,%VAL(8)))}.
2951 Name of multi-entrypoint master function changed to incorporate
2952 the name of the primary entry point instead of a decimal
2953 value, so the name of the master function for @samp{SUBROUTINE X}
2954 with alternate entry points is now @samp{__g77_masterfun_x}.
2957 Remove redundant message about zero-step-count @code{DO} loops.
2960 Clean up diagnostic messages, shortening many of them.
2963 Fix typo in @code{g77} man page.
2966 Clarify implications of constant-handling bugs in @file{f/BUGS}.
2969 Generate better code for @samp{**} operator with a right-hand operand of
2970 type @code{INTEGER}.
2973 Generate better code for @code{SQRT()} and @code{DSQRT()},
2974 also when @samp{-ffast-math}
2975 specified, enable better code generation for @code{SIN()} and @code{COS()}.
2978 Generate better code for some kinds of array references.
2981 Speed up lexing somewhat (this makes the compilation phase noticeably