1 @c Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002
2 @c Free Software Foundation, Inc.
3 @c This is part of the G77 manual.
4 @c For copying conditions, see the file g77.texi.
6 @c The text of this file appears in the file NEWS
7 @c in the G77 distribution, as well as in the G77 manual.
9 @c Keep this the same as the dates above, since it's used
10 @c in the standalone derivations of this file (e.g. NEWS).
11 @set copyrights-news 1995,1996,1997,1998,1999,2000,2001,2002
13 @set last-update-news 2002-11-19
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 @command{g77} that are visible
44 to the programmers who actually write and maintain Fortran
45 code they compile with @command{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
49 users use @command{g77}.
50 @xref{News,,News About GNU Fortran}, for information on
51 such changes to @command{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 @command{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, as appropriate.
100 Note that all such variants are obsolete @emph{as of July 1999} -
101 the information is retained here only for its historical value.
103 Therefore, @code{egcs} versions sometimes have multiple listings
104 to help clarify how they differ from other versions,
105 though this can make getting a complete picture
106 of what a particular @code{egcs} version contains
107 somewhat more difficult.
110 For information on bugs in the @value{which-g77} version of @command{g77},
111 see @ref{Known Bugs,,Known Bugs In GNU Fortran}.
115 For information on bugs in the @value{which-g77} version of @command{g77},
116 see @file{@value{path-g77}/BUGS}.
120 @emph{Warning:} The information below is still under development,
121 and might not accurately reflect the @command{g77} code base
122 of which it is a part.
123 Efforts are made to keep it somewhat up-to-date,
124 but they are particularly concentrated
125 on any version of this information
126 that is distributed as part of a @emph{released} @command{g77}.
128 In particular, while this information is intended to apply to
129 the @value{which-g77} version of @command{g77},
130 only an official @emph{release} of that version
131 is expected to contain documentation that is
132 most consistent with the @command{g77} product in that version.
134 Nevertheless, information on @emph{previous} releases of @command{g77}, below,
135 is likely to be more up-to-date and accurate
136 than the equivalent information that accompanied
138 assuming the last-updated date of the information below
139 is later than the dates of those releases.
141 That's due to attempts to keep this development version
142 of news about previous @command{g77} versions up-to-date.
146 An online, ``live'' version of this document
147 (derived directly from the mainline, development version
148 of @command{g77} within @command{gcc})
150 @uref{http://gcc.gnu.org/onlinedocs/g77/News.html}.
153 The following information was last updated on @value{last-update-news}:
155 @heading In @code{GCC} 3.3 versus @code{GCC} 3.2:
158 Problem Reports fixed (in chronological order of submission):
161 g77 generates code which is rejected by GAS if COFF debugging info is
164 Broken links on web pages
166 (libf2c) multiple repeat counts confuse namelist read into array
168 Logical operations error on logicals when using -fugly-logint
170 Generation of C++ Prototype for FORTRAN and extern "C"
172 Failure of g77.f-torture/execute/f90-intrinsic-bit.f -Os on irix6.5
174 (libf2c) OPEN(...,RECL=nnn,...) without ACCESS='DIRECT' should assume a direct
177 g77 "bug"; the executable misbehave (use of options -O2 -fno-automatic
180 (libf2c) DATE_AND_TIME milliseconds field inactive on Windows
182 Incorrect output with 0-based array of characters
184 Double complex zero ** double precision number -> NaN instead of zero
187 Richard Henderson (@email{rth@@redhat.com}) analyzed and improved the handling
188 of (no-)aliasing information for dummy arguments and improved the optimization
189 of induction variables in unrolled loops.
192 @heading In @code{GCC} 3.2 versus @code{GCC} 3.1:
195 Problem Reports fixed (in chronological order of submission):
198 gcc-3.x does not compile files with suffix .r (RATFOR) [Fixed in 3.2.1]
202 @heading In @code{GCC} 3.1 (formerly known as g77-0.5.27) versus @code{GCC} 3.0:
205 Problem Reports fixed (in chronological order of submission):
208 Data statement initialization with subscript of kind INTEGER*2
210 Reference to intrinsic `ISHFT' invalid
212 Function BESJN(integer,double) problems
214 g77 -pipe -xf77-cpp-input sends output to stdout
216 g77 -h" gives bogus output
218 ICE on valid input using CALL EXIT(%VAL(...))
220 g77 -v -c -xf77-version /dev/null -xnone causes ice
222 BACKSPACE example that doesn't work as of gcc/g77-3.0.x
224 g77 rejects accepted use of INTEGER*2 as type of DATA statement loop index
226 ICE on compiling source with 540 000 000 REAL array
228 ICE on BESJN(integer*8,real)
230 bug in loop unrolling
232 sparc-sun-solaris2.7 gcc-3.1 extra g77 testsuite failures w/-m64
234 Incorrect acces of integer*1 variables on PA
236 Failure of LAPACK test dtest on irix6.5 with -mabi=64 -O2
240 @command{g77} now has its man page generated from the texinfo documentation,
241 to guarantee that it remains up to date.
244 @command{g77} used to reject the following program on 32-bit targets:
247 DIMENSION A(140 000 000)
252 prog.f: In program `prog':
254 DIMENSION A(140 000 000)
256 Array `a' at (^) is too large to handle
258 because 140 000 000 REALs is larger than the largest bit-extent that can be
259 expressed in 32 bits. However, bit-sizes never play a role after offsets
260 have been converted to byte addresses. Therefore this check has been removed,
261 and the limit is now 2 Gbyte of memory (around 530 000 000 REALs).
262 Note: On GNU/Linux systems one has to compile and link programs that occupy
263 more than 1 Gbyte statically, i.e.@: @code{g77 -static ...}.
266 Based on work done by Juergen Pfeifer (@email{juergen.pfeifer@@gmx.net})
267 libf2c is now a shared library. One can still link in all objects with
268 the program by specifying the @option{-static} option.
271 Robert Anderson (@email{rwa@@alumni.princeton.edu}) thought up a two
272 line change that enables g77 to compile such code as:
276 DIMENSION A(N(1),N(2))
280 Note the use of array elements in the bounds of the adjustable array A.
283 George Helffrich (@email{george@@geo.titech.ac.jp}) implemented a change
284 in substring index checking (when specifying @option{-fbounds-check})
285 that permits the use of zero length substrings of the form
289 Based on code developed by Pedro Vazquez (@email{vazquez@@penelope.iqm.unicamp.br}),
290 the @code{libf2c} library is now able to read and write files larger than
291 2 Gbyte on 32-bit target machines, if the operating system supports this.
294 @heading In 0.5.26, @code{GCC} 3.0 versus @code{GCC} 2.95:
297 When a REWIND was issued after a WRITE statement on an unformatted
298 file, the implicit truncation was performed by copying the truncated
299 file to /tmp and copying the result back. This has been fixed by using
300 the @code{ftruncate} OS function. Thanks go to the GAMESS developers
301 for bringing this to our attention.
304 Using options @option{-g}, @option{-ggdb} or @option{-gdwarf[-2]} (where
305 appropriate for your target) now also enables debugging information
306 for COMMON BLOCK and EQUIVALENCE items to be emitted.
307 Thanks go to Andrew Vaught (@email{andy@@xena.eas.asu.edu}) and
308 George Helffrich (@email{george@@geology.bristol.ac.uk}) for
309 fixing this longstanding problem.
312 It is not necessary anymore to use the option @option{-femulate-complex}
313 to compile Fortran code using COMPLEX arithmetic, even on 64-bit machines
314 (like the Alpha). This will improve code generation.
317 INTRINSIC arithmetic functions are now treated as routines that do not
318 depend on anything but their argument(s). This enables further instruction
319 scheduling, because it is known that they cannot read or modify arbitrary
324 Upgrade to @code{libf2c} as of 2000-12-05.
326 This fixes a bug where a namelist containing initialization of LOGICAL
327 items and a variable starting with T or F would be read incorrectly.
330 The @code{TtyNam} intrinsics now set @var{Name} to all spaces (at run time)
331 if the system has no @code{ttyname} implementation available.
334 Upgrade to @code{libf2c} as of 1999-06-28.
336 This fixes a bug whereby
337 input to a @code{NAMELIST} read involving a repeat count,
338 such as @samp{K(5)=10*3},
339 was not properly handled by @code{libf2c}.
340 The first item was written to @samp{K(5)},
341 but the remaining nine were written elsewhere (still within the array),
342 not necessarily starting at @samp{K(6)}.
346 @heading In 0.5.25, @code{GCC} 2.95 (@code{EGCS} 1.2) versus @code{EGCS} 1.1.2:
350 @command{g77} no longer generates bad code for assignments,
351 or other conversions,
352 of @code{REAL} or @code{COMPLEX} constant expressions
353 to type @code{INTEGER(KIND=2)}
354 (often referred to as @code{INTEGER*8}).
356 For example, @samp{INTEGER*8 J; J = 4E10} now works as documented.
361 @command{g77} no longer truncates @code{INTEGER(KIND=2)}
362 (usually @code{INTEGER*8})
363 subscript expressions when evaluating array references
364 on systems with pointers widers than @code{INTEGER(KIND=1)}
370 @command{g77} no longer generates bad code
371 for an assignment to a @code{COMPLEX} variable or array
372 that partially overlaps one or more of the sources
373 of the same assignment
374 (a very rare construction).
375 It now assigns through a temporary,
376 in cases where such partial overlap is deemed possible.
381 @code{libg2c} (@code{libf2c}) no longer loses track
382 of the file being worked on
383 during a @code{BACKSPACE} operation.
388 @code{libg2c} (@code{libf2c}) fixes a bug whereby
389 input to a @code{NAMELIST} read involving a repeat count,
390 such as @samp{K(5)=10*3},
391 was not properly handled by @code{libf2c}.
392 The first item was written to @samp{K(5)},
393 but the remaining nine were written elsewhere (still within the array),
394 not necessarily starting at @samp{K(6)}.
399 @c Tim Prince reported this, regarding the TEST_FPU benchmark.
400 Automatic arrays now seem to be working on HP-UX systems.
405 The @code{Date} intrinsic now returns the correct result
406 on big-endian systems.
411 Fix @command{g77} so it no longer crashes when compiling
412 I/O statements using keywords that define @code{INTEGER} values,
413 such as @samp{IOSTAT=@var{j}},
414 where @var{j} is other than default @code{INTEGER}
415 (such as @code{INTEGER*2}).
416 Instead, it issues a diagnostic.
421 Fix @command{g77} so it properly handles @samp{DATA A/@var{rpt}*@var{val}/},
422 where @var{rpt} is not default @code{INTEGER}, such as @code{INTEGER*2},
423 instead of producing a spurious diagnostic.
424 Also fix @samp{DATA (A(I),I=1,N)},
425 where @samp{N} is not default @code{INTEGER}
426 to work instead of crashing @command{g77}.
431 The @option{-ax} option is now obeyed when compiling Fortran programs.
432 (It is passed to the @file{f771} driver.)
436 The new @option{-fbounds-check} option
437 causes @command{g77} to compile run-time bounds checks
438 of array subscripts, as well as of substring start and end points.
441 @code{libg2c} now supports building as multilibbed library,
442 which provides better support for systems
443 that require options such as @option{-mieee}
447 Source file names with the suffixes @samp{.FOR} and @samp{.FPP}
448 now are recognized by @command{g77}
449 as if they ended in @samp{.for} and @samp{.fpp}, respectively.
452 The order of arguments to the @emph{subroutine} forms of the
453 @code{CTime}, @code{DTime}, @code{ETime}, and @code{TtyNam}
454 intrinsics has been swapped.
455 The argument serving as the returned value
456 for the corresponding function forms
457 now is the @emph{second} argument,
458 making these consistent with the other subroutine forms
459 of @code{libU77} intrinsics.
462 @command{g77} now warns about a reference to an intrinsic
463 that has an interface that is not Year 2000 (Y2K) compliant.
464 Also, @code{libg2c} has been changed to increase the likelihood
465 of catching references to the implementations of these intrinsics
466 using the @code{EXTERNAL} mechanism
467 (which would avoid the new warnings).
470 @xref{Year 2000 (Y2K) Problems}, for more information.
475 @command{g77} now warns about a reference to a function
476 when the corresponding @emph{subsequent} function program unit
477 disagrees with the reference concerning the type of the function.
481 @option{-fno-emulate-complex} is now the default option.
482 This should result in improved performance
483 of code that uses the @code{COMPLEX} data type.
486 @cindex double-precision performance
487 @cindex -malign-double
489 The @option{-malign-double} option
490 now reliably aligns @emph{all} double-precision variables and arrays
491 on Intel x86 targets.
495 Even without the @option{-malign-double} option,
496 @command{g77} reliably aligns local double-precision variables
497 that are not in @code{EQUIVALENCE} areas
498 and not @code{SAVE}'d.
503 @command{g77} now open-codes (``inlines'') division of @code{COMPLEX} operands
504 instead of generating a run-time call to
505 the @code{libf2c} routines @code{c_div} or @code{z_div},
506 unless the @option{-Os} option is specified.
510 @command{g77} no longer generates code to maintain @code{errno},
511 a C-language concept,
512 when performing operations such as the @code{SqRt} intrinsic.
516 @command{g77} developers can temporarily use
517 the @option{-fflatten-arrays} option
518 to compare how the compiler handles code generation
519 using C-like constructs as compared to the
520 Fortran-like method constructs normally used.
525 A substantial portion of the @command{g77} front end's code-generation component
527 It now generates code using facilities more robustly supported
528 by the @command{gcc} back end.
529 One effect of this rewrite is that some codes no longer produce
530 a spurious ``label @var{lab} used before containing binding contour''
535 Support for the @option{-fugly} option has been removed.
539 Improve documentation and indexing,
540 including information on Year 2000 (Y2K) compliance,
541 and providing more information on internals of the front end.
546 Upgrade to @code{libf2c} as of 1999-05-10.
550 @heading In 0.5.24 versus 0.5.23:
552 There is no @command{g77} version 0.5.24 at this time,
554 0.5.24 is the version number designated for bug fixes and,
555 perhaps, some new features added,
557 Version 0.5.23 requires @command{gcc} 2.8.1,
558 as 0.5.24 was planned to require.
560 Due to @code{EGCS} becoming @code{GCC}
561 (which is now an acronym for ``GNU Compiler Collection''),
562 and @code{EGCS} 1.2 becoming officially designated @code{GCC} 2.95,
563 there seems to be no need for an actual 0.5.24 release.
565 To reduce the confusion already resulting from use of 0.5.24
566 to designate @command{g77} versions within @code{EGCS} versions 1.0 and 1.1,
567 as well as in versions of @command{g77} documentation and notices
569 ``mainline'' @command{g77} version numbering resumes
570 at 0.5.25 with @code{GCC} 2.95 (@code{EGCS} 1.2),
571 skipping over 0.5.24 as a placeholder version number.
573 To repeat, there is no @command{g77} 0.5.24, but there is now a 0.5.25.
574 Please remain calm and return to your keypunch units.
576 @c 1999-03-15: EGCS 1.1.2 released.
577 @heading In @code{EGCS} 1.1.2 versus @code{EGCS} 1.1.1:
581 Fix the @code{IDate} intrinsic (VXT) (in @code{libg2c})
582 so the returned year is in the documented, non-Y2K-compliant range
584 instead of being returned as 100 in the year 2000.
587 @xref{IDate Intrinsic (VXT)},
588 for more information.
592 Fix the @code{Date_and_Time} intrinsic (in @code{libg2c})
593 to return the milliseconds value properly
597 Fix the @code{LStat} intrinsic (in @code{libg2c})
598 to return device-ID information properly
602 Improve documentation.
606 @c 1998-12-04: EGCS 1.1.1 released.
607 @heading In @code{EGCS} 1.1.1 versus @code{EGCS} 1.1:
611 Fix @code{libg2c} so it performs an implicit @code{ENDFILE} operation
613 whenever a @code{REWIND} is done.
615 (This bug was introduced in 0.5.23 and @code{egcs} 1.1 in
616 @command{g77}'s version of @code{libf2c}.)
619 Fix @code{libg2c} so it no longer crashes with a spurious diagnostic
620 upon doing any I/O following a direct formatted write.
622 (This bug was introduced in 0.5.23 and @code{egcs} 1.1 in
623 @command{g77}'s version of @code{libf2c}.)
626 Fix @command{g77} so it no longer crashes compiling references
627 to the @code{Rand} intrinsic on some systems.
630 Fix @command{g77} portion of installation process so it works
631 better on some systems
632 (those with shells requiring @samp{else true} clauses
633 on @code{if} constructs
634 for the completion code to be set properly).
638 @c 1998-09-03: EGCS 1.1 released.
639 @heading In @code{EGCS} 1.1 versus @code{EGCS} 1.0.3:
643 Fix bugs in the @code{libU77} intrinsic @code{HostNm}
644 that wrote one byte beyond the end of its @code{CHARACTER}
646 and in the @code{libU77} intrinsics
647 @code{GMTime} and @code{LTime}
648 that overwrote their arguments.
653 Assumed arrays with negative bounds
654 (such as @samp{REAL A(-1:*)})
655 no longer elicit spurious diagnostics from @command{g77},
656 even on systems with pointers having
657 different sizes than integers.
659 This bug is not known to have existed in any
660 recent version of @command{gcc}.
661 It was introduced in an early release of @code{egcs}.
666 Valid combinations of @code{EXTERNAL},
667 passing that external as a dummy argument
668 without explicitly giving it a type,
669 and, in a subsequent program unit,
670 referencing that external as
671 an external function with a different type
672 no longer crash @command{g77}.
677 @code{CASE DEFAULT} no longer crashes @command{g77}.
682 The @option{-Wunused} option no longer issues a spurious
683 warning about the ``master'' procedure generated by
684 @command{g77} for procedures containing @code{ENTRY} statements.
688 Support @samp{FORMAT(I<@var{expr}>)} when @var{expr} is a
689 compile-time constant @code{INTEGER} expression.
692 Fix @command{g77} @option{-g} option so procedures that
693 use @code{ENTRY} can be stepped through, line by line,
697 Allow any @code{REAL} argument to intrinsics
698 @code{Second} and @code{CPU_Time}.
701 Use @code{tempnam}, if available, to open scratch files
702 (as in @samp{OPEN(STATUS='SCRATCH')})
703 so that the @code{TMPDIR} environment variable,
707 @command{g77}'s version of @code{libf2c} separates out
708 the setting of global state
709 (such as command-line arguments and signal handling)
710 from @file{main.o} into distinct, new library
713 This should make it easier to write portable applications
714 that have their own (non-Fortran) @code{main()} routine
715 properly set up the @code{libf2c} environment, even
716 when @code{libf2c} (now @code{libg2c}) is a shared library.
720 @command{g77} no longer installs the @file{f77} command
721 and @file{f77.1} man page
722 in the @file{/usr} or @file{/usr/local} hierarchy,
723 even if the @file{f77-install-ok} file exists
724 in the source or build directory.
725 See the installation documentation for more information.
730 @command{g77} no longer installs the @file{libf2c.a} library
731 and @file{f2c.h} include file
732 in the @file{/usr} or @file{/usr/local} hierarchy,
733 even if the @file{f2c-install-ok} or @file{f2c-exists-ok} files exist
734 in the source or build directory.
735 See the installation documentation for more information.
740 The @file{libf2c.a} library produced by @command{g77} has been
741 renamed to @file{libg2c.a}.
742 It is installed only in the @command{gcc} ``private''
743 directory hierarchy, @file{gcc-lib}.
744 This allows system administrators and users to choose which
745 version of the @code{libf2c} library from @code{netlib} they
746 wish to use on a case-by-case basis.
747 See the installation documentation for more information.
752 The @file{f2c.h} include (header) file produced by @command{g77}
753 has been renamed to @file{g2c.h}.
754 It is installed only in the @command{gcc} ``private''
755 directory hierarchy, @file{gcc-lib}.
756 This allows system administrators and users to choose which
757 version of the include file from @code{netlib} they
758 wish to use on a case-by-case basis.
759 See the installation documentation for more information.
763 The @command{g77} command now expects the run-time library
764 to be named @code{libg2c.a} instead of @code{libf2c.a},
765 to ensure that a version other than the one built and
766 installed as part of the same @command{g77} version is picked up.
770 During the configuration and build process,
771 @command{g77} creates subdirectories it needs only as it
773 Other cleaning up of the configuration and build process
774 has been performed as well.
779 @code{install-info} now used to update the directory of
780 Info documentation to contain an entry for @command{g77}
781 (during installation).
785 Some diagnostics have been changed from warnings to errors,
786 to prevent inadvertent use of the resulting, probably buggy,
788 These mostly include diagnostics about use of unsupported features
789 in the @code{OPEN}, @code{INQUIRE}, @code{READ}, and
790 @code{WRITE} statements,
791 and about truncations of various sorts of constants.
795 Improve compilation of @code{FORMAT} expressions so that
796 a null byte is appended to the last operand if it
798 This provides a cleaner run-time diagnostic as provided
799 by @code{libf2c} for statements like @samp{PRINT '(I1', 42}.
804 Improve documentation and indexing.
809 The upgrade to @code{libf2c} as of 1998-06-18
810 should fix a variety of problems, including
811 those involving some uses of the @code{T} format
812 specifier, and perhaps some build (porting) problems
817 @c 1998-09-03: EGCS 1.1 released.
818 @heading In @code{EGCS} 1.1 versus @command{g77} 0.5.23:
822 @cindex stack, 387 coprocessor
826 Fix a code-generation bug that afflicted
827 Intel x86 targets when @option{-O2} was specified
828 compiling, for example, an old version of
829 the @code{DNRM2} routine.
831 The x87 coprocessor stack was being
832 mismanaged in cases involving assigned @code{GOTO}
838 @command{g77} no longer produces incorrect code
840 for @code{EQUIVALENCE} and @code{COMMON}
841 aggregates that, due to ``unnatural'' ordering of members
842 vis-a-vis their types, require initial padding.
847 Fix @command{g77} crash compiling code
848 containing the construct @samp{CMPLX(0.)} or similar.
853 @command{g77} no longer crashes when compiling code
854 containing specification statements such as
855 @samp{INTEGER(KIND=7) PTR}.
860 @command{g77} no longer crashes when compiling code
861 such as @samp{J = SIGNAL(1, 2)}.
865 @command{g77} now treats @samp{%LOC(@var{expr})} and
866 @samp{LOC(@var{expr})} as ``ordinary'' expressions
867 when they are used as arguments in procedure calls.
868 This change applies only to global (filewide) analysis,
869 making it consistent with
870 how @command{g77} actually generates code
873 Previously, @command{g77} treated these expressions
874 as denoting special ``pointer'' arguments
875 for the purposes of filewide analysis.
879 Fix @command{g77} crash
880 (or apparently infinite run-time)
881 when compiling certain complicated expressions
882 involving @code{COMPLEX} arithmetic
883 (especially multiplication).
887 @cindex double-precision performance
888 @cindex -malign-double
890 Align static double-precision variables and arrays
892 regardless of whether @option{-malign-double} is specified.
894 Generally, this affects only local variables and arrays
895 having the @code{SAVE} attribute
896 or given initial values via @code{DATA}.
899 The @command{g77} driver now ensures that @option{-lg2c}
900 is specified in the link phase prior to any
901 occurrence of @option{-lm}.
902 This prevents accidentally linking to a routine
903 in the SunOS4 @option{-lm} library
904 when the generated code wants to link to the one
905 in @code{libf2c} (@code{libg2c}).
908 @command{g77} emits more debugging information when
911 This new information allows, for example,
912 @kbd{which __g77_length_a} to be used in @command{gdb}
913 to determine the type of the phantom length argument
914 supplied with @code{CHARACTER} variables.
916 This information pertains to internally-generated
917 type, variable, and other information,
918 not to the longstanding deficiencies vis-a-vis
919 @code{COMMON} and @code{EQUIVALENCE}.
922 The F90 @code{Date_and_Time} intrinsic now is
926 The F90 @code{System_Clock} intrinsic allows
927 the optional arguments (except for the @code{Count}
928 argument) to be omitted.
932 Upgrade to @code{libf2c} as of 1998-06-18.
937 Improve documentation and indexing.
942 @heading In previous versions:
944 Information on previous versions is not provided
945 in this @file{@value{path-g77}/NEWS} file,
947 See @file{@value{path-g77}/news.texi},
948 or any of its other derivations
949 (Info, HTML, dvi forms)
950 for such information.
954 @c 1998-05-20: 0.5.23 released.
955 @heading In 0.5.23 versus 0.5.22:
958 This release contains several regressions against
959 version 0.5.22 of @command{g77}, due to using the
960 ``vanilla'' @command{gcc} back end instead of patching
961 it to fix a few bugs and improve performance in a
964 Features that have been dropped from this version
965 of @command{g77} due to their being implemented
966 via @command{g77}-specific patches to the @command{gcc}
967 back end in previous releases include:
971 Support for @code{__restrict__} keyword,
972 the options @option{-fargument-alias}, @option{-fargument-noalias},
973 and @option{-fargument-noalias-global},
974 and the corresponding alias-analysis code.
976 (@code{egcs} has the alias-analysis
977 code, but not the @code{__restrict__} keyword.
978 @code{egcs} @command{g77} users benefit from the alias-analysis
979 code despite the lack of the @code{__restrict__} keyword,
980 which is a C-language construct.)
983 Support for the GNU compiler options
984 @option{-fmove-all-movables},
985 @option{-freduce-all-givs},
986 and @option{-frerun-loop-opt}.
988 (@code{egcs} supports these options.
989 @command{g77} users of @code{egcs} benefit from them even if
990 they are not explicitly specified,
991 because the defaults are optimized for @command{g77} users.)
994 Support for the @option{-W} option warning about
995 integer division by zero.
998 The Intel x86-specific option @option{-malign-double}
999 applying to stack-allocated data
1000 as well as statically-allocate data.
1003 @ifclear USERVISONLY
1004 Note that the @file{gcc/f/gbe/} subdirectory has been removed
1005 from this distribution as a result of @command{g77} no longer
1006 including patches for the @command{gcc} back end.
1009 @ifclear USERVISONLY
1011 Fix bugs in the @code{libU77} intrinsic @code{HostNm}
1012 that wrote one byte beyond the end of its @code{CHARACTER}
1014 and in the @code{libU77} intrinsics
1015 @code{GMTime} and @code{LTime}
1016 that overwrote their arguments.
1020 Support @command{gcc} version 2.8,
1021 and remove support for prior versions of @command{gcc}.
1023 @cindex -@w{}-driver option
1024 @cindex @command{g77} options, -@w{}-driver
1025 @cindex options, -@w{}-driver
1027 Remove support for the @option{--driver} option,
1028 as @command{g77} now does all the driving,
1029 just like @command{gcc}.
1031 @ifclear USERVISONLY
1033 @code{CASE DEFAULT} no longer crashes @command{g77}.
1036 @ifclear USERVISONLY
1038 Valid combinations of @code{EXTERNAL},
1039 passing that external as a dummy argument
1040 without explicitly giving it a type,
1041 and, in a subsequent program unit,
1042 referencing that external as
1043 an external function with a different type
1044 no longer crash @command{g77}.
1047 @ifclear USERVISONLY
1049 @command{g77} no longer installs the @file{f77} command
1050 and @file{f77.1} man page
1051 in the @file{/usr} or @file{/usr/local} hierarchy,
1052 even if the @file{f77-install-ok} file exists
1053 in the source or build directory.
1054 See the installation documentation for more information.
1057 @ifclear USERVISONLY
1059 @command{g77} no longer installs the @file{libf2c.a} library
1060 and @file{f2c.h} include file
1061 in the @file{/usr} or @file{/usr/local} hierarchy,
1062 even if the @file{f2c-install-ok} or @file{f2c-exists-ok} files exist
1063 in the source or build directory.
1064 See the installation documentation for more information.
1067 @ifclear USERVISONLY
1069 The @file{libf2c.a} library produced by @command{g77} has been
1070 renamed to @file{libg2c.a}.
1071 It is installed only in the @command{gcc} ``private''
1072 directory hierarchy, @file{gcc-lib}.
1073 This allows system administrators and users to choose which
1074 version of the @code{libf2c} library from @code{netlib} they
1075 wish to use on a case-by-case basis.
1076 See the installation documentation for more information.
1079 @ifclear USERVISONLY
1081 The @file{f2c.h} include (header) file produced by @command{g77}
1082 has been renamed to @file{g2c.h}.
1083 It is installed only in the @command{gcc} ``private''
1084 directory hierarchy, @file{gcc-lib}.
1085 This allows system administrators and users to choose which
1086 version of the include file from @code{netlib} they
1087 wish to use on a case-by-case basis.
1088 See the installation documentation for more information.
1092 The @command{g77} command now expects the run-time library
1093 to be named @code{libg2c.a} instead of @code{libf2c.a},
1094 to ensure that a version other than the one built and
1095 installed as part of the same @command{g77} version is picked up.
1097 @ifclear USERVISONLY
1099 The @option{-Wunused} option no longer issues a spurious
1100 warning about the ``master'' procedure generated by
1101 @command{g77} for procedures containing @code{ENTRY} statements.
1105 @command{g77}'s version of @code{libf2c} separates out
1106 the setting of global state
1107 (such as command-line arguments and signal handling)
1108 from @file{main.o} into distinct, new library
1111 This should make it easier to write portable applications
1112 that have their own (non-Fortran) @code{main()} routine
1113 properly set up the @code{libf2c} environment, even
1114 when @code{libf2c} (now @code{libg2c}) is a shared library.
1116 @ifclear USERVISONLY
1118 During the configuration and build process,
1119 @command{g77} creates subdirectories it needs only as it
1120 needs them, thus avoiding unnecessary creation of, for example,
1121 @file{stage1/f/runtime} when doing a non-bootstrap build.
1122 Other cleaning up of the configuration and build process
1123 has been performed as well.
1126 @ifclear USERVISONLY
1128 @code{install-info} now used to update the directory of
1129 Info documentation to contain an entry for @command{g77}
1130 (during installation).
1134 Some diagnostics have been changed from warnings to errors,
1135 to prevent inadvertent use of the resulting, probably buggy,
1137 These mostly include diagnostics about use of unsupported features
1138 in the @code{OPEN}, @code{INQUIRE}, @code{READ}, and
1139 @code{WRITE} statements,
1140 and about truncations of various sorts of constants.
1142 @ifclear USERVISONLY
1144 Improve documentation and indexing.
1147 @ifclear USERVISONLY
1149 Upgrade to @code{libf2c} as of 1998-04-20.
1151 This should fix a variety of problems, including
1152 those involving some uses of the @code{T} format
1153 specifier, and perhaps some build (porting) problems
1158 @c 1998-03-16: 0.5.22 released.
1159 @heading In 0.5.22 versus 0.5.21:
1161 @ifclear USERVISONLY
1163 Fix code generation for iterative @code{DO} loops that
1164 have one or more references to the iteration variable,
1165 or to aliases of it, in their control expressions.
1166 For example, @samp{DO 10 J=2,J} now is compiled correctly.
1169 @ifclear USERVISONLY
1171 @cindex stack, 387 coprocessor
1175 Fix a code-generation bug that afflicted
1176 Intel x86 targets when @option{-O2} was specified
1177 compiling, for example, an old version of
1178 the @code{DNRM2} routine.
1180 The x87 coprocessor stack was being
1181 mismanaged in cases involving assigned @code{GOTO}
1185 @ifclear USERVISONLY
1187 Fix @code{DTime} intrinsic so as not to truncate
1188 results to integer values (on some systems).
1192 Fix @code{Signal} intrinsic so it offers portable
1193 support for 64-bit systems (such as Digital Alphas
1196 @ifclear USERVISONLY
1198 Fix run-time crash involving @code{NAMELIST} on 64-bit
1199 machines such as Alphas.
1202 @ifclear USERVISONLY
1204 Fix @command{g77} version of @code{libf2c} so it no longer
1205 produces a spurious @samp{I/O recursion} diagnostic at run time
1206 when an I/O operation (such as @samp{READ *,I}) is interrupted
1207 in a manner that causes the program to be terminated
1208 via the @code{f_exit} routine (such as via @kbd{C-c}).
1211 @ifclear USERVISONLY
1213 Fix @command{g77} crash triggered by @code{CASE} statement with
1214 an omitted lower or upper bound.
1217 @ifclear USERVISONLY
1219 Fix @command{g77} crash compiling references to @code{CPU_Time}
1223 @ifclear USERVISONLY
1225 Fix @command{g77} crash
1226 (or apparently infinite run-time)
1227 when compiling certain complicated expressions
1228 involving @code{COMPLEX} arithmetic
1229 (especially multiplication).
1232 @ifclear USERVISONLY
1234 Fix @command{g77} crash on statements such as
1235 @samp{PRINT *, (REAL(Z(I)),I=1,2)}, where
1236 @samp{Z} is @code{DOUBLE COMPLEX}.
1239 @ifclear USERVISONLY
1241 Fix a @command{g++} crash.
1245 Support @samp{FORMAT(I<@var{expr}>)} when @var{expr} is a
1246 compile-time constant @code{INTEGER} expression.
1249 Fix @command{g77} @option{-g} option so procedures that
1250 use @code{ENTRY} can be stepped through, line by line,
1253 @ifclear USERVISONLY
1255 Fix a profiling-related bug in @command{gcc} back end for
1256 Intel x86 architecture.
1260 Allow any @code{REAL} argument to intrinsics
1261 @code{Second} and @code{CPU_Time}.
1264 Allow any numeric argument to intrinsics
1265 @code{Int2} and @code{Int8}.
1268 Use @code{tempnam}, if available, to open scratch files
1269 (as in @samp{OPEN(STATUS='SCRATCH')})
1270 so that the @code{TMPDIR} environment variable,
1271 if present, is used.
1274 Rename the @command{gcc} keyword @code{restrict} to
1275 @code{__restrict__}, to avoid rejecting valid, existing,
1277 Support for @code{restrict} is now more like support
1280 @ifclear USERVISONLY
1282 Fix @option{-fpedantic} to not reject procedure invocations
1283 such as @samp{I=J()} and @samp{CALL FOO()}.
1287 Fix @option{-fugly-comma} to affect invocations of
1288 only external procedures.
1289 Restore rejection of gratuitous trailing omitted
1290 arguments to intrinsics, as in @samp{I=MAX(3,4,,)}.
1293 Fix compiler so it accepts @option{-fgnu-intrinsics-*} and
1294 @option{-fbadu77-intrinsics-*} options.
1296 @ifclear USERVISONLY
1298 Improve diagnostic messages from @code{libf2c}
1299 so it is more likely that the printing of the
1300 active format string is limited to the string,
1301 with no trailing garbage being printed.
1303 (Unlike @command{f2c}, @command{g77} did not append
1304 a null byte to its compiled form of every
1305 format string specified via a @code{FORMAT} statement.
1306 However, @command{f2c} would exhibit the problem
1307 anyway for a statement like @samp{PRINT '(I)garbage', 1}
1308 by printing @samp{(I)garbage} as the format string.)
1311 @ifclear USERVISONLY
1313 Improve compilation of @code{FORMAT} expressions so that
1314 a null byte is appended to the last operand if it
1316 This provides a cleaner run-time diagnostic as provided
1317 by @code{libf2c} for statements like @samp{PRINT '(I1', 42}.
1320 @ifclear USERVISONLY
1322 Fix various crashes involving code with diagnosed errors.
1325 @ifclear USERVISONLY
1327 Fix cross-compilation bug when configuring @code{libf2c}.
1330 @ifclear USERVISONLY
1332 Improve diagnostics.
1335 @ifclear USERVISONLY
1337 Improve documentation and indexing.
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 1998-03-18: EGCS 1.0.2 released.
1350 @heading In @code{EGCS} 1.0.2 versus @code{EGCS} 1.0.1:
1352 @ifclear USERVISONLY
1354 Fix @command{g77} crash triggered by @code{CASE} statement with
1355 an omitted lower or upper bound.
1358 @ifclear USERVISONLY
1360 Fix @command{g77} crash on statements such as
1361 @samp{PRINT *, (REAL(Z(I)),I=1,2)}, where
1362 @samp{Z} is @code{DOUBLE COMPLEX}.
1365 @ifclear USERVISONLY
1367 @cindex support, ELF
1368 @cindex -fPIC option
1369 @cindex options, -fPIC
1371 Fix @option{-fPIC} (such as compiling for ELF targets)
1372 on the Intel x86 architecture target
1373 so invalid assembler code is no longer produced.
1376 @ifclear USERVISONLY
1378 Fix @option{-fpedantic} to not reject procedure invocations
1379 such as @samp{I=J()} and @samp{CALL FOO()}.
1382 @ifclear USERVISONLY
1384 Fix @option{-fugly-comma} to affect invocations of
1385 only external procedures.
1386 Restore rejection of gratuitous trailing omitted
1387 arguments to intrinsics, as in @samp{I=MAX(3,4,,)}.
1391 Fix compiler so it accepts @option{-fgnu-intrinsics-*} and
1392 @option{-fbadu77-intrinsics-*} options.
1395 @c 1998-01-06: EGCS 1.0.1 released.
1396 @heading In @code{EGCS} 1.0.1 versus @code{EGCS} 1.0:
1397 @ifclear USERVISONLY
1400 Fix run-time crash involving @code{NAMELIST} on 64-bit
1401 machines such as Alphas.
1405 @c 1997-12-03: EGCS 1.0 released.
1406 @heading In @code{EGCS} 1.0 versus @command{g77} 0.5.21:
1409 Version 1.0 of @code{egcs}
1410 contains several regressions against
1411 version 0.5.21 of @command{g77},
1413 ``vanilla'' @command{gcc} back end instead of patching
1414 it to fix a few bugs and improve performance in a
1417 Features that have been dropped from this version
1418 of @command{g77} due to their being implemented
1419 via @command{g77}-specific patches to the @command{gcc}
1420 back end in previous releases include:
1424 Support for the C-language @code{restrict} keyword.
1427 Support for the @option{-W} option warning about
1428 integer division by zero.
1431 The Intel x86-specific option @option{-malign-double}
1432 applying to stack-allocated data
1433 as well as statically-allocate data.
1436 @ifclear USERVISONLY
1437 Note that the @file{gcc/f/gbe/} subdirectory has been removed
1438 from this distribution as a result of @command{g77}
1439 being fully integrated with
1440 the @code{egcs} variant of the @command{gcc} back end.
1443 @ifclear USERVISONLY
1445 Fix code generation for iterative @code{DO} loops that
1446 have one or more references to the iteration variable,
1447 or to aliases of it, in their control expressions.
1448 For example, @samp{DO 10 J=2,J} now is compiled correctly.
1451 @ifclear USERVISONLY
1453 Fix @code{DTime} intrinsic so as not to truncate
1454 results to integer values (on some systems).
1457 @ifclear USERVISONLY
1459 @c Toon Moene discovered these.
1460 Some Fortran code, miscompiled
1461 by @command{g77} built on @command{gcc} version 2.8.1
1462 on m68k-next-nextstep3 configurations
1463 when using the @option{-O2} option,
1464 is now compiled correctly.
1465 It is believed that a C function known to miscompile
1466 on that configuration
1467 when using the @samp{-O2 -funroll-loops} options
1468 also is now compiled correctly.
1471 @ifclear USERVISONLY
1473 Remove support for non-@code{egcs} versions of @command{gcc}.
1476 @cindex -@w{}-driver option
1477 @cindex @command{g77} options, -@w{}-driver
1478 @cindex options, -@w{}-driver
1480 Remove support for the @option{--driver} option,
1481 as @command{g77} now does all the driving,
1482 just like @command{gcc}.
1485 Allow any numeric argument to intrinsics
1486 @code{Int2} and @code{Int8}.
1488 @ifclear USERVISONLY
1490 Improve diagnostic messages from @code{libf2c}
1491 so it is more likely that the printing of the
1492 active format string is limited to the string,
1493 with no trailing garbage being printed.
1495 (Unlike @command{f2c}, @command{g77} did not append
1496 a null byte to its compiled form of every
1497 format string specified via a @code{FORMAT} statement.
1498 However, @code{f2c} would exhibit the problem
1499 anyway for a statement like @samp{PRINT '(I)garbage', 1}
1500 by printing @samp{(I)garbage} as the format string.)
1503 @ifclear USERVISONLY
1505 Upgrade to @code{libf2c} as of 1997-09-23.
1506 This fixes a formatted-I/O bug that afflicted
1507 64-bit systems with 32-bit integers
1508 (such as Digital Alpha running GNU/Linux).
1512 @c 1997-09-09: 0.5.21 released.
1515 @ifclear USERVISONLY
1517 Fix a code-generation bug introduced by 0.5.20
1518 caused by loop unrolling (by specifying
1519 @option{-funroll-loops} or similar).
1520 This bug afflicted all code compiled by
1521 version 2.7.2.2.f.2 of @command{gcc} (C, C++,
1522 Fortran, and so on).
1525 @ifclear USERVISONLY
1527 Fix a code-generation bug manifested when
1528 combining local @code{EQUIVALENCE} with a
1529 @code{DATA} statement that follows
1530 the first executable statement (or is
1531 treated as an executable-context statement
1532 as a result of using the @option{-fpedantic}
1536 @ifclear USERVISONLY
1538 Fix a compiler crash that occurred when an
1539 integer division by a constant zero is detected.
1540 Instead, when the @option{-W} option is specified,
1541 the @command{gcc} back end issues a warning about such a case.
1542 This bug afflicted all code compiled by
1543 version 2.7.2.2.f.2 of @command{gcc} (C, C++,
1544 Fortran, and so on).
1548 When the @option{-W} option is specified, @command{gcc}, @command{g77},
1549 and other GNU compilers that incorporate the @command{gcc}
1550 back end as modified by @command{g77}, issue
1551 a warning about integer division by constant zero.
1554 @ifclear USERVISONLY
1556 Fix a compiler crash that occurred in some cases
1557 of procedure inlining.
1558 (Such cases became more frequent in 0.5.20.)
1561 @ifclear USERVISONLY
1563 Fix a compiler crash resulting from using @code{DATA}
1564 or similar to initialize a @code{COMPLEX} variable or
1568 @ifclear USERVISONLY
1570 Fix compiler crashes involving use of @code{AND}, @code{OR},
1571 or @code{XOR} intrinsics.
1574 @ifclear USERVISONLY
1576 Fix compiler bug triggered when using a @code{COMMON}
1577 or @code{EQUIVALENCE} variable
1578 as the target of an @code{ASSIGN}
1579 or assigned-@code{GOTO} statement.
1582 @ifclear USERVISONLY
1584 Fix compiler crashes due to using the name of a some
1585 non-standard intrinsics (such as @code{FTell} or
1586 @code{FPutC}) as such and as the name of a procedure
1588 Such dual use of a name in a program is allowed by
1592 @c @command{g77}'s version of @code{libf2c} has been modified
1593 @c so that the external names of library's procedures do not
1594 @c conflict with names used for Fortran procedures compiled
1595 @c by @command{g77}.
1596 @c An additional layer of jacket procedures has been added
1597 @c to @code{libf2c} to map the old names to the new names,
1598 @c for automatic use by programs that interface to the
1599 @c library procedures via the external-procedure mechanism.
1601 @c For example, the intrinsic @code{FPUTC} previously was
1602 @c implemented by @command{g77} as a call to the @code{libf2c}
1603 @c routine @code{fputc_}.
1604 @c This would conflict with a Fortran procedure named @code{FPUTC}
1605 @c (using default compiler options), and this conflict
1606 @c would cause a crash under certain circumstances.
1608 @c Now, the intrinsic @code{FPUTC} calls @code{G77_fputc_0},
1609 @c which does not conflict with the @code{fputc_} external
1610 @c that implements a Fortran procedure named @code{FPUTC}.
1612 @c Programs that refer to @code{FPUTC} as an external procedure
1613 @c without supplying their own implementation will link to
1614 @c the new @code{libf2c} routine @code{fputc_}, which is
1615 @c simply a jacket routine that calls @code{G77_fputc_0}.
1617 @ifclear USERVISONLY
1619 Place automatic arrays on the stack, even if
1620 @code{SAVE} or the @option{-fno-automatic} option
1622 This avoids a compiler crash in some cases.
1625 @ifclear USERVISONLY
1627 The @option{-malign-double} option now reliably aligns
1628 @code{DOUBLE PRECISION} optimally on Pentium and
1629 Pentium Pro architectures (586 and 686 in @command{gcc}).
1633 New option @option{-Wno-globals} disables warnings
1634 about ``suspicious'' use of a name both as a global
1635 name and as the implicit name of an intrinsic, and
1636 warnings about disagreements over the number or natures of
1637 arguments passed to global procedures, or the
1638 natures of the procedures themselves.
1640 The default is to issue such warnings, which are
1641 new as of this version of @command{g77}.
1644 New option @option{-fno-globals} disables diagnostics
1645 about potentially fatal disagreements
1646 analysis problems, such as disagreements over the
1647 number or natures of arguments passed to global
1648 procedures, or the natures of those procedures themselves.
1650 The default is to issue such diagnostics and flag
1651 the compilation as unsuccessful.
1652 With this option, the diagnostics are issued as
1653 warnings, or, if @option{-Wno-globals} is specified,
1654 are not issued at all.
1656 This option also disables inlining of global procedures,
1657 to avoid compiler crashes resulting from coding errors
1658 that these diagnostics normally would identify.
1660 @ifclear USERVISONLY
1662 Diagnose cases where a reference to a procedure
1663 disagrees with the type of that procedure, or
1664 where disagreements about the number or nature
1666 This avoids a compiler crash.
1669 @ifclear USERVISONLY
1671 Fix parsing bug whereby @command{g77} rejected a
1672 second initialization specification immediately
1673 following the first's closing @samp{/} without
1674 an intervening comma in a @code{DATA} statement,
1675 and the second specification was an implied-DO list.
1678 @ifclear USERVISONLY
1680 Improve performance of the @command{gcc} back end so
1681 certain complicated expressions involving @code{COMPLEX}
1682 arithmetic (especially multiplication) don't appear to
1683 take forever to compile.
1686 @ifclear USERVISONLY
1688 Fix a couple of profiling-related bugs in @command{gcc}
1692 @ifclear USERVISONLY
1694 Integrate GNU Ada's (GNAT's) changes to the back end,
1695 which consist almost entirely of bug fixes.
1696 These fixes are circa version 3.10p of GNAT.
1699 @ifclear USERVISONLY
1701 Include some other @command{gcc} fixes that seem useful in
1702 @command{g77}'s version of @command{gcc}.
1703 (See @file{gcc/ChangeLog} for details---compare it
1704 to that file in the vanilla @code{gcc-2.7.2.3.tar.gz}
1709 Fix @code{libU77} routines that accept file and other names
1710 to strip trailing blanks from them, for consistency
1711 with other implementations.
1712 Blanks may be forcibly appended to such names by
1713 appending a single null character (@samp{CHAR(0)})
1714 to the significant trailing blanks.
1717 Fix @code{CHMOD} intrinsic to work with file names
1718 that have embedded blanks, commas, and so on.
1721 Fix @code{SIGNAL} intrinsic so it accepts an
1722 optional third @code{Status} argument.
1724 @ifclear USERVISONLY
1726 Fix @code{IDATE()} intrinsic subroutine (VXT form)
1727 so it accepts arguments in the correct order.
1728 Documentation fixed accordingly, and for
1729 @code{GMTIME()} and @code{LTIME()} as well.
1733 Make many changes to @code{libU77} intrinsics to
1734 support existing code more directly.
1736 Such changes include allowing both subroutine and
1737 function forms of many routines, changing @code{MCLOCK()}
1738 and @code{TIME()} to return @code{INTEGER(KIND=1)} values,
1739 introducing @code{MCLOCK8()} and @code{TIME8()} to
1740 return @code{INTEGER(KIND=2)} values,
1741 and placing functions that are intended to perform
1742 side effects in a new intrinsic group, @code{badu77}.
1744 @ifclear USERVISONLY
1746 Improve @code{libU77} so it is more portable.
1750 Add options @option{-fbadu77-intrinsics-delete},
1751 @option{-fbadu77-intrinsics-hide}, and so on.
1753 @ifclear USERVISONLY
1755 Fix crashes involving diagnosed or invalid code.
1758 @ifclear USERVISONLY
1760 @command{g77} and @command{gcc} now do a somewhat better
1761 job detecting and diagnosing arrays that are too
1762 large to handle before these cause diagnostics
1763 during the assembler or linker phase, a compiler
1764 crash, or generation of incorrect code.
1767 @ifclear USERVISONLY
1769 Make some fixes to alias analysis code.
1772 @ifclear USERVISONLY
1774 Add support for @code{restrict} keyword in @command{gcc}
1778 @ifclear USERVISONLY
1780 Support @command{gcc} version 2.7.2.3
1781 (modified by @command{g77} into version 2.7.2.3.f.1),
1783 support for prior versions of @command{gcc}.
1786 @ifclear USERVISONLY
1788 Incorporate GNAT's patches to the @command{gcc} back
1789 end into @command{g77}'s, so GNAT users do not need
1790 to apply GNAT's patches to build both GNAT and @command{g77}
1791 from the same source tree.
1794 @ifclear USERVISONLY
1796 Modify @command{make} rules and related code so that
1797 generation of Info documentation doesn't require
1798 compilation using @command{gcc}.
1799 Now, any ANSI C compiler should be adequate to
1800 produce the @command{g77} documentation (in particular,
1801 the tables of intrinsics) from scratch.
1805 Add @code{INT2} and @code{INT8} intrinsics.
1808 Add @code{CPU_TIME} intrinsic.
1811 Add @code{ALARM} intrinsic.
1814 @code{CTIME} intrinsic now accepts any @code{INTEGER}
1815 argument, not just @code{INTEGER(KIND=2)}.
1817 @ifclear USERVISONLY
1819 Warn when explicit type declaration disagrees with
1820 the type of an intrinsic invocation.
1823 @ifclear USERVISONLY
1825 Support @samp{*f771} entry in @command{gcc} @file{specs} file.
1828 @ifclear USERVISONLY
1830 Fix typo in @command{make} rule @command{g77-cross}, used only for
1834 @ifclear USERVISONLY
1836 Fix @code{libf2c} build procedure to re-archive library
1837 if previous attempt to archive was interrupted.
1840 @ifclear USERVISONLY
1842 Change @command{gcc} to unroll loops only during the last
1843 invocation (of as many as two invocations) of loop
1847 @ifclear USERVISONLY
1849 Improve handling of @option{-fno-f2c} so that code that
1850 attempts to pass an intrinsic as an actual argument,
1851 such as @samp{CALL FOO(ABS)}, is rejected due to the fact
1852 that the run-time-library routine is, effectively,
1853 compiled with @option{-ff2c} in effect.
1856 @ifclear USERVISONLY
1858 Fix @command{g77} driver to recognize @option{-fsyntax-only}
1859 as an option that inhibits linking, just like @option{-c} or
1860 @option{-S}, and to recognize and properly handle the
1861 @option{-nostdlib}, @option{-M}, @option{-MM}, @option{-nodefaultlibs},
1862 and @option{-Xlinker} options.
1865 @ifclear USERVISONLY
1867 Upgrade to @code{libf2c} as of 1997-08-16.
1870 @ifclear USERVISONLY
1872 Modify @code{libf2c} to consistently and clearly diagnose
1873 recursive I/O (at run time).
1877 @command{g77} driver now prints version information (such as produced
1878 by @kbd{g77 -v}) to @code{stderr} instead of @code{stdout}.
1881 The @samp{.r} suffix now designates a Ratfor source file,
1882 to be preprocessed via the @command{ratfor} command, available
1885 @ifclear USERVISONLY
1887 Fix some aspects of how @command{gcc} determines what kind of
1888 system is being configured and what kinds are supported.
1889 For example, GNU Linux/Alpha ELF systems now are directly
1893 @ifclear USERVISONLY
1895 Improve diagnostics.
1898 @ifclear USERVISONLY
1900 Improve documentation and indexing.
1903 @ifclear USERVISONLY
1905 Include all pertinent files for @code{libf2c} that come
1906 from @code{netlib.bell-labs.com}; give any such files
1907 that aren't quite accurate in @command{g77}'s version of
1908 @code{libf2c} the suffix @samp{.netlib}.
1911 @ifclear USERVISONLY
1913 Reserve @code{INTEGER(KIND=0)} for future use.
1917 @c 1997-02-28: 0.5.20 released.
1921 The @option{-fno-typeless-boz} option is now the default.
1923 This option specifies that non-decimal-radix
1924 constants using the prefixed-radix form (such as @samp{Z'1234'})
1925 are to be interpreted as @code{INTEGER(KIND=1)} constants.
1926 Specify @option{-ftypeless-boz} to cause such
1927 constants to be interpreted as typeless.
1929 (Version 0.5.19 introduced @option{-fno-typeless-boz} and
1933 @xref{Fortran Dialect Options,,Options Controlling Fortran Dialect},
1934 for information on the @option{-ftypeless-boz} option.
1938 Options @option{-ff90-intrinsics-enable} and
1939 @option{-fvxt-intrinsics-enable} now are the
1942 Some programs might use names that clash with
1943 intrinsic names defined (and now enabled) by these
1944 options or by the new @code{libU77} intrinsics.
1945 Users of such programs might need to compile them
1946 differently (using, for example, @option{-ff90-intrinsics-disable})
1947 or, better yet, insert appropriate @code{EXTERNAL}
1948 statements specifying that these names are not intended
1949 to be names of intrinsics.
1952 The @code{ALWAYS_FLUSH} macro is no longer defined when
1953 building @code{libf2c}, which should result in improved
1954 I/O performance, especially over NFS.
1956 @emph{Note:} If you have code that depends on the behavior
1957 of @code{libf2c} when built with @code{ALWAYS_FLUSH} defined,
1958 you will have to modify @code{libf2c} accordingly before
1959 building it from this and future versions of @command{g77}.
1962 @xref{Output Assumed To Flush}, for more information.
1966 Dave Love's implementation of @code{libU77} has been
1967 added to the version of @code{libf2c} distributed with
1968 and built as part of @command{g77}.
1969 @command{g77} now knows about the routines in this library
1973 New option @option{-fvxt} specifies that the
1974 source file is written in VXT Fortran, instead of GNU Fortran.
1977 @xref{VXT Fortran}, for more information on the constructs
1978 recognized when the @option{-fvxt} option is specified.
1982 The @option{-fvxt-not-f90} option has been deleted,
1983 along with its inverse, @option{-ff90-not-vxt}.
1985 If you used one of these deleted options, you should
1986 re-read the pertinent documentation to determine which
1987 options, if any, are appropriate for compiling your
1988 code with this version of @command{g77}.
1991 @xref{Other Dialects}, for more information.
1995 The @option{-fugly} option now issues a warning, as it
1996 likely will be removed in a future version.
1998 (Enabling all the @option{-fugly-*} options is unlikely
1999 to be feasible, or sensible, in the future,
2000 so users should learn to specify only those
2001 @option{-fugly-*} options they really need for a
2002 particular source file.)
2005 The @option{-fugly-assumed} option, introduced in
2006 version 0.5.19, has been changed to
2007 better accommodate old and new code.
2010 @xref{Ugly Assumed-Size Arrays}, for more information.
2013 @ifclear USERVISONLY
2015 Make a number of fixes to the @command{g77} front end and
2016 the @command{gcc} back end to better support Alpha (AXP)
2018 This includes providing at least one bug-fix to the
2019 @command{gcc} back end for Alphas.
2023 Related to supporting Alpha (AXP) machines, the @code{LOC()}
2024 intrinsic and @code{%LOC()} construct now return
2025 values of @code{INTEGER(KIND=0)} type,
2026 as defined by the GNU Fortran language.
2028 This type is wide enough
2029 (holds the same number of bits)
2030 as the character-pointer type on the machine.
2032 On most machines, this won't make a difference,
2033 whereas, on Alphas and other systems with 64-bit pointers,
2034 the @code{INTEGER(KIND=0)} type is equivalent to @code{INTEGER(KIND=2)}
2035 (often referred to as @code{INTEGER*8})
2036 instead of the more common @code{INTEGER(KIND=1)}
2037 (often referred to as @code{INTEGER*4}).
2040 Emulate @code{COMPLEX} arithmetic in the @command{g77} front
2041 end, to avoid bugs in @code{complex} support in the
2042 @command{gcc} back end.
2043 New option @option{-fno-emulate-complex}
2044 causes @command{g77} to revert the 0.5.19 behavior.
2046 @ifclear USERVISONLY
2048 Fix bug whereby @samp{REAL A(1)}, for example, caused
2049 a compiler crash if @option{-fugly-assumed} was in effect
2050 and @var{A} was a local (automatic) array.
2051 That case is no longer affected by the new
2052 handling of @option{-fugly-assumed}.
2055 @ifclear USERVISONLY
2057 Fix @command{g77} command driver so that @samp{g77 -o foo.f}
2058 no longer deletes @file{foo.f} before issuing other
2059 diagnostics, and so the @option{-x} option is properly
2063 @ifclear USERVISONLY
2065 Enable inlining of subroutines and functions by the @command{gcc}
2067 This works as it does for @command{gcc} itself---program units
2068 may be inlined for invocations that follow them in the same
2069 program unit, as long as the appropriate compile-time
2070 options are specified.
2074 Dummy arguments are no longer assumed to potentially alias
2076 other dummy arguments or @code{COMMON} areas when any of
2077 these are defined (assigned to) by Fortran code.
2079 This can result in faster and/or smaller programs when
2080 compiling with optimization enabled, though on some
2081 systems this effect is observed only when @option{-fforce-addr}
2084 New options @option{-falias-check}, @option{-fargument-alias},
2085 @option{-fargument-noalias},
2086 and @option{-fno-argument-noalias-global} control the
2087 way @command{g77} handles potential aliasing.
2090 @xref{Aliasing Assumed To Work}, for detailed information on why the
2091 new defaults might result in some programs no longer working the way they
2092 did when compiled by previous versions of @command{g77}.
2095 @ifclear USERVISONLY
2097 The @code{CONJG()} and @code{DCONJG()} intrinsics now
2098 are compiled in-line.
2101 @ifclear USERVISONLY
2103 The bug-fix for 0.5.19.1 has been re-done.
2104 The @command{g77} compiler has been changed back to
2105 assume @code{libf2c} has no aliasing problems in
2106 its implementations of the @code{COMPLEX} (and
2107 @code{DOUBLE COMPLEX}) intrinsics.
2108 The @code{libf2c} has been changed to have no such
2111 As a result, 0.5.20 is expected to offer improved performance
2112 over 0.5.19.1, perhaps as good as 0.5.19 in most
2113 or all cases, due to this change alone.
2115 @emph{Note:} This change requires version 0.5.20 of
2116 @code{libf2c}, at least, when linking code produced
2117 by any versions of @command{g77} other than 0.5.19.1.
2118 Use @samp{g77 -v} to determine the version numbers
2119 of the @code{libF77}, @code{libI77}, and @code{libU77}
2120 components of the @code{libf2c} library.
2121 (If these version numbers are not printed---in
2122 particular, if the linker complains about unresolved
2123 references to names like @samp{g77__fvers__}---that
2124 strongly suggests your installation has an obsolete
2125 version of @code{libf2c}.)
2129 New option @option{-fugly-assign} specifies that the
2130 same memory locations are to be used to hold the
2131 values assigned by both statements @samp{I = 3} and
2132 @samp{ASSIGN 10 TO I}, for example.
2133 (Normally, @command{g77} uses a separate memory location
2134 to hold assigned statement labels.)
2137 @xref{Ugly Assigned Labels}, for more information.
2141 @code{FORMAT} and @code{ENTRY} statements now are allowed to
2142 precede @code{IMPLICIT NONE} statements.
2144 @ifclear USERVISONLY
2146 Produce diagnostic for unsupported @code{SELECT CASE} on
2147 @code{CHARACTER} type, instead of crashing, at compile time.
2150 @ifclear USERVISONLY
2152 Fix crashes involving diagnosed or invalid code.
2155 @ifclear USERVISONLY
2157 Change approach to building @code{libf2c} archive
2158 (@file{libf2c.a}) so that members are added to it
2159 only when truly necessary, so the user that installs
2160 an already-built @command{g77} doesn't need to have write
2161 access to the build tree (whereas the user doing the
2162 build might not have access to install new software
2166 @ifclear USERVISONLY
2168 Support @command{gcc} version 2.7.2.2
2169 (modified by @command{g77} into version 2.7.2.2.f.2),
2171 support for prior versions of @command{gcc}.
2174 @ifclear USERVISONLY
2176 Upgrade to @code{libf2c} as of 1997-02-08, and
2177 fix up some of the build procedures.
2180 @ifclear USERVISONLY
2182 Improve general build procedures for @command{g77},
2183 fixing minor bugs (such as deletion of any file
2184 named @file{f771} in the parent directory of @code{gcc/}).
2188 Enable full support of @code{INTEGER(KIND=2)}
2189 (often referred to as @code{INTEGER*8})
2191 @code{libf2c} and @file{f2c.h} so that @command{f2c} users
2192 may make full use of its features via the @command{g77}
2193 version of @file{f2c.h} and the @code{INTEGER(KIND=2)}
2194 support routines in the @command{g77} version of @code{libf2c}.
2197 Improve @command{g77} driver and @code{libf2c} so that @samp{g77 -v}
2198 yields version information on the library.
2201 The @code{SNGL} and @code{FLOAT} intrinsics now are
2202 specific intrinsics, instead of synonyms for the
2203 generic intrinsic @code{REAL}.
2206 New intrinsics have been added.
2207 These are @code{REALPART}, @code{IMAGPART},
2209 @code{LONG}, and @code{SHORT}.
2212 A new group of intrinsics, @code{gnu}, has been added
2213 to contain the new @code{REALPART}, @code{IMAGPART},
2214 and @code{COMPLEX} intrinsics.
2215 An old group, @code{dcp}, has been removed.
2218 Complain about industry-wide ambiguous references
2219 @samp{REAL(@var{expr})} and @samp{AIMAG(@var{expr})},
2220 where @var{expr} is @code{DOUBLE COMPLEX} (or any
2221 complex type other than @code{COMPLEX}), unless
2222 @option{-ff90} option specifies Fortran 90 interpretation
2223 or new @option{-fugly-complex} option, in conjunction with
2224 @option{-fnot-f90}, specifies @command{f2c} interpretation.
2226 @ifclear USERVISONLY
2228 Make improvements to diagnostics.
2231 @ifclear USERVISONLY
2233 Speed up compiler a bit.
2236 @ifclear USERVISONLY
2238 Improvements to documentation and indexing, including
2239 a new chapter containing information on one, later
2240 more, diagnostics that users are directed to pull
2241 up automatically via a message in the diagnostic itself.
2243 (Hence the menu item @code{M} for the node
2244 @code{Diagnostics} in the top-level menu of
2245 the Info documentation.)
2249 @ifclear DOC-OLDNEWS
2250 @heading In previous versions:
2252 Information on previous versions is archived
2253 in @file{@value{path-g77}/news.texi}
2254 following the test of the @code{DOC-OLDNEWS} macro.
2258 @c 1997-02-01: 0.5.19.1 released.
2259 @heading In 0.5.19.1:
2262 Code-generation bugs afflicting operations on complex
2263 data have been fixed.
2265 These bugs occurred when assigning the result of an
2266 operation to a complex variable (or array element)
2267 that also served as an input to that operation.
2269 The operations affected by this bug were: @code{CONJG()},
2270 @code{DCONJG()}, @code{CCOS()}, @code{CDCOS()},
2271 @code{CLOG()}, @code{CDLOG()}, @code{CSIN()}, @code{CDSIN()},
2272 @code{CSQRT()}, @code{CDSQRT()}, complex division, and
2273 raising a @code{DOUBLE COMPLEX} operand to an @code{INTEGER}
2275 (The related generic and @samp{Z}-prefixed intrinsics,
2276 such as @code{ZSIN()}, also were affected.)
2278 For example, @samp{C = CSQRT(C)}, @samp{Z = Z/C}, and @samp{Z = Z**I}
2279 (where @samp{C} is @code{COMPLEX} and @samp{Z} is
2280 @code{DOUBLE COMPLEX}) have been fixed.
2283 @c 1996-12-07: 0.5.19 released.
2287 Fix @code{FORMAT} statement parsing so negative values for
2288 specifiers such as @code{P} (e.g. @samp{FORMAT(-1PF8.1)})
2289 are correctly processed as negative.
2292 Fix @code{SIGNAL} intrinsic so it once again accepts a
2293 procedure as its second argument.
2296 A temporary kludge option provides bare-bones information on
2297 @code{COMMON} and @code{EQUIVALENCE} members at debug time.
2300 New @option{-fonetrip} option specifies FORTRAN-66-style
2301 one-trip @code{DO} loops.
2304 New @option{-fno-silent} option causes names of program units
2305 to be printed as they are compiled, in a fashion similar to
2306 UNIX @command{f77} and @command{f2c}.
2309 New @option{-fugly-assumed} option specifies that arrays
2310 dimensioned via @samp{DIMENSION X(1)}, for example, are to be
2311 treated as assumed-size.
2314 New @option{-fno-typeless-boz} option specifies that non-decimal-radix
2315 constants using the prefixed-radix form (such as @samp{Z'1234'})
2316 are to be interpreted as @code{INTEGER} constants.
2319 New @option{-ff66} option is a ``shorthand'' option that specifies
2320 behaviors considered appropriate for FORTRAN 66 programs.
2323 New @option{-ff77} option is a ``shorthand'' option that specifies
2324 behaviors considered appropriate for UNIX @command{f77} programs.
2327 New @option{-fugly-comma} and @option{-fugly-logint} options provided
2328 to perform some of what @option{-fugly} used to do.
2329 @option{-fugly} and @option{-fno-ugly} are now ``shorthand'' options,
2330 in that they do nothing more than enable (or disable) other
2331 @option{-fugly-*} options.
2334 Fix parsing of assignment statements involving targets that
2335 are substrings of elements of @code{CHARACTER} arrays having
2336 names such as @samp{READ}, @samp{WRITE}, @samp{GOTO}, and
2337 @samp{REALFUNCTIONFOO}.
2340 Fix crashes involving diagnosed code.
2343 Fix handling of local @code{EQUIVALENCE} areas so certain cases
2344 of valid Fortran programs are not misdiagnosed as improperly
2345 extending the area backwards.
2348 Support @command{gcc} version 2.7.2.1.
2351 Upgrade to @code{libf2c} as of 1996-09-26, and
2352 fix up some of the build procedures.
2355 Change code generation for list-directed I/O so it allows
2356 for new versions of @code{libf2c} that might return nonzero
2357 status codes for some operations previously assumed to always
2360 This change not only affects how @code{IOSTAT=} variables
2361 are set by list-directed I/O, it also affects whether
2362 @code{END=} and @code{ERR=} labels are reached by these
2366 Add intrinsic support for new @code{FTELL} and @code{FSEEK}
2367 procedures in @code{libf2c}.
2370 Modify @code{fseek_()} in @code{libf2c} to be more portable
2371 (though, in practice, there might be no systems where this
2372 matters) and to catch invalid @code{whence} arguments.
2375 Some useless warnings from the @option{-Wunused} option have
2379 Fix a problem building the @file{f771} executable
2380 on AIX systems by linking with the @option{-bbigtoc} option.
2383 Abort configuration if @command{gcc} has not been patched
2384 using the patch file provided in the @file{gcc/f/gbe/}
2388 Add options @option{--help} and @option{--version} to the
2389 @command{g77} command, to conform to GNU coding guidelines.
2390 Also add printing of @command{g77} version number when
2391 the @option{--verbose} (@option{-v}) option is used.
2394 Change internally generated name for local @code{EQUIVALENCE}
2395 areas to one based on the alphabetically sorted first name
2396 in the list of names for entities placed at the beginning
2400 Improvements to documentation and indexing.
2403 @c 1996-04-01: 0.5.18 released.
2407 Add some rudimentary support for @code{INTEGER*1},
2408 @code{INTEGER*2}, @code{INTEGER*8},
2409 and their @code{LOGICAL} equivalents.
2410 (This support works on most, maybe all, @command{gcc} targets.)
2412 Thanks to Scott Snyder (@email{snyder@@d0sgif.fnal.gov})
2413 for providing the patch for this!
2415 Among the missing elements from the support for these
2416 features are full intrinsic support and constants.
2419 Add some rudimentary support for the @code{BYTE} and
2420 @code{WORD} type-declaration statements.
2421 @code{BYTE} corresponds to @code{INTEGER*1},
2422 while @code{WORD} corresponds to @code{INTEGER*2}.
2424 Thanks to Scott Snyder (@email{snyder@@d0sgif.fnal.gov})
2425 for providing the patch for this!
2428 The compiler code handling intrinsics has been largely
2429 rewritten to accommodate the new types.
2430 No new intrinsics or arguments for existing
2431 intrinsics have been added, so there is, at this
2432 point, no intrinsic to convert to @code{INTEGER*8},
2436 Support automatic arrays in procedures.
2439 Reduce space/time requirements for handling large
2440 @emph{sparsely} initialized aggregate arrays.
2441 This improvement applies to only a subset of
2442 the general problem to be addressed in 0.6.
2445 Treat initial values of zero as if they weren't
2446 specified (in DATA and type-declaration statements).
2447 The initial values will be set to zero anyway, but the amount
2448 of compile time processing them will be reduced,
2449 in some cases significantly (though, again, this
2450 is only a subset of the general problem to be
2453 A new option, @option{-fzeros}, is introduced to
2454 enable the traditional treatment of zeros as any
2458 With @option{-ff90} in force, @command{g77} incorrectly
2459 interpreted @samp{REAL(Z)} as returning a @code{REAL}
2460 result, instead of as a @code{DOUBLE PRECISION}
2462 (Here, @samp{Z} is @code{DOUBLE COMPLEX}.)
2464 With @option{-fno-f90} in force, the interpretation remains
2465 unchanged, since this appears to be how at least some
2466 F77 code using the @code{DOUBLE COMPLEX} extension expected
2469 Essentially, @samp{REAL(Z)} in F90 is the same as
2470 @samp{DBLE(Z)}, while in extended F77, it appears to
2471 be the same as @samp{REAL(REAL(Z))}.
2474 An expression involving exponentiation, where both operands
2475 were type @code{INTEGER} and the right-hand operand
2476 was negative, was erroneously evaluated.
2479 Fix bugs involving @code{DATA} implied-@code{DO} constructs
2480 (these involved an errant diagnostic and a crash, both on good
2481 code, one involving subsequent statement-function definition).
2484 Close @code{INCLUDE} files after processing them, so compiling source
2485 files with lots of @code{INCLUDE} statements does not result in
2486 being unable to open @code{INCLUDE} files after all the available
2487 file descriptors are used up.
2490 Speed up compiling, especially of larger programs, and perhaps
2491 slightly reduce memory utilization while compiling (this is
2492 @emph{not} the improvement planned for 0.6 involving large aggregate
2493 areas)---these improvements result from simply turning
2494 off some low-level code to do self-checking that hasn't been
2495 triggered in a long time.
2498 Introduce three new options that
2499 implement optimizations in the @command{gcc} back end (GBE).
2500 These options are @option{-fmove-all-movables}, @option{-freduce-all-givs},
2501 and @option{-frerun-loop-opt}, which are enabled, by default,
2502 for Fortran compilations.
2503 These optimizations are intended to help toon Fortran programs.
2506 Patch the GBE to do a better job optimizing certain
2507 kinds of references to array elements.
2510 Due to patches to the GBE, the version number of @command{gcc}
2511 also is patched to make it easier to manage installations,
2512 especially useful if it turns out a @command{g77} change to the
2515 The @command{g77}-modified version number is the @command{gcc}
2516 version number with the string @samp{.f.@var{n}} appended,
2517 where @samp{f} identifies the version as enhanced for
2518 Fortran, and @var{n} is @samp{1} for the first Fortran
2519 patch for that version of @command{gcc}, @samp{2} for the
2522 So, this introduces version 2.7.2.f.1 of @command{gcc}.
2525 Make several improvements and fixes to diagnostics, including
2526 the removal of two that were inappropriate or inadequate.
2529 Warning about two successive arithmetic operators, produced
2530 by @option{-Wsurprising}, now produced @emph{only} when both
2531 operators are, indeed, arithmetic (not relational/boolean).
2534 @option{-Wsurprising} now warns about the remaining cases
2535 of using non-integral variables for implied-@code{DO}
2536 loops, instead of these being rejected unless @option{-fpedantic}
2537 or @option{-fugly} specified.
2540 Allow @code{SAVE} of a local variable or array, even after
2541 it has been given an initial value via @code{DATA}, for example.
2544 Introduce an Info version of @command{g77} documentation, which
2545 supersedes @file{gcc/f/CREDITS}, @file{gcc/f/DOC}, and
2546 @file{gcc/f/PROJECTS}.
2547 These files will be removed in a future release.
2548 The files @file{gcc/f/BUGS}, @file{gcc/f/INSTALL}, and
2549 @file{gcc/f/NEWS} now are automatically built from
2550 the texinfo source when distributions are made.
2552 This effort was inspired by a first pass at translating
2553 @file{g77-0.5.16/f/DOC} that was contributed to Craig by
2554 David Ronis (@email{ronis@@onsager.chem.mcgill.ca}).
2557 New @option{-fno-second-underscore} option to specify
2558 that, when @option{-funderscoring} is in effect, a second
2559 underscore is not to be appended to Fortran names already
2560 containing an underscore.
2563 Change the way iterative @code{DO} loops work to follow
2565 In particular, calculation of the iteration count is
2566 still done by converting the start, end, and increment
2567 parameters to the type of the @code{DO} variable, but
2568 the result of the calculation is always converted to
2569 the default @code{INTEGER} type.
2571 (This should have no effect on existing code compiled
2572 by @command{g77}, but code written to assume that use
2573 of a @emph{wider} type for the @code{DO} variable
2574 will result in an iteration count being fully calculated
2575 using that wider type (wider
2576 than default @code{INTEGER}) must be rewritten.)
2579 Support @command{gcc} version 2.7.2.
2582 Upgrade to @code{libf2c} as of 1996-03-23, and
2583 fix up some of the build procedures.
2585 Note that the email addresses related to @command{f2c}
2586 have changed---the distribution site now is
2587 named @code{netlib.bell-labs.com}, and the
2588 maintainer's new address is @email{dmg@@bell-labs.com}.
2591 @c 1995-11-18: 0.5.17 released.
2595 @strong{Fix serious bug} in @samp{g77 -v} command that can cause removal of a
2596 system's @file{/dev/null} special file if run by user @code{root}.
2598 @strong{All users} of version 0.5.16 should ensure that
2599 they have not removed @file{/dev/null} or replaced it with an ordinary
2600 file (e.g. by comparing the output of @samp{ls -l /dev/null} with
2601 @samp{ls -l /dev/zero}.
2602 If the output isn't basically the
2603 same, contact your system
2604 administrator about restoring @file{/dev/null} to its proper status).
2606 This bug is particularly insidious because removing @file{/dev/null} as
2607 a special file can go undetected for quite a while, aside from
2608 various applications and programs exhibiting sudden, strange
2611 I sincerely apologize for not realizing the
2612 implications of the fact that when @samp{g77 -v} runs the @command{ld} command
2613 with @samp{-o /dev/null} that @command{ld} tries to @emph{remove} the executable
2614 it is supposed to build (especially if it reports unresolved
2615 references, which it should in this case)!
2618 Fix crash on @samp{CHARACTER*(*) FOO} in a main or block data program unit.
2621 Fix crash that can occur when diagnostics given outside of any
2622 program unit (such as when input file contains @samp{@@foo}).
2625 Fix crashes, infinite loops (hangs), and such involving diagnosed code.
2628 Fix @code{ASSIGN}'ed variables so they can be @code{SAVE}'d or dummy arguments,
2629 and issue clearer error message in cases where target of @code{ASSIGN}
2630 or @code{ASSIGN}ed @code{GOTO}/@code{FORMAT} is too small (which should
2634 Make @code{libf2c} build procedures work on more systems again by
2635 eliminating unnecessary invocations of @samp{ld -r -x} and @command{mv}.
2638 Fix omission of @option{-funix-intrinsics-@dots{}} options in list of permitted
2639 options to compiler.
2642 Fix failure to always diagnose missing type declaration for
2643 @code{IMPLICIT NONE}.
2646 Fix compile-time performance problem (which could sometimes
2647 crash the compiler, cause a hang, or whatever, due to a bug
2648 in the back end) involving exponentiation with a large @code{INTEGER}
2649 constant for the right-hand operator (e.g. @samp{I**32767}).
2652 Fix build procedures so cross-compiling @command{g77} (the @command{fini}
2653 utility in particular) is properly built using the host compiler.
2656 Add new @option{-Wsurprising} option to warn about constructs that are
2657 interpreted by the Fortran standard (and @command{g77}) in ways that
2658 are surprising to many programmers.
2661 Add @code{ERF()} and @code{ERFC()} as generic intrinsics mapping to existing
2662 @code{ERF}/@code{DERF} and @code{ERFC}/@code{DERFC} specific intrinsics.
2664 @emph{Note:} You should
2665 specify @samp{INTRINSIC ERF,ERFC} in any code where you might use
2666 these as generic intrinsics, to improve likelihood of diagnostics
2667 (instead of subtle run-time bugs) when using a compiler that
2668 doesn't support these as intrinsics (e.g. @command{f2c}).
2671 Remove from @option{-fno-pedantic} the diagnostic about @code{DO}
2672 with non-@code{INTEGER} index variable; issue that under
2673 @option{-Wsurprising} instead.
2676 Clarify some diagnostics that say things like ``ignored'' when that's
2680 Clarify diagnostic on use of @code{.EQ.}/@code{.NE.} on @code{LOGICAL}
2684 Minor improvements to code generation for various operations on
2685 @code{LOGICAL} operands.
2688 Minor improvement to code generation for some @code{DO} loops on some
2692 Support @command{gcc} version 2.7.1.
2695 Upgrade to @code{libf2c} as of 1995-11-15.
2698 @c 1995-08-30: 0.5.16 released.
2702 Fix a code-generation bug involving complicated @code{EQUIVALENCE} statements
2703 not involving @code{COMMON}.
2706 Fix code-generation bugs involving invoking ``gratis'' library procedures
2707 in @code{libf2c} from code compiled with @option{-fno-f2c} by making these
2708 procedures known to @command{g77} as intrinsics (not affected by -fno-f2c).
2709 This is known to fix code invoking @code{ERF()}, @code{ERFC()},
2710 @code{DERF()}, and @code{DERFC()}.
2713 Update @code{libf2c} to include netlib patches through 1995-08-16, and
2714 @code{#define} @code{WANT_LEAD_0} to 1 to make @command{g77}-compiled code more
2715 consistent with other Fortran implementations by outputting
2716 leading zeros in formatted and list-directed output.
2719 Fix a code-generation bug involving adjustable dummy arrays with high
2720 bounds whose primaries are changed during procedure execution, and
2721 which might well improve code-generation performance for such arrays
2722 compared to @command{f2c} plus @command{gcc} (but apparently only when using
2723 @file{gcc-2.7.0} or later).
2726 Fix a code-generation bug involving invocation of @code{COMPLEX} and
2727 @code{DOUBLE COMPLEX} @code{FUNCTION}s and doing @code{COMPLEX} and
2728 @code{DOUBLE COMPLEX} divides, when the result
2729 of the invocation or divide is assigned directly to a variable
2730 that overlaps one or more of the arguments to the invocation or divide.
2733 Fix crash by not generating new optimal code for @samp{X**I} if @samp{I} is
2734 nonconstant and the expression is used to dimension a dummy
2735 array, since the @command{gcc} back end does not support the necessary
2736 mechanics (and the @command{gcc} front end rejects the equivalent
2737 construct, as it turns out).
2740 Fix crash on expressions like @samp{COMPLEX**INTEGER}.
2743 Fix crash on expressions like @samp{(1D0,2D0)**2}, i.e. raising a
2744 @code{DOUBLE COMPLEX} constant to an @code{INTEGER} constant power.
2747 Fix crashes and such involving diagnosed code.
2750 Diagnose, instead of crashing on, statement function definitions
2751 having duplicate dummy argument names.
2754 Fix bug causing rejection of good code involving statement function
2758 Fix bug resulting in debugger not knowing size of local equivalence
2759 area when any member of area has initial value (via @code{DATA},
2763 Fix installation bug that prevented installation of @command{g77} driver.
2764 Provide for easy selection of whether to install copy of @command{g77}
2765 as @command{f77} to replace the broken code.
2768 Fix @command{gcc} driver (affects @command{g77} thereby) to not
2769 gratuitously invoke the
2770 @code{f771} program (e.g. when @option{-E} is specified).
2773 Fix diagnostic to point to correct source line when it immediately
2774 follows an @code{INCLUDE} statement.
2777 Support more compiler options in @command{gcc}/@command{g77} when
2778 compiling Fortran files.
2779 These options include @option{-p}, @option{-pg}, @option{-aux-info}, @option{-P},
2780 correct setting of version-number macros for preprocessing, full
2781 recognition of @option{-O0}, and
2782 automatic insertion of configuration-specific linker specs.
2785 Add new intrinsics that interface to existing routines in @code{libf2c}:
2786 @code{ABORT}, @code{DERF}, @code{DERFC}, @code{ERF}, @code{ERFC}, @code{EXIT},
2787 @code{FLUSH}, @code{GETARG}, @code{GETENV}, @code{IARGC},
2788 @code{SIGNAL}, and @code{SYSTEM}.
2789 Note that @code{ABORT}, @code{EXIT}, @code{FLUSH}, @code{SIGNAL}, and
2790 @code{SYSTEM} are intrinsic subroutines, not functions (since they
2791 have side effects), so to get the return values from @code{SIGNAL}
2792 and @code{SYSTEM}, append a final argument specifying an @code{INTEGER}
2793 variable or array element (e.g. @samp{CALL SYSTEM('rm foo',ISTAT)}).
2796 Add new intrinsic group named @code{unix} to contain the new intrinsics,
2797 and by default enable this new group.
2800 Move @code{LOC()} intrinsic out of the @code{vxt} group to the new
2804 Improve @command{g77} so that @samp{g77 -v} by itself (or with
2805 certain other options, including @option{-B}, @option{-b}, @option{-i},
2806 @option{-nostdlib}, and @option{-V}) reports lots more useful
2807 version info, and so that long-form options @command{gcc} accepts are
2808 understood by @command{g77} as well (even in truncated, unambiguous forms).
2811 Add new @command{g77} option @option{--driver=name} to specify driver when
2812 default, @command{gcc}, isn't appropriate.
2815 Add support for @samp{#} directives (as output by the preprocessor) in the
2816 compiler, and enable generation of those directives by the
2817 preprocessor (when compiling @samp{.F} files) so diagnostics and debugging
2818 info are more useful to users of the preprocessor.
2821 Produce better diagnostics, more like @command{gcc}, with info such as
2822 @samp{In function `foo':} and @samp{In file included from...:}.
2825 Support @command{gcc}'s @option{-fident} and @option{-fno-ident} options.
2828 When @option{-Wunused} in effect, don't warn about local variables used as
2829 statement-function dummy arguments or @code{DATA} implied-@code{DO} iteration
2830 variables, even though, strictly speaking, these are not uses
2831 of the variables themselves.
2834 When @samp{-W -Wunused} in effect, don't warn about unused dummy arguments
2835 at all, since there's no way to turn this off for individual
2836 cases (@command{g77} might someday start warning about these)---applies
2837 to @command{gcc} versions 2.7.0 and later, since earlier versions didn't
2838 warn about unused dummy arguments.
2841 New option @option{-fno-underscoring} that inhibits transformation of names
2842 (by appending one or two underscores) so users may experiment
2843 with implications of such an environment.
2846 Minor improvement to @file{gcc/f/info} module to make it easier to build
2847 @command{g77} using the native (non-@command{gcc}) compiler on certain machines
2848 (but definitely not all machines nor all non-@command{gcc} compilers).
2850 do not report bugs showing problems compilers have with
2851 macros defined in @file{gcc/f/target.h} and used in places like
2852 @file{gcc/f/expr.c}.
2855 Add warning to be printed for each invocation of the compiler
2856 if the target machine @code{INTEGER}, @code{REAL}, or @code{LOGICAL} size
2858 since @command{g77} is known to not work well for such cases.
2861 Lots of new documentation (though work is still needed to put it into
2862 canonical GNU format).
2865 Build @code{libf2c} with @option{-g0}, not @option{-g2}, in effect
2866 (by default), to produce
2867 smaller library without lots of debugging clutter.
2870 @c 1995-05-19: 0.5.15 released.
2874 Fix bad code generation involving @samp{X**I} and temporary, internal variables
2875 generated by @command{g77} and the back end (such as for @code{DO} loops).
2878 Fix crash given @samp{CHARACTER A;DATA A/.TRUE./}.
2881 Replace crash with diagnostic given @samp{CHARACTER A;DATA A/1.0/}.
2884 Fix crash or other erratic behavior when null character constant
2885 (@samp{''}) is encountered.
2888 Fix crash or other erratic behavior involving diagnosed code.
2891 Fix code generation for external functions returning type @code{REAL} when
2892 the @option{-ff2c} option is in force (which it is by default) so that
2893 @command{f2c} compatibility is indeed provided.
2896 Disallow @samp{COMMON I(10)} if @samp{I} has previously been specified
2897 with an array declarator.
2900 New @option{-ffixed-line-length-@var{n}} option, where @var{n} is the
2902 of a typical fixed-form line, defaulting to 72 columns, such
2903 that characters beyond column @var{n} are ignored, or @var{n} is @samp{none},
2904 meaning no characters are ignored.
2905 does not affect lines
2906 with @samp{&} in column 1, which are always processed as if
2907 @option{-ffixed-line-length-none} was in effect.
2910 No longer generate better code for some kinds of array references,
2911 as @command{gcc} back end is to be fixed to do this even better, and it
2912 turned out to slow down some code in some cases after all.
2915 In @code{COMMON} and @code{EQUIVALENCE} areas with any members given initial
2916 values (e.g. via @code{DATA}), uninitialized members now always
2917 initialized to binary zeros (though this is not required by
2918 the standard, and might not be done in future versions
2920 Previously, in some @code{COMMON}/@code{EQUIVALENCE} areas
2921 (essentially those with members of more than one type), the
2922 uninitialized members were initialized to spaces, to
2923 cater to @code{CHARACTER} types, but it seems no existing code expects
2924 that, while much existing code expects binary zeros.
2930 Don't emit bad code when low bound of adjustable array is nonconstant
2931 and thus might vary as an expression at run time.
2934 Emit correct code for calculation of number of trips in @code{DO} loops
2936 where the loop should not execute at all.
2937 (This bug affected cases
2938 where the difference between the begin and end values was less
2939 than the step count, though probably not for floating-point cases.)
2942 Fix crash when extra parentheses surround item in
2943 @code{DATA} implied-@code{DO} list.
2946 Fix crash over minor internal inconsistencies in handling diagnostics,
2947 just substitute dummy strings where necessary.
2950 Fix crash on some systems when compiling call to @code{MVBITS()} intrinsic.
2953 Fix crash on array assignment @samp{TYPE@var{ddd}(@dots{})=@dots{}}, where @var{ddd}
2954 is a string of one or more digits.
2957 Fix crash on @code{DCMPLX()} with a single @code{INTEGER} argument.
2960 Fix various crashes involving code with diagnosed errors.
2963 Support @option{-I} option for @code{INCLUDE} statement, plus @command{gcc}'s
2964 @file{header.gcc} facility for handling systems like MS-DOS.
2967 Allow @code{INCLUDE} statement to be continued across multiple lines,
2968 even allow it to coexist with other statements on the same line.
2971 Incorporate Bellcore fixes to @code{libf2c} through 1995-03-15---this
2972 fixes a bug involving infinite loops reading EOF with empty list-directed
2976 Remove all the @command{g77}-specific auto-configuration scripts, code,
2978 except for temporary substitutes for bsearch() and strtoul(), as
2979 too many configure/build problems were reported in these areas.
2980 People will have to fix their systems' problems themselves, or at
2981 least somewhere other than @command{g77}, which expects a working ANSI C
2982 environment (and, for now, a GNU C compiler to compile @command{g77} itself).
2985 Complain if initialized common redeclared as larger in subsequent program
2989 Warn if blank common initialized, since its size can vary and hence
2990 related warnings that might be helpful won't be seen.
2993 New @option{-fbackslash} option, on by default, that causes @samp{\}
2994 within @code{CHARACTER}
2995 and Hollerith constants to be interpreted a la GNU C.
2997 this behavior is somewhat different from @command{f2c}'s, which supports only
2998 a limited subset of backslash (escape) sequences.
3001 Make @option{-fugly-args} the default.
3004 New @option{-fugly-init} option, on by default, that allows typeless/Hollerith
3005 to be specified as initial values for variables or named constants
3006 (@code{PARAMETER}), and also allows character<->numeric conversion in
3007 those contexts---turn off via @option{-fno-ugly-init}.
3010 New @option{-finit-local-zero} option to initialize
3011 local variables to binary zeros.
3012 This does not affect whether they are @code{SAVE}d, i.e. made
3013 automatic or static.
3016 New @option{-Wimplicit} option to warn about implicitly typed variables, arrays,
3018 (Basically causes all program units to default to @code{IMPLICIT NONE}.)
3021 @option{-Wall} now implies @option{-Wuninitialized} as with @command{gcc}
3022 (i.e. unless @option{-O} not specified, since @option{-Wuninitialized}
3023 requires @option{-O}), and implies @option{-Wunused} as well.
3026 @option{-Wunused} no longer gives spurious messages for unused
3027 @code{EXTERNAL} names (since they are assumed to refer to block data
3028 program units, to make use of libraries more reliable).
3031 Support @code{%LOC()} and @code{LOC()} of character arguments.
3034 Support null (zero-length) character constants and expressions.
3037 Support @command{f2c}'s @code{IMAG()} generic intrinsic.
3040 Support @code{ICHAR()}, @code{IACHAR()}, and @code{LEN()} of
3041 character expressions that are valid in assignments but
3042 not normally as actual arguments.
3045 Support @command{f2c}-style @samp{&} in column 1 to mean continuation line.
3048 Allow @code{NAMELIST}, @code{EXTERNAL}, @code{INTRINSIC}, and @code{VOLATILE}
3049 in @code{BLOCK DATA}, even though these are not allowed by the standard.
3052 Allow @code{RETURN} in main program unit.
3055 Changes to Hollerith-constant support to obey Appendix C of the
3060 Now padded on the right with zeros, not spaces.
3063 Hollerith ``format specifications'' in the form of arrays of
3064 non-character allowed.
3067 Warnings issued when non-space truncation occurs when converting
3071 When specified as actual argument, now passed
3072 by reference to @code{INTEGER} (padded on right with spaces if constant
3073 too small, otherwise fully intact if constant wider the @code{INTEGER}
3074 type) instead of by value.
3077 @strong{Warning:} @command{f2c} differs on the
3078 interpretation of @samp{CALL FOO(1HX)}, which it treats exactly the
3079 same as @samp{CALL FOO('X')}, but which the standard and @command{g77} treat
3080 as @samp{CALL FOO(%REF('X '))} (padded with as many spaces as necessary
3081 to widen to @code{INTEGER}), essentially.
3084 Changes and fixes to typeless-constant support:
3088 Now treated as a typeless double-length @code{INTEGER} value.
3091 Warnings issued when overflow occurs.
3094 Padded on the left with zeros when converting
3098 Should be properly aligned and ordered on
3099 the target machine for whatever type it is turned into.
3102 When specified as actual argument, now passed as reference to
3103 a default @code{INTEGER} constant.
3107 @code{%DESCR()} of a non-@code{CHARACTER} expression now passes a pointer to
3108 the expression plus a length for the expression just as if
3109 it were a @code{CHARACTER} expression.
3110 For example, @samp{CALL FOO(%DESCR(D))}, where
3111 @samp{D} is @code{REAL*8}, is the same as @samp{CALL FOO(D,%VAL(8)))}.
3114 Name of multi-entrypoint master function changed to incorporate
3115 the name of the primary entry point instead of a decimal
3116 value, so the name of the master function for @samp{SUBROUTINE X}
3117 with alternate entry points is now @samp{__g77_masterfun_x}.
3120 Remove redundant message about zero-step-count @code{DO} loops.
3123 Clean up diagnostic messages, shortening many of them.
3126 Fix typo in @command{g77} man page.
3129 Clarify implications of constant-handling bugs in @file{f/BUGS}.
3132 Generate better code for @samp{**} operator with a right-hand operand of
3133 type @code{INTEGER}.
3136 Generate better code for @code{SQRT()} and @code{DSQRT()},
3137 also when @option{-ffast-math}
3138 specified, enable better code generation for @code{SIN()} and @code{COS()}.
3141 Generate better code for some kinds of array references.
3144 Speed up lexing somewhat (this makes the compilation phase noticeably