1 @c Copyright (C) 1995-1999 Free Software Foundation, Inc.
2 @c This is part of the G77 manual.
3 @c For copying conditions, see the file g77.texi.
5 @c When changing the above copyright notice, be sure to
6 @c change the one in `news0.texi' accordingly, to effect
7 @c the change in the derived file `NEWS'.
9 @c The text of this file appears in the file BUGS
10 @c in the G77 distribution, as well as in the G77 manual.
16 @chapter News About GNU Fortran
18 @cindex versions, recent
19 @cindex recent versions
21 Changes made to recent versions of GNU Fortran are listed
22 below, with the most recent version first.
24 The changes are generally listed in order:
28 Code-generation and run-time-library bug-fixes
31 Compiler and run-time-library crashes involving valid code
38 Fixes and enhancements to existing features
50 This order is not strict---for example, some items
51 involve a combination of these elements.
53 Note that two variants of @code{g77} are tracked below.
54 The @code{egcs} variant is described vis-a-vis
55 previous versions of @code{egcs} and/or
56 an official FSF version,
59 Therefore, @code{egcs} versions sometimes have multiple listings
60 to help clarify how they differ from other versions,
61 though this can make getting a complete picture
62 of what a particular @code{egcs} version contains
63 somewhat more difficult.
65 @heading In @code{egcs} 1.2:
68 Fix @code{g77} so it no longer crashes when compiling
69 I/O statements using keywords that define @code{INTEGER} values,
70 such as @samp{IOSTAT=@var{j}},
71 where @var{j} is other than default @code{INTEGER}
72 (such as @code{INTEGER*2}).
73 Instead, it issues a diagnostic.
76 The @samp{-ax} option is now obeyed when compiling Fortran programs.
77 (It is passed to the @file{f771} driver.)
80 Source file names with the suffixes @samp{.FOR} and @samp{.FPP}
81 now are recognized by @code{g77}
82 as if they ended in @samp{.for} and @samp{.fpp}, respectively.
85 @code{g77} now warns about a reference to a function
86 when the corresponding @emph{subsequent} function program unit
87 disagrees with the reference concerning the type of the function.
90 Improve documentation and indexing,
91 including information on Year 2000 (Y2K) compliance.
94 @heading In 0.5.24 and @code{egcs} 1.1.2 (versus 0.5.23 and 1.1.1):
97 Fix the @code{IDate} Intrinsic (VXT)
98 so the returned year is in the documented, non-Y2K-compliant range
100 instead of being returned as 100 in the year 2000.
103 @xref{IDate Intrinsic (VXT)},
104 for more information.
108 Fix the @samp{Date_and_Time} intrinsic (in @code{libg2c})
109 to return the milliseconds value properly
113 Fix the @samp{LStat} intrinsic (in @code{libg2c})
114 to return device-ID information properly
118 Improve documentation.
121 @heading In 0.5.24 and @code{egcs} 1.1.1 (versus 0.5.23 and 1.1):
124 Fix @code{libg2c} so it performs an implicit @code{ENDFILE} operation
126 whenever a @code{REWIND} is done.
128 (This bug was introduced in 0.5.23 and @code{egcs} 1.1 in
129 @code{g77}'s version of @code{libf2c}.)
132 Fix @code{libg2c} so it no longer crashes with a spurious diagnostic
133 upon doing any I/O following a direct formatted write.
135 (This bug was introduced in 0.5.23 and @code{egcs} 1.1 in
136 @code{g77}'s version of @code{libf2c}.)
139 Fix @code{g77} so it no longer crashes compiling references
140 to the @samp{Rand} intrinsic on some systems.
143 Fix @code{g77} portion of installation process so it works
144 better on some systems
145 (those with shells requiring @samp{else true} clauses
146 on @samp{if} constructs
147 for the completion code to be set properly).
150 @heading In @code{egcs} 1.1 (versus 0.5.24):
153 Fix @code{g77} crash compiling code
154 containing the construct @samp{CMPLX(0.)} or similar.
158 (or apparently infinite run-time)
159 when compiling certain complicated expressions
160 involving @code{COMPLEX} arithmetic
161 (especially multiplication).
164 @cindex stack, 387 coprocessor
168 Fix a code-generation bug that afflicted
169 Intel x86 targets when @samp{-O2} was specified
170 compiling, for example, an old version of
171 the @samp{DNRM2} routine.
173 The x87 coprocessor stack was being
174 mismanaged in cases involving assigned @code{GOTO}
178 @cindex double-precision performance
179 @cindex -malign-double
181 Align static double-precision variables and arrays
183 regardless of whether @samp{-malign-double} is specified.
185 Generally, this affects only local variables and arrays
186 having the @code{SAVE} attribute
187 or given initial values via @code{DATA}.
190 @c 1998-09-01: egcs-1.1 released.
191 @heading In @code{egcs} 1.1 (versus @code{egcs} 1.0.3):
194 Fix bugs in the @code{libU77} intrinsic @samp{HostNm}
195 that wrote one byte beyond the end of its @samp{CHARACTER}
197 and in the @code{libU77} intrinsics
198 @samp{GMTime} and @samp{LTime}
199 that overwrote their arguments.
202 Assumed arrays with negative bounds
203 (such as @samp{REAL A(-1:*)})
204 no longer elicit spurious diagnostics from @code{g77},
205 even on systems with pointers having
206 different sizes than integers.
208 This bug is not known to have existed in any
209 recent version of @code{gcc}.
210 It was introduced in an early release of @code{egcs}.
213 Valid combinations of @code{EXTERNAL},
214 passing that external as a dummy argument
215 without explicitly giving it a type,
216 and, in a subsequent program unit,
217 referencing that external as
218 an external function with a different type
219 no longer crash @code{g77}.
222 @code{CASE DEFAULT} no longer crashes @code{g77}.
225 The @samp{-Wunused} option no longer issues a spurious
226 warning about the ``master'' procedure generated by
227 @code{g77} for procedures containing @code{ENTRY} statements.
230 Support @samp{FORMAT(I<@var{expr}>)} when @var{expr} is a
231 compile-time constant @code{INTEGER} expression.
234 Fix @code{g77} @samp{-g} option so procedures that
235 use @samp{ENTRY} can be stepped through, line by line,
239 Allow any @code{REAL} argument to intrinsics
240 @code{Second} and @code{CPU_Time}.
243 Use @code{tempnam}, if available, to open scratch files
244 (as in @samp{OPEN(STATUS='SCRATCH')})
245 so that the @code{TMPDIR} environment variable,
249 @code{g77}'s version of @code{libf2c} separates out
250 the setting of global state
251 (such as command-line arguments and signal handling)
252 from @file{main.o} into distinct, new library
255 This should make it easier to write portable applications
256 that have their own (non-Fortran) @code{main()} routine
257 properly set up the @code{libf2c} environment, even
258 when @code{libf2c} (now @code{libg2c}) is a shared library.
261 @code{g77} no longer installs the @file{f77} command
262 and @file{f77.1} man page
263 in the @file{/usr} or @file{/usr/local} heirarchy,
264 even if the @file{f77-install-ok} file exists
265 in the source or build directory.
266 See the installation documentation for more information.
269 @code{g77} no longer installs the @file{libf2c.a} library
270 and @file{f2c.h} include file
271 in the @file{/usr} or @file{/usr/local} heirarchy,
272 even if the @file{f2c-install-ok} or @file{f2c-exists-ok} files exist
273 in the source or build directory.
274 See the installation documentation for more information.
277 The @file{libf2c.a} library produced by @code{g77} has been
278 renamed to @file{libg2c.a}.
279 It is installed only in the @code{gcc} ``private''
280 directory heirarchy, @file{gcc-lib}.
281 This allows system administrators and users to choose which
282 version of the @code{libf2c} library from @code{netlib} they
283 wish to use on a case-by-case basis.
284 See the installation documentation for more information.
287 The @file{f2c.h} include (header) file produced by @code{g77}
288 has been renamed to @file{g2c.h}.
289 It is installed only in the @code{gcc} ``private''
290 directory heirarchy, @file{gcc-lib}.
291 This allows system administrators and users to choose which
292 version of the include file from @code{netlib} they
293 wish to use on a case-by-case basis.
294 See the installation documentation for more information.
297 The @code{g77} command now expects the run-time library
298 to be named @code{libg2c.a} instead of @code{libf2c.a},
299 to ensure that a version other than the one built and
300 installed as part of the same @code{g77} version is picked up.
303 During the configuration and build process,
304 @code{g77} creates subdirectories it needs only as it
306 Other cleaning up of the configuration and build process
307 has been performed as well.
310 @code{install-info} now used to update the directory of
311 Info documentation to contain an entry for @code{g77}
312 (during installation).
315 Some diagnostics have been changed from warnings to errors,
316 to prevent inadvertent use of the resulting, probably buggy,
318 These mostly include diagnostics about use of unsupported features
319 in the @code{OPEN}, @code{INQUIRE}, @code{READ}, and
320 @code{WRITE} statements,
321 and about truncations of various sorts of constants.
324 Improve compilation of @code{FORMAT} expressions so that
325 a null byte is appended to the last operand if it
327 This provides a cleaner run-time diagnostic as provided
328 by @code{libf2c} for statements like @samp{PRINT '(I1', 42}.
331 Improve documentation and indexing.
334 The upgrade to @code{libf2c} as of 1998-06-18
335 should fix a variety of problems, including
336 those involving some uses of the @samp{T} format
337 specifier, and perhaps some build (porting) problems
341 @heading In 0.5.24 and @code{egcs} 1.1 (versus 0.5.23):
344 @code{g77} no longer produces incorrect code
346 for @samp{EQUIVALENCE} and @samp{COMMON}
347 aggregates that, due to ``unnatural'' ordering of members
348 vis-a-vis their types, require initial padding.
351 @code{g77} no longer crashes when compiling code
352 containing specification statements such as
353 @samp{INTEGER(KIND=7) PTR}.
356 @code{g77} no longer crashes when compiling code
357 such as @samp{J = SIGNAL(1, 2)}.
360 @code{g77} now treats @samp{%LOC(@var{expr})} and
361 @samp{LOC(@var{expr})} as ``ordinary'' expressions
362 when they are used as arguments in procedure calls.
363 This change applies only to global (filewide) analysis,
364 making it consistent with
365 how @code{g77} actually generates code
368 Previously, @code{g77} treated these expressions
369 as denoting special ``pointer'' arguments
370 for the purposes of filewide analysis.
373 The @code{g77} driver now ensures that @samp{-lg2c}
374 is specified in the link phase prior to any
375 occurrence of @samp{-lm}.
376 This prevents accidentally linking to a routine
377 in the SunOS4 @samp{-lm} library
378 when the generated code wants to link to the one
379 in @code{libf2c} (@code{libg2c}).
382 @code{g77} emits more debugging information when
385 This new information allows, for example,
386 @kbd{which __g77_length_a} to be used in @code{gdb}
387 to determine the type of the phantom length argument
388 supplied with @samp{CHARACTER} variables.
390 This information pertains to internally-generated
391 type, variable, and other information,
392 not to the longstanding deficiencies vis-a-vis
393 @samp{COMMON} and @samp{EQUIVALENCE}.
396 The F90 @samp{Date_and_Time} intrinsic now is
400 The F90 @samp{System_Clock} intrinsic allows
401 the optional arguments (except for the @samp{Count}
402 argument) to be omitted.
405 Upgrade to @code{libf2c} as of 1998-06-18.
408 Improve documentation and indexing.
411 @c 1998-05-20: 0.5.23 released.
412 @heading In 0.5.23 (versus 0.5.22):
415 This release contains several regressions against
416 version 0.5.22 of @code{g77}, due to using the
417 ``vanilla'' @code{gcc} back end instead of patching
418 it to fix a few bugs and improve performance in a
422 See the page on Fortran news
425 @xref{Actual Bugs,,Actual Bugs We Haven't Fixed Yet},
426 available in plain-text format in @code{gcc/f/BUGS},
428 for information on the known bugs in this version,
429 including the regressions.
431 Features that have been dropped from this version
432 of @code{g77} due to their being implemented
433 via @code{g77}-specific patches to the @code{gcc}
434 back end in previous releases include:
438 Support for @code{__restrict__} keyword,
439 the options @samp{-fargument-alias}, @samp{-fargument-noalias},
440 and @samp{-fargument-noalias-global},
441 and the corresponding alias-analysis code.
443 (@code{egcs} has the alias-analysis
444 code, but not the @code{__restrict__} keyword.
445 @code{egcs} @code{g77} users benefit from the alias-analysis
446 code despite the lack of the @code{__restrict__} keyword,
447 which is a C-language construct.)
450 Support for the GNU compiler options
451 @samp{-fmove-all-movables},
452 @samp{-freduce-all-givs},
453 and @samp{-frerun-loop-opt}.
455 (@code{egcs} supports these options.
456 @code{g77} users of @code{egcs} benefit from them even if
457 they are not explicitly specified,
458 because the defaults are optimized for @code{g77} users.)
461 Support for the @samp{-W} option warning about
462 integer division by zero.
465 The Intel x86-specific option @samp{-malign-double}
466 applying to stack-allocated data
467 as well as statically-allocate data.
470 Note that the @file{gcc/f/gbe/} subdirectory has been removed
471 from this distribution as a result of @code{g77} no longer
472 including patches for the @code{gcc} back end.
475 Fix bugs in the @code{libU77} intrinsic @samp{HostNm}
476 that wrote one byte beyond the end of its @samp{CHARACTER}
478 and in the @code{libU77} intrinsics
479 @samp{GMTime} and @samp{LTime}
480 that overwrote their arguments.
483 Support @code{gcc} version 2.8,
484 and remove support for prior versions of @code{gcc}.
486 @cindex -@w{}-driver option
487 @cindex g77 options, -@w{}-driver
488 @cindex options, -@w{}-driver
490 Remove support for the @samp{--driver} option,
491 as @code{g77} now does all the driving,
492 just like @code{gcc}.
495 @code{CASE DEFAULT} no longer crashes @code{g77}.
498 Valid combinations of @code{EXTERNAL},
499 passing that external as a dummy argument
500 without explicitly giving it a type,
501 and, in a subsequent program unit,
502 referencing that external as
503 an external function with a different type
504 no longer crash @code{g77}.
507 @code{g77} no longer installs the @file{f77} command
508 and @file{f77.1} man page
509 in the @file{/usr} or @file{/usr/local} heirarchy,
510 even if the @file{f77-install-ok} file exists
511 in the source or build directory.
512 See the installation documentation for more information.
515 @code{g77} no longer installs the @file{libf2c.a} library
516 and @file{f2c.h} include file
517 in the @file{/usr} or @file{/usr/local} heirarchy,
518 even if the @file{f2c-install-ok} or @file{f2c-exists-ok} files exist
519 in the source or build directory.
520 See the installation documentation for more information.
523 The @file{libf2c.a} library produced by @code{g77} has been
524 renamed to @file{libg2c.a}.
525 It is installed only in the @code{gcc} ``private''
526 directory heirarchy, @file{gcc-lib}.
527 This allows system administrators and users to choose which
528 version of the @code{libf2c} library from @code{netlib} they
529 wish to use on a case-by-case basis.
530 See the installation documentation for more information.
533 The @file{f2c.h} include (header) file produced by @code{g77}
534 has been renamed to @file{g2c.h}.
535 It is installed only in the @code{gcc} ``private''
536 directory heirarchy, @file{gcc-lib}.
537 This allows system administrators and users to choose which
538 version of the include file from @code{netlib} they
539 wish to use on a case-by-case basis.
540 See the installation documentation for more information.
543 The @code{g77} command now expects the run-time library
544 to be named @code{libg2c.a} instead of @code{libf2c.a},
545 to ensure that a version other than the one built and
546 installed as part of the same @code{g77} version is picked up.
549 The @samp{-Wunused} option no longer issues a spurious
550 warning about the ``master'' procedure generated by
551 @code{g77} for procedures containing @code{ENTRY} statements.
554 @code{g77}'s version of @code{libf2c} separates out
555 the setting of global state
556 (such as command-line arguments and signal handling)
557 from @file{main.o} into distinct, new library
560 This should make it easier to write portable applications
561 that have their own (non-Fortran) @code{main()} routine
562 properly set up the @code{libf2c} environment, even
563 when @code{libf2c} (now @code{libg2c}) is a shared library.
566 During the configuration and build process,
567 @code{g77} creates subdirectories it needs only as it
568 needs them, thus avoiding unnecessary creation of, for example,
569 @file{stage1/f/runtime} when doing a non-bootstrap build.
570 Other cleaning up of the configuration and build process
571 has been performed as well.
574 @code{install-info} now used to update the directory of
575 Info documentation to contain an entry for @code{g77}
576 (during installation).
579 Some diagnostics have been changed from warnings to errors,
580 to prevent inadvertent use of the resulting, probably buggy,
582 These mostly include diagnostics about use of unsupported features
583 in the @code{OPEN}, @code{INQUIRE}, @code{READ}, and
584 @code{WRITE} statements,
585 and about truncations of various sorts of constants.
588 Improve documentation and indexing.
591 Upgrade to @code{libf2c} as of 1998-04-20.
593 This should fix a variety of problems, including
594 those involving some uses of the @samp{T} format
595 specifier, and perhaps some build (porting) problems
599 @c 1998-03-16: 0.5.22 released.
600 @heading In 0.5.22 (versus 0.5.21):
603 Fix code generation for iterative @code{DO} loops that
604 have one or more references to the iteration variable,
605 or to aliases of it, in their control expressions.
606 For example, @samp{DO 10 J=2,J} now is compiled correctly.
609 @cindex stack, 387 coprocessor
613 Fix a code-generation bug that afflicted
614 Intel x86 targets when @samp{-O2} was specified
615 compiling, for example, an old version of
616 the @samp{DNRM2} routine.
618 The x87 coprocessor stack was being
619 mismanaged in cases involving assigned @code{GOTO}
623 Fix @code{DTime} intrinsic so as not to truncate
624 results to integer values (on some systems).
627 Fix @code{Signal} intrinsic so it offers portable
628 support for 64-bit systems (such as Digital Alphas
632 Fix run-time crash involving @code{NAMELIST} on 64-bit
633 machines such as Alphas.
636 Fix @code{g77} version of @code{libf2c} so it no longer
637 produces a spurious @samp{I/O recursion} diagnostic at run time
638 when an I/O operation (such as @samp{READ *,I}) is interrupted
639 in a manner that causes the program to be terminated
640 via the @samp{f_exit} routine (such as via @kbd{C-c}).
643 Fix @code{g77} crash triggered by @code{CASE} statement with
644 an omitted lower or upper bound.
647 Fix @code{g77} crash compiling references to @code{CPU_Time}
652 (or apparently infinite run-time)
653 when compiling certain complicated expressions
654 involving @code{COMPLEX} arithmetic
655 (especially multiplication).
658 Fix @code{g77} crash on statements such as
659 @samp{PRINT *, (REAL(Z(I)),I=1,2)}, where
660 @samp{Z} is @code{DOUBLE COMPLEX}.
663 Fix a @code{g++} crash.
666 Support @samp{FORMAT(I<@var{expr}>)} when @var{expr} is a
667 compile-time constant @code{INTEGER} expression.
670 Fix @code{g77} @samp{-g} option so procedures that
671 use @samp{ENTRY} can be stepped through, line by line,
675 Fix a profiling-related bug in @code{gcc} back end for
676 Intel x86 architecture.
679 Allow any @code{REAL} argument to intrinsics
680 @code{Second} and @code{CPU_Time}.
683 Allow any numeric argument to intrinsics
684 @code{Int2} and @code{Int8}.
687 Use @code{tempnam}, if available, to open scratch files
688 (as in @samp{OPEN(STATUS='SCRATCH')})
689 so that the @code{TMPDIR} environment variable,
693 Rename the @code{gcc} keyword @code{restrict} to
694 @code{__restrict__}, to avoid rejecting valid, existing,
696 Support for @code{restrict} is now more like support
700 Fix @samp{-fpedantic} to not reject procedure invocations
701 such as @samp{I=J()} and @samp{CALL FOO()}.
704 Fix @samp{-fugly-comma} to affect invocations of
705 only external procedures.
706 Restore rejection of gratuitous trailing omitted
707 arguments to intrinsics, as in @samp{I=MAX(3,4,,)}.
710 Fix compiler so it accepts @samp{-fgnu-intrinsics-*} and
711 @samp{-fbadu77-intrinsics-*} options.
714 Improve diagnostic messages from @code{libf2c}
715 so it is more likely that the printing of the
716 active format string is limited to the string,
717 with no trailing garbage being printed.
719 (Unlike @code{f2c}, @code{g77} did not append
720 a null byte to its compiled form of every
721 format string specified via a @code{FORMAT} statement.
722 However, @code{f2c} would exhibit the problem
723 anyway for a statement like @samp{PRINT '(I)garbage', 1}
724 by printing @samp{(I)garbage} as the format string.)
727 Improve compilation of @code{FORMAT} expressions so that
728 a null byte is appended to the last operand if it
730 This provides a cleaner run-time diagnostic as provided
731 by @code{libf2c} for statements like @samp{PRINT '(I1', 42}.
734 Fix various crashes involving code with diagnosed errors.
737 Fix cross-compilation bug when configuring @code{libf2c}.
743 Improve documentation and indexing.
746 Upgrade to @code{libf2c} as of 1997-09-23.
747 This fixes a formatted-I/O bug that afflicted
748 64-bit systems with 32-bit integers
749 (such as Digital Alpha running GNU/Linux).
752 @c 1998-03-15: egcs-1.0.2 released.
753 @heading In @code{egcs} 1.0.2 (versus @code{egcs} 1.0.1):
756 Fix @code{g77} crash triggered by @code{CASE} statement with
757 an omitted lower or upper bound.
760 Fix @code{g77} crash on statements such as
761 @samp{PRINT *, (REAL(Z(I)),I=1,2)}, where
762 @samp{Z} is @code{DOUBLE COMPLEX}.
767 @cindex options, -fPIC
769 Fix @samp{-fPIC} (such as compiling for ELF targets)
770 on the Intel x86 architecture target
771 so invalid assembler code is no longer produced.
774 Fix @samp{-fpedantic} to not reject procedure invocations
775 such as @samp{I=J()} and @samp{CALL FOO()}.
778 Fix @samp{-fugly-comma} to affect invocations of
779 only external procedures.
780 Restore rejection of gratuitous trailing omitted
781 arguments to intrinsics, as in @samp{I=MAX(3,4,,)}.
784 Fix compiler so it accepts @samp{-fgnu-intrinsics-*} and
785 @samp{-fbadu77-intrinsics-*} options.
788 @c 1998-01-02: egcs-1.0.1 released.
789 @heading In @code{egcs} 1.0.1 (versus @code{egcs} 1.0):
792 Fix run-time crash involving @code{NAMELIST} on 64-bit
793 machines such as Alphas.
796 @c 1997-12-03: egcs-1.0 released.
797 @heading In @code{egcs} 1.0 (versus 0.5.21):
800 Version 1.0 of @code{egcs}
801 contains several regressions against
802 version 0.5.21 of @code{g77},
804 ``vanilla'' @code{gcc} back end instead of patching
805 it to fix a few bugs and improve performance in a
809 See the page on Fortran news
812 @xref{Actual Bugs,,Actual Bugs We Haven't Fixed Yet},
813 available in plain-text format in @code{gcc/f/BUGS},
815 for information on the known bugs in this version,
816 including the regressions.
818 Features that have been dropped from this version
819 of @code{g77} due to their being implemented
820 via @code{g77}-specific patches to the @code{gcc}
821 back end in previous releases include:
825 Support for the C-language @code{restrict} keyword.
828 Support for the @samp{-W} option warning about
829 integer division by zero.
832 The Intel x86-specific option @samp{-malign-double}
833 applying to stack-allocated data
834 as well as statically-allocate data.
837 Note that the @file{gcc/f/gbe/} subdirectory has been removed
838 from this distribution as a result of @code{g77}
839 being fully integrated with
840 the @code{egcs} variant of the @code{gcc} back end.
843 Fix code generation for iterative @code{DO} loops that
844 have one or more references to the iteration variable,
845 or to aliases of it, in their control expressions.
846 For example, @samp{DO 10 J=2,J} now is compiled correctly.
849 Fix @code{DTime} intrinsic so as not to truncate
850 results to integer values (on some systems).
853 @c Toon Moene discovered these.
854 Some Fortran code, miscompiled
855 by @code{g77} built on @code{gcc} version 2.8.1
856 on m68k-next-nextstep3 configurations
857 when using the @samp{-O2} option,
858 is now compiled correctly.
859 It is believed that a C function known to miscompile
860 on that configuration
861 when using the @samp{-O2 -funroll-loops} options
862 also is now compiled correctly.
865 Remove support for non-@code{egcs} versions of @code{gcc}.
867 @cindex -@w{}-driver option
868 @cindex g77 options, -@w{}-driver
869 @cindex options, -@w{}-driver
871 Remove support for the @samp{--driver} option,
872 as @code{g77} now does all the driving,
873 just like @code{gcc}.
876 Allow any numeric argument to intrinsics
877 @code{Int2} and @code{Int8}.
880 Improve diagnostic messages from @code{libf2c}
881 so it is more likely that the printing of the
882 active format string is limited to the string,
883 with no trailing garbage being printed.
885 (Unlike @code{f2c}, @code{g77} did not append
886 a null byte to its compiled form of every
887 format string specified via a @code{FORMAT} statement.
888 However, @code{f2c} would exhibit the problem
889 anyway for a statement like @samp{PRINT '(I)garbage', 1}
890 by printing @samp{(I)garbage} as the format string.)
893 Upgrade to @code{libf2c} as of 1997-09-23.
894 This fixes a formatted-I/O bug that afflicted
895 64-bit systems with 32-bit integers
896 (such as Digital Alpha running GNU/Linux).
899 @c 1997-09-09: 0.5.21 released.
903 Fix a code-generation bug introduced by 0.5.20
904 caused by loop unrolling (by specifying
905 @samp{-funroll-loops} or similar).
906 This bug afflicted all code compiled by
907 version 2.7.2.2.f.2 of @code{gcc} (C, C++,
911 Fix a code-generation bug manifested when
912 combining local @code{EQUIVALENCE} with a
913 @code{DATA} statement that follows
914 the first executable statement (or is
915 treated as an executable-context statement
916 as a result of using the @samp{-fpedantic}
920 Fix a compiler crash that occured when an
921 integer division by a constant zero is detected.
922 Instead, when the @samp{-W} option is specified,
923 the @code{gcc} back end issues a warning about such a case.
924 This bug afflicted all code compiled by
925 version 2.7.2.2.f.2 of @code{gcc} (C, C++,
929 Fix a compiler crash that occurred in some cases
930 of procedure inlining.
931 (Such cases became more frequent in 0.5.20.)
934 Fix a compiler crash resulting from using @code{DATA}
935 or similar to initialize a @code{COMPLEX} variable or
939 Fix compiler crashes involving use of @code{AND}, @code{OR},
940 or @code{XOR} intrinsics.
943 Fix compiler bug triggered when using a @code{COMMON}
944 or @code{EQUIVALENCE} variable
945 as the target of an @code{ASSIGN}
946 or assigned-@code{GOTO} statement.
949 Fix compiler crashes due to using the name of a some
950 non-standard intrinsics (such as @samp{FTELL} or
951 @samp{FPUTC}) as such and as the name of a procedure
953 Such dual use of a name in a program is allowed by
956 @c @code{g77}'s version of @code{libf2c} has been modified
957 @c so that the external names of library's procedures do not
958 @c conflict with names used for Fortran procedures compiled
960 @c An additional layer of jacket procedures has been added
961 @c to @code{libf2c} to map the old names to the new names,
962 @c for automatic use by programs that interface to the
963 @c library procedures via the external-procedure mechanism.
965 @c For example, the intrinsic @code{FPUTC} previously was
966 @c implemented by @code{g77} as a call to the @code{libf2c}
967 @c routine @samp{fputc_}.
968 @c This would conflict with a Fortran procedure named @code{FPUTC}
969 @c (using default compiler options), and this conflict
970 @c would cause a crash under certain circumstances.
972 @c Now, the intrinsic @code{FPUTC} calls @samp{G77_fputc_0},
973 @c which does not conflict with the @samp{fputc_} external
974 @c that implements a Fortran procedure named @code{FPUTC}.
976 @c Programs that refer to @code{FPUTC} as an external procedure
977 @c without supplying their own implementation will link to
978 @c the new @code{libf2c} routine @samp{fputc_}, which is
979 @c simply a jacket routine that calls @samp{G77_fputc_0}.
982 Place automatic arrays on the stack, even if
983 @code{SAVE} or the @samp{-fno-automatic} option
985 This avoids a compiler crash in some cases.
988 The @samp{-malign-double} option now reliably aligns
989 @code{DOUBLE PRECISION} optimally on Pentium and
990 Pentium Pro architectures (586 and 686 in @code{gcc}).
993 New option @samp{-Wno-globals} disables warnings
994 about ``suspicious'' use of a name both as a global
995 name and as the implicit name of an intrinsic, and
996 warnings about disagreements over the number or natures of
997 arguments passed to global procedures, or the
998 natures of the procedures themselves.
1000 The default is to issue such warnings, which are
1001 new as of this version of @code{g77}.
1004 New option @samp{-fno-globals} disables diagnostics
1005 about potentially fatal disagreements
1006 analysis problems, such as disagreements over the
1007 number or natures of arguments passed to global
1008 procedures, or the natures of those procedures themselves.
1010 The default is to issue such diagnostics and flag
1011 the compilation as unsuccessful.
1012 With this option, the diagnostics are issued as
1013 warnings, or, if @samp{-Wno-globals} is specified,
1014 are not issued at all.
1016 This option also disables inlining of global procedures,
1017 to avoid compiler crashes resulting from coding errors
1018 that these diagnostics normally would identify.
1021 Diagnose cases where a reference to a procedure
1022 disagrees with the type of that procedure, or
1023 where disagreements about the number or nature
1025 This avoids a compiler crash.
1028 Fix parsing bug whereby @code{g77} rejected a
1029 second initialization specification immediately
1030 following the first's closing @samp{/} without
1031 an intervening comma in a @code{DATA} statement,
1032 and the second specification was an implied-DO list.
1035 Improve performance of the @code{gcc} back end so
1036 certain complicated expressions involving @code{COMPLEX}
1037 arithmetic (especially multiplication) don't appear to
1038 take forever to compile.
1041 Fix a couple of profiling-related bugs in @code{gcc}
1045 Integrate GNU Ada's (GNAT's) changes to the back end,
1046 which consist almost entirely of bug fixes.
1047 These fixes are circa version 3.10p of GNAT.
1050 Include some other @code{gcc} fixes that seem useful in
1051 @code{g77}'s version of @code{gcc}.
1052 (See @file{gcc/ChangeLog} for details---compare it
1053 to that file in the vanilla @code{gcc-2.7.2.3.tar.gz}
1057 Fix @code{libU77} routines that accept file and other names
1058 to strip trailing blanks from them, for consistency
1059 with other implementations.
1060 Blanks may be forcibly appended to such names by
1061 appending a single null character (@samp{CHAR(0)})
1062 to the significant trailing blanks.
1065 Fix @code{CHMOD} intrinsic to work with file names
1066 that have embedded blanks, commas, and so on.
1069 Fix @code{SIGNAL} intrinsic so it accepts an
1070 optional third @samp{Status} argument.
1073 Fix @code{IDATE()} intrinsic subroutine (VXT form)
1074 so it accepts arguments in the correct order.
1075 Documentation fixed accordingly, and for
1076 @code{GMTIME()} and @code{LTIME()} as well.
1079 Make many changes to @code{libU77} intrinsics to
1080 support existing code more directly.
1082 Such changes include allowing both subroutine and
1083 function forms of many routines, changing @code{MCLOCK()}
1084 and @code{TIME()} to return @code{INTEGER(KIND=1)} values,
1085 introducing @code{MCLOCK8()} and @code{TIME8()} to
1086 return @code{INTEGER(KIND=2)} values,
1087 and placing functions that are intended to perform
1088 side effects in a new intrinsic group, @code{badu77}.
1091 Improve @code{libU77} so it is more portable.
1094 Add options @samp{-fbadu77-intrinsics-delete},
1095 @samp{-fbadu77-intrinsics-hide}, and so on.
1098 Fix crashes involving diagnosed or invalid code.
1101 @code{g77} and @code{gcc} now do a somewhat better
1102 job detecting and diagnosing arrays that are too
1103 large to handle before these cause diagnostics
1104 during the assembler or linker phase, a compiler
1105 crash, or generation of incorrect code.
1108 Make some fixes to alias analysis code.
1111 Add support for @code{restrict} keyword in @code{gcc}
1115 Support @code{gcc} version 2.7.2.3
1116 (modified by @code{g77} into version 2.7.2.3.f.1),
1118 support for prior versions of @code{gcc}.
1121 Incorporate GNAT's patches to the @code{gcc} back
1122 end into @code{g77}'s, so GNAT users do not need
1123 to apply GNAT's patches to build both GNAT and @code{g77}
1124 from the same source tree.
1127 Modify @code{make} rules and related code so that
1128 generation of Info documentation doesn't require
1129 compilation using @code{gcc}.
1130 Now, any ANSI C compiler should be adequate to
1131 produce the @code{g77} documentation (in particular,
1132 the tables of intrinsics) from scratch.
1135 Add @code{INT2} and @code{INT8} intrinsics.
1138 Add @code{CPU_TIME} intrinsic.
1141 Add @code{ALARM} intrinsic.
1144 @code{CTIME} intrinsic now accepts any @code{INTEGER}
1145 argument, not just @code{INTEGER(KIND=2)}.
1148 Warn when explicit type declaration disagrees with
1149 the type of an intrinsic invocation.
1152 Support @samp{*f771} entry in @code{gcc} @file{specs} file.
1155 Fix typo in @code{make} rule @samp{g77-cross}, used only for
1159 Fix @code{libf2c} build procedure to re-archive library
1160 if previous attempt to archive was interrupted.
1163 Change @code{gcc} to unroll loops only during the last
1164 invocation (of as many as two invocations) of loop
1168 Improve handling of @samp{-fno-f2c} so that code that
1169 attempts to pass an intrinsic as an actual argument,
1170 such as @samp{CALL FOO(ABS)}, is rejected due to the fact
1171 that the run-time-library routine is, effectively,
1172 compiled with @samp{-ff2c} in effect.
1175 Fix @code{g77} driver to recognize @samp{-fsyntax-only}
1176 as an option that inhibits linking, just like @samp{-c} or
1177 @samp{-S}, and to recognize and properly handle the
1178 @samp{-nostdlib}, @samp{-M}, @samp{-MM}, @samp{-nodefaultlibs},
1179 and @samp{-Xlinker} options.
1182 Upgrade to @code{libf2c} as of 1997-08-16.
1185 Modify @code{libf2c} to consistently and clearly diagnose
1186 recursive I/O (at run time).
1189 @code{g77} driver now prints version information (such as produced
1190 by @kbd{g77 -v}) to @code{stderr} instead of @code{stdout}.
1193 The @samp{.r} suffix now designates a Ratfor source file,
1194 to be preprocessed via the @code{ratfor} command, available
1198 Fix some aspects of how @code{gcc} determines what kind of
1199 system is being configured and what kinds are supported.
1200 For example, GNU Linux/Alpha ELF systems now are directly
1204 Improve diagnostics.
1207 Improve documentation and indexing.
1210 Include all pertinent files for @code{libf2c} that come
1211 from @code{netlib.bell-labs.com}; give any such files
1212 that aren't quite accurate in @code{g77}'s version of
1213 @code{libf2c} the suffix @samp{.netlib}.
1216 Reserve @code{INTEGER(KIND=0)} for future use.
1219 @c 1997-02-28: 0.5.20 released.
1223 The @samp{-fno-typeless-boz} option is now the default.
1225 This option specifies that non-decimal-radix
1226 constants using the prefixed-radix form (such as @samp{Z'1234'})
1227 are to be interpreted as @code{INTEGER} constants.
1228 Specify @samp{-ftypeless-boz} to cause such
1229 constants to be interpreted as typeless.
1231 (Version 0.5.19 introduced @samp{-fno-typeless-boz} and
1235 Options @samp{-ff90-intrinsics-enable} and
1236 @samp{-fvxt-intrinsics-enable} now are the
1239 Some programs might use names that clash with
1240 intrinsic names defined (and now enabled) by these
1241 options or by the new @code{libU77} intrinsics.
1242 Users of such programs might need to compile them
1243 differently (using, for example, @samp{-ff90-intrinsics-disable})
1244 or, better yet, insert appropriate @code{EXTERNAL}
1245 statements specifying that these names are not intended
1246 to be names of intrinsics.
1249 The @samp{ALWAYS_FLUSH} macro is no longer defined when
1250 building @code{libf2c}, which should result in improved
1251 I/O performance, especially over NFS.
1253 @emph{Note:} If you have code that depends on the behavior
1254 of @code{libf2c} when built with @samp{ALWAYS_FLUSH} defined,
1255 you will have to modify @code{libf2c} accordingly before
1256 building it from this and future versions of @code{g77}.
1259 Dave Love's implementation of @code{libU77} has been
1260 added to the version of @code{libf2c} distributed with
1261 and built as part of @code{g77}.
1262 @code{g77} now knows about the routines in this library
1266 New option @samp{-fvxt} specifies that the
1267 source file is written in VXT Fortran, instead of GNU Fortran.
1270 The @samp{-fvxt-not-f90} option has been deleted,
1271 along with its inverse, @samp{-ff90-not-vxt}.
1273 If you used one of these deleted options, you should
1274 re-read the pertinent documentation to determine which
1275 options, if any, are appropriate for compiling your
1276 code with this version of @code{g77}.
1279 The @samp{-fugly} option now issues a warning, as it
1280 likely will be removed in a future version.
1282 (Enabling all the @samp{-fugly-*} options is unlikely
1283 to be feasible, or sensible, in the future,
1284 so users should learn to specify only those
1285 @samp{-fugly-*} options they really need for a
1286 particular source file.)
1289 The @samp{-fugly-assumed} option, introduced in
1290 version 0.5.19, has been changed to
1291 better accommodate old and new code.
1294 Make a number of fixes to the @code{g77} front end and
1295 the @code{gcc} back end to better support Alpha (AXP)
1297 This includes providing at least one bug-fix to the
1298 @code{gcc} back end for Alphas.
1301 Related to supporting Alpha (AXP) machines, the @code{LOC()}
1302 intrinsic and @code{%LOC()} construct now return
1303 values of integer type that is the same width (holds
1304 the same number of bits) as the pointer type on the
1307 On most machines, this won't make a difference, whereas
1308 on Alphas, the type these constructs return is
1309 @code{INTEGER*8} instead of the more common @code{INTEGER*4}.
1312 Emulate @code{COMPLEX} arithmetic in the @code{g77} front
1313 end, to avoid bugs in @code{complex} support in the
1314 @code{gcc} back end.
1315 New option @samp{-fno-emulate-complex}
1316 causes @code{g77} to revert the 0.5.19 behavior.
1319 Fix bug whereby @samp{REAL A(1)}, for example, caused
1320 a compiler crash if @samp{-fugly-assumed} was in effect
1321 and @var{A} was a local (automatic) array.
1322 That case is no longer affected by the new
1323 handling of @samp{-fugly-assumed}.
1326 Fix @code{g77} command driver so that @samp{g77 -o foo.f}
1327 no longer deletes @file{foo.f} before issuing other
1328 diagnostics, and so the @samp{-x} option is properly
1332 Enable inlining of subroutines and functions by the @code{gcc}
1334 This works as it does for @code{gcc} itself---program units
1335 may be inlined for invocations that follow them in the same
1336 program unit, as long as the appropriate compile-time
1337 options are specified.
1340 Dummy arguments are no longer assumed to potentially alias
1342 other dummy arguments or @code{COMMON} areas when any of
1343 these are defined (assigned to) by Fortran code.
1345 This can result in faster and/or smaller programs when
1346 compiling with optimization enabled, though on some
1347 systems this effect is observed only when @samp{-fforce-addr}
1350 New options @samp{-falias-check}, @samp{-fargument-alias},
1351 @samp{-fargument-noalias},
1352 and @samp{-fno-argument-noalias-global} control the
1353 way @code{g77} handles potential aliasing.
1356 The @code{CONJG()} and @code{DCONJG()} intrinsics now
1357 are compiled in-line.
1360 The bug-fix for 0.5.19.1 has been re-done.
1361 The @code{g77} compiler has been changed back to
1362 assume @code{libf2c} has no aliasing problems in
1363 its implementations of the @code{COMPLEX} (and
1364 @code{DOUBLE COMPLEX}) intrinsics.
1365 The @code{libf2c} has been changed to have no such
1368 As a result, 0.5.20 is expected to offer improved performance
1369 over 0.5.19.1, perhaps as good as 0.5.19 in most
1370 or all cases, due to this change alone.
1372 @emph{Note:} This change requires version 0.5.20 of
1373 @code{libf2c}, at least, when linking code produced
1374 by any versions of @code{g77} other than 0.5.19.1.
1375 Use @samp{g77 -v} to determine the version numbers
1376 of the @code{libF77}, @code{libI77}, and @code{libU77}
1377 components of the @code{libf2c} library.
1378 (If these version numbers are not printed---in
1379 particular, if the linker complains about unresolved
1380 references to names like @samp{g77__fvers__}---that
1381 strongly suggests your installation has an obsolete
1382 version of @code{libf2c}.)
1385 New option @samp{-fugly-assign} specifies that the
1386 same memory locations are to be used to hold the
1387 values assigned by both statements @samp{I = 3} and
1388 @samp{ASSIGN 10 TO I}, for example.
1389 (Normally, @code{g77} uses a separate memory location
1390 to hold assigned statement labels.)
1393 @code{FORMAT} and @code{ENTRY} statements now are allowed to
1394 precede @code{IMPLICIT NONE} statements.
1397 Produce diagnostic for unsupported @code{SELECT CASE} on
1398 @code{CHARACTER} type, instead of crashing, at compile time.
1401 Fix crashes involving diagnosed or invalid code.
1404 Change approach to building @code{libf2c} archive
1405 (@file{libf2c.a}) so that members are added to it
1406 only when truly necessary, so the user that installs
1407 an already-built @code{g77} doesn't need to have write
1408 access to the build tree (whereas the user doing the
1409 build might not have access to install new software
1413 Support @code{gcc} version 2.7.2.2
1414 (modified by @code{g77} into version 2.7.2.2.f.2),
1416 support for prior versions of @code{gcc}.
1419 Upgrade to @code{libf2c} as of 1997-02-08, and
1420 fix up some of the build procedures.
1423 Improve general build procedures for @code{g77},
1424 fixing minor bugs (such as deletion of any file
1425 named @file{f771} in the parent directory of @code{gcc/}).
1428 Enable full support of @code{INTEGER*8} available in
1429 @code{libf2c} and @file{f2c.h} so that @code{f2c} users
1430 may make full use of its features via the @code{g77}
1431 version of @file{f2c.h} and the @code{INTEGER*8}
1432 support routines in the @code{g77} version of @code{libf2c}.
1435 Improve @code{g77} driver and @code{libf2c} so that @samp{g77 -v}
1436 yields version information on the library.
1439 The @code{SNGL} and @code{FLOAT} intrinsics now are
1440 specific intrinsics, instead of synonyms for the
1441 generic intrinsic @code{REAL}.
1444 New intrinsics have been added.
1445 These are @code{REALPART}, @code{IMAGPART},
1447 @code{LONG}, and @code{SHORT}.
1450 A new group of intrinsics, @samp{gnu}, has been added
1451 to contain the new @code{REALPART}, @code{IMAGPART},
1452 and @code{COMPLEX} intrinsics.
1453 An old group, @samp{dcp}, has been removed.
1456 Complain about industry-wide ambiguous references
1457 @samp{REAL(@var{expr})} and @samp{AIMAG(@var{expr})},
1458 where @var{expr} is @code{DOUBLE COMPLEX} (or any
1459 complex type other than @code{COMPLEX}), unless
1460 @samp{-ff90} option specifies Fortran 90 interpretation
1461 or new @samp{-fugly-complex} option, in conjunction with
1462 @samp{-fnot-f90}, specifies @code{f2c} interpretation.
1465 Make improvements to diagnostics.
1468 Speed up compiler a bit.
1471 Improvements to documentation and indexing, including
1472 a new chapter containing information on one, later
1473 more, diagnostics that users are directed to pull
1474 up automatically via a message in the diagnostic itself.
1476 (Hence the menu item @samp{M} for the node
1477 @samp{Diagnostics} in the top-level menu of
1478 the Info documentation.)
1481 @c 1997-02-01: 0.5.19.1 released.
1482 @heading In 0.5.19.1:
1485 Code-generation bugs afflicting operations on complex
1486 data have been fixed.
1488 These bugs occurred when assigning the result of an
1489 operation to a complex variable (or array element)
1490 that also served as an input to that operation.
1492 The operations affected by this bug were: @samp{CONJG()},
1493 @samp{DCONJG()}, @samp{CCOS()}, @samp{CDCOS()},
1494 @samp{CLOG()}, @samp{CDLOG()}, @samp{CSIN()}, @samp{CDSIN()},
1495 @samp{CSQRT()}, @samp{CDSQRT()}, complex division, and
1496 raising a @code{DOUBLE COMPLEX} operand to an @code{INTEGER}
1498 (The related generic and @samp{Z}-prefixed intrinsics,
1499 such as @samp{ZSIN()}, also were affected.)
1501 For example, @samp{C = CSQRT(C)}, @samp{Z = Z/C}, and @samp{Z = Z**I}
1502 (where @samp{C} is @code{COMPLEX} and @samp{Z} is
1503 @code{DOUBLE COMPLEX}) have been fixed.
1506 @c 1996-12-07: 0.5.19 released.
1510 Fix @code{FORMAT} statement parsing so negative values for
1511 specifiers such as @samp{P} (e.g. @samp{FORMAT(-1PF8.1)})
1512 are correctly processed as negative.
1515 Fix @code{SIGNAL} intrinsic so it once again accepts a
1516 procedure as its second argument.
1519 A temporary kludge option provides bare-bones information on
1520 @code{COMMON} and @code{EQUIVALENCE} members at debug time.
1523 New @samp{-fonetrip} option specifies FORTRAN-66-style
1524 one-trip @code{DO} loops.
1527 New @samp{-fno-silent} option causes names of program units
1528 to be printed as they are compiled, in a fashion similar to
1529 UNIX @code{f77} and @code{f2c}.
1532 New @samp{-fugly-assumed} option specifies that arrays
1533 dimensioned via @samp{DIMENSION X(1)}, for example, are to be
1534 treated as assumed-size.
1537 New @samp{-fno-typeless-boz} option specifies that non-decimal-radix
1538 constants using the prefixed-radix form (such as @samp{Z'1234'})
1539 are to be interpreted as @code{INTEGER} constants.
1542 New @samp{-ff66} option is a ``shorthand'' option that specifies
1543 behaviors considered appropriate for FORTRAN 66 programs.
1546 New @samp{-ff77} option is a ``shorthand'' option that specifies
1547 behaviors considered appropriate for UNIX @code{f77} programs.
1550 New @samp{-fugly-comma} and @samp{-fugly-logint} options provided
1551 to perform some of what @samp{-fugly} used to do.
1552 @samp{-fugly} and @samp{-fno-ugly} are now ``shorthand'' options,
1553 in that they do nothing more than enable (or disable) other
1554 @samp{-fugly-*} options.
1557 Fix parsing of assignment statements involving targets that
1558 are substrings of elements of @code{CHARACTER} arrays having
1559 names such as @samp{READ}, @samp{WRITE}, @samp{GOTO}, and
1560 @samp{REALFUNCTIONFOO}.
1563 Fix crashes involving diagnosed code.
1566 Fix handling of local @code{EQUIVALENCE} areas so certain cases
1567 of valid Fortran programs are not misdiagnosed as improperly
1568 extending the area backwards.
1571 Support @code{gcc} version 2.7.2.1.
1574 Upgrade to @code{libf2c} as of 1996-09-26, and
1575 fix up some of the build procedures.
1578 Change code generation for list-directed I/O so it allows
1579 for new versions of @code{libf2c} that might return non-zero
1580 status codes for some operations previously assumed to always
1583 This change not only affects how @code{IOSTAT=} variables
1584 are set by list-directed I/O, it also affects whether
1585 @code{END=} and @code{ERR=} labels are reached by these
1589 Add intrinsic support for new @code{FTELL} and @code{FSEEK}
1590 procedures in @code{libf2c}.
1593 Modify @code{fseek_()} in @code{libf2c} to be more portable
1594 (though, in practice, there might be no systems where this
1595 matters) and to catch invalid @samp{whence} arguments.
1598 Some useless warnings from the @samp{-Wunused} option have
1602 Fix a problem building the @file{f771} executable
1603 on AIX systems by linking with the @samp{-bbigtoc} option.
1606 Abort configuration if @code{gcc} has not been patched
1607 using the patch file provided in the @samp{gcc/f/gbe/}
1611 Add options @samp{--help} and @samp{--version} to the
1612 @code{g77} command, to conform to GNU coding guidelines.
1613 Also add printing of @code{g77} version number when
1614 the @samp{--verbose} (@samp{-v}) option is used.
1617 Change internally generated name for local @code{EQUIVALENCE}
1618 areas to one based on the alphabetically sorted first name
1619 in the list of names for entities placed at the beginning
1623 Improvements to documentation and indexing.
1626 @c 1996-04-01: 0.5.18 released.
1630 Add some rudimentary support for @code{INTEGER*1},
1631 @code{INTEGER*2}, @code{INTEGER*8},
1632 and their @code{LOGICAL} equivalents.
1633 (This support works on most, maybe all, @code{gcc} targets.)
1635 Thanks to Scott Snyder (@email{snyder@@d0sgif.fnal.gov})
1636 for providing the patch for this!
1638 Among the missing elements from the support for these
1639 features are full intrinsic support and constants.
1642 Add some rudimentary support for the @code{BYTE} and
1643 @code{WORD} type-declaration statements.
1644 @code{BYTE} corresponds to @code{INTEGER*1},
1645 while @code{WORD} corresponds to @code{INTEGER*2}.
1647 Thanks to Scott Snyder (@email{snyder@@d0sgif.fnal.gov})
1648 for providing the patch for this!
1651 The compiler code handling intrinsics has been largely
1652 rewritten to accommodate the new types.
1653 No new intrinsics or arguments for existing
1654 intrinsics have been added, so there is, at this
1655 point, no intrinsic to convert to @code{INTEGER*8},
1659 Support automatic arrays in procedures.
1662 Reduce space/time requirements for handling large
1663 @emph{sparsely} initialized aggregate arrays.
1664 This improvement applies to only a subset of
1665 the general problem to be addressed in 0.6.
1668 Treat initial values of zero as if they weren't
1669 specified (in DATA and type-declaration statements).
1670 The initial values will be set to zero anyway, but the amount
1671 of compile time processing them will be reduced,
1672 in some cases significantly (though, again, this
1673 is only a subset of the general problem to be
1676 A new option, @samp{-fzeros}, is introduced to
1677 enable the traditional treatment of zeros as any
1681 With @samp{-ff90} in force, @code{g77} incorrectly
1682 interpreted @samp{REAL(Z)} as returning a @code{REAL}
1683 result, instead of as a @code{DOUBLE PRECISION}
1685 (Here, @samp{Z} is @code{DOUBLE COMPLEX}.)
1687 With @samp{-fno-f90} in force, the interpretation remains
1688 unchanged, since this appears to be how at least some
1689 F77 code using the @code{DOUBLE COMPLEX} extension expected
1692 Essentially, @samp{REAL(Z)} in F90 is the same as
1693 @samp{DBLE(Z)}, while in extended F77, it appears to
1694 be the same as @samp{REAL(REAL(Z))}.
1697 An expression involving exponentiation, where both operands
1698 were type @code{INTEGER} and the right-hand operand
1699 was negative, was erroneously evaluated.
1702 Fix bugs involving @code{DATA} implied-@code{DO} constructs
1703 (these involved an errant diagnostic and a crash, both on good
1704 code, one involving subsequent statement-function definition).
1707 Close @code{INCLUDE} files after processing them, so compiling source
1708 files with lots of @code{INCLUDE} statements does not result in
1709 being unable to open @code{INCLUDE} files after all the available
1710 file descriptors are used up.
1713 Speed up compiling, especially of larger programs, and perhaps
1714 slightly reduce memory utilization while compiling (this is
1715 @emph{not} the improvement planned for 0.6 involving large aggregate
1716 areas)---these improvements result from simply turning
1717 off some low-level code to do self-checking that hasn't been
1718 triggered in a long time.
1721 Introduce three new options that
1722 implement optimizations in the @code{gcc} back end (GBE).
1723 These options are @samp{-fmove-all-movables}, @samp{-freduce-all-givs},
1724 and @samp{-frerun-loop-opt}, which are enabled, by default,
1725 for Fortran compilations.
1726 These optimizations are intended to help toon Fortran programs.
1729 Patch the GBE to do a better job optimizing certain
1730 kinds of references to array elements.
1733 Due to patches to the GBE, the version number of @code{gcc}
1734 also is patched to make it easier to manage installations,
1735 especially useful if it turns out a @code{g77} change to the
1738 The @code{g77}-modified version number is the @code{gcc}
1739 version number with the string @samp{.f.@var{n}} appended,
1740 where @samp{f} identifies the version as enhanced for
1741 Fortran, and @var{n} is @samp{1} for the first Fortran
1742 patch for that version of @code{gcc}, @samp{2} for the
1745 So, this introduces version 2.7.2.f.1 of @code{gcc}.
1748 Make several improvements and fixes to diagnostics, including
1749 the removal of two that were inappropriate or inadequate.
1752 Warning about two successive arithmetic operators, produced
1753 by @samp{-Wsurprising}, now produced @emph{only} when both
1754 operators are, indeed, arithmetic (not relational/boolean).
1757 @samp{-Wsurprising} now warns about the remaining cases
1758 of using non-integral variables for implied-@code{DO}
1759 loops, instead of these being rejected unless @samp{-fpedantic}
1760 or @samp{-fugly} specified.
1763 Allow @code{SAVE} of a local variable or array, even after
1764 it has been given an initial value via @code{DATA}, for example.
1767 Introduce an Info version of @code{g77} documentation, which
1768 supercedes @file{gcc/f/CREDITS}, @file{gcc/f/DOC}, and
1769 @file{gcc/f/PROJECTS}.
1770 These files will be removed in a future release.
1771 The files @file{gcc/f/BUGS}, @file{gcc/f/INSTALL}, and
1772 @file{gcc/f/NEWS} now are automatically built from
1773 the texinfo source when distributions are made.
1775 This effort was inspired by a first pass at translating
1776 @file{g77-0.5.16/f/DOC} that was contributed to Craig by
1777 David Ronis (@email{ronis@@onsager.chem.mcgill.ca}).
1780 New @samp{-fno-second-underscore} option to specify
1781 that, when @samp{-funderscoring} is in effect, a second
1782 underscore is not to be appended to Fortran names already
1783 containing an underscore.
1786 Change the way iterative @code{DO} loops work to follow
1788 In particular, calculation of the iteration count is
1789 still done by converting the start, end, and increment
1790 parameters to the type of the @code{DO} variable, but
1791 the result of the calculation is always converted to
1792 the default @code{INTEGER} type.
1794 (This should have no effect on existing code compiled
1795 by @code{g77}, but code written to assume that use
1796 of a @emph{wider} type for the @code{DO} variable
1797 will result in an iteration count being fully calculated
1798 using that wider type (wider
1799 than default @code{INTEGER}) must be rewritten.)
1802 Support @code{gcc} version 2.7.2.
1805 Upgrade to @code{libf2c} as of 1996-03-23, and
1806 fix up some of the build procedures.
1808 Note that the email addresses related to @code{f2c}
1809 have changed---the distribution site now is
1810 named @code{netlib.bell-labs.com}, and the
1811 maintainer's new address is @email{dmg@@bell-labs.com}.
1814 @c 1995-11-18: 0.5.17 released.
1818 @strong{Fix serious bug} in @samp{g77 -v} command that can cause removal of a
1819 system's @file{/dev/null} special file if run by user @samp{root}.
1821 @strong{All users} of version 0.5.16 should ensure that
1822 they have not removed @file{/dev/null} or replaced it with an ordinary
1823 file (e.g. by comparing the output of @samp{ls -l /dev/null} with
1824 @samp{ls -l /dev/zero}.
1825 If the output isn't basically the
1826 same, contact your system
1827 administrator about restoring @file{/dev/null} to its proper status).
1829 This bug is particularly insidious because removing @file{/dev/null} as
1830 a special file can go undetected for quite a while, aside from
1831 various applications and programs exhibiting sudden, strange
1834 I sincerely apologize for not realizing the
1835 implications of the fact that when @samp{g77 -v} runs the @code{ld} command
1836 with @samp{-o /dev/null} that @code{ld} tries to @emph{remove} the executable
1837 it is supposed to build (especially if it reports unresolved
1838 references, which it should in this case)!
1841 Fix crash on @samp{CHARACTER*(*) FOO} in a main or block data program unit.
1844 Fix crash that can occur when diagnostics given outside of any
1845 program unit (such as when input file contains @samp{@@foo}).
1848 Fix crashes, infinite loops (hangs), and such involving diagnosed code.
1851 Fix @code{ASSIGN}'ed variables so they can be @code{SAVE}'d or dummy arguments,
1852 and issue clearer error message in cases where target of @code{ASSIGN}
1853 or @code{ASSIGN}ed @code{GOTO}/@code{FORMAT} is too small (which should
1857 Make @code{libf2c} build procedures work on more systems again by
1858 eliminating unnecessary invocations of @samp{ld -r -x} and @samp{mv}.
1861 Fix omission of @samp{-funix-intrinsics-@dots{}} options in list of permitted
1862 options to compiler.
1865 Fix failure to always diagnose missing type declaration for
1866 @code{IMPLICIT NONE}.
1869 Fix compile-time performance problem (which could sometimes
1870 crash the compiler, cause a hang, or whatever, due to a bug
1871 in the back end) involving exponentiation with a large @code{INTEGER}
1872 constant for the right-hand operator (e.g. @samp{I**32767}).
1875 Fix build procedures so cross-compiling @code{g77} (the @code{fini}
1876 utility in particular) is properly built using the host compiler.
1879 Add new @samp{-Wsurprising} option to warn about constructs that are
1880 interpreted by the Fortran standard (and @code{g77}) in ways that
1881 are surprising to many programmers.
1884 Add @code{ERF()} and @code{ERFC()} as generic intrinsics mapping to existing
1885 @code{ERF}/@code{DERF} and @code{ERFC}/@code{DERFC} specific intrinsics.
1887 @emph{Note:} You should
1888 specify @samp{INTRINSIC ERF,ERFC} in any code where you might use
1889 these as generic intrinsics, to improve likelihood of diagnostics
1890 (instead of subtle run-time bugs) when using a compiler that
1891 doesn't support these as intrinsics (e.g. @code{f2c}).
1894 Remove from @samp{-fno-pedantic} the diagnostic about @code{DO}
1895 with non-@code{INTEGER} index variable; issue that under
1896 @samp{-Wsurprising} instead.
1899 Clarify some diagnostics that say things like ``ignored'' when that's
1903 Clarify diagnostic on use of @code{.EQ.}/@code{.NE.} on @code{LOGICAL}
1907 Minor improvements to code generation for various operations on
1908 @code{LOGICAL} operands.
1911 Minor improvement to code generation for some @code{DO} loops on some
1915 Support @code{gcc} version 2.7.1.
1918 Upgrade to @code{libf2c} as of 1995-11-15.
1921 @c 1995-08-30: 0.5.16 released.
1925 Fix a code-generation bug involving complicated @code{EQUIVALENCE} statements
1926 not involving @code{COMMON}.
1929 Fix code-generation bugs involving invoking ``gratis'' library procedures
1930 in @code{libf2c} from code compiled with @samp{-fno-f2c} by making these
1931 procedures known to @code{g77} as intrinsics (not affected by -fno-f2c).
1932 This is known to fix code invoking @code{ERF()}, @code{ERFC()},
1933 @code{DERF()}, and @code{DERFC()}.
1936 Update @code{libf2c} to include netlib patches through 1995-08-16, and
1937 @code{#define} @samp{WANT_LEAD_0} to 1 to make @code{g77}-compiled code more
1938 consistent with other Fortran implementations by outputting
1939 leading zeros in formatted and list-directed output.
1942 Fix a code-generation bug involving adjustable dummy arrays with high
1943 bounds whose primaries are changed during procedure execution, and
1944 which might well improve code-generation performance for such arrays
1945 compared to @code{f2c} plus @code{gcc} (but apparently only when using
1946 @file{gcc-2.7.0} or later).
1949 Fix a code-generation bug involving invocation of @code{COMPLEX} and
1950 @code{DOUBLE COMPLEX} @code{FUNCTION}s and doing @code{COMPLEX} and
1951 @code{DOUBLE COMPLEX} divides, when the result
1952 of the invocation or divide is assigned directly to a variable
1953 that overlaps one or more of the arguments to the invocation or divide.
1956 Fix crash by not generating new optimal code for @samp{X**I} if @samp{I} is
1957 nonconstant and the expression is used to dimension a dummy
1958 array, since the @code{gcc} back end does not support the necessary
1959 mechanics (and the @code{gcc} front end rejects the equivalent
1960 construct, as it turns out).
1963 Fix crash on expressions like @samp{COMPLEX**INTEGER}.
1966 Fix crash on expressions like @samp{(1D0,2D0)**2}, i.e. raising a
1967 @code{DOUBLE COMPLEX} constant to an @code{INTEGER} constant power.
1970 Fix crashes and such involving diagnosed code.
1973 Diagnose, instead of crashing on, statement function definitions
1974 having duplicate dummy argument names.
1977 Fix bug causing rejection of good code involving statement function
1981 Fix bug resulting in debugger not knowing size of local equivalence
1982 area when any member of area has initial value (via @code{DATA},
1986 Fix installation bug that prevented installation of @code{g77} driver.
1987 Provide for easy selection of whether to install copy of @code{g77}
1988 as @code{f77} to replace the broken code.
1991 Fix @code{gcc} driver (affects @code{g77} thereby) to not
1992 gratuitously invoke the
1993 @code{f771} program (e.g. when @samp{-E} is specified).
1996 Fix diagnostic to point to correct source line when it immediately
1997 follows an @code{INCLUDE} statement.
2000 Support more compiler options in @code{gcc}/@code{g77} when
2001 compiling Fortran files.
2002 These options include @samp{-p}, @samp{-pg}, @samp{-aux-info}, @samp{-P},
2003 correct setting of version-number macros for preprocessing, full
2004 recognition of @samp{-O0}, and
2005 automatic insertion of configuration-specific linker specs.
2008 Add new intrinsics that interface to existing routines in @code{libf2c}:
2009 @code{ABORT}, @code{DERF}, @code{DERFC}, @code{ERF}, @code{ERFC}, @code{EXIT},
2010 @code{FLUSH}, @code{GETARG}, @code{GETENV}, @code{IARGC},
2011 @code{SIGNAL}, and @code{SYSTEM}.
2012 Note that @code{ABORT}, @code{EXIT}, @code{FLUSH}, @code{SIGNAL}, and
2013 @code{SYSTEM} are intrinsic subroutines, not functions (since they
2014 have side effects), so to get the return values from @code{SIGNAL}
2015 and @code{SYSTEM}, append a final argument specifying an @code{INTEGER}
2016 variable or array element (e.g. @samp{CALL SYSTEM('rm foo',ISTAT)}).
2019 Add new intrinsic group named @samp{unix} to contain the new intrinsics,
2020 and by default enable this new group.
2023 Move @code{LOC()} intrinsic out of the @samp{vxt} group to the new
2027 Improve @code{g77} so that @samp{g77 -v} by itself (or with
2028 certain other options, including @samp{-B}, @samp{-b}, @samp{-i},
2029 @samp{-nostdlib}, and @samp{-V}) reports lots more useful
2030 version info, and so that long-form options @code{gcc} accepts are
2031 understood by @code{g77} as well (even in truncated, unambiguous forms).
2034 Add new @code{g77} option @samp{--driver=name} to specify driver when
2035 default, @code{gcc}, isn't appropriate.
2038 Add support for @samp{#} directives (as output by the preprocessor) in the
2039 compiler, and enable generation of those directives by the
2040 preprocessor (when compiling @samp{.F} files) so diagnostics and debugging
2041 info are more useful to users of the preprocessor.
2044 Produce better diagnostics, more like @code{gcc}, with info such as
2045 @samp{In function `foo':} and @samp{In file included from...:}.
2048 Support @code{gcc}'s @samp{-fident} and @samp{-fno-ident} options.
2051 When @samp{-Wunused} in effect, don't warn about local variables used as
2052 statement-function dummy arguments or @code{DATA} implied-@code{DO} iteration
2053 variables, even though, strictly speaking, these are not uses
2054 of the variables themselves.
2057 When @samp{-W -Wunused} in effect, don't warn about unused dummy arguments
2058 at all, since there's no way to turn this off for individual
2059 cases (@code{g77} might someday start warning about these)---applies
2060 to @code{gcc} versions 2.7.0 and later, since earlier versions didn't
2061 warn about unused dummy arguments.
2064 New option @samp{-fno-underscoring} that inhibits transformation of names
2065 (by appending one or two underscores) so users may experiment
2066 with implications of such an environment.
2069 Minor improvement to @file{gcc/f/info} module to make it easier to build
2070 @code{g77} using the native (non-@code{gcc}) compiler on certain machines
2071 (but definitely not all machines nor all non-@code{gcc} compilers).
2073 do not report bugs showing problems compilers have with
2074 macros defined in @file{gcc/f/target.h} and used in places like
2075 @file{gcc/f/expr.c}.
2078 Add warning to be printed for each invocation of the compiler
2079 if the target machine @code{INTEGER}, @code{REAL}, or @code{LOGICAL} size
2081 since @code{g77} is known to not work well for such cases (to be
2082 fixed in Version 0.6---@pxref{Actual Bugs,,Actual Bugs We Haven't Fixed Yet}).
2085 Lots of new documentation (though work is still needed to put it into
2086 canonical GNU format).
2089 Build @code{libf2c} with @samp{-g0}, not @samp{-g2}, in effect
2090 (by default), to produce
2091 smaller library without lots of debugging clutter.
2094 @c 1995-05-19: 0.5.15 released.
2098 Fix bad code generation involving @samp{X**I} and temporary, internal variables
2099 generated by @code{g77} and the back end (such as for @code{DO} loops).
2102 Fix crash given @samp{CHARACTER A;DATA A/.TRUE./}.
2105 Replace crash with diagnostic given @samp{CHARACTER A;DATA A/1.0/}.
2108 Fix crash or other erratic behavior when null character constant
2109 (@samp{''}) is encountered.
2112 Fix crash or other erratic behavior involving diagnosed code.
2115 Fix code generation for external functions returning type @code{REAL} when
2116 the @samp{-ff2c} option is in force (which it is by default) so that
2117 @code{f2c} compatibility is indeed provided.
2120 Disallow @samp{COMMON I(10)} if @samp{I} has previously been specified
2121 with an array declarator.
2124 New @samp{-ffixed-line-length-@var{n}} option, where @var{n} is the
2126 of a typical fixed-form line, defaulting to 72 columns, such
2127 that characters beyond column @var{n} are ignored, or @var{n} is @samp{none},
2128 meaning no characters are ignored.
2129 does not affect lines
2130 with @samp{&} in column 1, which are always processed as if
2131 @samp{-ffixed-line-length-none} was in effect.
2134 No longer generate better code for some kinds of array references,
2135 as @code{gcc} back end is to be fixed to do this even better, and it
2136 turned out to slow down some code in some cases after all.
2139 In @code{COMMON} and @code{EQUIVALENCE} areas with any members given initial
2140 values (e.g. via @code{DATA}), uninitialized members now always
2141 initialized to binary zeros (though this is not required by
2142 the standard, and might not be done in future versions
2144 Previously, in some @code{COMMON}/@code{EQUIVALENCE} areas
2145 (essentially those with members of more than one type), the
2146 uninitialized members were initialized to spaces, to
2147 cater to @code{CHARACTER} types, but it seems no existing code expects
2148 that, while much existing code expects binary zeros.
2154 Don't emit bad code when low bound of adjustable array is nonconstant
2155 and thus might vary as an expression at run time.
2158 Emit correct code for calculation of number of trips in @code{DO} loops
2160 where the loop should not execute at all.
2161 (This bug affected cases
2162 where the difference between the begin and end values was less
2163 than the step count, though probably not for floating-point cases.)
2166 Fix crash when extra parentheses surround item in
2167 @code{DATA} implied-@code{DO} list.
2170 Fix crash over minor internal inconsistencies in handling diagnostics,
2171 just substitute dummy strings where necessary.
2174 Fix crash on some systems when compiling call to @code{MVBITS()} intrinsic.
2177 Fix crash on array assignment @samp{TYPE@var{ddd}(@dots{})=@dots{}}, where @var{ddd}
2178 is a string of one or more digits.
2181 Fix crash on @code{DCMPLX()} with a single @code{INTEGER} argument.
2184 Fix various crashes involving code with diagnosed errors.
2187 Support @samp{-I} option for @code{INCLUDE} statement, plus @code{gcc}'s
2188 @file{header.gcc} facility for handling systems like MS-DOS.
2191 Allow @code{INCLUDE} statement to be continued across multiple lines,
2192 even allow it to coexist with other statements on the same line.
2195 Incorporate Bellcore fixes to @code{libf2c} through 1995-03-15---this
2196 fixes a bug involving infinite loops reading EOF with empty list-directed
2200 Remove all the @code{g77}-specific auto-configuration scripts, code,
2202 except for temporary substitutes for bsearch() and strtoul(), as
2203 too many configure/build problems were reported in these areas.
2204 People will have to fix their systems' problems themselves, or at
2205 least somewhere other than @code{g77}, which expects a working ANSI C
2206 environment (and, for now, a GNU C compiler to compile @code{g77} itself).
2209 Complain if initialized common redeclared as larger in subsequent program
2213 Warn if blank common initialized, since its size can vary and hence
2214 related warnings that might be helpful won't be seen.
2217 New @samp{-fbackslash} option, on by default, that causes @samp{\}
2218 within @code{CHARACTER}
2219 and Hollerith constants to be interpreted a la GNU C.
2221 this behavior is somewhat different from @code{f2c}'s, which supports only
2222 a limited subset of backslash (escape) sequences.
2225 Make @samp{-fugly-args} the default.
2228 New @samp{-fugly-init} option, on by default, that allows typeless/Hollerith
2229 to be specified as initial values for variables or named constants
2230 (@code{PARAMETER}), and also allows character<->numeric conversion in
2231 those contexts---turn off via @samp{-fno-ugly-init}.
2234 New @samp{-finit-local-zero} option to initialize
2235 local variables to binary zeros.
2236 This does not affect whether they are @code{SAVE}d, i.e. made
2237 automatic or static.
2240 New @samp{-Wimplicit} option to warn about implicitly typed variables, arrays,
2242 (Basically causes all program units to default to @code{IMPLICIT NONE}.)
2245 @samp{-Wall} now implies @samp{-Wuninitialized} as with @code{gcc}
2246 (i.e. unless @samp{-O} not specified, since @samp{-Wuninitialized}
2247 requires @samp{-O}), and implies @samp{-Wunused} as well.
2250 @samp{-Wunused} no longer gives spurious messages for unused
2251 @code{EXTERNAL} names (since they are assumed to refer to block data
2252 program units, to make use of libraries more reliable).
2255 Support @code{%LOC()} and @code{LOC()} of character arguments.
2258 Support null (zero-length) character constants and expressions.
2261 Support @code{f2c}'s @code{IMAG()} generic intrinsic.
2264 Support @code{ICHAR()}, @code{IACHAR()}, and @code{LEN()} of
2265 character expressions that are valid in assignments but
2266 not normally as actual arguments.
2269 Support @code{f2c}-style @samp{&} in column 1 to mean continuation line.
2272 Allow @code{NAMELIST}, @code{EXTERNAL}, @code{INTRINSIC}, and @code{VOLATILE}
2273 in @code{BLOCK DATA}, even though these are not allowed by the standard.
2276 Allow @code{RETURN} in main program unit.
2279 Changes to Hollerith-constant support to obey Appendix C of the
2284 Now padded on the right with zeros, not spaces.
2287 Hollerith ``format specifications'' in the form of arrays of
2288 non-character allowed.
2291 Warnings issued when non-space truncation occurs when converting
2295 When specified as actual argument, now passed
2296 by reference to @code{INTEGER} (padded on right with spaces if constant
2297 too small, otherwise fully intact if constant wider the @code{INTEGER}
2298 type) instead of by value.
2301 @strong{Warning:} @code{f2c} differs on the
2302 interpretation of @samp{CALL FOO(1HX)}, which it treats exactly the
2303 same as @samp{CALL FOO('X')}, but which the standard and @code{g77} treat
2304 as @samp{CALL FOO(%REF('X '))} (padded with as many spaces as necessary
2305 to widen to @code{INTEGER}), essentially.
2308 Changes and fixes to typeless-constant support:
2312 Now treated as a typeless double-length @code{INTEGER} value.
2315 Warnings issued when overflow occurs.
2318 Padded on the left with zeros when converting
2322 Should be properly aligned and ordered on
2323 the target machine for whatever type it is turned into.
2326 When specified as actual argument, now passed as reference to
2327 a default @code{INTEGER} constant.
2331 @code{%DESCR()} of a non-@code{CHARACTER} expression now passes a pointer to
2332 the expression plus a length for the expression just as if
2333 it were a @code{CHARACTER} expression.
2334 For example, @samp{CALL FOO(%DESCR(D))}, where
2335 @samp{D} is @code{REAL*8}, is the same as @samp{CALL FOO(D,%VAL(8)))}.
2338 Name of multi-entrypoint master function changed to incorporate
2339 the name of the primary entry point instead of a decimal
2340 value, so the name of the master function for @samp{SUBROUTINE X}
2341 with alternate entry points is now @samp{__g77_masterfun_x}.
2344 Remove redundant message about zero-step-count @code{DO} loops.
2347 Clean up diagnostic messages, shortening many of them.
2350 Fix typo in @code{g77} man page.
2353 Clarify implications of constant-handling bugs in @file{f/BUGS}.
2356 Generate better code for @samp{**} operator with a right-hand operand of
2357 type @code{INTEGER}.
2360 Generate better code for @code{SQRT()} and @code{DSQRT()},
2361 also when @samp{-ffast-math}
2362 specified, enable better code generation for @code{SIN()} and @code{COS()}.
2365 Generate better code for some kinds of array references.
2368 Speed up lexing somewhat (this makes the compilation phase noticeably