| blob | 8678e79d3a9177792458a51ad3b57e2f5811de5a |
1 Sat Jun 13 03:13:18 1998 Craig Burley <burley@gnu.org>
3 * g77spec.c (lang_specific_driver): Print out egcs
4 version info first, to be compatible with what some
5 test facilities expect.
7 Wed Jun 10 13:17:32 1998 Dave Brolley <brolley@cygnus.com>
9 * top.h (ffe_decode_option): New argc/argv interface.
10 * top.c (ffe_decode_option): New argc/argv interface.
11 * parse.c (yyparse): New argc/argv interface for ffe_decode_option.
12 * com.c (lang_decode_option): New argc/argv interface.
14 Mon Jun 1 19:37:42 1998 Craig Burley <burley@gnu.org>
16 * com.c (ffecom_init_0): Fix setup of INTEGER(KIND=7)
17 pointer type.
18 * info.c (ffeinfo_type): Don't crash on null type.
19 * expr.c (ffeexpr_fulfill_call_): Don't special-case
20 %LOC(expr) or LOC(expr).
21 Delete FFEGLOBAL_argsummaryPTR.
22 * global.c, global.h: Delete FFEGLOBAL_argsummaryPTR.
24 Thu May 28 21:32:18 1998 Craig Burley <burley@gnu.org>
26 Restore circa-0.5.22 capabilities of `g77' driver:
27 * Make-lang.in (g77spec.o): Depend on f/version.h.
28 (g77version.o): New rule to compile g77 version info.
29 (g77$(exeext)): Depend on and link in g77version.o.
30 * g77spec.c: Rewrite to be more like 0.5.22 version
31 of g77.c, making filtering of command line smarter
32 so mixed Fortran and C (etc.) can be compiled, verbose
33 version info can be obtained, etc.
34 * lang-specs.h (f77-version): New "language" to support
35 "g77 -v" command under new gcc 2.8 regime.
36 * lex.c (ffelex_file_fixed): If -fnull-version, just
37 substitute a "source file" that prints out version info.
38 * top.c, top.h: Support -fnull-version.
40 * lang-specs.h: Use "%O" instead of OO macro to specify
41 object extension. Remove old stringizing cruft.
43 * Make-lang.in (g77.c, g77spec.o, g77.o, g77$(exeext),
44 g77-cross$(exeext), f771,
45 $(srcdir)/f/g77.info, $(srcdir)/f/g77.dvi,
46 $(srcdir)/f/intdoc.texi,
47 f77.install-common, f77.install-info, f77.install-man,
48 f77.uninstall, $(G77STAGESTUFF), f77.stage1, f77.stage2,
49 f77.stage3, f77.stage4, f77.distdir): Don't do anything
50 unless user specified "f77" or "F77" in $LANGUAGES either
51 during configuration or explicitly. For convenience of
52 various tests and to work around lack of the assignment
53 "LANGUAGES=$(BOOT_LANGUAGES)" in the "make stage1" command
54 of "make bootstrap" in gcc, use a touch file named "lang-f77"
55 to communicate whether this is the case.
57 * Make-lang.in (F77_FLAGS_TO_PASS): Delete this macro,
58 replace with minimal expansion of its former self in
59 each of the two instances where it was used.
61 * Makefile.in (HOST_CC): Delete this definition.
63 * com.c (index, rindex): Delete these declarations.
65 * proj.h: (isascii): Delete this.
67 * Make-lang.in (f77.install-common): Warn if `f77-install-ok'
68 flag-file exists, since it no longer triggers any activity.
70 Rename libf2c.a and f2c.h to libg2c.a and g2c.h,
71 normalize and simplify g77/libg2c build process:
72 * Make-lang.in: Remove all support for overwriting
73 /usr/bin/f77 etc., or whatever the actual names are
74 via $(prefix) and $(local_prefix). (g++ overwrites
75 /usr/bin/c++, but then it's often the only C++ compiler
76 on the system; f77 often exists on systems that are
77 installing g77.)
78 (f77.realclean): Remove obsolete target.
79 (g77.c, g77$(exeext)): Minor changes to look more like g++'s
80 stuff.
81 (f771): Now built with srcdir=gcc/f, not srcdir=gcc, to be
82 more like g++ and such.
83 (f/Makefile): Removed, as g++ doesn't need this rule.
84 (f77.install-common): No longer install f77, etc.
85 (f77.install-man): No longer install f77.1.
86 (f77.uninstall): No longer uninstall f77, f77.1, etc.
87 (f77.stage1, f77.stage2, f77.stage3, f77.stage4): Do work
88 only if "f77" appears in $(LANGUAGES).
89 (Note: gcc's Makefile.in's bootstrap target should set
90 LANGUAGES=$(BOOT_LANGUAGES) when making the stage1 target.)
91 * Makefile.in: Update vis-a-vis gcc/cp/Makefile.in.
92 (none): Remove.
93 (g77-only): Relocate.
94 (all.indirect, f771, *.o): Now assumes current directory
95 is this dir (gcc/f), not the parent directory.
96 (TAGS): Remove "echo 'parse.y,0' >> TAGS ;" line.
97 * config-lang.in: Delete commented-out code.
98 Fix stagestuff definition. Add more stuff to
99 diff_excludes definition. Don't create any directories.
100 Set outputs to f/Makefile, to get variable substition
101 to happen (what does that really do, anyway?!).
102 * g77spec.c: Rename libf2c to libg2c.
104 * com.h: Remove all of the gcc back-end decls,
105 since egcs should have all of them correct.
107 * com.c: Include "proj.h" before anything else,
108 as that's how things are supposed to work.
109 * ste.c: Ditto.
111 * bad.c: Include "flags.j" here, since some diagnostics
112 check flag_pedantic_errors.
114 * Makefile.in (f/*.o): Rebuild dependencies via
115 deps-kinda.
117 * output.j: New source file.
118 * Make-lang.in (F77_SRCS): Update accordingly.
119 * Makefile.in (OUTPUT_H): Ditto.
120 (deps-kinda): Ditto.
121 * com.c: Include "output.j" here.
122 * lex.c: Ditto.
124 Mon May 25 03:34:42 1998 Craig Burley <burley@gnu.org>
126 * com.c (ffecom_expr_): Fix D**I and Z**I cases to
127 not convert (DOUBLE PRECISION) D and (DOUBLE COMPLEX) Z
128 to INTEGER. (This is dead code here anyway.)
130 Sat May 23 06:32:52 1998 Craig Burley <burley@gnu.org>
132 * com.c (ffecom_finish_symbol_transform_): Don't transform
133 statement (nested) functions, to avoid gcc compiling them
134 and thus producing linker errors if they refer to undefined
135 external functions. But warn if they're unused and -Wunused.
136 * bad.def (FFEBAD_SFUNC_UNUSED): New diagnostic.
138 Tue May 19 14:52:41 1998 Craig Burley <burley@gnu.org>
140 * bad.def (FFEBAD_OPEN_UNSUPPORTED, FFEBAD_INQUIRE_UNSUPPORTED,
141 FFEBAD_READ_UNSUPPORTED, FFEBAD_WRITE_UNSUPPORTED,
142 FFEBAD_QUAD_UNSUPPORTED, FFEBAD_BLOCKDATA_STMT,
143 FFEBAD_TRUNCATING_CHARACTER, FFEBAD_TRUNCATING_HOLLERITH,
144 FFEBAD_TRUNCATING_NUMERIC, FFEBAD_TRUNCATING_TYPELESS,
145 FFEBAD_TYPELESS_OVERFLOW): Change these from warnings
146 to errors.
148 Tue May 19 14:51:59 1998 Craig Burley <burley@gnu.org>
150 * Make-lang.in (f77.install-info, f77.uninstall):
151 Use install-info as appropriate.
153 Tue May 19 12:56:54 1998 Craig Burley <burley@gnu.org>
155 * com.c (ffecom_init_0): Rename xargc to f__xargc,
156 in accord with same-dated change to f/runtime.
158 Fri May 15 10:52:49 1998 Craig Burley <burley@gnu.org>
160 * com.c (ffecom_convert_narrow_, ffecom_convert_widen_):
161 Be even more persnickety in checking for internal bugs.
162 Also, if precision isn't changing, just return the expr.
164 * expr.c (ffeexpr_token_number_): Call
165 ffeexpr_make_float_const_ to make an integer.
166 (ffeexpr_make_float_const_): Handle making an integer.
168 * intrin.c (ffeintrin_init_0): Distinguish between
169 crashes on bad arg base and kind types.
171 Thu May 14 13:30:59 1998 Craig Burley <burley@gnu.org>
173 * Make-lang.in (f/expr.c): Now depends on f/stamp-str.
174 * expr.c: Use ffestrOther in place of ffeexprDotdot_.
175 * str-ot.fin: Add more keywords for expr.c.
177 * intdoc.c (dumpimp): Trivial fix.
179 * com.c (ffecom_expr_): Add ltkt variable for clarity.
181 Wed May 13 13:05:34 1998 Craig Burley <burley@gnu.org>
183 * Make-lang.in (G77STAGESTUFF): Add g77.o, g77spec.o,
184 and g77version.o.
185 (f77.clean): Add removal of g77.c, g77.o, g77spec.o,
186 and g77version.o.
187 (f77.distclean): Delete removal of g77.c.
189 Tue Apr 28 09:23:10 1998 Craig Burley <burley@gnu.org>
191 Fix 980427-0.f:
192 * global.c (ffeglobal_ref_progunit_): When transitioning
193 from EXT to FUNC, discard hook, since the decl, if any, is
194 probably wrong.
196 Sun Apr 26 09:05:50 1998 Craig Burley <burley@gnu.org>
198 * com.c (ffecom_char_enhance_arg_): Wrap the upper bound
199 (the PARM_DECL specifying the length of the CHARACTER*(*)
200 dummy arg) in a variable_size invocation, to prevent
201 dwarf2out.c crashing when compiling code with -g.
203 Sat Apr 18 05:03:21 1998 Craig Burley <burley@gnu.org>
205 * com.c (ffecom_check_size_overflow_): Ignore overflow
206 as well if dummy argument.
208 Fri Apr 17 17:18:04 1998 Craig Burley <burley@gnu.org>
210 * version.h: Get rid of the overly large headers
211 here too, as done in version.c.
213 Tue Apr 14 14:40:40 1998 Craig Burley <burley@gnu.org>
215 * com.c (ffecom_start_progunit_): Mark function decl
216 as used, to avoid spurious warning (-Wunused) for ENTRY.
218 Tue Apr 14 14:19:34 1998 Craig Burley <burley@gnu.org>
220 * sta.c (ffesta_second_): Check for CASE DEFAULT
221 as well as CASE, or it won't be recognized.
223 Mon Mar 23 21:20:35 1998 Craig Burley <burley@gnu.org>
225 * version.c: Reduce to a one-line file, like
226 gcc's version.c, since there's really no content
227 there.
229 Mon Mar 23 11:58:43 1998 Craig Burley <burley@gnu.org>
231 * bugs.texi: Various updates.
233 * com.c (ffecom_tree_canonize_ptr_): Fix up spacing a bit.
235 Mon Mar 16 21:20:35 1998 Craig Burley <burley@gnu.org>
237 * expr.c (ffeexpr_sym_impdoitem_): Don't blindly
238 reset symbol info after calling ffesymbol_error,
239 to avoid crash.
241 Thu Mar 12 13:50:21 1998 Craig Burley <burley@gnu.org>
243 * lang-options.h: Sort all -f*-intrinsics-* options,
244 for consistency with other g77 versions.
246 Mon Mar 16 15:38:50 1998 Craig Burley <burley@gnu.org>
248 * Version 0.5.22 released.
250 Mon Mar 16 14:36:02 1998 Craig Burley <burley@gnu.org>
252 Make -g work better for ENTRY:
253 * com.c (ffecom_start_progunit_): Master function
254 for ENTRY-laden procedure is not really invented,
255 so it can be debugged.
256 (ffecom_do_entry_): Push/set/pop lineno for each
257 entry point.
259 Sun Mar 15 05:48:49 1998 Craig Burley <burley@gnu.org>
261 * intrin.def: Fix spelling of mixed-case form
262 of `CPU_Time' (was `Cpu_Time').
264 1998-03-09 Dave Love <d.love@dl.ac.uk>
266 * Make-lang.in: Set CONFIG_SITE to a non-existent file since
267 /dev/null loses with bash 2.0/autoconf 2.12. Put
268 F77_FLAGS_TO_PASS before CC.
270 Sun Mar 8 16:35:34 1998 Craig Burley <burley@gnu.org>
272 * intrin.def: Use tabs instead of blanks more
273 consistently (excepting DEFGEN section for now).
275 Sat Feb 28 15:24:38 1998 Craig Burley <burley@gnu.org>
277 * intrin.def: Make CPU_TIME's arg generic real to be just
278 like SECOND_subr.
280 Fri Feb 20 12:45:53 1998 Craig Burley <burley@gnu.org>
282 * expr.c (ffeexpr_token_arguments_): Make sure
283 outer exprstack isn't null.
285 1998-02-16 Dave Love <d.love@dl.ac.uk>
287 * Makefile.in (f/fini): Don't use -W -Wall with HOST_CC.
289 Sun Jan 11 02:14:47 1998 Craig Burley <burley@gnu.org>
291 Support FORMAT(I<1+2>) (constant variable-FORMAT
292 expressions):
293 * bad.def (FFEBAD_FORMAT_VARIABLE): New diagnostic.
294 * std.c (ffestd_R1001rtexpr_): New function.
295 (ffestd_R1001dump_, ffestd_R1001dump_1005_1_,
296 ffestd_R1001dump_1005_2_, ffestd_R1001dump_1005_3_,
297 ffestd_R1001dump_1005_4_, ffestd_R1001dump_1005_5_,
298 ffestd_R1001dump_1010_2_, ffestd_R1001dump_1010_3_,
299 ffestd_R1001dump_1010_4_, ffestd_R1001dump_1010_5_):
300 Use new function instead of ffestd_R1001error_.
302 * stb.c (ffestb_R10014_, ffestb_R10016_, ffestb_R10018_,
303 ffestb_R100110_): Restructure `for' loop for style.
305 Fix 970626-2.f by not doing most back-end processing
306 when current_function_decl is an ERROR_MARK, and by
307 making that the case when its type would be an ERROR_MARK:
308 * com.c (ffecom_start_progunit_, finish_function,
309 lang_printable_name, start_function,
310 ffecom_finish_symbol_transform_): Test for ERROR_MARK.
311 * std.c (ffestd_stmt_pass_): Don't do any downstream
312 processing if ERROR_MARK.
314 * Make-lang.in (f77.install-common): Don't install, and
315 don't uninstall existing, Info files if f/g77.info
316 doesn't exit. (This is a somewhat modified version
317 of an egcs patch on 1998-01-07 12:05:51 by Bruno Haible
318 <bruno@linuix.mathematik.uni-karlsruhe.de>.)
320 Fri Jan 9 19:09:07 1998 Craig Burley <burley@gnu.org>
322 Fix -fpedantic combined with `F()' invocation,
323 also -fugly-comma combined with `IARGC()' invocation:
324 * bad.def (FFEBAD_NULL_ARGUMENT_W): New diagnostic.
325 * expr.c (ffeexpr_finished_): Don't reject null expressions
326 in the argument-expression context -- let outer context
327 handle that.
328 (ffeexpr_token_arguments_): Warn about null expressions
329 here if -fpedantic (as appropriate).
330 Obey -fugly-comma for only external-procedure invocations.
331 * intrin.c (ffeintrin_check_): No longer ignore explicit
332 omitted trailing args.
334 Tue Dec 23 14:58:04 1997 Craig Burley <burley@gnu.org>
336 * intrin.c (ffeintrin_fulfill_generic): Don't generate
337 FFEBAD_INTRINSIC_TYPE for CHARACTER*(*) intrinsic.
339 * com.c (ffecom_gfrt_basictype):
340 (ffecom_gfrt_kindtype):
341 (ffecom_make_gfrt_):
342 (FFECOM_rttypeVOIDSTAR_): New return type `void *', for
343 the SIGNAL intrinsic.
344 * com-rt.def (FFECOM_rttypeSIGNAL): Now returns `void *'.
345 * intdoc.c: Replace `p' kind specifier with `7'.
346 * intrin.c (ffeintrin_check_, ffeintrin_init_0): Replace
347 `p' kind specifier with `7'.
348 * intrin.def (FFEINTRIN_impLOC, FFEINTRIN_impSIGNAL_func,
349 FFEINTRIN_impSIGNAL_subr): Replace `p' specifier with `7'.
350 Also, SIGNAL now returns a `void *' status, not `int'.
352 Improve run-time diagnostic for "PRINT '(I1', 42":
353 * com.c (ffecom_char_args_x_): Renamed from ffecom_char_args_,
354 which is now a macro (to avoid lots of changes to other code)
355 with new arg, ffecom_char_args_with_null_ being another new
356 macro to call same function with different value for new arg.
357 This function now appends a null byte to opCONTER expression
358 if the new arg is TRUE.
359 (ffecom_arg_ptr_to_expr): Support NULL length pointer.
360 * ste.c (ffeste_io_cilist_):
361 (ffeste_io_icilist_): Pass NULL length ptr for
362 FORMAT expression, so null byte gets appended where
363 feasible.
364 * target.c (ffetarget_character1):
365 (ffetarget_concatenate_character1):
366 (ffetarget_substr_character1):
367 (ffetarget_convert_character1_character1):
368 (ffetarget_convert_character1_hollerith):
369 (ffetarget_convert_character1_integer4):
370 (ffetarget_convert_character1_logical4):
371 (ffetarget_convert_character1_typeless):
372 (ffetarget_hollerith): Append extra phantom null byte as
373 part of FFETARGET-NULL-BYTE kludge.
375 * intrin.def (FFEINTRIN_impCPU_TIME): Point to
376 FFECOM_gfrtSECOND as primary run-time routine.
378 Mon Dec 22 12:41:07 1997 Craig Burley <burley@gnu.org>
380 * intrin.c (ffeintrin_init_0): Remove duplicate
381 check for `!'.
383 Sun Dec 14 02:49:58 1997 Craig Burley <burley@gnu.org>
385 * intrin.c (ffeintrin_init_0): Fix up indentation a bit.
386 Fix bug that prevented checking of arguments other
387 than the first.
389 * intdoc.c: Fix up indentation a bit.
391 Tue Dec 9 16:20:57 1997 Richard Henderson <rth@cygnus.com>
393 * com.c (ffecom_type_vardesc_): Vardesc.dims is a `ftnlen*'.
395 Mon Dec 1 19:12:36 1997 Craig Burley <burley@gnu.org>
397 * intrin.c (ffeintrin_check_): Fix up indentation a bit more.
399 Mon Dec 1 16:21:08 1997 Craig Burley <burley@gnu.org>
401 * com.c (ffecom_arglist_expr_): Crash if non-supplied
402 optional arg isn't passed as an address.
403 Pass null pointer explicitly, instead of via ffecom routine.
404 If incoming argstring is NULL, substitute pointer to "0".
405 Recognize '0' as ending the usual arg stuff, just like '\0'.
407 Sun Nov 30 22:22:22 1997 Craig Burley <burley@gnu.org>
409 * intdoc.c: Minor fix-ups.
411 * intrin.c (ffeintrin_check_): Fix up indentation a bit.
413 * intrin.def: Fix up spacing a bit.
415 1997-11-17 Dave Love <d.love@dl.ac.uk>
417 * com.c (ffecom_arglist_expr_): Pass null pointers for optional
418 args which aren't supplied.
420 Sun Nov 16 21:45:43 1997 H.J. Lu (hjl@gnu.ai.mit.edu)
422 * Make-lang.in (f77.install-info): Depend on f77.info.
424 1997-11-06 Dave Love <d.love@dl.ac.uk>
426 * intrin.def: Allow non-integer args for INT2 and INT8 (per
427 documentation).
429 Tue Oct 28 02:21:25 1997 Craig Burley <burley@gnu.ai.mit.edu>
431 * lang-options.h: Add -fgnu-intrinsics-* and
432 -fbadu77-intrinsics-* options.
434 Sun Oct 26 02:36:21 1997 Craig Burley <burley@gnu.ai.mit.edu>
436 * com.c (lang_print_error_function): Fix to more
437 reliably notice when the diagnosed region changes.
439 Sat Oct 25 23:43:36 1997 Craig Burley <burley@gnu.ai.mit.edu>
441 Fix 950327-0.f:
442 * sta.c, sta.h (ffesta_outpooldisp): New function.
443 * std.c (ffestd_stmt_pass_): Don't kill NULL pool.
444 (ffestd_R842): If pool already preserved, save NULL
445 for pool, because it should be killed only once.
447 * malloc.c [MALLOC_DEBUG]: Put initializer for `name'
448 component in braces, to avoid compiler warning.
450 Fri Oct 10 13:00:48 1997 Craig Burley <burley@gnu.ai.mit.edu>
452 * ste.c (ffeste_begin_iterdo_): Fix loop setup so iteration
453 variable is modified only after the #iterations is calculated;
454 otherwise if the iteration variable is aliased to any of the
455 operands in the start, end, or increment expressions, the
456 wrong #iterations might be calculated.
458 * com.c (ffecom_save_tree): Fix indentation.
460 1997-10-05 Dave Love <d.love@dl.ac.uk>
462 * intrin.def: Make SECOND_subr's arg generic real for people
463 porting from Cray and making everything double precision.
465 Mon Sep 29 16:18:21 1997 Craig Burley <burley@gnu.ai.mit.edu>
467 * stu.c (ffestu_list_exec_transition_,
468 ffestu_dummies_transition_): Specify `bool' type for
469 `in_progress' variables.
471 * com.h (assemble_string): Declare this routine (instead
472 of #include'ing "output.h" from gcc) to eliminate warnings
473 from lex.c.
475 Fri Sep 19 01:12:27 1997 Craig Burley <burley@gnu.ai.mit.edu>
477 * expr.c (ffeexpr_reduced_eqop2_):
478 (ffeexpr_reduced_relop2_): Minor fixes to diagnostic code.
480 * fini.c (main): Change return type to `int'.
482 Wed Sep 17 10:47:08 1997 Craig Burley <burley@gnu.ai.mit.edu>
484 * com-rt.def (FFECOM_gfrtDSIGN, FFECOM_gfrtISIGN,
485 FFECOM_gfrtSIGN): Add second argument.
487 * expr.c (ffeexpr_cb_comma_c_): Trivial fixes.
489 Tue Sep 9 01:59:35 1997 Craig Burley <burley@gnu.ai.mit.edu>
491 * Version 0.5.21 released.
493 Tue Sep 9 00:31:01 1997 Craig Burley <burley@gnu.ai.mit.edu>
495 * intdoc.c (dumpem): Put appropriate commentary in
496 output file, so readers know it isn't source.
498 Wed Aug 27 08:08:25 1997 Craig Burley <burley@gnu.ai.mit.edu>
500 * proj.h: Always #include "config.j" first, to pick up
501 gcc's configuration.
502 * com.c: Change bcopy() and bzero() calls to memcpy()
503 and memset() calls, to make more of g77 ANSI C.
505 1997-08-26 Dave Love <d.love@dl.ac.uk>
507 * Make-lang.in ($(srcdir)/f/runtime/configure,
508 $(srcdir)/f/runtime/libU77/configure): Fix for when srcdir isn't
509 relative.
511 Tue Aug 26 05:59:21 1997 Craig Burley <burley@gnu.ai.mit.edu>
513 * ansify.c (main): Make sure readers of stdout know
514 it's derived from stdin; omit comment text; get source
515 line numbers in future stderr output to be correct.
517 Tue Aug 26 01:36:01 1997 Craig Burley <burley@gnu.ai.mit.edu>
519 Fix 970825-0.f:
520 * stb.c (ffestb_R5284_): Allow OPEN_PAREN after closing
521 SLASH as well as NAME.
523 Mon Aug 25 23:48:17 1997 Craig Burley <burley@gnu.ai.mit.edu>
525 Changes to allow g77 docs to be built entirely from scratch
526 using any ANSI C compiler, not requiring GNU C:
527 * Make-lang.in ($(srcdir)/f/intdoc.texi): "Pipe" new
528 location of intrinsic documentation data base, f/intdoc.in,
529 through new `ansify' program to append `\n\' to quoted
530 newlines, into f/intdoc.h0. Do appropriate cleanups. Explain.
531 (f77.mostlyclean): Add f/ansify and f/intdoc.h0 to cleanups.
532 * f/ansify.c: New program.
533 * f/intdoc.c: Fix so it conforms to ANSI C.
534 #include f/intdoc.h0 instead of f/intdoc.h.
535 Avoid some warnings.
536 * f/intdoc.h, f/intdoc.in: Rename the former to the latter; no
537 changes made to the content in this patch!
538 * f/intrin.h (ffeintrinFamily): Fix to conform to ANSI C.
540 Sun Aug 24 06:52:48 1997 Craig Burley <burley@gnu.ai.mit.edu>
542 Fix up g77 compiler data base for libf2c routines:
543 * com-rt.def (FFECOM_gfrtSIGNAL): Change return type to
544 FTNINT to match actual code.
546 * com.c (ffecomRttype_): Replace FFECOM_rttypeINT_ with
547 FFECOM_rttypeFTNINT_.
548 Add and fix up comments.
549 (ffecom_make_gfrt_, ffecom_gfrt_basictype,
550 ffecom_gfrt_kindtype): Replace FFECOM_rttypeINT_ with
551 FFECOM_rttypeFTNINT_; add FFECOM_rttypeDOUBLEREAL_.
553 Wed Aug 20 17:18:40 1997 Craig Burley <burley@gnu.ai.mit.edu>
555 * global.c (ffeglobal_ref_progunit_): It's okay to have
556 a different CHARACTER*n length for a reference if the
557 existing length is for another reference, not a definition.
559 Mon Aug 18 14:27:18 1997 Craig Burley <burley@gnu.ai.mit.edu>
561 Fix 970814-0.f:
562 * global.c (ffeglobal_new_progunit_): Distinguish
563 between previously defined, versus inferred, filewide
564 when it comes to diagnostics.
566 Fix 970816-1.f:
567 * global.c (ffeglobal_ref_progunit_): Change BDATA into EXT
568 right at the beginning, so EXTERNAL FOO followed later
569 by SUBROUTINE FOO is not diagnosed.
571 Fix 970813-0.f:
572 * com-rt.def (FFECOM_gfrtALARM): Returns `integer', not
573 `void'.
575 Sun Aug 17 03:32:44 1997 Craig Burley <burley@gnu.ai.mit.edu>
577 Fix up problems when virtual memory exhausted:
578 * malloc.c (malloc_new_): Use gcc's xmalloc(), so we
579 print a nicer message when malloc returns no memory.
580 (malloc_resize_): Ditto for xrealloc().
582 * Make-lang.in, Makefile.in: Comment out lines containing
583 just formfeeds.
585 Sat Aug 16 19:41:33 1997 Craig Burley <burley@gnu.ai.mit.edu>
587 * com.c (ffecom_make_gfrt_): For rttypeREAL_F2C_, return
588 double_type_node; for rttypeREAL_GNU_, return
589 _real_type_node.
591 1997-08-13 Dave Love <d.love@dl.ac.uk>
593 * config-lang.in (diff_excludes): Add some hints about known
594 problematic platforms.
596 1997-08-13 Dave Love <d.love@dl.ac.uk>
598 * intdoc.h: Document `alarm'.
600 Mon Aug 11 21:19:22 1997 Craig Burley <burley@gnu.ai.mit.edu>
602 * Make-lang.in ($(RUNTIMESTAGESTUFF)): Add
603 f/runtime/stamp-lib.
605 Mon Aug 11 01:52:03 1997 Craig Burley <burley@gnu.ai.mit.edu>
607 * com.c (ffecom_build_complex_constant_): Go with the
608 new build_complex() approach used in gcc-2.8.
610 * com.c (ffecom_sym_transform_): Don't set
611 DECL_IN_SYSTEM_HEADER for a tree node that isn't
612 a VAR_DECL, which happens when var is in common!
614 * com.c (ffecom_expr_intrinsic_) (case FFEINTRIN_impALARM):
615 No need to test codegen_imp -- there's only one valid here.
617 * intrin.def (FFEINTRIN_impALARM): Specify `Status' argument
618 as write-only.
620 Fri Aug 8 05:40:23 1997 Craig Burley <burley@gnu.ai.mit.edu>
622 Substantial changes to accommodate distinctions among
623 run-time routines that support intrinsics, and between
624 routines that compute and return the same type vs. those
625 that compute one type and return another (or `void'):
626 * com-rt.def: Specify new return type REAL_F2C_ instead
627 of many DOUBLE_, COMPLEX_F2C_ instead of COMPLEX_, and
628 so on.
629 Clear up the *BES* routines "once and for all".
630 * com.c: New return types.
631 (ffecom_convert_narrow_, ffecom_convert_widen_):
632 New functions that are "safe" variants of convert(),
633 to catch errors that ffecom_expr_intrinsic_() now
634 no longer catches.
635 (ffecom_arglist_expr_): Ensure arguments are not
636 converted to narrower types.
637 (ffecom_call_): Ensure return value is not converted
638 to a wider type.
639 (ffecom_char_args_): Use new ffeintrin_gfrt_direct()
640 routine.
641 (ffecom_expr_intrinsic_): Simplify how run-time
642 routine is selected (via `gfrt' only now; lose the
643 redundant `ix' variable).
644 Eliminate the `library' label; any code that doesn't
645 return directly just `break's out now with `gfrt'
646 set appropriately.
647 Set `gfrt' to default choice initially, either a
648 fast direct form or, if not available, a slower
649 indirect-callable form.
650 (ffecom_make_gfrt_): No longer need to do special
651 check for complex; it's built into the new return-type
652 regime.
653 (ffecom_ptr_to_expr): Use new ffeintrin_gfrt_indirect()
654 routine.
655 * intrin.c, intrin.h: `gfrt' field replaced with three fields,
656 so it is easier to provide faster direct-callable and
657 GNU-convention indirect-callable routines in the future.
658 DEFIMP macro adjusted accordingly, along with all its uses.
659 (ffeintrin_gfrt_direct): New function.
660 (ffeintrin_gfrt_indirect): Ditto.
661 (ffeintrin_is_actualarg): If `-fno-f2c' is in effect,
662 require a GNU-callable version of intrinsic instead of
663 an f2c-callable version, so indirect calling is still checked.
664 * intrin.def: Replace one GFRT field with the three new fields,
665 as appropriate for each DEFIMP intrinsic.
667 * com.c (ffecom_stabilize_aggregate_,
668 ffecom_convert_to_complex_): Make these `static'.
670 Thu Aug 7 11:24:34 1997 Craig Burley <burley@gnu.ai.mit.edu>
672 Provide means for front end to determine actual
673 "standard" return type for an intrinsic if it is
674 passed as an actual argument:
675 * com.h, com.c (ffecom_gfrt_basictype,
676 ffecom_gfrt_kindtype): New functions.
677 (ffecom_gfrt_kind_type_): Replaced with new function.
678 All callers updated.
679 (ffecom_make_gfrt_): No longer need do anything
680 with kind type.
682 * intrin.c (ffeintrin_basictype, ffeintrin_kindtype):
683 Now returns correct type info for specific intrinsic
684 (based on type of run-time-library implementation).
686 Wed Aug 6 23:08:46 1997 Craig Burley <burley@gnu.ai.mit.edu>
688 * global.c (ffeglobal_ref_progunit_): Don't reset
689 number of arguments just due to new type info,
690 so useful warnings can be issued.
692 1997-08-06 Dave Love <d.love@dl.ac.uk>
694 * intrin.def: Fix IDATE_vxt argument order.
695 * intdoc.h: Likewise.
697 Thu Jul 31 22:22:03 1997 Craig Burley <burley@gnu.ai.mit.edu>
699 * global.c (ffeglobal_proc_ref_arg): If REF/DESCR
700 disagreement, DESCR is CHARACTER, and types disagree,
701 pretend the argsummary agrees so the message ends up
702 being about type disagreement.
703 (ffeglobal_proc_def_arg): Ditto.
705 * expr.c (ffeexpr_token_first_rhs_3_): Set info for LABTOK
706 to NONE of everything, to avoid misdiagnosing filewide
707 usage of alternate returns.
709 Sun Jul 20 23:07:47 1997 Craig Burley <burley@gnu.ai.mit.edu>
711 * com.c (ffecom_sym_transform_): If type gets set
712 to error_mark_node, just return that for transformed symbol.
713 (ffecom_member_phase2_): If type gets set to error_mark_node,
714 just return.
715 (ffecom_check_size_overflow_): Add `dummy' argument to
716 flag that type is for a dummy, update all callers.
718 Sun Jul 13 17:40:53 1997 Craig Burley <burley@gnu.ai.mit.edu>
720 Fix 970712-1.f:
721 * where.c (ffewhere_set_from_track): If start point
722 is too large, just use initial start point. 0.6 should
723 fix all this properly.
725 Fix 970712-2.f:
726 * com.c (ffecom_sym_transform_): Preserve error_mark_node for type.
727 (ffecom_type_localvar_): Ditto.
728 (ffecom_sym_transform_): If type is error_mark_node,
729 don't error-check decl size, because back end responds by
730 setting that to an integer 0 instead of error_mark_node.
731 (ffecom_transform_common_): Same as earlier fix to _transform_
732 in that size is checked by dividing BITS_PER_UNIT instead of
733 multiplying.
734 (ffecom_transform_equiv_): Ditto.
736 Fix 970712-3.f:
737 * stb.c (ffestb_R10014_): Fix flaky fall-through in error
738 test for FFELEX_typeCONCAT by just replicating the code,
739 and do FFELEX_typeCOLONCOLON while at it.
741 1997-07-07 Dave Love <d.love@dl.ac.uk>
743 * intdoc.h: Add various missing pieces; correct GMTIME, LTIME
744 result ordering.
746 * intrin.def, com-rt.def: Add alarm.
748 * com.c (ffecom_expr_intrinsic_): Add case for alarm.
750 Thu Jun 26 04:19:40 1997 Craig Burley <burley@gnu.ai.mit.edu>
752 Fix 970302-3.f:
753 * com.c (ffecom_sym_transform_): For sanity-check compare
754 of gbe size of local variable to g77 expectation,
755 use varasm.c/assemble_variable technique of dividing
756 BITS_PER_UNIT out of gbe info instead of multiplying
757 g77 info up, to avoid crash when size in bytes is very
758 large, and overflows an `int' or similar when multiplied.
760 Fix 970626-2.f:
761 * com.c (ffecom_finish_symbol_transform_): Don't bother
762 transforming a dummy argument, to avoid a crash.
763 * ste.c (ffeste_R1227): Don't return a value if the
764 result decl, or its type, is error_mark_node.
766 Fix 970626-4.f:
767 * lex.c (ffelex_splice_tokens): `-fdollar-ok' is
768 irrelevant to whether a DOLLAR token should be made
769 from an initial character of `$'.
771 Fix 970626-6.f:
772 * stb.c (ffestb_do3_): DO iteration variable is an
773 lhs, not rhs, expression.
775 Fix 970626-7.f and 970626-8.f:
776 * expr.c (ffeexpr_cb_comma_i_1_): Set IMPDO expression
777 to have clean info, because undefined rank, for example,
778 caused crash on mangled source on UltraSPARC but not
779 on Alpha for a series of weird reasons.
780 (ffeexpr_cb_close_paren_): If not CLOSE_PAREN, push
781 opANY expression onto stack instead of attempting
782 to mimic what program might have wanted.
783 (ffeexpr_cb_close_paren_): Don't wrap opPAREN around
784 opIMPDO, just warn that it's gratuitous.
785 * bad.def (FFEBAD_IMPDO_PAREN): New warning.
787 Fix 970626-9.f:
788 * expr.c (ffeexpr_declare_parenthesized_): Must shut down
789 parsing in kindANY case, otherwise the parsing engine might
790 decide there's an ambiguity.
791 (ffeexpr_token_name_rhs_): Eliminate parentypeSUBROUTINE_
792 case, so we crash right away if it comes through.
793 * st.c, st.h, sta.c, sta.h (ffest_shutdown, ffesta_shutdown):
794 New functions.
796 Tue Jun 24 19:47:29 1997 Craig Burley <burley@gnu.ai.mit.edu>
798 * com.c (ffecom_check_size_overflow_): New function
799 catches some cases of the size of a type getting
800 too large. varasm.c must catch the rest.
801 (ffecom_sym_transform_): Use new function.
802 (ffecom_type_localvar_): Ditto.
804 Mon Jun 23 01:09:28 1997 Craig Burley <burley@gnu.ai.mit.edu>
806 * global.c (ffeglobal_proc_def_arg): Fix comparison
807 of argno to #args.
808 (ffeglobal_proc_ref_arg): Ditto.
810 * lang-options.h, top.c: Rename `-fdebug' to `-fxyzzy',
811 since it's an unsupported internals option and some
812 poor user might guess that it does something.
814 * bad.def: Make a warning for each filewide diagnostic.
815 Put all filewides together.
816 * com.c (ffecom_sym_transform_): Don't substitute
817 known global tree for global entities when `-fno-globals'.
818 * global.c (ffeglobal_new_progunit_): Don't produce
819 fatal diagnostics about globals when `-fno-globals'.
820 Instead, produce equivalent warning when `-Wglobals'.
821 (ffeglobal_proc_ref_arg): Ditto.
822 (ffeglobal_proc_ref_nargs): Ditto.
823 (ffeglobal_ref_progunit_): Ditto.
824 * lang-options.h, top.c, top.h: New `-fno-globals' option.
826 Sat Jun 21 12:32:54 1997 Craig Burley <burley@gnu.ai.mit.edu>
828 * expr.c (ffeexpr_fulfill_call_): Set array variable
829 to avoid warning about uninitialized variable.
831 * Make-lang.in: Get rid of any setting of HOST_* macros,
832 since these will break gcc's build!
833 * makefile: New file to make building derived files
834 easier.
836 Thu Jun 19 18:19:28 1997 Craig Burley <burley@gnu.ai.mit.edu>
838 * g77.c (main): Install Emilio Lopes' patch to support
839 Ratfor, and to fix the printing of the version string
840 to go to stderr, not stdout.
841 * lang-specs.h: Install Emilio Lopes' patch to support
842 Ratfor, and patch the result to support picking up
843 `*f771' from the `specs' file.
845 Thu Jun 12 14:36:25 1997 Craig Burley <burley@gnu.ai.mit.edu>
847 * storag.c (ffestorag_update_init, ffestorag_update_save):
848 Also update parent, in case equivalence processing
849 has already eliminated pointers to it via the
850 local equivalence info.
852 Tue Jun 10 14:08:26 1997 Craig Burley <burley@gnu.ai.mit.edu>
854 * intdoc.c: Add cross-reference to end of description
855 of any generic intrinsic pointing to other intrinsics
856 with the same name.
858 Warn about explicit type declaration for intrinsic
859 that disagrees with invocation:
860 * expr.c (ffeexpr_paren_rhs_let_): Preserve type info
861 for intrinsic functions.
862 (ffeexpr_token_funsubstr_): Ditto.
863 * intrin.c (ffeintrin_fulfill_generic): Warn if type
864 info of fulfilled intrinsic invocation disagrees with
865 explicit type info given symbol.
866 (ffeintrin_fulfill_specific): Ditto.
867 * stc.c (ffestc_R1208_item): Preserve type info
868 for intrinsics.
869 (ffestc_R501_item): Ditto.
871 Mon Jun 9 17:45:44 1997 Craig Burley <burley@gnu.ai.mit.edu>
873 * com.c (ffecom_expr_intrinsic_): Fix several of the
874 libU77/libF77-unix handlers to properly convert their
875 arguments.
877 * com-rt.def (FFECOM_gfrtFSTAT): Append missing "i" to
878 arg string.
880 Fri Jun 6 14:37:30 1997 Craig Burley <burley@gnu.ai.mit.edu>
882 * com.c (ffecom_expr_intrinsic_): Have a case statement
883 for every intrinsic implementation, so missing ones
884 are caught via gcc warnings.
885 Don't call ffeintrin_codegen_imp anymore.
886 * intrin.c (ffeintrin_fulfill_generic): Remove cg_imp
887 stuff from here.
888 (ffeintrin_codegen_imp): Delete this function.
889 * intrin.def, intrin.h: Remove DEFIMQ stuff from here
890 as well.
892 Thu Jun 5 13:03:07 1997 Craig Burley <burley@gnu.ai.mit.edu>
894 * top.c (ffe_decode_option): New -fbadu77-intrinsics-*
895 options.
896 * top.h: Ditto.
897 * intrin.h: New BADU77 family.
898 * intrin.c (ffeintrin_state_family): Ditto.
900 Implement new scheme to track intrinsic names vs. forms:
901 * intrin.c (ffeintrin_fulfill_generic),
902 (ffeintrin_fulfill_specific), (ffeintrin_is_intrinsic),
903 intrin.def: The documented name is now either in the
904 generic info or, if no generic, in the specific info.
905 For a generic, the specific info contains merely the
906 distinguishing form (usually "function" or "subroutine"),
907 used for diagnostics about ambiguous references and
908 in the documentation.
910 * intrin.def: Clean up formatting of DEFNAME block.
911 Convert many libU77 intrinsics into generics that
912 support both subroutine and function forms.
913 Put the function forms of side-effect routines into
914 the new BADU77 family.
915 Make MCLOCK and TIME return INTEGER*4 again, and add
916 INTEGER*8 equivalents called MCLOCK8 and TIME8.
917 Fix up more status return values to be written and
918 insist on them being I1 as well.
919 * com.c (ffecom_expr_intrinsic_): Lots of changes to
920 support new libU77 intrinsic interfaces.
922 Mon Jun 2 00:37:53 1997 Craig Burley <burley@gnu.ai.mit.edu>
924 * com.c (ffecom_init_0): Pointer type is now INTEGER(KIND=7),
925 not INTEGER(KIND=0), since we want to reserve KIND=0 for
926 future use.
928 Thu May 29 14:30:33 1997 Craig Burley <burley@gnu.ai.mit.edu>
930 Fix bugs preventing CTIME(I*4) from working correctly:
931 * com.c (ffecom_char_args_): For FUNCREF case, process
932 args to intrinsic just as they would be in
933 ffecom_expr_intrinsic_.
934 * com-rt.def (FFECOM_gfrtCTIME, FFECOM_gfrtTTYNAM): Fix
935 argument decls to specify `&'.
937 Wed May 28 22:19:49 1997 Craig Burley <burley@gnu.ai.mit.edu>
939 Fix gratuitous warnings exposed by dophot aka 970528-1:
940 * global.c (ffeglobal_proc_def_arg, ffeglobal_proc_ref_arg):
941 Support distinct function/subroutine arguments instead of
942 just procedures.
943 * global.h: Ditto.
944 * expr.c (ffeexpr_fulfill_call_): A SYMTER with kindNONE
945 also is a procedure (either function or subroutine).
947 Mon May 26 20:25:31 1997 Craig Burley <burley@gnu.ai.mit.edu>
949 * bad.def: Have several lexer diagnostics refer to
950 documentation for people who need more info on what Fortran
951 source code is supposed to look like.
953 * expr.c (ffeexpr_reduced_bool1_), bad.def: New diagnostics
954 specific to .NOT. now mention only one operand instead
955 of two.
957 * g77.c: Recognize -fsyntax-only, similar to -c etc.
958 (lookup_option): Fix bug that prevented non-`--' options
959 from being recognized.
961 Sun May 25 04:29:04 1997 Craig Burley <burley@gnu.ai.mit.edu>
963 * intrin.def (FFEINTRIN_impCTIME): Accept `I*' expression
964 for STime instead of requiring `I2'.
966 Tue May 20 16:14:40 1997 Craig Burley <burley@gnu.ai.mit.edu>
968 * symbol.c (ffesymbol_reference): All references to
969 standard intrinsics are considered explicit, so as
970 to avoid generating basically useless warnings.
971 * intrin.c, intrin.h (ffeintrin_is_standard): Returns TRUE
972 if intrinsic is standard.
974 Sun May 18 21:14:59 1997 Craig Burley <burley@gnu.ai.mit.edu>
976 * com-rt.def: Changed all external names of the
977 form `"\([a-z0-9]*\)_' to `"G77_\1_0"' so as to
978 allow any name valid as an intrinsic to be used
979 as such and as a user-defined external procedure
980 name or common block as well.
982 Thu May 8 13:07:10 1997 Craig Burley <burley@gnu.ai.mit.edu>
984 * expr.c (ffeexpr_cb_end_notloc_): For %VAL, %REF, and
985 %DESCR, copy arg info into new node.
987 Mon May 5 14:42:17 1997 Craig Burley <burley@gnu.ai.mit.edu>
989 From Uwe F. Mayer <mayer@math.Vanderbilt.Edu>:
990 * Make-lang.in (g77-cross): Fix typo in g77.c path.
992 From Brian McIlwrath <bkm@star.rl.ac.uk>:
993 * lang-specs.h: Have g77 pick up options from a section
994 labeled `*f771' of the `specs' file.
996 Sat May 3 02:46:08 1997 Craig Burley <burley@gnu.ai.mit.edu>
998 * intrin.def (FFEINTRIN_defSIGNAL): Add optional `Status'
999 argument that com.c already expects (per Dave Love).
1001 More changes to support better tracking of (filewide)
1002 globals, in particular, the arguments to procedures:
1003 * bad.def (FFEBAD_FILEWIDE_NARGS, FFEBAD_FILEWIDE_NARGS_W,
1004 FFEBAD_FILEWIDE_ARG, FFEBAD_FILEWIDE_ARG_W): New diagnostics.
1005 * expr.c (ffebad_fulfill_call_): Provide info on each
1006 argument to ffeglobal.
1007 * global.c, global.h (ffeglobal_proc_def_arg,
1008 ffeglobal_proc_def_nargs, ffeglobal_proc_ref_arg,
1009 ffeglobal_proc_ref_args): New functions.
1010 (ffeglobalArgSummary, ffeglobalArgInfo_): New types.
1012 Tue Apr 29 18:35:41 1997 Craig Burley <burley@gnu.ai.mit.edu>
1014 More changes to support better tracking of (filewide)
1015 globals:
1016 * expr.c (ffeexpr_fulfill_call_): New function.
1017 (ffeexpr_token_name_lhs_): Call after building procedure
1018 reference expression. Also leave info field for ANY-ized
1019 expression alone.
1020 (ffeexpr_token_arguments_): Ditto.
1022 Mon Apr 28 20:04:18 1997 Craig Burley <burley@gnu.ai.mit.edu>
1024 Changes to support better tracking of (filewide)
1025 globals, mainly to avoid crashes due to inlining:
1026 * bad.def: Go back to quoting intrinsic names,
1027 (FFEBAD_FILEWIDE_DISAGREEMENT, FFEBAD_FILEWIDE_TIFF,
1028 FFEBAD_FILEWIDE_TYPE_MISMATCH): New diagnostics.
1029 (FFEBAD_INTRINSIC_EXPIMP, FFEBAD_INTRINSIC_GLOBAL): Reword
1030 for clarity.
1031 * com.c (ffecom_do_entry_, ffecom_start_progunit_,
1032 ffecom_sym_transform_): Accommodate new FFEGLOBAL_typeEXT
1033 possibility.
1034 * expr.c (ffeexpr_sym_lhs_call_, ffeexpr_sym_lhs_extfunc_,
1035 ffeexpr_sym_rhs_actualarg_, ffeexpr_declare_parenthesized_,
1036 ffeexpr_paren_rhs_let_, ffeexpr_token_funsubstr_):
1037 Fill in real kind info instead of leaving NONE where
1038 appropriate.
1039 Register references to intrinsics and globals with ffesymbol
1040 using new ffesymbol_reference function instead of
1041 ffesymbol_globalize.
1042 * global.c (ffeglobal_type_string_): New array for
1043 new diagnostics.
1044 * global.h, global.c:
1045 Replace ->init mechanism with ->tick mechanism.
1046 Move other common-related members into a substructure of
1047 a union, so the proc substructure can be introduced
1048 to include members related to externals other than commons.
1049 Don't complain about ANY-ized globals; ANY-ize globals
1050 once they're complained about, in any case where code
1051 generation could become a problem.
1052 Handle global entries that have NONE type (seen as
1053 intrinsics), EXT type (seen as EXTERNAL), and so on.
1054 Keep track of kind and type of externals, both via
1055 definition and via reference.
1056 Diagnose disagreements about kind or type of externals
1057 (such as functions).
1058 (ffeglobal_ref_intrinsic, ffeglobal_ref_progunit_): New
1059 functions.
1060 * stc.c (ffestc_R1207_item, ffestc_R1208_item,
1061 ffestc_R1219, ffestc_R1226):
1062 Call ffesymbol_reference, not ffesymbol_globalize.
1063 * stu.c (ffestu_sym_end_transition,
1064 ffestu_sym_exec_transition):
1065 Call ffesymbol_reference, not ffesymbol_globalize.
1066 * symbol.c (ffesymbol_globalize): Removed...
1067 (ffesymbol_reference): ...to this new function,
1068 which more generally registers references to symbols,
1069 globalizes globals, and calls on the ffeglobal module
1070 to check globals filewide.
1072 * global.h, global.c: Rename some macros and functions
1073 to more clearly distinguish common from other globals.
1074 All callers changed.
1076 * com.c (ffecom_sym_transform_): Trees describing
1077 filewide globals must be allocated on permanent obstack.
1079 * expr.c (ffeexpr_token_name_lhs_): Don't generate
1080 gratuitous diagnostics for FFEINFO_whereANY case.
1082 Thu Apr 17 03:27:18 1997 Craig Burley <burley@gnu.ai.mit.edu>
1084 * global.c: Add support for flagging intrinsic/global
1085 confusion via warnings.
1086 * bad.def (FFEBAD_INTRINSIC_EXPIMP,
1087 FFEBAD_INTRINSIC_GLOBAL): New diagnostics.
1088 * expr.c (ffeexpr_token_funsubstr_): Ditto.
1089 (ffeexpr_sym_lhs_call_): Ditto.
1090 (ffeexpr_paren_rhs_let_): Ditto.
1091 * stc.c (ffestc_R1208_item): Ditto.
1093 Wed Apr 16 22:40:56 1997 Craig Burley <burley@gnu.ai.mit.edu>
1095 * expr.c (ffeexpr_declare_parenthesized_): INCLUDE
1096 context can't be an intrinsic invocation either.
1098 Fri Mar 28 10:43:28 1997 Craig Burley <burley@gnu.ai.mit.edu>
1100 * expr.c (ffeexpr_token_arguments_): Make sure top of
1101 exprstack is operand before dereferencing operand field.
1103 * lex.c (ffelex_prepare_eos_): Fill up truncated
1104 hollerith token, so crash on null ->text field doesn't
1105 happen later.
1107 * stb.c (ffestb_R10014_): If NAMES isn't recognized (or
1108 the recognized part is followed in the token by a
1109 non-digit), don't try and collect digits, as there
1110 might be more than FFEWHERE_indexMAX letters to skip
1111 past to do so -- and the code is diagnosed anyway.
1113 Thu Mar 27 00:02:48 1997 Craig Burley <burley@gnu.ai.mit.edu>
1115 * com.c (ffecom_sym_transform_): Force local
1116 adjustable array onto stack.
1118 * stc.c (ffestc_R547_item_object): Don't actually put
1119 the symbol in COMMON if the symbol has already been
1120 EQUIVALENCE'd to a different COMMON area.
1122 * equiv.c (ffeequiv_add): Don't actually do anything
1123 if there's a disagreement over which COMMON area is
1124 involved.
1126 Tue Mar 25 03:35:19 1997 Craig Burley <burley@gnu.ai.mit.edu>
1128 * com.c (ffecom_transform_common_): If no explicit init
1129 of COMMON area, don't actually init it even though
1130 storage area suggests it.
1132 Mon Mar 24 12:10:08 1997 Craig Burley <burley@gnu.ai.mit.edu>
1134 * lex.c (ffelex_image_char_): Avoid overflowing the
1135 column counter itself, as well as the card image.
1137 * where.c (ffewhere_line_new): Cast ffelex_line_length()
1138 to (size_t) so 255 doesn't overflow to 0!
1140 * stc.c (ffestc_labeldef_notloop_begin_): Don't gratuitously
1141 terminate loop before processing statement, so block
1142 doesn't disappear out from under EXIT/CYCLE processing.
1143 (ffestc_labeldef_notloop_): Has old code from above
1144 function, instead of just calling it.
1146 * expr.c (ffeexpr_cb_comma_i_4_): Don't skip over
1147 arbitrary token (such as EOS).
1149 * com.c (ffecom_init_zero_): Handle RECORD_TYPE and
1150 UNION_TYPE so -fno-zeros works with -femulated-complex.
1152 1997-03-12 Dave Love <d.love@dl.ac.uk>
1154 * intrin.def: New intrinsics INT2, INT8, CPU_TIME. Fix AND, OR,
1155 XOR. [Integrated by burley, AND/OR/XOR already fixed, INT8
1156 implementation changed/fixed.]
1158 Wed Mar 12 10:40:08 1997 Craig Burley <burley@gnu.ai.mit.edu>
1160 * Make-lang.in ($(srcdir)/f/intdoc.texi): Simplify rules
1161 so building f/intdoc is not always necessary; remove
1162 f/intdoc after running it if it is built.
1164 Tue Mar 11 23:42:00 1997 Craig Burley <burley@gnu.ai.mit.edu>
1166 * intrin.def (FFEINTRIN_impAND, FFEINTRIN_impOR,
1167 FFEINTRIN_impXOR): Use the IAND, IOR, and IEOR implementations
1168 of these, instead of crashing in ffecom_expr_intrinsic_
1169 or adding case labels there.
1171 Mon Mar 10 22:51:23 1997 Craig Burley <burley@gnu.ai.mit.edu>
1173 * intdoc.c: Fix so any C compiler can compile this.
1175 Fri Feb 28 13:16:50 1997 Craig Burley <burley@gnu.ai.mit.edu>
1177 * Version 0.5.20 released.
1179 Fri Feb 28 01:45:25 1997 Craig Burley <burley@gnu.ai.mit.edu>
1181 * Make-lang.in (RUNTIMESTAGESTUFF, LIBU77STAGESTUFF):
1182 Move some files incorrectly in the former to the latter,
1183 and add another file or two to the latter.
1185 New meanings for (KIND=n), and new denotations in the
1186 little language describing intrinsics:
1187 * com.c (ffecom_init_0): Assign new meanings.
1188 * intdoc.c: Document new meanings.
1189 Support the new denotations.
1190 * intrin.c: Employ new meanings, mapping them to internal
1191 values (which are the same as they ever were for now).
1192 Support the new denotations.
1193 * intrin.def: Switch DEFIMP table to the new denotations.
1195 * intrin.c (ffeintrin_check_): Fix bug that was leaving
1196 LOC() and %LOC() returning INTEGER*4 on systems where
1197 it should return INTEGER*8.
1199 * type.c: Canonicalize function definitions, for etags
1200 and such.
1202 Wed Feb 26 20:43:03 1997 Craig Burley <burley@gnu.ai.mit.edu>
1204 * com.c (ffecom_init_0): Choose INTEGER(KIND=n) types,
1205 where n is 2, 3, and 4, according to the new docs
1206 instead of according to the old C correspondences
1207 (which seem less useful at this point).
1209 * equiv.c (ffeequiv_destroy_): New function.
1210 (ffeequiv_layout_local_): Use this new function
1211 whenever the laying out of a local equivalence chain
1212 is aborted for any reason.
1213 Otherwise ensure that symbols no longer reference
1214 the stale ffeequiv entries that result when they
1215 are killed off in this procedure.
1216 Also, the rooted symbol is one that has storage,
1217 it really is irrelevant whether it has an equiv entry
1218 at this point (though the code to remove the equiv
1219 entry was put in at the end, just in case).
1220 (ffeequiv_kill): When doing internal checks, make
1221 sure the victim isn't named by any symbols it points
1222 to. Not as complete a check as looking through the
1223 entire symbol table (which does matter, since some
1224 code in equiv.c used to remove symbols from the lists
1225 for an ffeequiv victim but not remove that victim as the
1226 symbol's equiv info), but this check did find some
1227 real bugs in the code (that were fixed).
1229 Mon Feb 24 16:42:13 1997 Craig Burley <burley@gnu.ai.mit.edu>
1231 * com.c (ffecom_expr_intrinsic_): Fix a couple of
1232 warnings about uninitialized variables.
1233 * intrin.c (ffeintrin_check_): Ditto, but there were
1234 a couple of _real_ uninitialized-variable _bugs_ here!
1235 (ffeintrin_fulfill_specific): Ditto, no real bug here.
1237 Sun Feb 23 15:01:20 1997 Craig Burley <burley@gnu.ai.mit.edu>
1239 Clean up diagnostics (especially about intrinsics):
1240 * bad.def (FFEBAD_UNIMPL_STMT): Remove.
1241 (FFEBAD_INTRINSIC_*, FFEBAD_NEED_INTRINSIC): Clean these
1242 up so they're friendlier.
1243 (FFEBAD_INTRINSIC_CMPAMBIG): New.
1244 * intrin.c (ffeintrin_fulfill_generic,
1245 ffeintrin_fulfill_specific, ffeintrin_is_intrinsic):
1246 Always choose
1247 generic or specific name text (which is for doc purposes
1248 anyway) over implementation name text (which is for
1249 internal use).
1250 * intrin.def: Use more descriptive name texts for generics
1251 and specifics in cases where the names themselves are not
1252 enough (e.g. IDATE, which has two forms).
1254 Fix some intrinsic mappings:
1255 * intrin.def (FFEINTRIN_specIDINT, FFEINTRIN_specAND,
1256 FFEINTRIN_specDFLOAT, FFEINTRIN_specDREAL, FFEINTRIN_specOR,
1257 FFEINTRIN_specXOR): Now have their own implementations,
1258 instead of borrowing from others.
1259 (FFEINTRIN_specAJMAX0, FFEINTRIN_specAJMIN0, FFEINTRIN_specBJTEST,
1260 FFEINTRIN_specDFLOTJ, FFEINTRIN_specFLOATJ, FFEINTRIN_specJIABS,
1261 FFEINTRIN_specJIAND, FFEINTRIN_specJIBCLR, FFEINTRIN_specJIBITS,
1262 FFEINTRIN_specJIBSET, FFEINTRIN_specJIDIM, FFEINTRIN_specJIDINT,
1263 FFEINTRIN_specJIDNNT, FFEINTRIN_specJIEOR, FFEINTRIN_specJIFIX,
1264 FFEINTRIN_specJINT, FFEINTRIN_specJIOR, FFEINTRIN_specJISHFT,
1265 FFEINTRIN_specJISHFTC, FFEINTRIN_specJISIN, FFEINTRIN_specJMAX0,
1266 FFEINTRIN_specJMAX1, FFEINTRIN_specJMIN0, FFEINTRIN_specJMIN1,
1267 FFEINTRIN_specJMOD, FFEINTRIN_specJNINT, FFEINTRIN_specJNOT,):
1268 Turn these implementations off, since it's not clear
1269 just what types they expect in the context of portable Fortran.
1270 (DFLOAT): Now in FVZ family, since f2c supports them
1272 Support intrinsic inquiry functions (BIT_SIZE, LEN):
1273 * intrin.c: Allow `i' in <arg_extra>.
1274 * intrin.def (FFEINTRIN_impBIT_SIZE, FFEINTRIN_impLEN):
1275 Mark args with `i'.
1277 Sat Feb 22 13:34:09 1997 Craig Burley <burley@gnu.ai.mit.edu>
1279 Only warn, don't error, for reference to unimplemented
1280 intrinsic:
1281 * bad.def (FFEBAD_INTRINSIC_UNIMPLW): Warning version
1282 of _UNIMPL.
1283 * intrin.c (ffeintrin_is_intrinsic): Use new warning
1284 version of _UNIMPL (FFEBAD_INTRINSIC_UNIMPLW).
1286 Complain about REAL(Z) and AIMAG(Z) (Z is DOUBLE COMPLEX):
1287 * bad.def (FFEBAD_INTRINSIC_CMPAMBIG): New diagnostic.
1288 * expr.c: Needed #include "intrin.h" anyway.
1289 (ffeexpr_token_intrincheck_): New function handles delayed
1290 diagnostic for "REAL(REAL(expr)" if next token isn't ")".
1291 (ffeexpr_token_arguments_): Do most of the actual checking here.
1292 * intrin.h, intrin.c (ffeintrin_fulfill_specific): New
1293 argument, check_intrin, to tell caller that intrin is REAL(Z)
1294 or AIMAG(Z). All callers updated, mostly to pass NULL in
1295 for this.
1296 (ffeintrin_check_): Also has new arg check_intrin for same
1297 purpose. All callers updated the same way.
1298 * intrin.def (FFEINTRIN_impAIMAG): Change return type
1299 from "R0" to "RC", to accommodate f2c (and perhaps other
1300 non-F90 F77 compilers).
1301 * top.h, top.c: New option -fugly-complex.
1303 New GNU intrinsics REALPART, IMAGPART, and COMPLEX:
1304 * com.c (ffecom_expr_intrinsic_): Implement impCOMPLEX
1305 and impREALPART here. (specIMAGPART => specAIMAG.)
1306 * intrin.def: Add the intrinsics here.
1308 Rename implementations of VXTIDATE and VXTTIME to IDATEVXT
1309 and TIMEVXT, so they sort more consistently:
1310 * com.c (ffecom_expr_intrinsic_):
1311 * intrin.def:
1313 Delete intrinsic group `dcp', add `gnu', etc.:
1314 * intrin.c (ffeintrin_state_family): FFEINTRIN_familyGNU
1315 replaces FFEINTRIN_familyDCP, and gets state from `gnu'
1316 group.
1317 Get rid of FFEINTRIN_familyF2Z, nobody needs it.
1318 Move FFEINTRIN_specDCMPLX from DCP family to FVZ family,
1319 as f2c has it.
1320 Move FFEINTRIN_specDFLOAT from F2C family to FVZ family.
1321 (FFEINTRIN_specZABS, FFEINTRIN_specZCOS, FFEINTRIN_specZEXP,
1322 FFEINTRIN_specZLOG, FFEINTRIN_specZSIN, FFEINTRIN_specZSQRT):
1323 Move these from F2Z family to F2C family.
1324 * intrin.h (FFEINTRIN_familyF2Z, FFEINTRIN_familyDCP): Remove.
1325 (FFEINTRIN_familyGNU): Add.
1326 * top.h, top.c: Replace `dcp' with `gnu'.
1328 * com.c (ffecom_expr_intrinsic_): Clean up by collecting
1329 simple conversions into one nice, conceptual place.
1330 Fix up some intrinsic subroutines (MVBITS, KILL, UMASK) to
1331 properly push and pop call temps, to avoid wasting temp
1332 registers.
1334 * g77.c (doit): Toon says variables should be defined
1335 before being referenced. Spoilsport.
1337 * intrin.c (ffeintrin_check_): Now Dave's worried about
1338 warnings about uninitialized variables. Okay, so for
1339 basic return values 'g' and 's', they _were_
1340 uninitialized -- is determinism really _that_ useful?
1342 * intrin.def (FFEINTRIN_impFGETC): Fix STATUS argument
1343 so that it is INTENT(OUT) instead of INTENT(IN).
1345 1997-02-21 Dave Love <d.love@dl.ac.uk>
1347 * intrin.def, com.c: Support Sun-type `short' and `long'
1348 intrinsics. Perhaps should also do Microcruft-style `int2'.
1350 Thu Feb 20 15:16:53 1997 Craig Burley <burley@gnu.ai.mit.edu>
1352 * com.c (ffecom_expr_intrinsic_): Clean up indentation.
1353 Support SECONDSUBR intrinsic implementation.
1354 Rename SECOND to SECONDFUNC for direct support via library.
1356 * g77.c: Fix to return proper status value to shell,
1357 by obtaining it from processes it spawns.
1359 * intdoc.c: Fix minor typo.
1361 * intrin.def: Turn SECOND into generic that maps into
1362 function and subroutine forms.
1364 * intrin.def: Make FLOAT and SNGL into specific intrinsics.
1366 * intrin.def, intrin.h: Change the way DEFGEN and DEFSPEC
1367 macros work, to save on verbage.
1369 Mon Feb 17 02:08:04 1997 Craig Burley <burley@gnu.ai.mit.edu>
1371 New subsystem to automatically generate documentation
1372 on intrinsics:
1373 * Make-lang.in ($(srcdir)/f/g77.info,
1374 $(srcdir)/f/g77.dvi): Move g77 doc rules around.
1375 Add to g77 doc rules the new subsystem.
1376 (f77.mostlyclean, f77.maintainer-clean): Also clean up
1377 after new doc subsystem.
1378 * intdoc.c, intdoc.h: New doc subsystem code.
1379 * intrin.h [FFEINTRIN_DOC]: When 1, don't pull in
1380 stuff not needed by doc subsystem.
1382 Improve on intrinsics mechanism to both be more
1383 self-documenting and to catch more user errors:
1384 * intrin.c (ffeintrin_check_): Recognize new arg-len
1385 and arg-rank information, and check it.
1386 Move goto and signal indicators to the basic type.
1387 Permit reference to arbitrary argument number, not
1388 just first argument (for BESJN and BESYN).
1389 (ffeintrin_init_0): Check and accept new notations.
1390 * intrin.c, intrin.def: Value in COL now identifies
1391 arguments starting with number 0 being the first.
1393 Some minor intrinsics cleanups (resulting from doc work):
1394 * com.c (ffecom_expr_intrinsic_): Implement FLUSH
1395 directly once again, handle its optional argument,
1396 so it need not be a generic (awkward to handle in docs).
1397 * intrin.def (BESJ0, BESJ1, BESJN, BESY0, BESY1, BESYN,
1398 CHDIR, CHMOD, CTIME, DBESJ0, DBESJ1, DBESJN, DBESY0,
1399 DBESY1, DBESYN, DDIM, ETIME, FGETC, FNUM, FPUTC, FSTAT,
1400 GERROR, GETCWD, GETGID, GETLOG, GETPID, GETUID, GMTIME,
1401 HOSTNM, IDATE, IERRNO, IIDINT, IRAND, ISATTY, ITIME, JIDINT,
1402 LNBLNK, LSTAT, LTIME, MCLOCK, PERROR, SRAND, SYMLNK, TTYNAM,
1403 UMASK): Change capitalization of initcaps (official) name
1404 to be consistent with Burley's somewhat arbitrary rules.
1405 (BESJN, BESYN): These have return arguments of same type
1406 as their _second_ argument.
1407 (FLUSH): Now a specific, not generic, intrinsic, with one
1408 optional argument.
1409 (FLUSH1): Eliminated.
1410 Add arg-len and arg-rank info to several intrinsics.
1411 (ITIME): Change argument type from REAL to INTEGER.
1413 Tue Feb 11 14:04:42 1997 Craig Burley <burley@gnu.ai.mit.edu>
1415 * Make-lang.in (f771): Invocation of Makefile now done
1416 with $(srcdir)=gcc to go along with $(VPATH)=gcc.
1417 ($(srcdir)/f/runtime/configure,
1418 $(srcdir)/f/runtime/libU77/configure): Break these out
1419 so spurious triggers of this rule don't happen (as when
1420 configure.in is more recent than libU77/configure).
1421 (f77.rebuilt): Distinguish source versus build files,
1422 so this target can be invoked from build directory and
1423 still work.
1424 * Makefile.in: This now expects $(srcdir) to be the gcc
1425 source directory, not gcc/f, to agree with $(VPATH).
1426 Accordingly, $(INCLUDES) has been fixed, various cruft
1427 removed, the removal of f771 has been fixed to remove
1428 the _real_ f771 (not the one in gcc's parent directory),
1429 and so on.
1431 * lex.c: Part of ffelex_finish_statement_() now done
1432 by new function ffelex_prepare_eos_(), so that, in one
1433 popular case, the EOS can be prepared while the pointer
1434 is at the end of the non-continued line instead of the
1435 end of the line that marks no continuation. This improves
1436 the appearance of diagnostics substantially.
1438 Mon Feb 10 12:44:06 1997 Craig Burley <burley@gnu.ai.mit.edu>
1440 * Make-lang.in: runtime Makefile's, and include/f2c.h,
1441 also depend on f/runtime/configure and f/runtime/libU77/configure.
1443 Fix various libU77 routines:
1444 * com-rt.def (FFECOM_gfrtCTIME, FFECOM_gfrtMCLOCK,
1445 FFECOM_gfrtTIME): These now use INTEGER*8 for time values,
1446 for compatibility with systems like Alpha.
1447 (FFECOM_gfrtSYSTEM_CLOCK, FFECOM_gfrtTTYNAM): Delete incorrect
1448 trailing underscore in routine names.
1449 * intrin.c, intrin.def: Support INTEGER*8 return values and
1450 arguments ('4'). Change FFEINTRIN_impCTIME, FFEINTRIN_impMCLOCK,
1451 and FFEINTRIN_impTIME accordingly.
1452 (ffeintrin_is_intrinsic): Don't give caller a clue about
1453 form of intrinsic -- shouldn't be needed at this point.
1455 Cope with generic intrinsics that are subroutines and functions:
1456 * com.c (ffecom_finish_symbol_transform_, ffecom_expr_transform_):
1457 Don't transform an intrinsic that is not known to be a subroutine
1458 or a function. (Maybe someday have to avoid transforming
1459 any intrinsic with an undecided or unknown implementation.)
1460 * expr.c (ffeexpr_declare_unadorned_,
1461 ffeexpr_declare_parenthesized_): Ok to invoke generic
1462 intrinsic that has at least one subroutine form as a
1463 subroutine.
1464 Ok to pass intrinsic as actual arg if it has a known specific
1465 intrinsic form that is valid as actual arg.
1466 (ffeexpr_declare_parenthesized_): An unknown kind of
1467 intrinsic has a paren_type chosen based on context.
1468 (ffeexpr_token_arguments_): Build funcref/subrref based
1469 on context, not on kind of procedure being called.
1470 * intrin.h, intrin.c (ffeintrin_is_intrinsic): Undo changes of
1471 Tue Feb 4 23:12:04 1997 by me, change all callers to leave
1472 intrinsics as FFEINFO_kindNONE at this point. (Some callers
1473 also had unused variables deleted as a result.)
1475 Enable all intrinsic groups (especially f90 and vxt):
1476 * target.h (FFETARGET_defaultSTATE_DCP, FFETARGET_defaultSTATE_F2C,
1477 FFETARGET_defaultSTATE_F90, FFETARGET_defaultSTATE_MIL,
1478 FFETARGET_defaultSTATE_UNIX, FFETARGET_defaultSTATE_VXT):
1479 Delete these macros, let top.c set them directly.
1480 * top.c (ffeintrinsic_state_dcp_, ffe_intrinsic_state_f2c_,
1481 ffe_intrinsic_state_f90_, ffe_intrinsic_state_mil_,
1482 ffe_intrinsic_state_unix_, ffe_intrinsic_state_vxt_):
1483 Enable all these directly.
1485 Sat Feb 8 03:21:50 1997 Craig Burley <burley@gnu.ai.mit.edu>
1487 * g77.c: Incorporate recent changes to ../gcc.c.
1488 For version magic (e.g. `g77 -v'), instead of compiling
1489 /dev/null, write, compile, run, and then delete a small
1490 program that prints the version numbers of the three
1491 components of libf2c (libF77, libI77, and libU77),
1492 so we get this info with bug reports.
1493 Also, this change reduces the chances of accidentally
1494 linking to an old (complex-alias-problem) libf2c.
1495 Fix `-L' so the argument is expected in `-Larg'.
1497 * com.h (FFECOM_f2cLONGINT): For INTEGER*8 support in f2c.h,
1498 dynamically determine proper type here, instead of
1499 assuming `long long int' is correct.
1501 Tue Feb 4 23:12:04 1997 Craig Burley <burley@gnu.ai.mit.edu>
1503 Add libU77 library from Dave Love <d.love@dl.ac.uk>:
1504 * Make-lang.in (f77-runtime): Depend on new Makefile.
1505 (f/runtime/libU77/Makefile): New rule.
1506 Also configure libU77.
1507 ($(srcdir)/f/runtime/configure: Use Makefile.in,
1508 so configuration doesn't have to have happened.
1509 (f77.mostlyclean, f77.clean, f77.distclean,
1510 f77.maintainer-clean): Some fixups here, but more work
1511 needed.
1512 (RUNTIMESTAGESTUFF): Add libU77's config.status.
1513 (LIBU77STAGESTUFF, f77.stage1, f77.stage2, f77.stage3,
1514 f77.stage4): New macro, appropriate uses added.
1515 * com-rt.def: Add libU77 procedures.
1516 * com.c (ffecom_f2c_ptr_to_integer_type_node,
1517 ffecom_f2c_ptr_to_real_type_node): New type nodes.
1518 (FFECOM_rttypeCHARACTER_): New type of run-time function.
1519 (ffecom_char_args_): Handle CHARACTER*n intrinsics
1520 where n != 1 here, instead of in ffecom_expr_intrinsic_.
1521 (ffecom_expr_intrinsic_): New code to handle new
1522 intrinsics.
1523 In particular, change how FFEINTRIN_impFLUSH is handled.
1524 (ffecom_make_gfrt_): Handle new type of run-time function.
1525 (ffecom_init_0): Initialize new type nodes.
1526 * config-lang.in: New libU77 directory.
1527 * intrin.h, intrin.c (ffeintrin_is_intrinsic): Handle
1528 potential generic for subroutine _and_ function
1529 specifics via two new arguments. All callers changed.
1530 Properly ignore deleted/disabled intrinsics in resolving
1531 generics.
1532 (ffeintrin_check_, ffeintrin_init_0): Handle CHARACTER intrinsics of (*)
1533 length.
1534 * intrin.def: Permission granted by FSF to place this in
1535 public domain, which will allow it to serve as source
1536 for both g77 program and its documentation.
1537 Add libU77 intrinsics.
1538 (FLUSH): Now a generic, not specific, intrinsic.
1539 (DEFIMP): Now support return modifier for CHARACTER intrinsics.
1541 * com-rt.def (FFECOM_gfrtDIM, FFECOM_gfrtERF,
1542 FFECOM_gfrtERFC, FFECOM_gfrtEXP, FFECOM_gfrtSIGN,
1543 FFECOM_gfrtSIN, FFECOM_gfrtSINH, FFECOM_gfrtTAN,
1544 FFECOM_gfrtTANH, FFECOM_gfrtPOW_RI): Change "&r" to "&f".
1546 Sat Feb 1 12:15:09 1997 Craig Burley <burley@gnu.ai.mit.edu>
1548 * Version 0.5.19.1 released.
1550 * com.c (ffecom_expr_, ffecom_expr_intrinsic_,
1551 ffecom_tree_divide_): FFECOM_gfrtPOW_ZI,
1552 FFECOM_gfrtCONJG, FFECOM_gfrtDCONJG,
1553 FFECOM_gfrtCCOS, FFECOM_gfrtCDCOS,
1554 FFECOM_gfrtCLOG, FFECOM_gfrtCDLOG,
1555 FFECOM_gfrtCSIN, FFECOM_gfrtCDSIN,
1556 FFECOM_gfrtCSQRT, FFECOM_gfrtCDSQRT,
1557 FFECOM_gfrtDIV_CC, FFECOM_gfrtDIV_ZZ: These all require
1558 result to _not_ overlap one or more inputs.
1560 Sat Feb 1 00:25:55 1997 Craig Burley <burley@gnu.ai.mit.edu>
1562 * com.c (ffecom_init_0): Do internal checks only if
1563 -fset-g77-defaults not specified.
1565 Fix %LOC(), LOC() to return sufficiently wide type:
1566 * com.h, com.c (ffecom_pointer_kind_, ffecom_label_kind_,
1567 ffecom_pointer_kind(), ffecom_label_kind()): New globals
1568 and accessor macros hold kind for integer pointers on target
1569 machine.
1570 (ffecom_init_0): Determine narrowest INTEGER type that
1571 can hold a pointer (usually INTEGER*4 or INTEGER*8),
1572 store it in ffecom_pointer_kind_, etc.
1573 * expr.c (ffeexpr_cb_end_loc_): Use right type for %LOC().
1574 * intrin.c (ffeintrin_check_, ffeintrin_init_0): Support
1575 new 'p' kind for type of intrinsic.
1576 * intrin.def (FFEINTRIN_impLOC): Returns "Ip" instead of "I1",
1577 so LOC() type is correct for target machine.
1579 Support -fugly-assign:
1580 * lang-options.h, top.h, top.c (ffe_decode_option):
1581 Accept -fugly-assign and -fno-ugly-assign.
1582 * com.c (ffecom_expr_): Handle -fugly-assign.
1583 * expr.c (ffeexpr_finished_): Check right type for ASSIGN
1584 contexts.
1586 Fri Jan 31 14:30:00 1997 Craig Burley <burley@gnu.ai.mit.edu>
1588 Remove last vestiges of -fvxt-not-f90:
1589 * stb.c (ffestb_R10012_, ffestb_R10014_, ffestb_V0201_):
1590 top.c, top.h:
1592 Fri Jan 31 02:13:54 1997 Craig Burley <burley@gnu.ai.mit.edu>
1594 * top.c (ffe_decode_option): Warn if -fugly is specified,
1595 it'll go away soon.
1597 * symbol.h: No need to #include "bad.h".
1599 Reorganize features from -fvxt-not-f90 to -fvxt:
1600 * lang-options.h, top.h, top.c:
1601 Accept -fvxt and -fno-vxt, but not -fvxt-not-f90 or -ff90-not-vxt.
1602 Warn if the latter two are used.
1603 * expr.c (ffeexpr_nil_rhs_): Double-quote means octal constant.
1604 (ffeexpr_token_rhs_): Double-quote means octal constant.
1605 * target.h (FFETARGET_defaultIS_VXT_NOT_90): Delete macro
1606 definition, no longer needed.
1608 Make some -ff90 features the default:
1609 * data.c (ffedata_value): DATA implies SAVE.
1610 * src.h (ffesrc_is_name_noninit): Underscores always okay.
1612 Fix up some more #error directives by quoting their text:
1613 * bld.c (ffebld_constant_is_zero):
1614 * target.h:
1616 Sat Jan 18 18:22:09 1997 Craig Burley <burley@gnu.ai.mit.edu>
1618 * g77.c (lookup_option, main): Recognize `-Xlinker',
1619 `-Wl,', `-l', `-L', `--library-directory', `-o',
1620 `--output'.
1621 (lookup_option): Don't depend on SWITCH_TAKES_ARG
1622 being correct, it might or might not have `-x' in
1623 it depending on host.
1624 Return NULL argument if it would be an empty string.
1625 (main): If no input files (by gcc.c's definition)
1626 but `-o' or `--output' specified, produce diagnostic
1627 to avoid overwriting output via gcc.
1628 Recognize C++ `+e' options.
1629 Treat -L as another non-magical option (like -B).
1630 Don't append_arg `-x' twice.
1632 Fri Jan 10 23:36:00 1997 Craig Burley <burley@gnu.ai.mit.edu>
1634 * top.c [BUILT_FOR_270] (ffe_decode_option): Make
1635 -fargument-noalias-global the default.
1637 Fri Jan 10 07:42:27 1997 Craig Burley <burley@gnu.ai.mit.edu>
1639 Enable inlining of previously-compiled program units:
1640 * com.c (ffecom_do_entry_, ffecom_start_progunit_):
1641 Register new public function in ffeglobal database.
1642 (ffecom_sym_transform_): Any GLOBAL or potentially GLOBAL
1643 symbol should be looked up in ffeglobal database and
1644 that tree node used, if found. That way, gcc knows
1645 the references are to those earlier definitions, so it
1646 can emit shorter branches/calls, inline, etc.
1647 (ffecom_transform_common_): Minor change for clarity.
1648 * expr.c (ffeexpr_sym-lhs_call_, ffeexpr_sym_lhs_extfunc_,
1649 ffeexpr_sym_rhs_actualarg_, ffeexpr_paren_rhs_let_,
1650 ffeexpr_token_funsubstr_): Globalize symbol as needed.
1651 * global.c (ffeglobal_promoted): New function to look up
1652 existing local symbol in ffeglobal database.
1653 * global.h: Declare new function.
1654 * name.h (ffename_token): New macro, plus alphabetize.
1655 * stc.c (ffestc_R1207_item): Globalize EXTERNAL symbol.
1656 * stu.c (ffestu_sym_end_transition, ffestu_sym_exec_transition):
1657 Globalize symbol as needed.
1658 * symbol.h, symbol.c (ffesymbol_globalize): New function.
1660 Thu Jan 9 14:20:00 1997 Craig Burley <burley@gnu.ai.mit.edu>
1662 * ste.c (ffeste_R809): Produce a diagnostic for SELECT CASE
1663 on CHARACTER type, instead of crashing.
1665 Thu Jan 9 00:52:45 1997 Craig Burley <burley@gnu.ai.mit.edu>
1667 * stc.c (ffestc_order_entry_, ffestc_order_format_,
1668 ffestc_R1226): Allow ENTRY and FORMAT before IMPLICIT
1669 NONE, by having them transition only to state 1 instead
1670 of state 2 (which is disallowed by IMPLICIT NONE).
1672 Mon Jan 6 22:44:53 1997 Craig Burley <burley@gnu.ai.mit.edu>
1674 Fix AXP bug found by Rick Niles (961201-1.f):
1675 * com.c (ffecom_init_0): Undo my 1996-05-14 change, as
1676 it is incorrect and prevented easily finding this bug.
1677 * target.h [__alpha__] (ffetargetReal1, ffetargetReal2):
1678 Use int instead of long.
1679 (ffetarget_cvt_r1_to_rv_, ffetarget_cvt_rv_to_r1_,
1680 ffetarget_cvt_r2_to_rv_, ffetarget_cvt_rv_to_r2_):
1681 New functions that intercede for callers of
1682 REAL_VALUE_(TO|UNTO)_TARGET_(SINGLE|DOUBLE).
1683 All callers changed, and damaging casts to (long *) removed.
1685 Sun Jan 5 03:26:11 1997 Craig Burley <burley@gnu.ai.mit.edu>
1687 * Make-lang.in (g77, g77-cross): Depend on both g77.c and
1688 zzz.c, in $(srcdir)/f/.
1690 Better design for -fugly-assumed:
1691 * stc.c (ffestc_R501_item, ffestc_R524_item,
1692 ffestc_R547_item_object): Pass new is_ugly_assumed flag.
1693 * stt.c, stt.h (ffestt_dimlist_as_expr,
1694 ffestt_dimlist_type): New is_ugly_assumed flag now
1695 controls whether "1" is treated as "*".
1696 Don't treat "2-1" or other collapsed constants as "*".
1698 Sat Jan 4 15:26:22 1997 Craig Burley <burley@gnu.ai.mit.edu>
1700 * stb.c (ffestb_R10012_): Don't confirm on FORMAT(A,)
1701 or even FORMAT(A,,B), as R1229 only warns about the
1702 former currently, and this seems reasonable.
1704 Improvements to diagnostics:
1705 * sta.c (ffesta_second_): Don't add any ffestb parsers
1706 unless they're specifically called for.
1707 Set up ffesta_tokens[0] before calling ffestc_exec_transition,
1708 else stale info might get used.
1709 (ffesta_save_): Do a better job picking which parser to run
1710 after running all parsers with no confirmed possibles.
1711 (FFESTA_maxPOSSIBLES_): Decrease from 100 now that so few
1712 possibles are ever on the list at a given time.
1713 (struct _ffesta_possible): Add named attribute.
1714 (ffesta_add_possible_exec_, ffesta_add_possible_nonexec_):
1715 Make these into macros that call a single function that now
1716 sets the named attribute.
1717 (ffesta_add_possible_unnamed_exec_,
1718 ffeseta_add_possible_unnamed_nonexec_): New macros.
1719 (ffesta_second_): Designate unnamed possibles as
1720 appropriate.
1721 * stb.c (ffestb_R1229, ffestb_R12291_): Use more general
1722 diagnostic, so things like "POINTER (FOO, BAR)" are
1723 diagnosed as unrecognized statements, not invalid statement
1724 functions.
1725 * stb.h, stb.c (ffestb_unimplemented): Remove function.
1727 1996-12-30 Dave Love <d.love@dl.ac.uk>
1729 * com.c: #include libU77/config.h
1730 (ffecom_f2c_ptr_to_integer_type_node,
1731 ffecom_f2c_ptr_to_integer_type_node): New variables.
1732 (ffecom_init_0): Use them.
1733 (ffecom_expr_intrinsic_): Many news cases for libU77 intrinsics.
1735 * com-rt.def: New definitions for libU77.
1736 * intrin.def: Likewise. Also correct ftell arg spec.
1738 * Makefile.in (f/runtime/libU77/config.h): New target for com.c
1739 dependency.
1740 * Make-lang.in (f771): Depend on f/runtime/Makefile for the above.
1742 Sat Dec 28 12:28:29 1996 Craig Burley <burley@gnu.ai.mit.edu>
1744 * stt.c (ffestt_dimlist_type): Treat ([...,]1) in dimlist
1745 as ([...,]*) if -fugly-assumed, so assumed-size array
1746 detected early enough.
1748 Thu Dec 19 14:01:57 1996 Craig Burley <burley@gnu.ai.mit.edu>
1750 * target.h (FFETARGET_REAL_VALUE_FROM_INT_): Conditionalize
1751 definition on BUILT_FOR_280, not BUILT_WITH_280, since
1752 the name of the macro was (properly) changed since 0.5.19.
1754 Fix warnings/errors resulting from ffetargetOffset becoming
1755 `long long int' instead of `unsigned long' as of 0.5.19,
1756 while ffebitCount remains `unsigned long':
1757 * bld.c (ffebld_constantarray_dump): Avoid warnings by
1758 using loop var of appropriate type, and using casts.
1759 * com.c (ffecom_expr_): Use right type for loop var.
1760 (ffecom_sym_transform_, ffecom_transform_equiv_):
1761 Cast to right type in assertions.
1762 * data.c (ffedata_gather_, ffedata_value_): Cast to right
1763 type in assertions and comparisons.
1765 Wed Dec 18 12:07:11 1996 Craig Burley <burley@gnu.ai.mit.edu>
1767 Patch from Alexandre Oliva <oliva@dcc.unicamp.br>:
1768 * Makefile.in (all.indirect): Don't pass -bbigtoc option
1769 to GNU ld.
1771 Cope with new versions of gcc:
1772 * com.h (BUILT_FOR_280): New macro.
1773 * com.c (ffecom_ptr_to_expr): Conditionalize test of
1774 OFFSET_REF.
1775 (ffecom_build_complex_constant_): Conditionalize calling
1776 sequence for build_complex.
1778 Sat Dec 7 07:15:17 1996 Craig Burley <burley@gnu.ai.mit.edu>
1780 * Version 0.5.19 released.
1782 Fri Dec 6 12:23:55 1996 Craig Burley <burley@gnu.ai.mit.edu>
1784 * g77.c: Default to assuming "f77" is in $LANGUAGES, since
1785 the LANGUAGE_F77 macro isn't defined by anyone anymore (but
1786 might as well leave the no-f77 code in just in case).
1787 * Make-lang.in (g77, g77-cross): Don't define LANGUAGE_F77
1788 anymore.
1790 1996-12-06 Dave Love <d.love@dl.ac.uk>
1792 * Make-lang.in (g77, g77-cross): Revert to building `g77' or not
1793 conditional on `f77' in LANGUAGES.
1795 Wed Dec 4 13:08:44 1996 Craig Burley <burley@gnu.ai.mit.edu>
1797 * Make-lang.in (g77, g77-cross): No libs or lib dependencies
1798 in case where "f77" is not in $LANGUAGES.
1800 * lex.c (ffelex_image_char_, ffelex_file_fixed,
1801 ffelex_file_free): Fixes to properly handle lines with
1802 null character, and too-long lines as well.
1804 * lex.c: Call ffebad_start_msg_lex instead of
1805 ffebad_start_msg throughout.
1807 Sun Dec 1 21:19:55 1996 Craig Burley <burley@gnu.ai.mit.edu>
1809 Fix-up for 1996-11-25 changes:
1810 * com.c (ffecom_member_phase2_): Subtract out 0 offset for
1811 elegance and consistency with EQUIVALENCE aggregates.
1812 (ffecom_sym_transform_): Ditto for LOCAL/COMMON, and
1813 ensure we get the same parent storage area.
1814 * data.c (ffedata_gather_, ffedata_value_): Subtract out
1815 aggregate offset.
1817 Wed Nov 27 13:55:57 1996 Craig Burley <burley@gnu.ai.mit.edu>
1819 * proj.h: Quote the text of the #error message, to avoid
1820 strange-looking diagnostics from non-gcc ANSI compilers.
1822 * top.c: Make -fno-debug-kludge the default.
1824 Mon Nov 25 20:13:45 1996 Craig Burley <burley@gnu.ai.mit.edu>
1826 Provide more info on EQUIVALENCE mismatches:
1827 * bad.def (FFEBAD_EQUIV_MISMATCH): More detailed message.
1828 * equiv.c (ffeequiv_layout_local_, ffeequiv_layout_cblock):
1829 More details for FFEBAD_EQUIV_MISMATCH.
1831 Fix problem with EQUIVALENCE handling:
1832 * equiv.c (ffeequiv_layout_local_): Redesign algorithm --
1833 old one was broken, resulting in rejection of good code.
1834 (ffeequiv_offset_): Add argument, change callers.
1835 Clean up the code, fix up the (probably unused) negative-value
1836 case for SYMTER.
1837 * com.c (ffecom_sym_transform_): For local EQUIVALENCE
1838 member, subtract out aggregate offset (which is <= 0).
1840 Thu Nov 21 12:44:56 1996 Craig Burley <burley@gnu.ai.mit.edu>
1842 Change type of ffetargetOffset from `unsigned long' to `long long':
1843 * bld.c (ffebld_constantarray_dump): Change printf formats.
1844 * storag.c (ffestorag_dump): Ditto.
1845 * symbol.c (ffesymbol_report): Ditto.
1846 * target.h (ffetargetOffset_f): Ditto and change type itself.
1848 Handle situation where list of languages does not include f77:
1849 * Make-lang.in: Define LANGUAGE_F77 to 1 only if `f77' is in
1850 the $LANGUAGES macro for the build.
1851 * g77.c: Compile to a (nearly) no-op program if LANGUAGE_F77
1852 is not defined to 1.
1854 Fixes to delay confirmation of READ, WRITE, and GOTO statements
1855 so the corresponding assignments to same-named CHAR*(*) arrays
1856 work:
1857 * stb.c (ffestb_R90915_, ffestb_91014_): New functions.
1858 (ffestb_goto3_, ffestb_goto5_): Move confirmation from 3 to 5
1859 for the OPEN_PAREN case.
1860 (ffestb_R9091_, ffestb_R9094_, ffestb_R90913_, ffestb_R90914_,
1861 ffestb_R91012_, ffestb_R91013_): Use new functions, and confirm
1862 except for the OPEN_PAREN case.
1864 Fixes to not confirm declarations with an open paren where
1865 an equal sign or other assignment-like token might be, so the
1866 corresponding assignments to same-named CHAR*(*) arrays work:
1867 (ffestb_decl_entsp_5_): Move assertion so we crash on that first,
1868 if it turns out to be wrong, before the less-debuggable crash
1869 on mistaken confirmation.
1870 (ffestb_decl_entsp_6_, ffestb_decl_entsp_7_, ffestb_decl_entsp_8_):
1871 Include OPEN_PAREN in list of assignment-only tokens.
1873 Fix more diagnosed-crash bugs:
1874 * stu.c (ffestu_sym_end_transition): ANY-ize an adjustable array
1875 with bad dimension expressions even if still stateUNCERTAIN.
1876 (ffestu_symter_end_transition_, ffestu_symter_exec_transition_):
1877 Return TRUE for opANY as well.
1878 For code elegance, move opSYMTER case into first switch.
1880 1996-11-17 Dave Love <d.love@dl.ac.uk>
1882 * lex.c: Fix last change.
1884 1996-11-14 Dave Love <d.love@dl.ac.uk>
1886 * Make-lang.in, config-lang.in: Remove the (broken) libU77 stuff,
1887 pending 0.5.20.
1889 Thu Nov 14 15:40:59 1996 Craig Burley <burley@gnu.ai.mit.edu>
1891 * bad.def (FFEBAD_UNIMPL_STMT): Explain that invalid
1892 intrinsic references can trigger this message, too.
1894 1996-11-12 Dave Love <d.love@dl.ac.uk>
1896 * lex.c: Declare dwarfout routines.
1898 * config-lang.in: Sink grep o/p.
1900 Mon Nov 11 14:21:13 1996 Craig Burley <burley@gnu.ai.mit.edu>
1902 * g77.c (main): Might as well print version number
1903 for --verbose as well.
1905 Thu Nov 7 18:41:41 1996 Craig Burley <burley@gnu.ai.mit.edu>
1907 * expr.c, lang-options.h, target.h, top.c, top.h: Split out
1908 remaining -fugly stuff into -fugly-logint and -fugly-comma,
1909 leaving -fugly as simply a `macro' that expands into other
1910 options, and eliminate defaults for some of the ugly stuff
1911 in target.h.
1913 * Make-lang.in (gcc-cross): Compile zzz.c, not version.o (!),
1914 in to get version info for this target.
1916 * config-lang.in: Test for GBE patch application based
1917 on whether 2.6.x or 2.7.x GBE is detected.
1919 Wed Nov 6 14:19:45 1996 Craig Burley <burley@gnu.ai.mit.edu>
1921 * Make-lang.in (g77): Compile zzz.c in to get version info.
1922 * g77.c: Add support for --help and --version.
1924 * g77.c (lookup_option): Short-circuit long-winded tests
1925 when second char is not hyphen, just to save a spot of time.
1927 Sat Nov 2 13:50:31 1996 Craig Burley <burley@gnu.ai.mit.edu>
1929 * intrin.def: Add FTELL and FSEEK intrinsics, plus new
1930 `g' codes for alternate-return (GOTO) arguments.
1931 * intrin.c (ffeintrin_check_): Support `g' codes.
1932 * com-rt.def: Add ftell_() and fseek_() to database.
1933 * com.c (ffecom_expr_intrinsic_): Ditto. Also, let each
1934 subroutine intrinsic decide for itself what to do with
1935 tree_type, the default being NULL_TREE once again (so
1936 ffecom_call_ doesn't think it's supposed to cast the
1937 function call to the type in the fall-through case).
1939 * ste.c (ffeste_R909_finish): Don't special-case list-directed
1940 I/O, now that libf2c can return non-zero status codes.
1941 (ffeste_R910_finish): Ditto.
1942 (ffeste_io_call_): Simplify logic.
1943 (ffeste_io_impdo_):
1944 (ffeste_subr_beru_):
1945 (ffeste_R904):
1946 (ffeste_R907):
1947 (ffeste_R909_start):
1948 (ffeste_R909_item):
1949 (ffeste_R909_finish):
1950 (ffeste_R910_start):
1951 (ffeste_R910_item):
1952 (ffeste_R910_finish):
1953 (ffeste_R911_start):
1954 (ffeste_R923A): Ditto all the above.
1956 Thu Oct 31 20:56:28 1996 Craig Burley <burley@gnu.ai.mit.edu>
1958 * config-lang.in, Make-lang.in: Rename flag file
1959 build-u77 to build-libu77, for consistency with
1960 install-libf2c and such.
1962 * config-lang.in: Don't complain about failure to patch
1963 if pre-2.7.0 gcc is involved (since our patch for that
1964 doesn't add support for tooning).
1966 Sat Oct 26 05:56:51 1996 Craig Burley <burley@gnu.ai.mit.edu>
1968 * bad.def (FFEBAD_TYPELESS_TOO_LARGE): Remove this
1969 unused and redundant diagnostic.
1971 Sat Oct 26 00:45:42 1996 Craig Burley <burley@gnu.ai.mit.edu>
1973 * target.c (ffetarget_integerhex): Fix dumb bug.
1975 1996-10-20 Dave Love <d.love@dl.ac.uk>
1977 * gbe/2.7.2.1.diff: New file.
1979 * Makefile.in (F771_LDFLAGS): Add -bbigtoc for AIX4.1 up, suggested by
1980 endo@material.tohoku.ac.jp [among others!].
1982 Sat Oct 19 03:11:14 1996 Craig Burley <burley@gnu.ai.mit.edu>
1984 * bad.def, bld.c, bld.h, expr.c, lang-options.h, target.c,
1985 target.h, top.c, top.h (ffebld_constant_new_integerbinary,
1986 ffebld_constant_new_integerhex, ffebld_constant_new_integeroctal,
1987 ffeexpr_token_name_apos_name_, ffetarget_integerbinary,
1988 ffetarget_integerhex, ffetarget_integeroctal): Support
1989 new -fno-typeless-boz option with new functions, mods to
1990 existing octal-handling functions, new macros, new error
1991 messages, and so on.
1993 * com.c, lang-options.h, top.c, top.h (ffecom_notify_primary_entry):
1994 Print program unit name on stderr if -fno-silent (new option).
1996 * lang-options.h, top.c, top.h, stt.c (ffestt_dimlist_as_expr):
1997 Treat ([...,]1) in dimlist as ([...,]*) if -fugly-assumed
1998 (new option).
2000 * lang-options.h: Comment out options duplicated in gcc/toplev.c,
2001 because, somehow, having them commented in and building on my
2002 DEC Alpha results in a cc1 that always segfaults, and gdb that
2003 also segfaults whenever it debugs it up to init_lex() calling
2004 xmalloc() or so.
2006 Thu Oct 17 00:39:27 1996 Craig Burley <burley@gnu.ai.mit.edu>
2008 * stb.c (ffestb_R10013_): Don't change meaning of .sign until
2009 after previous meaning/value used to set sign of value
2010 (960507-1.f).
2012 Sun Oct 13 22:15:23 1996 Craig Burley <burley@gnu.ai.mit.edu>
2014 * top.c (ffe_decode_option): Don't set back-end flags
2015 that are nonexistent prior to gcc 2.7.0.
2017 Sun Oct 13 12:48:45 1996 Craig Burley <burley@gnu.ai.mit.edu>
2019 * com.c (convert): Don't convert emulated complex expr to
2020 real (via REALPART_EXPR) if the target type is (emulated)
2021 complex.
2023 Wed Oct 2 21:57:12 1996 Craig Burley <burley@gnu.ai.mit.edu>
2025 * com.c (ffecom_debug_kludge_): Set DECL_IN_SYSTEM_HEADER so
2026 -Wunused doesn't complain about these manufactured decls.
2027 (ffecom_expr_): Ditto, for original (non-ASSIGN'ed) variable.
2028 (ffecom_transform_equiv_): Clear DECL_IGNORED_P for aggregate
2029 area so it shows up as a debug-accessible symbol.
2030 (pushdecl): Default for "invented" identifiers (a g77-specific
2031 concept for now) is that they are artificial, in system header,
2032 ignored for debugging purposes, used, and (for types) suppressed.
2033 This ought to be overkill.
2035 Fri Sep 27 23:13:07 1996 Craig Burley <burley@gnu.ai.mit.edu>
2037 * ste.c (ffeste_begin_iterdo_, ffeste_end_iterdo_): Support
2038 one-trip DO loops (F66-style).
2039 * lang-options.h, top.c, top.h (-fonetrip): New option.
2041 Thu Sep 26 00:18:40 1996 Craig Burley <burley@gnu.ai.mit.edu>
2043 * com.c (ffecom_debug_kludge_): New function.
2044 (ffecom_sym_transform_): Use new function for COMMON and EQUIVALENCE
2045 members.
2047 * lang-options.h, top.c, top.h (-fno-debug-kludge):
2048 New option.
2050 1996-09-24 Dave Love <d.love@dl.ac.uk>
2052 * Make-lang.in (include/f2c.h):
2053 Remove dependencies on xmake_file and tmake_file.
2054 They expand inconsistently in 2.8 c.f. 2.7; $(GCC_PARTS) depends on
2055 them anyhow.
2057 1996-09-22 Dave Love <d.love@dl.ac.uk>
2059 * config-lang.in: Add --enable-libu77 option handling.
2061 * Make-lang.in:
2062 Conditionally add --enable-libu77 when running runtime configure.
2063 Define LIBU77STAGESTUFF and use it in relevant rules.
2065 1996-08-21 Dave Love <d.love@dl.ac.uk>
2067 * Make-lang.in (f77-runtime):
2068 `stmp-hdrs' should have been `stmp-headers'.
2070 1996-08-20 Dave Love <d.love@dl.ac.uk>
2072 * Make-lang.in (f77-runtime):
2073 Depend on stmp-hdrs, not stmp-int-hdrs, since libF77
2074 needs float.h.
2076 Sat Jun 22 18:17:11 1996 Craig Burley <burley@gnu.ai.mit.edu>
2078 * com.c (ffecom_tree_divide_): Fix RECORD_TYPE case to
2079 look at type of first field, properly, to determine
2080 whether to call c_div or z_div.
2082 Tue Jun 4 04:27:18 1996 Craig Burley <burley@gnu.ai.mit.edu>
2084 * com.c (ffecom_build_complex_constant_): Explicitly specify
2085 TREE_PURPOSE.
2086 (ffecom_expr_): Fix thinko.
2087 (ffecom_2): For COMPLEX_EXPR, explicitly specify TREE_PURPOSE.
2089 Mon May 27 16:23:43 1996 Craig Burley <burley@gnu.ai.mit.edu>
2091 Changes to optionally avoid gcc's back-end complex support:
2092 * com.c (ffecom_stabilize_aggregate_): New function.
2093 (ffecom_convert_to_complex_): New function.
2094 (ffecom_make_complex_type_): New function.
2095 (ffecom_build_complex_constant_): New function.
2096 (ffecom_expr_): For opCONVERT of non-COMPLEX to COMPLEX,
2097 don't bother explicitly converting to the subtype first,
2098 because gcc does that anyway, and more code would have
2099 to be added to find the subtype for the emulated-complex
2100 case.
2101 (ffecom_f2c_make_type_): Use ffecom_make_complex_type_
2102 instead of make_node etc. to make a complex type.
2103 (ffecom_1, ffecom_2): Translate operations on COMPLEX operands
2104 to appropriate operations when emulating complex.
2105 (ffecom_constantunion): Use ffecom_build_complex_constant_
2106 instead of build_complex to build a complex constant.
2107 (ffecom_init_0): Change point at which types are laid out
2108 for improved consistency.
2109 Use ffecom_make_complex_type_ instead of make_node etc.
2110 to make a complex type.
2111 Always calculate storage sizes from TYPE_SIZE, never TYPE_PRECISION.
2112 (convert): Use e, not expr, since we've copied into that anyway.
2113 For RECORD_TYPE cases, do emulated-complex conversions.
2114 (ffecom_f2c_set_lio_code_): Always calculate storage sizes
2115 from TYPE_SIZE, never TYPE_PRECISION.
2116 (ffecom_tree_divide_): Allow RECORD_TYPE to also be handled
2117 by run-time library.
2118 (ffecom_expr_intrinsic_): Handle possible RECORD_TYPE as argument
2119 to AIMAG intrinsic.
2121 * top.h, top.c, lang-options.h: Support new -f(no-)emulate-complex option.
2123 * com.c (ffecom_sym_transform_): Clarify and fix typos in comments.
2125 Mon May 20 02:06:27 1996 Craig Burley <burley@gnu.ai.mit.edu>
2127 * target.h: Use new REAL_VALUE_UNTO_TARGET_* macros instead
2128 of REAL_VALUE_FROM_TARGET_DOUBLE and _SINGLE.
2129 Explicitly use long instead of HOST_WIDE_INT for emulation
2130 of ffetargetReal1 and ffetargetReal2.
2132 1996-05-20 Dave Love <d.love@dl.ac.uk>
2134 * config-lang.in:
2135 Test for patch being applied with flag_move_all_movables in toplev.c.
2137 * install.texi (Patching GNU Fortran):
2138 Mention overriding X_CFLAGS rather than
2139 editing proj.h on SunOS4.
2141 * Make-lang.in (F77_FLAGS_TO_PASS):
2142 Add X_CFLAGS (convenient for SunOS4 kluge, in
2143 particular).
2144 (f77.{,mostly,dist}clean): Reorder things, in particular not to delete
2145 Makefiles too early.
2147 * g77.c (DEFAULT_SWITCH_TAKES_ARG): Define a la gcc.c in the
2148 current GCC snapshot.
2150 Tue May 14 00:24:07 1996 Craig Burley <burley@gnu.ai.mit.edu>
2152 Changes for DEC Alpha AXP support:
2153 * com.c (ffecom_init_0): REAL_ARITHMETIC means internal
2154 REAL/DOUBLE PRECISION might well have a different size
2155 than the compiled type, so don't crash if this is the
2156 case.
2157 * target.h: Use `int' for ffetargetInteger1,
2158 ffetargetLogical1, and magical tests. Set _f format
2159 strings accordingly.
2161 Tue Apr 16 14:08:28 1996 Craig Burley <burley@gnu.ai.mit.edu>
2163 * top.c (ffe_decode_option): -Wall no longer implies
2164 -Wsurprising.
2166 Sat Apr 13 14:50:06 1996 Craig Burley <burley@gnu.ai.mit.edu>
2168 * com.c (ffecom_char_args_): If item is error_mark_node,
2169 set *length that way, too.
2171 * com.c (ffecom_expr_power_integer_): If either operand
2172 is error_mark_node, return that.
2174 * com.c (ffecom_intrinsic_len_): If item is error_mark_node,
2175 return that for length.
2177 * expr.c (ffeexpr_declare_unadorned_,
2178 ffeexpr_declare_parenthesized_): Instead of crashing
2179 on unexpected contexts, produce a diagnostic.
2181 * intrin.c (ffeintrin_check_), intrin.def (impSIGNAL):
2182 Allow procedure as second arg to SIGNAL intrinsic.
2184 * stu.c (ffestu_symter_end_transition_): New function.
2185 (ffestu_symter_exec_transition_): Return bool arg.
2186 Always transition symbol (don't inhibit when !whereNONE).
2187 (ffestu_sym_end_transition): If DUMMY/LOCAL arg has any
2188 opANY exprs in its dimlist, diagnose it so it doesn't
2189 make it through to later stages that try to deal with
2190 dimlist stuff.
2191 (ffestu_sym_exec_transition): If sym has any opANY exprs
2192 in its dimlist, diagnose it so it becomes opANY itself.
2194 * symbol.c (ffesymbol_error): If token arg is NULL,
2195 just ANY-ize the symbol -- don't produce diagnostic.
2197 Mon Apr 1 10:14:02 1996 Craig Burley <burley@gnu.ai.mit.edu>
2199 * Version 0.5.18 released.
2201 Mon Mar 25 20:52:24 1996 Craig Burley <burley@gnu.ai.mit.edu>
2203 * com.c (ffecom_expr_power_integer_): Don't generate code
2204 that compares COMPLEX (or, as it happens, REAL) via "LT_EXPR",
2205 since the back end crashes on that. (This code would never
2206 be executed anyway, but the test that avoids it has now been
2207 translated to control whether the code gets generated at all.)
2208 Fixes 960323-3.f.
2210 * com.c (ffecom_type_localvar_): Handle variable-sized
2211 dimension bounds expressions here, so they get calculated
2212 and saved on procedure entry. Fixes 960323-4.f.
2214 * com.c (ffecom_notify_init_symbol): Symbol has no init
2215 info at all if only zeros have been used to initialize it.
2216 Fixes 960324-0.f.
2218 * expr.c, expr.h (ffeexpr_type_combine): Renamed from
2219 ffeexpr_type_combine_ and now a public procedure; last arg now
2220 a token, instead of an internal structure used to extract a token.
2221 Now allows the outputs to be aliased with the inputs.
2222 Now allows a NULL token to mean "don't report error".
2223 (ffeexpr_reduced_bool2_, ffeexpr_reduced_eqop2_,
2224 ffeexpr_reduced_math2_, ffeexpr_reduced_power_,
2225 ffeexpr_reduced_relop2_): Handle new calling sequence for
2226 ffeexpr_type_combine.
2227 * (ffeexpr_convert): Don't put an opCONVERT node
2228 in just because the size is unknown; all downstream code
2229 should be able to deal without it being there anyway, and
2230 getting rid of it allows new intrinsic code to more easily
2231 combine types and such without generating bad code.
2232 * info.c, info.h (ffeinfo_kindtype_max): Rewrite to do
2233 proper comparison of size of types, not just comparison
2234 of their internal kind numbers (so I2.eq.I1 doesn't promote
2235 I1 to I2, rather the other way around).
2236 * intrin.c (ffeintrin_check_): Combine types of arguments
2237 in COL a la expression handling, for greater flexibility
2238 and permissiveness (though, someday, -fpedantic should
2239 report use of this kind of thing).
2240 Make sure Hollerith/typeless where CHARACTER expected is
2241 rejected. This all fixes 960323-2.f.
2243 * ste.c (ffeste_begin_iterdo_): Fix some more type conversions
2244 so INTEGER*2-laden DO loops don't crash at compile time on
2245 certain machines. Believed to fix 960323-1.f.
2247 * stu.c (ffestu_sym_end_transition): Certainly reject
2248 whereDUMMY not in any dummy list, whether stateUNCERTAIN
2249 or stateUNDERSTOOD. Fixes 960323-0.f.
2251 Tue Mar 19 13:12:40 1996 Craig Burley <burley@gnu.ai.mit.edu>
2253 * data.c (ffedata_value): Fix crash on opANY, and simplify
2254 the code at the same time.
2256 * Make-lang.in (f77-runtime): Also depends on lib[FI]77/Makefile...
2257 (include/f2c.h...): ...which in turn depend on */Makefile.in.
2258 (f77.rebuilt): Rebuild runtime stuff too.
2260 * intrin.c (ffeintrin_check_): Accommodate TYPELESS/HOLLERITH
2261 types, convert args as necessary, etc.
2263 * expr.c (ffeexpr_convert): Fix test for TYPELESS/HOLLERITH
2264 to obey the docs; crash if no source token when error.
2265 (ffeexpr_collapse_convert): Crash if no token when error.
2267 Mon Mar 18 15:51:30 1996 Craig Burley <burley@gnu.ai.mit.edu>
2269 * com.c (ffecom_init_zero_): Renamed from
2270 ffecom_init_local_zero_; now handles top-level
2271 (COMMON) initializations too.
2273 * bld.c (ffebld_constant_is_zero):
2274 * com.c (ffecom_symbol_transform_, ffecom_sym_transform_assign_,
2275 ffecom_transform_common_, ffecom_transform_equiv_):
2276 * data.c:
2277 * equiv.c:
2278 * equiv.h:
2279 * lang-options.h:
2280 * stc.c:
2281 * storag.c:
2282 * storag.h:
2283 * symbol.c:
2284 * symbol.h:
2285 * target.c:
2286 * target.h:
2287 * top.c:
2288 * top.h: All of this is mostly housekeeping-type changes
2289 to support -f(no-)zeros, i.e. not always stuff zero
2290 values into the initializer fields of symbol/storage objects,
2291 but still track that they have been given initial values.
2293 * bad.def: Fix wording for DATA-related diagnostics.
2295 * com.c (ffecom_sym_transform_assign_): Don't check
2296 any EQUIVALENCE stuff for local ASSIGN, the check was
2297 bad (crashing), and it's not necessary, anyway.
2299 * com.c (ffecom_expr_intrinsic_): For MAX and MIN,
2300 ignore null arguments as far arg[123], and fix handling
2301 of ANY arguments. (New intrinsic support now allows
2302 spurious trailing null arguments.)
2304 * com.c (ffecom_init_0): Add HOLLERITH (unsigned)
2305 equivalents for INTEGER*2, *4, and *8, so shift intrinsics
2306 and other things that need unsigned versions of signed
2307 types work.
2309 Sat Mar 16 12:11:40 1996 Craig Burley <burley@gnu.ai.mit.edu>
2311 * storag.c (ffestorag_exec_layout): Treat adjustable
2312 local array like dummy -- don't create storage object.
2313 * com.c (ffecom_sym_transform_): Allow for NULL storage
2314 object in LOCAL case (adjustable array).
2316 Fri Mar 15 13:09:41 1996 Craig Burley <burley@gnu.ai.mit.edu>
2318 * com.c (ffecom_sym_transform_): Allow local symbols
2319 with nonconstant sizes (adjustable local arrays).
2320 (ffecom_type_localvar_): Allow dimensions with nonconstant
2321 component (adjustable local arrays).
2322 * expr.c: Various minor changes to handle adjustable
2323 local arrays (a new case of stateUNCERTAIN).
2324 * stu.c (ffestu_sym_end_transition,
2325 ffestu_sym_exec_transition): Ditto.
2326 * symbol.def: Update docs to reflect these changes.
2328 * com.c (ffecom_expr_): Reduce space/time needed for
2329 opACCTER case by handling it here instead of converting
2330 it to opARRTER earlier on.
2331 (ffecom_notify_init_storage): Don't convert ACCTER to ARRTER.
2332 (ffecom_notify_init_symbol): Ditto.
2334 * com.c (ffecom_init_0): Crash and burn if any of the types'
2335 sizes, according to the GBE, disagrees with the sizes of
2336 the FFE's internal implementation. This might catch
2337 Alpha/SGI bugs earlier.
2339 Fri Mar 15 01:09:41 1996 Craig Burley <burley@gnu.ai.mit.edu>
2341 * com-rt.def, com.c, com.h: Changes for rewrite of intrinsic
2342 handling.
2343 * com.c (ffecom_arglist_expr_): New function.
2344 (ffecom_widest_expr_type_): New function.
2345 (ffecom_expr_intrinsic_): Reorganize, some rewriting.
2346 (ffecom_f2c_make_type_): Layout complex types.
2347 (ffecom_gfrt_args_): New function.
2348 (ffecom_list_expr): Trivial change for consistency.
2350 * expr.c (ffeexpr_token_name_rhs_): Go back to getting
2351 type from specific, not implementation, info.
2352 (ffeexpr_token_funsubstr_): Set intrinsic implementation too!
2353 * intrin.c: Major rewrite of most portions.
2354 * intrin.def: Major rearchitecting of tables.
2355 * intrin.h (ffeintrin_basictype, ffeintrin_kindtype):
2356 Now (once again) take ffeintrinSpec as arg, not ffeintrinImp;
2357 for now, these return NONE, since they're not really needed
2358 and adding the necessary info to the tables is not trivial.
2359 (ffeintrin_codegen_imp): New function.
2360 * stc.c (ffestc_R1208_item): Change way ffeintrin funcs called,
2361 back to original per above; but comment out the code anyway.
2363 * intrin.c (ffe_init_0): Do internal checks only if
2364 -fset-g77-defaults not specified.
2366 * lang-options.h: Add -fset-g77-defaults option.
2367 * lang-specs.h: Always pass -fset-g77-defaults.
2368 * top.c, top.h: New option.
2370 Sat Mar 9 17:49:50 1996 Craig Burley <burley@gnu.ai.mit.edu>
2372 * Make-lang.in (stmp-int-hdrs): Use --no-validate when
2373 generating the f77.rebuilt files (BUGS, INSTALL, NEWS)
2374 so cross-references can work properly in g77.info
2375 without a lot of hassle. Users can probably deal with
2376 the way they end up looking in the f77.rebuilt files.
2378 * bld.c (ffebld_constant_new_integer4_val): INTEGER*8
2379 support -- new function.
2380 (ffebld_constant_new_logical4_val): New function.
2381 * com.c (ffecom_f2c_longint_type_node): New type.
2382 (FFECOM_rttypeLONGINT_): New return type code.
2383 (ffecom_expr_): Add code to invoke pow_qq instead
2384 of pow_ii for INTEGER4 (INTEGER*8) case.
2385 If ffecom_expr_power_integer_ returns NULL_TREE, just do
2386 the usual work.
2387 (ffecom_make_gfrt_): Handle new type.
2388 (ffecom_expr_power_integer_): Let caller do the work if in
2389 dummy-transforming case, since
2390 caller now knows about INTEGER*8 and such, by returning
2391 NULL_TREE.
2392 * expr.c (ffeexpr_reduced_power_): Complain about non-INTEGER
2393 raised to INTEGER4 (INTEGER*8) power.
2395 * target.c (ffetarget_power_integerdefault_integerdefault):
2396 Fix any**negative.
2397 * com.c (ffecom_expr_power_integer_): Fix (-1)**(-8) and similar
2398 to ABS() the integral result if the exponent is negative
2399 and even.
2401 * ste.c (ffeste_begin_iterdo_): Clean up a type ref.
2402 Always convert iteration count to _default_ INTEGER.
2404 * sta.c (ffesta_second_): Add BYTE and WORD type/stmts;
2405 changes by Scott Snyder <snyder@d0sgif.fnal.gov>.
2406 * stb.c (ffestb_decl_recursive): Ditto.
2407 (ffestb_decl_recursive): Ditto.
2408 (ffestb_decl_entsp_2_): Ditto.
2409 (ffestb_decl_entsp_3_): Ditto.
2410 (ffestb_decl_funcname_2_): Ditto.
2411 (ffestb_decl_R539): Ditto.
2412 (ffestb_decl_R5395_): Ditto.
2413 * stc.c (ffestc_establish_declstmt_): Ditto.
2414 * std.c (ffestd_R539item): Ditto.
2415 (ffestd_R1219): Ditto.
2416 * stp.h: Ditto.
2417 * str-1t.fin: Ditto.
2418 * str-2t.fin: Ditto.
2420 * expr.c (ffeexpr_finished_): For DO loops, allow
2421 any INTEGER type; convert LOGICAL (assuming -fugly)
2422 to corresponding INTEGER type instead of always default
2423 INTEGER; let later phases do conversion of DO start,
2424 end, incr vars for implied-DO; change checks for non-integral
2425 DO vars to be -Wsurprising warnings.
2426 * ste.c (ffeste_io_impdo_): Convert start, end, and incr
2427 to type of DO variable.
2429 * com.c (ffecom_init_0): Add new types for [IL][234],
2430 much of which was done by Scott Snyder <snyder@d0sgif.fnal.gov>.
2431 * target.c: Ditto.
2432 * target.h: Ditto.
2434 Wed Mar 6 14:08:45 1996 Craig Burley <burley@gnu.ai.mit.edu>
2436 * top.c (ffe_init_gbe_): Make -frerun-loop-opt the default.
2438 Mon Mar 4 12:27:00 1996 Craig Burley <burley@gnu.ai.mit.edu>
2440 * expr.c (ffeexpr_exprstack_push_unary_): Really warn only
2441 about two successive _arithmetic_ operators.
2443 * stc.c (ffestc_R522item_object): Allow SAVE of (understood)
2444 local entity.
2446 * top.c (ffe_decode_option): New -f(no-)second-underscore options.
2447 * top.h: New options.
2448 * com.c (ffecom_get_external_identifier_, ffecom_get_identifier_):
2449 New options.
2451 * Make-lang.in (f77.maintainer-clean): Clean f/BUGS, f/INSTALL,
2452 f/NEWS.
2453 ($(srcdir)/f/BUGS, $(srcdir)/f/INSTALL, $(srcdir)/f/NEWS):
2454 New rules.
2455 ($(srcdir)/f/g77.info, $(srcdir)/f/g77.dvi): Depend on
2456 f/bugs.texi and f/news.texi.
2457 (f77.install-man): Install f77 man pages (if enabled).
2458 (f77.uninstall): Uninstall info docs, f77 man pages (if enabled).
2460 * top.c (ffe_init_gbe_): New function.
2461 (ffe_decode_option, ffe_file): Call ffe_init_gbe_ to
2462 set defaults for gcc options.
2464 Sat Jan 20 13:57:19 1996 Craig Burley <burley@gnu.ai.mit.edu>
2466 * com.c (ffecom_get_identifier_): Eliminate needless
2467 comparison of results of strchr.
2469 Tue Dec 26 11:41:56 1995 Craig Burley <burley@gnu.ai.mit.edu>
2471 * Make-lang.in: Add rules for new files g77.texi, g77.info,
2472 and g77.dvi.
2473 Reorganize the *clean rules to more closely parallel gcc's.
2475 * config-lang.in: Exclude g77.info from diffs.
2477 Sun Dec 10 02:29:13 1995 Craig Burley <burley@gnu.ai.mit.edu>
2479 * expr.c (ffeexpr_declare_unadorned_,
2480 ffeexpr_declare_parenthesized_): Break out handling of
2481 contextDATAIMPDO[INDEX,CTRL] so it's independent of symbol state.
2482 Don't exec-transition these here (let ffeexpr_sym_impdoitem_
2483 handle that when appropriate). Don't "declare" them twice.
2485 Tue Dec 5 06:48:26 1995 Craig Burley <burley@gnu.ai.mit.edu>
2487 * stc.c (ffestc_promote_sfdummy_): Allow whereNONE parent
2488 symbol, since it is not necessarily known whether it will
2489 become LOCAL or DUMMY.
2491 Mon Dec 4 03:46:55 1995 Craig Burley <burley@gnu.ai.mit.edu>
2493 * lex.c (ffelex_display_token, ffelex_type_string_): Resurrect
2494 these from their old versions and update them for possible invocation
2495 from debugger.
2496 * lex.h (ffelex_display_token): Declare this in case anyone
2497 else wants to call it.
2499 * lex.c (ffelex_total_tokens_): Have this reflect actual allocated
2500 tokens, no longer include outstanding "uses" of tokens.
2502 * malloc.c, malloc.h (MALLOC_DEBUG): New macro to control
2503 checking of whether callers follow rules, now defaults to 0
2504 for "no checking" to improve compile times.
2506 * malloc.c (malloc_pool_kill): Fix bug that could prevent
2507 subpool from actually being killed (wasn't setting its use
2508 count to 1).
2510 * proj.h, *.c (dmpout): Replace all occurrences of `stdout'
2511 and some of `stderr' with `dmpout', so where to dump debugging
2512 output can be easily controlled during build; add default
2513 for `dmpout' of `stderr' to proj.h.
2515 Sun Dec 3 00:56:29 1995 Craig Burley <burley@gnu.ai.mit.edu>
2517 * com.c (ffecom_return_expr): Eliminate attempt at warning
2518 about unset return values, since the back end does this better,
2519 with better wording, and is not triggered by clearly working
2520 (but spaghetti) code as easily as this test.
2522 Sat Dec 2 08:28:56 1995 Craig Burley <burley@gnu.ai.mit.edu>
2524 * target.c (ffetarget_power_*_integerdefault): Raising 0 to
2525 integer constant power should not be an error condition;
2526 if so, other code should catch 0 to any power, etc.
2528 * bad.def (FFEBAD_BAD_POWER): 0**integer now a warning instead
2529 of an error.
2531 Fri Dec 1 00:12:03 1995 Craig Burley (burley@gnu.ai.mit.edu)
2533 * bad.def: Clarify diagnostic regarding complex constant elements.
2534 * expr.c (ffeexpr_cb_comma_c_): Capitalize real/imaginary
2535 for clarified diagnostic.
2537 * com.c (ffecom_close_include_): Close the file!
2539 * lex.c (ffelex_file_fixed): Update line info if the line
2540 has any content, not just if it finishes a previous line
2541 or has a label.
2542 (ffelex_file_free): Clarify switch statement code.
2544 Sat Nov 18 19:37:22 1995 Craig Burley (burley@gnu.ai.mit.edu)
2546 * Version 0.5.17 released.
2548 Fri Nov 17 14:27:24 1995 Craig Burley (burley@gnu.ai.mit.edu)
2550 * Make-lang.in: Fix typo in comment.
2552 * Makefile.in (f/fini.o, f/proj-h.o): Don't use `$<' since
2553 not all makes support it (e.g. NeXT make), use explicit
2554 source name instead (with $(srcdir) and munging).
2555 (ASSERT_H): assert.h lives in source dir, not build dir.
2557 Thu Nov 16 12:47:50 1995 Craig Burley (burley@gnu.ai.mit.edu)
2559 * com.c (ffecom_init_0): Fix dumb bug in code to produce
2560 warning message about non-32-bit-systems.
2562 * stc.c (ffestc_R501_item): Parenthesize test to make
2563 warning go away (and perhaps fix bug).
2565 Thu Nov 16 03:43:33 1995 Craig Burley (burley@gnu.ai.mit.edu)
2567 * g77.c: Upgrade to 2.7.0's gcc.c.
2568 Fix -v to pass a temp name instead of "/dev/null" for "-o".
2570 Fri Nov 10 19:16:05 1995 Craig Burley (burley@gnu.ai.mit.edu)
2572 * ste.c (ffeste_begin_iterdo_): Add Toon's change to
2573 make loops faster on some machines (implement termination
2574 condition as "--i >= 0" instead of "i-- > 0").
2576 Thu Nov 2 03:58:17 1995 Craig Burley (burley@gnu.ai.mit.edu)
2578 * Make-lang.in: Remove unnecessary $(exeext) a la cp/Make-lang.in.
2580 * com.c (ffecom_expr_): Restore old strategy for assignp variant
2581 of opSYMTER case...always return the ASSIGN version of var.
2582 That way, `-O -Wuninitialized' will catch "I=3;GOTO I;END"
2583 (though the diagnostic will refer to `__g77_ASSIGN_i').
2585 * com.c (ffecom_expr_power_integer_): For constant rhs case,
2586 wrap every new eval of lhs in save_expr() so it is clear to
2587 back end that MULT_EXPR(lhs,lhs) has identical operands,
2588 otherwise for an rhs like 32767 it generates around 65K pseudo
2589 registers, with which stupid_life_analysis cannot cope
2590 (due to reg_renumber in regs.h being `short *' instead of
2591 `int *').
2593 * com.c (ffecom_expr_): Speed up implementation of LOGICAL
2594 versions of opNOT, opAND, opOR, opXOR/opNEQV, and opEQV by
2595 assuming the values actually are kosher LOGICAL bit patterns.
2596 Also simplify code that implements some of the INTEGER versions
2597 of these.
2599 * com.c (skip_redundant_dir_prefix, read_name_map,
2600 ffecom_open_include_, signed_type, unsigned_type): Fold in
2601 changes to cccp.c made from 2.7.0 through ss-950826.
2603 * equiv.c (ffeequiv_layout_local_): Kill the equiv list
2604 if no syms in list.
2606 * expr.c (ffeexpr_reduced_eqop2_): Issue specific diagnostic
2607 regarding usage of .EQV./.NEQV. in preference to .EQ./.NE..
2609 * intrin.c: Add ERF and ERFC as generic intrinsics.
2610 intrin.def: Same.
2612 * sta.c (ffesta_save_, ffesta_second_): Whoever calls
2613 ffestd_exec_begin must also set ffesta_seen_first_exec = TRUE,
2614 and anytime stc sees an exec transition, it must do both.
2615 stc.c (ffestc_eof): Same.
2617 * stc.c (ffestc_promote_sfdummy_): If failed implicit typing
2618 or CHARACTER*(*) arg, after calling ffesymbol_error, don't
2619 reset info to ENTITY/DUMMY, because ffecom_sym_transform_
2620 doesn't expect such a thing with ANY/ANY type.
2622 * target.h (*logical*): Change some of these so they parallel
2623 changes in com.c, e.g. for _eqv_, use (l)==(r) instead of
2624 !!(l)==!!(r), to get a more faithful result.
2626 Fri Oct 27 07:06:59 1995 Craig Burley (burley@gnu.ai.mit.edu)
2628 * com.c (ffecom_sym_transform_): Simplify code for local
2629 EQUIVALENCE case.
2631 * expr.c (ffeexpr_exprstack_push_unary_): Warn about two
2632 successive operators.
2633 (ffeexpr_exprstack_push_binary_): Warn about "surprising"
2634 operator precedence, as in "-2**2".
2636 * lang-options.h: Add -W(no-)surprising options.
2638 * parse.c (yyparse): Don't reset -fpedantic if not -pedantic.
2640 * top.c (ffe_decode_option): Support new -Wsurprising option.
2641 * top.h: Ditto.
2643 Mon Oct 23 09:14:15 1995 Craig Burley (burley@gnu.ai.mit.edu)
2645 * com.c (ffecom_finish_symbol_transform_): Don't transform
2646 NONE/NONE (CHARACTER*(*)) vars, as these don't mean anything
2647 in debugging terms, and can't be turned into anything
2648 in the back end (so ffecom_sym_transform_ crashes on them).
2650 * com.c (ffecom_expr_): Change strategy for assignp variant
2651 of opSYMTER case...always return the original var unless
2652 it is not wide enough.
2654 * ste.c (ffeste_io_cilist_): Clarify diagnostic for ASSIGN
2655 involving too-narrow variable. This shouldn't happen, though.
2656 (ffeste_io_icilist_): Ditto.
2657 (ffeste_R838): Ditto.
2658 (ffeste_R839): Ditto.
2660 Thu Oct 19 03:21:20 1995 Craig Burley (burley@gnu.ai.mit.edu)
2662 * com.c (ffecom_sym_transform_assign_): Set TREE_STATIC
2663 using the same decision-making process as used for their twin
2664 variables, so ASSIGN can last across RETURN/CALL as appropriate.
2666 Fri Sep 22 20:21:18 1995 Craig Burley (burley@gnu.ai.mit.edu)
2668 * Makefile.in: fini is a host program, so it needs a host-compiled
2669 version of proj.o, named proj-h.o. f/fini, f/fini.o, and
2670 f/proj-h.o targets updated accordingly.
2672 * com.c (__eprintf): New function.
2674 Wed Sep 20 02:26:36 1995 Craig Burley (burley@gnu.ai.mit.edu)
2676 * lang-options.h: Add omitted -funix-intrinsics-* options.
2678 * malloc.c (malloc_find_inpool_): Check for infinite
2679 loop, crash if detected (user reports encountering
2680 them in some large programs, this might help track
2681 down the bugs).
2683 Thu Sep 7 13:00:32 1995 Craig Burley (burley@gnu.ai.mit.edu)
2685 * com.c (lang_print_error_function): Don't dereference null
2686 pointer when outside any program unit.
2687 (ffecom_let_char_, ffecom_arg_ptr_to_expr): If catlist
2688 item or length ever error_mark_node, don't continue processing,
2689 since back-end functions like build_pointer_type crash on
2690 error_mark_node's (due to pushing bad obstacks, etc.).
2692 Wed Aug 30 15:58:35 1995 Craig Burley (burley@gnu.ai.mit.edu)
2694 * Version 0.5.16 released.
2696 Mon Aug 28 12:24:20 1995 Craig Burley (burley@gnu.ai.mit.edu)
2698 * bad.c (ffebad_finish): Fix botched message when no places
2699 are printed (due to unknown line info, etc.).
2701 * std.c (ffestd_subr_labels_): Do a better job finding
2702 line info in the case of typeANY and diagnostics.
2704 Fri Aug 25 15:19:29 1995 Craig Burley (burley@gnu.ai.mit.edu)
2706 * com.c (DECL_ARTIFICIAL): Surround all references to this
2707 macro with #if !BUILT_FOR_270 and #endif.
2708 (init_lex): Surround print_error_function decl with
2709 #if !BUILT_FOR_270 and #endif.
2710 (lang_init): Call new ffelex_hash_kludge function to solve
2711 problem with preprocessed files that have INCLUDE statements.
2713 * lex.c (ffelex_getc_): New function.
2714 (ffelex_cfelex_): Use ffelex_getc_ instead of getc in any
2715 paths of code that can be affected by ffelex_hash_kludge.
2716 Don't make an EOF token for unrecognized token; set token
2717 to NULL instead, to avoid problems when not initialized.
2718 (ffelex_hash_): Use ffelex_getc_ instead of getc in any
2719 paths of code that can be affected by ffelex_hash_kludge.
2720 Test token returned by ffelex_cfelex_ for NULL, meaning
2721 unrecognized token.
2722 Get rid of useless used_up variable.
2723 Don't do ffewhere stuff or kill any tokens if in
2724 ffelex_hash_kludge.
2725 (ffelex_file_fixed, ffelex_file_free): Use ffelex_getc_
2726 instead of getc in any paths of code that can be affected
2727 by ffelex_hash_kludge.
2728 (ffelex_hash_kludge): New function.
2730 * lex.h (ffelex_hash_kludge): New function.
2732 Wed Aug 23 15:17:40 1995 Craig Burley (burley@gnu.ai.mit.edu)
2734 * com.c: Implement -f(no-)underscoring options by always
2735 compiling in code to do it, and having that code inhibit
2736 itself when -fno-underscoring is in effect. This option
2737 overrides -f(no-)f2c for this purpose; -f(no-)f2c returns
2738 to it's <=0.5.15 behavior of affecting only how code
2739 is generated, not how/whether names are mangled.
2741 * target.h: Redo specification of appending underscores so
2742 the macros are named "_default" instead of "_is" and the
2743 two-underscore macro defaults to 1.
2745 * top.c, top.h (underscoring): Add appropriate stuff
2746 for the -f(no-)underscoring options.
2748 Tue Aug 22 10:25:01 1995 Craig Burley (burley@gnu.ai.mit.edu)
2750 * bad.c (ffebad_finish): Call report_error_function (in toplev.c)
2751 to better identify location of problem.
2752 Say "(continued):" instead of "(continued:)" for consistency.
2754 * com.c (ffecom_gen_sfuncdef_): Set and reset new
2755 ffecom_nested_entry_ variable to hold ffesymbol being compiled.
2756 (lang_print_error_function): New function from toplev.c.
2757 Use ffecom_nested_entry_ to help determine which name
2758 and kind-string to print.
2759 (ffecom_expr_intrinsic_): Handle EXIT and FLUSH invocations
2760 with different calling sequences than library functions.
2761 Have SIGNAL and SYSTEM push and pop calltemps, and convert
2762 their return values to the destination type (just in case).
2763 (FFECOM_rttypeINT_): New return type for `int', in case
2764 gcc/f/runtime/libF77/system_.c(system_) is really supposed
2765 to return `int' instead of `ftnint'.
2767 * com.h (report_error_function): Declare this.
2769 * equiv.c (ffeequiv_layout_local_): Don't forget to consider
2770 root variable itself as possible "first rooted variable",
2771 else might never set symbol and then crash later.
2773 * intrin.c (ffeintrin_check_exit_): Change to allow no args
2774 and rename to ffeintrin_check_int_1_o_ for `optional'.
2775 #define ffeintrin_check_exit_ and _flush_ to this new
2776 function, so intrin.def can refer to the appropriate names.
2778 * intrin.def (FFEINTRIN_impFLUSH): Validate using
2779 ffeintrin_check_flush_ so passing an INTEGER arg is allowed.
2781 * lex.c (ffelex_file_push_, ffelex_file_pop_): New functions
2782 to manage input_file_stack in gbe.
2783 (ffelex_hash_): Call new functions (instead of doing code).
2784 (ffelex_include_): Call new functions to update stack for
2785 INCLUDE (_hash_ handles cpp output of #include).
2787 Mon Aug 21 08:09:04 1995 Craig Burley (burley@gnu.ai.mit.edu)
2789 * Makefile.in: Put `-W' in front of every `-Wall', since
2790 2.7.0 requires that to engage `-Wunused' for parameters.
2792 * com.c: Mark all parameters as artificial, so
2793 `-W -Wunused' doesn't complain about unused ones (since
2794 there's no way right not to individually specify attributes
2795 like `unused').
2797 * proj.h: Don't #define UNUSED if already defined, regardless
2798 of host compiler.
2800 Sun Aug 20 16:03:56 1995 Craig Burley (burley@gnu.ai.mit.edu)
2802 * gbe/2.7.0.diff: Regenerate.
2804 * lang-options.h, lang-specs.h: If not __STDC__ (ANSI C),
2805 avoid doing anything, especially the stringizing in -specs.h.
2807 Thu Aug 17 03:36:12 1995 Craig Burley (burley@gnu.ai.mit.edu)
2809 * lang-specs.h: Remove useless optional settings of -traditional,
2810 since -traditional is always set anyway.
2812 Wed Aug 16 16:56:46 1995 Craig Burley (burley@gnu.ai.mit.edu)
2814 * Make-lang.in (F2C_INSTALL_FLAG, F2CLIBOK): More
2815 control over whether to install f2c-related stuff.
2816 (install-f2c-*): New targets to install f2c-related
2817 stuff in system, not just gcc, directories.
2819 * com.c: Change calls to ffecom_get_invented_identifier
2820 to use generally more predictable names.
2821 Change calls to build_range_type to ensure consistency
2822 of types of operands.
2823 (ffecom_get_external_identifier_): Change to accept
2824 symbol info, not just text, so it can use f2c flag for
2825 symbol to decide whether to append underscore(s).
2826 (ffecom_get_identifier_): Don't change names if f2c flag
2827 off for compilation.
2828 (ffecom_type_permanent_copy_): Use same type for new max as
2829 used for min.
2830 (ffecom_notify_init_storage): Offline fixups for stand-alone.
2832 * data.c (ffedata_gather): Explicitly test for common block,
2833 since it's no longer always the case that a local EQUIVALENCE
2834 group has no symbol ptr (it now can, if a user-predictable
2835 "rooted" symbol has been identified).
2837 * equiv.c: Add some debugging stuff.
2838 (ffeequiv_layout_local_): Set symbol ptr with user-predictable
2839 "rooted" symbol, for giving the invented aggregate a
2840 predictable name.
2842 * g77.c (append_arg): Allow for 20 extra args instead of 10.
2843 (main): For version-only case, add `-fnull-version' and, unless
2844 explicitly omitted, `-lf2c -lm'.
2846 * lang-options.h: New "-fnull-version" option.
2848 * lang-specs.h: Support ".fpp" suffix for preprocessed source
2849 (useful for OS/2, MS-DOS, other case-insensitive systems).
2851 * stc.c (ffestc_R544_equiv_): Swap way lists are merged so this
2852 is consistent with the order in which lists are built, making
2853 user predictability of invented aggregate name much higher.
2855 * storag.c, storag.h (FFESTORAG_typeDUMMY): Delete this enum.
2857 * top.c: Accept, but otherwise ignore, `-fnull-version'.
2859 Tue Aug 15 07:01:07 1995 Craig Burley (burley@gnu.ai.mit.edu)
2861 * DOC, INSTALL, PROJECTS: Extensive improvements to documentation.
2863 Sun Aug 13 01:55:18 1995 Craig Burley (burley@gnu.ai.mit.edu)
2865 * INSTALL (f77-install-ok): Document the use of this file.
2867 * Make-lang.in (F77_INSTALL_FLAG): New flag to control
2868 whether to install an `f77' command (based on whether
2869 a file named `f77-install-ok' exists in the source or
2870 build directory) to replace the broken attempt to use
2871 comment lines to avoid installing `f77' (broken in the
2872 sense that it prevented installation of `g77').
2874 Mon Aug 7 06:14:26 1995 Craig Burley (burley@gnu.ai.mit.edu)
2876 * DOC: Add new sections for g77 & gcc compiler options,
2877 source code form, and types, sizes and precisions.
2878 Remove lots of old "delta-version" info, or at least
2879 summarize it.
2881 * INSTALL: Add info here that used to be in DOC.
2882 Other changes.
2884 * g77.c (lookup_option, main): Check for --print-* options,
2885 so we avoid adding version-determining stuff.
2887 Wed Jul 26 15:51:03 1995 Craig Burley (burley@gnu.ai.mit.edu)
2889 * Make-lang.in, Makefile.in (input.j, INPUT_H): New file.
2890 Update dependencies accordingly.
2892 * bad.c (ffebad_here): Okay to use unknown line/col.
2894 * compilers.h (@f77-cpp-input): Remove -P option now that
2895 # directives are handled by f771. Update other options
2896 to be more consistent with @c in gcc/gcc.c. Don't run f771
2897 if -E specified, etc., a la @c.
2898 (@f77): Don't run f771 if -E specified, etc., a la @c.
2900 * config-lang.in: Avoid use of word "guaranteed".
2902 * input.j: New file to wrap around gcc/input.h.
2904 * lex.j: Add support for parsing # directives output by cpp.
2905 (ffelex_cfebackslash_): New function.
2906 (ffelex_cfelex_): New function.
2907 (ffelex_get_directive_line_): New function.
2908 (ffelex_hash_): New function.
2909 (ffelex_include_): Change to not use ffewhere_file_(begin|end).
2910 Also fix bug in pointing to next line (for diagnostics, &c)
2911 following successful INCLUDE.
2912 (ffelex_next_line_): New function that does chunk of code
2913 seen in several places elsewhere in the lexers.
2914 (ffelex_file_fixed): Delay finishing statement until source
2915 line is registered with ffewhere, so INCLUDE processing
2916 picks up the info correctly.
2917 Okay to kill or use unknown line/col objects now.
2918 Handle HASH (#) lines.
2919 Reorder tests for insubstantial lines to put most frequent
2920 occurrences at top, for possible minor speedup.
2921 Some general consolidation of code.
2922 (ffelex_file_free): Handle HASH (#) lines.
2923 Okay to kill or use unknown line/col objects now.
2924 Some general consolidation of code.
2925 (ffelex_init_1): Detect HASH (#) lines.
2926 (ffelex_set_expecting_hollerith): Okay to kill or use unknown
2927 line/col objects now.
2929 * lex.h (FFELEX_typeHASH): New enum.
2931 * options-lang.h (-fident, -fno-ident): New options.
2933 * stw.c (ffestw_update): Okay to kill unknown line/col objects
2934 now.
2936 * target.h (FFETARGET_okREALQUAD, FFETARGET_okCOMPLEXDOUBLE,
2937 FFETARGET_okCOMPLEXQUAD): #define these appropriately.
2939 * top.c: Include flag.j wrapper, not flags.h directly.
2940 (ffe_is_ident_): New flag.
2941 (ffe_decode_option): Handle -fident and -fno-ident.
2942 (ffe_file): Replace obsolete ffewhere_file_(begin|end) with
2943 ffewhere_file_set.
2945 * top.h (ffe_is_ident_, ffe_is_ident, ffe_set_is_ident):
2946 New flag and access functions.
2948 * where.c, where.h: Remove all tracking of parent file.
2949 (ffewhere_file_begin, ffewhere_file_end): Delete these.
2950 (ffewhere_line_use): Make it work with unknown line object.
2952 Mon Jul 17 03:04:09 1995 Craig Burley (burley@gnu.ai.mit.edu)
2954 * com.c (ffecom_sym_transform_): Set DECL_IN_SYSTEM_HEADER
2955 flag for any local vars used as stmtfunc dummies or DATA
2956 implied-DO iter vars, so no -Wunused warnings are produced
2957 for them (a la f2c).
2958 (ffecom_init_0): Do "extern int xargc;" for IARGC() intrinsic.
2959 Warn if target machine not 32 bits, since g77 isn't yet
2960 working on them at all well.
2962 * expr.c (ffeexpr_sym_lhs_call_, ffeexpr_sym_lhs_data_,
2963 ffeexpr_sym_lhs_extfunc_, ffeexpr_sym_rhs_actualarg_,
2964 ffeexpr_sym_rhs_let_, ffeexpr_paren_rhs_let_): Don't
2965 gratuitously set attr bits that don't apply just
2966 to avoid null set meaning error; instead, use explicit
2967 error flag, and allow null attr set, to
2968 fix certain bugs discovered by looking at this code.
2970 * g77.c: Major changes to improve support for gcc long options,
2971 to make `g77 -v' report more useful info, and so on.
2973 Mon Jul 3 14:49:16 1995 Craig Burley (burley@gnu.ai.mit.edu)
2975 * DOC, com.c, intrin.h, intrin.c, intrin.def, target.h, top.c,
2976 top.h: Add new `unix' group of intrinsics, which includes the
2977 newly added ERF, ERFC, EXIT, plus even newer ABORT, DERF, DERFC,
2978 FLUSH, GETARG, GETENV, SIGNAL, and SYSTEM.
2980 Tue Jun 27 23:01:05 1995 Craig Burley (burley@gnu.ai.mit.edu)
2982 * bld.c, bld.h (ffebld_constant_pool,
2983 ffebld_constant_character_pool): Use a single macro (the
2984 former) to access the pool for allocating constants, instead
2985 of latter in public and FFEBLD_CONSTANT_POOL_ internally
2986 in bld.c (which was the only one that was correct before
2987 these changes). Add verification of integrity of certain
2988 heap-allocated areas.
2990 * com.c (ffecom_overlap_, ffecom_args_overlap_,
2991 ffecom_tree_canonize_ptr_, ffecom_tree_canonize_ref_): New
2992 functions to optimize calling COMPLEX and, someday, CHARACTER
2993 functions requiring additional argument to be passed.
2994 (ffecom_call_, ffecom_call_binop_, ffecom_expr_,
2995 ffecom_expr_intrinsic_): Change calling
2996 sequences to include more info on possible destination.
2997 (ffecom_expr_intrinsic_): Add ERF(), ERFC(), and EXIT()
2998 intrinsic code.
2999 (ffecom_sym_transform_): For assumed-size arrays, set high
3000 bound to highest possible value instead of low bound, to
3001 improve validity of overlap checking.
3002 (duplicate_decls): If olddecl and newdecl are the same,
3003 don't do any munging, just return affirmative.
3005 * expr.c: Change ffecom_constant_character_pool() to
3006 ffecom_constant_pool().
3008 * info.c (ffeinfo_new): Compile this version if not being
3009 compiled by GNU C.
3011 * info.h (ffeinfo_new): Don't define macro if not being
3012 compiled by GNU C.
3014 * intrin.c, intrin.def: Add ERF(), ERFC(), and EXIT() intrinsics.
3015 (ffeintrin_check_exit_): New for EXIT() subroutine intrinsic.
3017 * malloc.c, malloc.h (malloc_verify_*): New functions to verify
3018 integrity of heap-storage areas.
3020 * stc.c (ffestc_R834, ffestc_R835): Handle possibility that
3021 an enclosing DO won't have a construct name even when the
3022 CYCLE/EXIT does (i.e. without dereferencing NULL).
3024 * target.c, target.h (ffetarget_verify_character1): New function
3025 to verify integrity of heap storage used to hold character constant.
3027 Thu Jun 22 15:36:39 1995 Howard Gordon (flash@super.org)
3029 * stp.h (ffestpVxtcodeIx): Fix typo in typedef for this.
3031 Mon May 29 15:22:31 1995 Craig Burley (burley@gnu.ai.mit.edu)
3033 * *: Make all sorts of changes to accommodate upcoming gcc-2.7.0.
3034 I didn't keep track of them, nor just when I made them, nor
3035 when I (much later, probably in early August 1995) modified
3036 them so they could properly handle both 2.7.0 and 2.6.x.
3038 * com.c (ffecom_expr_power_integer_): Don't expand_start_stmt_expr
3039 if transforming dummy args, because the back end cannot handle
3040 that (it's rejected by the gcc front end), just generate
3041 call to run-time library.
3042 Back out changes in 0.5.15 because more temporaries might be
3043 needed anyway (for COMPLEX**INTEGER).
3044 (ffecom_push_tempvar): Remove inhibitor.
3045 Around start_decl and finish_decl (in particular, arround
3046 expand_decl, which is called by them), push NULL_TREE into
3047 sequence_rtl_expr, an external published by gcc/function.c.
3048 This makes sure the temporary is truly in the function's
3049 context, not the inner context of a statement-valued expression.
3050 (I think the back end is inconsistent here, but am not
3051 interested in convincing the gbe maintainers about this now.)
3052 (pushdecl): Make sure that when pushing PARM_DECLs, nothing
3053 other than them are pushed, as happened for 0.5.15 and which,
3054 if done for other reasons not fixed here, might well indicate
3055 some other problem -- so crash if it happens.
3057 * equiv.c (ffeequiv_layout_local_): If the local equiv group
3058 has a non-nil COMMON field, it should mean that an error has
3059 occurred and been reported, so just trash the local equiv
3060 group and do nothing.
3062 * stc.c (ffestc_promote_sfdummy_): Set sfdummy arg state to
3063 UNDERSTOOD so above checking for duplicate args actually
3064 works, and so we don't crash later in pushdecl.
3066 * ste.c (ffeste_R1001): Set initial value only for VAR_DECLs,
3067 not for, e.g., LABEL_DECLs, which the FORMAT label can be
3068 if it was previously treated as an executable label.
3070 Sat May 20 01:53:53 1995 Craig Burley (burley@gnu.ai.mit.edu)
3072 * com.c (ffecom_sym_transform_): For adjustable arrays,
3073 pass high bound through variable_size in case its primaries
3074 are changed (dumb0.f, and this might also improve
3075 performance so it approaches f2c|gcc).
3077 Fri May 19 11:00:36 1995 Craig Burley (burley@gnu.ai.mit.edu)
3079 * Version 0.5.15 released.
3081 * com.c (ffecom_expr_power_integer_): Push temp vars
3082 before expanding a statement expression, since that seems
3083 to cause temp vars to be "forgotten" after the end of the
3084 expansion in the back end. Disallow more temp-var
3085 pushing during such an expansion, just in case.
3086 (ffecom_push_tempvar): Crash if a new variable needs to be
3087 pushed but cannot be at this point (should never happen).
3089 Wed May 17 12:26:16 1995 Craig Burley (burley@gnu.ai.mit.edu)
3091 * expr.c (ffeexpr_collapse_convert): Add code to convert
3092 LOGICAL to CHARACTER. Reject conversion of REAL or COMPLEX
3093 to CHARACTER entirely, as it cannot be supported with all
3094 configurations.
3096 * target.h, target.c (ffetarget_convert_character1_logical1):
3097 New function.
3099 Sun May 14 00:00:09 1995 Craig Burley (burley@gnu.ai.mit.edu)
3101 * com.c (ffecom_do_entry_, ffecom_gen_sfuncdef_,
3102 ffecom_start_progunit_, ffecom_sym_transform_,
3103 ffecom_init_0, start_function): Changes to have REAL
3104 external functions return same type as DOUBLE PRECISION
3105 external functions when -ff2c is in force; while at it,
3106 some code cleanups done.
3108 * stc.c (ffestc_R547_item_object): Disallow array declarator
3109 if one already exists for symbol.
3111 * ste.c (ffeste_R1227): Convert result variable to type
3112 of function result as seen by back end (e.g. for when REAL
3113 external function actually returns result as double).
3115 * target.h (FFETARGET_defaultFIXED_LINE_LENGTH): New
3116 macro for default for -ffixed-line-length-N option.
3118 * top.c (ffe_fixed_line_length_): Initialize this to new
3119 target.h macro instead of constant 72.
3121 Tue May 9 01:20:03 1995 Craig Burley (burley@gnu.ai.mit.edu)
3123 * lex.c (ffelex_send_token_): If sending CHARACTER token with
3124 null text field, put a single '\0' in it and set length/size
3125 fields to 0 (to fix 950508-0.f).
3126 (ffelex_image_char_): When setting ffelex_bad_line_ to TRUE,
3127 always "close" card image by appending a null char and setting
3128 ffelex_card_length_. As part of this, append useful text
3129 to identify the two kinds of problems that involve this.
3130 (ffelex_file_fixed): Reset ffelex_bad_line_ to FALSE after
3131 seeing a line with invalid first character (fixes 950508-1.f).
3132 If final nontab column is zero, assume tab seen in line.
3133 (ffelex_card_image_): Always make this array 8 characters
3134 longer than reflected by ffelex_card_size_.
3135 (ffelex_init_1): Get final nontab column info from top instead
3136 of assuming 72.
3138 * options-lang.h: Add -ffixed-line-length- prefix.
3140 * top.h: Add ffe_fixed_line_length() and _set_ version, plus
3141 corresponding extern.
3143 * top.c: Handle -ffixed-line-length- option prefix.
3145 Fri Apr 28 05:40:25 1995 Craig Burley (burley@gnu.ai.mit.edu)
3147 * Version 0.5.14 released.
3149 * Make-lang.in: Add assert.j.
3151 * Makefile.in: Add assert.j.
3153 * assert.j: New file.
3155 Thu Apr 27 16:24:22 1995 Craig Burley (burley@gnu.ai.mit.edu)
3157 * bad.h (ffebad_severity): New function.
3159 * bad.c (ffebad_severity): New function.
3161 * bad.def (FFEBAD_OPEN_INCLUDE): Change severity from SEVERE
3162 to FATAL, since processing continues, and that seems fine.
3164 * com.c: Add facility to handle -I.
3165 (ffecom_file, ffecom_close_include, ffecom_open_include,
3166 ffecom_decode_include_option): New global functions for -I.
3167 (ffecom_file_, ffecom_initialize_char_syntax_,
3168 ffecom_close_include_, ffecom_decode_include_option_,
3169 ffecom_open_include_, append_include_chain, open_include_file,
3170 print_containing_files, read_filename_string, file_name_map,
3171 savestring): New internal functions for -I.
3173 * compilers.h: Pass -I flag(s) to f771 (via "%{I*}").
3175 * lex.c (ffelex_include_): Call ffecom_close_include
3176 to close include file, for its tracking needs for -I,
3177 instead of using fclose.
3179 * options-lang.h: Add -I prefix.
3181 * parse.c (yyparse): Call ffecom_file for main input file,
3182 so -I handling works (diagnostics).
3184 * std.c (ffestd_S3P4): Have ffecom_open_include handle
3185 opening and diagnosing errors with INCLUDE files.
3187 * ste.c (ffeste_begin_iterdo_): Use correct algorithm for
3188 calculating # of iterations -- mathematically similar but
3189 computationally different algorithm was not handling cases
3190 like "DO I=6,5,2" correctly, because (5-6)/2+1 => 1, not 0.
3192 * top.c (ffe_decode_option): Allow -I, restructure a bit
3193 for clarity and, maybe, speed.
3195 Mon Apr 17 13:31:11 1995 Craig Burley (burley@gnu.ai.mit.edu)
3197 * g77.c: Remove -lc, turns out not all systems has it, but
3198 leave other changes in for clarity of code.
3200 Sun Apr 16 21:50:33 1995 Craig Burley (burley@gnu.ai.mit.edu)
3202 * com.c (ffecom_expr_): Implement ARRAY_EXPR as INDIRECT_REF
3203 of appropriate PLUS_EXPRs of ptr_to_expr of array, to see
3204 if this generates better code. (Conditional on
3205 FFECOM_FASTER_ARRAY_REFS.)
3207 Sun Apr 16 00:22:48 1995 Craig Burley (burley@gnu.ai.mit.edu)
3209 * Make-lang.in (F77_SRCS): Remove g77.c, since it doesn't
3210 contribute to building f771.
3212 * Makefile.in (dircheck): Remove/replace with f/Makefile, because
3213 phony targets that are referenced in other real targets get run
3214 when those targets are specified, which is a waste of time (e.g.
3215 when rebuilding and only g77.c has changed, f771 was being linked
3216 anyway).
3218 * g77.c: Include -lc between -lf2c and -lm throughout.
3220 * implic.c (ffeimplic_establish_symbol): If -Wimplicit, warn if
3221 implicit type given to symbol.
3223 * lex.c (ffelex_include_): Don't gratuitously increment line
3224 number here.
3226 * top.h, top.c (ffe_is_warn_implicit_): New global variable and
3227 related access macros.
3228 (ffe_decode_option): Handle -W options, including -Wall and
3229 -Wimplicit.
3231 * where.c (ffewhere_line_new): Don't muck with root line (was
3232 crashing on null input since lexer changes over the past week
3233 or so).
3235 Thu Apr 13 16:48:30 1995 Craig Burley (burley@gnu.ai.mit.edu)
3237 * com.c (ffecom_init_0): Register built-in functions for cos,
3238 sin, and sqrt.
3239 (ffecom_tree_fun_type_double): New variable.
3240 (ffecom_expr_intrinsic_): Update f2c input and output files
3241 to latest version of f2c (no important g77-related changes
3242 noted, just bug fixes to f2c and such).
3243 (builtin_function): New function from c-decl.c.
3245 * com-rt.def: Refer to built-in functions for cos, sin, and sqrt.
3247 Thu Apr 13 10:25:09 1995 Craig Burley (burley@gnu.ai.mit.edu)
3249 * com.c (ffecom_expr_intrinsic_): Convert 0. to appropriate
3250 type to keep DCMPLX(I) from crashing the compiler.
3251 (ffecom_expr_): Don't convert result from ffecom_tree_divide_.
3252 (ffecom_tree_divide_): Add tree_type argument, have all callers
3253 pass one, and don't convert right-hand operand to it (this is
3254 to make this new function work as much like the old in-line
3255 code used in ffecom_expr_ as possible).
3257 * lex.c: Maintain lineno and input_filename the way the gcc
3258 lexer does.
3260 * std.c (ffestd_exec_end): Save and restore lineno and
3261 input_filename around the second pass, which sets them
3262 appropriately for each saved statement.
3264 Wed Apr 12 09:44:45 1995 Craig Burley (burley@gnu.ai.mit.edu)
3266 * com.c (ffecom_expr_power_integer_): New function.
3267 (ffecom_expr_): Call new function for power op with integer second
3268 argument, for generating better code. Also replace divide
3269 code with call to new ffecom_tree_divide_ function.
3270 Canonicalize calls to ffecom_truth_value(_invert).
3271 (ffecom_tree_divide_): New function.
3273 Wed Apr 5 14:15:44 1995 Craig Burley (burley@gnu.ai.mit.edu)
3275 * lex.c: Change to allocate text for tokens only when actually
3276 needed, which should speed compilation up somewhat.
3277 Change to allow INCLUDE at any point where a statement
3278 can end, i.e. in ffelex_finish_statement_ or when a SEMICOLON
3279 token is sent.
3280 Remove some old, obsolete code.
3281 Clean up layout of entire file to improve formatting,
3282 readability, etc.
3283 (ffelex_set_expecting_hollerith): Remove include argument.
3285 Fri Mar 31 23:19:08 1995 Craig Burley (burley@gnu.ai.mit.edu)
3287 * bad.h, bad.c (ffebad_start_msg, ffebad_start_msg_lex):
3288 New functions to generate arbitrary messages.
3289 (FFEBAD_severityPEDANTIC): New severity, to correspond
3290 to toplev's pedwarn() function.
3292 * lex.c (ffelex_backslash_): New function to implement
3293 backslash processing.
3294 (ffelex_file_fixed, ffelex_file_free): Implement new
3295 backslash processing.
3297 * std.c (ffestd_R1001dump_): Don't assume CHARACTER and
3298 HOLLERITH tokens stop at '\0' characters, now that backslash
3299 processing is supported -- use their advertised lengths instead,
3300 and double up the '\002' character for libf2c.
3302 Mon Mar 27 17:10:33 1995 Craig Burley (burley@gnu.ai.mit.edu)
3304 * com.c (ffecom_init_local_zero_): Implement -finit-local-zero.
3305 (ffecom_sym_transform_): Same.
3306 (ffecom_transform_equiv_): Same.
3308 * options-lang.h: Add -f(no-)(init-local-zero,backslash,ugly-init).
3310 * stb.c (ffestb_V020): Reject "TYPEblah(...", which might be
3311 an array assignment.
3313 * target.h, top.h, top.c: Implement -finit-local-zero.
3315 Fri Mar 24 19:56:22 1995 Craig Burley (burley@gnu.ai.mit.edu)
3317 * Make-lang.in, Makefile.in: Remove conf-proj(.in) and
3318 proj.h(.in) rules, plus related config.log, config.cache,
3319 and config.status stuff.
3321 * com.c (ffecom_init_0): Change messages when atof(), bsearch(),
3322 or strtoul() do not work as expected in the start-up test.
3324 * conf-proj, conf-proj.in: Delete.
3326 * lex.c (ffelex_file_fixed): Allow f2c's '&' in column 1
3327 to mean continuation line.
3329 * options-lang.h: New file, #include'd by ../toplev.c.
3331 * proj.h.in: Rename back to proj.h.
3333 * proj.h (LAME_ASSERT): Remove.
3334 (LAME_STDIO): Remove.
3335 (NO_STDDEF): Remove.
3336 (NO_STDLIB): Remove.
3337 (NO_BSEARCH): Remove auto detection, rename to !FFEPROJ_BSEARCH.
3338 (NO_STRTOUL): Remove auto detection, rename to !FFEPROJ_STRTOUL.
3339 (USE_HOST_LIMITS): Remove (maybe still needed by stand-alone?).
3340 (STR, STRX): Do only ANSI C definitions.
3342 Mon Mar 13 10:46:13 1995 Craig Burley (burley@gnu.ai.mit.edu)
3344 * BUGS: Add item about g77 requiring gcc to compile it.
3346 * NEWS: New file listing user-visible changes in the release.
3348 * PROJECTS: Update to include a new item or two, and modify
3349 or delete items that are addressed in this or previous releases.
3351 * bad.c (ffebad_finish): Don't crash if missing string &c,
3352 just substitute obviously distressed string "[REPORT BUG!!]"
3353 for cases where the message/caller are fudgy.
3355 * bad.def: Clean up error messages in a major way, add new ones
3356 for use by changes in target.c.
3358 * com.c (ffecom_expr_): Handle opANY in opCONVERT.
3359 (ffecom_let_char_): Disregard destinations with ERROR_MARK.
3360 (ffecom_1, ffecom_1_fn, ffecom_2, ffecom_2s, ffecom_3,
3361 ffecom_3s, &c): Check all inputs for error_mark_node.
3362 (ffecom_start_progunit_): Don't transform all symbols
3363 in BLOCK DATA, since it never executes, and it is silly
3364 to, e.g., generate all the structures for NAMELIST.
3365 (ffecom_char_length_expr_): Rename to ffecom_intrinsic_len_.
3366 (ffecom_intrinsic_ichar_): New function to handle ICHAR of
3367 arbitrary expression with possible 0-length operands.
3368 (ffecom_expr_intrinsic_): Use ffecom_intrinsic_char_.
3369 For MVBITS, set tree_type to void_type_node.
3370 (ffecom_start_progunit_): Name master function for entry points
3371 after primary entry point so users can easily guess it while
3372 debugging.
3373 (ffecom_arg_ptr_to_expr): Change treatment of Hollerith,
3374 Typeless, and %DESCR.
3375 (ffecom_expr_): Change treatment of Hollerith.
3377 * data.c (ffedata_gather_): Handle opANY in opCONVERT.
3379 * expr.c (ffeexpr_token_apostrophe_): Issue FFEBAD_NULL_CHAR_CONST
3380 warning as necessary.
3381 (ffeexpr_token_name_rhs_): Set context for args to intrinsic
3382 so that assignment-like concatenation is allowed for ICHAR(),
3383 IACHAR(), and LEN() intrinsics.
3384 (ffeexpr_reduced_*_): Say "an array" instead of "an entity" in
3385 diagnostics, since it's more informative.
3386 (ffeexpr_finished_): For many contexts, check for null expression
3387 and array before trying to do a conversion, to avoid redundant
3388 diagnostics.
3390 * g77.1: Fix typo for preprocessed suffix (.F, not .f).
3392 * global.c (ffeglobal_init_common): Warn if initializing
3393 blank common.
3394 (ffeglobal_pad_common): Enable code to warn if initial
3395 padding needed.
3396 (ffeglobal_size_common): Complain if enlarging already-
3397 initialized common, since it won't work right anyway.
3399 * intrin.c: Add IMAG() intrinsic.
3400 (ffeintrin_check_loc_): Allow opSUBSTR in LOC().
3402 * intrin.def: Add IMAG() intrinsic.
3404 * lex.c: Don't report FFEBAD_NULL_CHAR_CONST errors.
3406 * sta.c, sta.h, stb.c: Changes to clean up error messages (see
3407 bad.def).
3409 * stb.c (ffestb_R100113_): Issue FFEBAD_NULL_CHAR_CONST
3410 warning as necessary.
3412 * stc.c (ffestc_shriek_do_): Don't try to reference doref_line
3413 stuff in ANY case, since it won't be valid.
3414 (ffestc_R1227): Allow RETURN in main program unit, with
3415 appropriate warnings/errors.
3416 (ffestc_subr_format_): Array of any type is a CHAREXPR (F77 C5).
3418 * ste.c (ffeste_begin_doiter_): Couple of fixes to accurately
3419 determine if loop never executes.
3421 * target.c (ffetarget_convert_*_hollerith_): Append spaces,
3422 not zeros, to follow F77 Appendix C, and to warn when
3423 truncation of non-blanks done.
3424 (ffetarget_convert_*_typeless): Rewrite to do typeless
3425 conversions properly, and warn when truncation done.
3426 (ffetarget_print_binary, ffetarget_print_octal,
3427 ffetarget_print_hex): Rewrite to use new implementation of
3428 typeless.
3429 (ffetarget_typeless_*): Rewrite to use new implementation
3430 of typeless, and to warn about overflow.
3432 * target.h (ffetargetTypeless): New implementation of
3433 this type.
3435 * type.h, type.c (ffetype_size_typeless): Remove (incorrect)
3436 implementation of this function and its extern.
3438 Sun Mar 5 18:46:42 1995 Craig Burley (burley@gnu.ai.mit.edu)
3440 * BUGS: Clarify that constant handling would also fix lack of
3441 adequate IEEE-754/854 support to some degree, and typeless
3442 and non-decimal constants.
3444 * com.c (ffecom_type_permanent_copy_): Comment out to avoid
3445 warnings.
3446 (duplicate_decls): New function a la gcc/c-decl.c.
3447 (pushdecl): Use duplicate_decls to decide whether to return
3448 existing decl or new one, instead of always returning existing
3449 decl.
3450 (ffecom_expr_): opPERCENT_LOC now supports CHARACTER arguments.
3451 (ffecom_init_0): Give f2c I/O code 0 for basictypeANY/kindtypeANY.
3452 (ffecom_sym_transform_): For adjustable arrays, pass low bound
3453 through variable_size in case its primaries are changed (950302-1.f).
3455 * com.h: More decls that belong in tree.h &c.
3457 * data.c (ffedata_eval_integer1_): Fix opPAREN case to not
3458 treat value of expression as an error code.
3460 * expr.c (ffeexpr_finished_): Allow opSUBSTR in contextLOC case.
3462 * proj.c: Add "const" as appropriate.
3464 Mon Feb 27 10:04:03 1995 Craig Burley (burley@gnu.ai.mit.edu)
3466 * bad.def (FFEBAD_BAD_SUBSTR): Fix bad grammar in message.
3468 Fri Feb 24 16:21:31 1995 Craig Burley (burley@gnu.ai.mit.edu)
3470 * Version 0.5.13 released.
3472 * INSTALL: Warn that f/zzz.o will compare differently between
3473 stages, since it puts the __TIME__ macro into a string.
3475 * com.c (ffecom_sym_transform_): Transform kindFUNCTION/whereDUMMY
3476 to pointer-to-function, not function.
3477 (ffecom_expr_): Use ffecom_arg_ptr_to_expr instead of
3478 ffecom_char_args_ to handle comparison between CHARACTER
3479 types, so either operand can be a CONCATENATE.
3480 (ffecom_transform_common_): Set size of initialized common area
3481 to global (largest-known) size, even though size of init might
3482 be smaller.
3484 * equiv.c (ffeequiv_offset_): Check symbol info for ANY.
3486 * expr.c (ffeexpr_find_close_paren_, ffeexpr_nil_*): New functions
3487 to handle following the contour of a rejected expression, so
3488 statements like "PRINT(I,I,I)=0" don't cause the PRINT statement
3489 code to get the second passed back to it as if there was a
3490 missing close-paren before it, the comma causing the PRINT code
3491 to confirm the statement, resulting in an ambiguity vis-a-vis
3492 the let statement code.
3493 Use the new ffecom_find_close_paren_ handler when an expected
3494 close-paren is missing.
3495 (ffeexpr_isdigits_): New function, use in all places that
3496 currently use isdigit in repetitive code.
3497 (ffeexpr_collapse_symter): Collapse to ANY if init-expr is ANY,
3498 so as to avoid having symbol get "transformed" if used to
3499 dimension an array.
3500 (ffeexpr_token_real_, ffeexpr_token_number_real_): Don't issue
3501 diagnostic about exponent, since it'll be passed along the
3502 handler path, resulting in a diagnostic anyway.
3503 (ffeexpr_token_apos_char_): Use consistent handler path
3504 regardless of whether diagnostics inhibited.
3505 (ffeexpr_token_name_apos_name_): Skip past closing quote/apos
3506 even if not a match or other diagnostic issued.
3507 (ffeexpr_sym_impdoitem_): Exec-transition local SEEN symbol.
3509 * lex.c (ffelex_image_char_): Set ffelex_saw_tab_ if TAB
3510 seen, not if anything other than TAB seen!
3512 * stc.c (ffestc_R537_item): If source is ANY but dest isn't,
3513 set dest symbol's init expr to ANY.
3514 (ffestc_R501_attrib, ffestc_R522, ffestc_R522start): Complain
3515 about conflict between "SAVE" by itself and other uses of
3516 SAVE only in pedantic mode.
3518 * ste.c (ffeste_R1212): Fix loop over labels to always
3519 increment caseno, to avoid pushcase returning 2 for duplicate
3520 values when one of the labels is invalid.
3522 Thu Feb 23 12:42:04 1995 Craig Burley (burley@gnu.ai.mit.edu)
3524 * Version 0.5.12 released.
3526 * Make-lang.in (f77.install-common): Add "else true;" before outer
3527 "fi" per Makefile.in patch.
3529 * Makefile.in (dircheck): Add "else true;" before "fi" per
3530 patch from chs1pm@surrey.ac.uk.
3532 * com.c (ffecom_push_tempvar): If type desired is ERROR_MARK,
3533 return error_mark_node, to avoid crash that results from
3534 making a VAR_DECL with error_mark_node as its type.
3536 * ste.c (ffeste_begin_iterdo_): Convert itercount to INTEGER
3537 anytime calculation of number of iterations ends up with type
3538 other than INTEGER (e.g. DOUBLE PRECISION, REAL).
3540 Thu Feb 23 02:48:38 1995 Craig Burley (burley@gnu.ai.mit.edu)
3542 * Version 0.5.11 released.
3544 * DOC: Explain -fugly-args.
3546 * bad.def (FFEBAD_ACTUALARG): Explain -fugly-args and how to
3547 rewrite code to not require it.
3549 * com.c (ffecom_vardesc_): Handle negative type code, just in
3550 case.
3551 (ffecom_arg_ptr_to_expr): Let ffecom_expr handle hollerith
3552 and typeless constants (move code to ffecom_constantunion).
3553 (ffecom_constantunion): Handle hollerith and typeless constants.
3555 * expr.c (ffecom_finished_): Check -fugly-args in actual-arg
3556 context where hollerith/typeless provided.
3558 * intrin.def (FFEINTRIN_genDFLOAT): Add FFEINTRIN_specDFLOAT.
3559 (FFEINTRIN_specDFLOAT): Add as f2c intrinsic.
3561 * target.h (ffetarget_convert_real[12]_integer,
3562 ffetarget_convert_complex[12]_integer): Pass -1 for high integer
3563 value if low part is negative.
3564 (FFETARGET_defaultIS_UGLY_ARGS): New macro.
3566 * top.c (ffe_is_ugly_args_): New variable.
3567 (ffe_decode_option): Handle -fugly-args and -fno-ugly-args.
3569 * top.h (ffe_is_ugly_args_, ffe_is_ugly_args(),
3570 ffe_set_is_ugly_args()): New variable and macros.
3572 Thu Feb 23 02:48:38 1995 Pedro A M Vazquez (vazquez@iqm.unicamp.br)
3574 * g77.c (sys_errlist): Use const for __FreeBSD__ systems
3575 as well.
3577 Wed Feb 22 13:33:43 1995 Craig Burley (burley@gnu.ai.mit.edu)
3579 * Version 0.5.10 released.
3581 * CREDITS: Add Rick Niles.
3583 * INSTALL: Note how to get around lack of makeinfo.
3585 * Make-lang.in (f/proj.h): Remove # comment.
3587 * Makefile.in (f/proj.h): Remove # comment.
3589 * com.c (ffecom_expr_): Simplify opFUNCREF/opSUBRREF conversion.
3590 (ffecom_sym_transform_): For whereGLOBAL and whereDUMMY
3591 kindFUNCTION, use ffecom_tree_fun_type[][] only for non-constant
3592 (non-statement-function) f2c functions.
3593 (ffecom_init_0): ffecom_tree_fun_type[][] and _ptr_to_*_* are
3594 really f2c-interface arrays, so use base type void for COMPLEX
3595 (like CHARACTER).
3597 Tue Feb 21 19:01:18 1995 Dave Love <d.love@dl.ac.uk>
3599 * Make-lang.in (f77.install-common): Expurgate the test for and
3600 possible installation of f2c in line with elsewhere. Seems to have
3601 been missing a semicolon anyhow!
3603 Tue Feb 21 11:45:25 1995 Craig Burley (burley@gnu.ai.mit.edu)
3605 * Version 0.5.9 released.
3607 * Make-lang.in (f/proj.h): touch file to register update,
3608 because the previous commands won't necessarily modify it.
3610 * Makefile.in (f/proj.h): touch file to register update,
3611 because the previous commands won't necessarily modify it.
3613 * Makefile.in (f/str-*.h, f/str-*.j): Explicitly specify
3614 output file names, so these targets go in build, not source,
3615 directory.
3617 * bits.c, bits.h: Switch to valid ANSI C replacement for
3618 ARRAY_ZERO.
3620 * com.c (ffecom_expr_): Add assignp arg to support ASSIGN better.
3621 If assignp is TRUE, use different tree for FFEBLD_opSYMTER case.
3622 (ffecom_sym_transform_assign_): New function.
3623 (ffecom_expr_assign): New function.
3624 (ffecom_expr_assign_w): New function.
3626 * com.c (ffecom_f2c_make_type_): Do make_signed_type instead
3627 of make_unsigned_type throughout.
3629 * com.c (ffecom_finish_symbol_transform_): Expand scope of
3630 commented-out code to probably produce faster compiler code.
3632 * com.c (ffecom_gen_sfuncdef_): Push/pop calltemps so
3633 COMPLEX works right.
3634 Remove obsolete comment.
3636 * com.c (ffecom_start_progunit_): If non-multi alt-entry
3637 COMPLEX function, primary (static) entry point returns result
3638 directory, not via extra arg -- to agree with ffecom_return_expr
3639 and others.
3640 Pretransform all symbols so statement functions are defined
3641 before any code emitted.
3643 * com.c (ffecom_finish_progunit): Don't posttransform all
3644 symbols here -- pretransform them instead.
3646 * com.c (ffecom_init_0): Don't warn about possible ASSIGN
3647 crash, as this shouldn't happen now.
3649 * com.c (ffecom_push_tempvar): Fix to handle temp vars
3650 pushed while context is a statement (nested) function, and
3651 add appropriate commentary.
3653 * com.c (ffecom_return_expr): Check TREE_USED to determine
3654 where return value is unset.
3656 * com.h (struct _ffecom_symbol_): Add note about length_tree
3657 now being used to keep tree for ASSIGN version of symbol.
3659 * com.h (ffecom_expr_assign, ffecom_expr_assign_rw): New decls.
3660 (error): Add this prototype for back-end function.
3662 * fini.c (main): Grab input, output, and include names
3663 directly off the command line instead of making the latter
3664 two out of the first.
3666 * lex.c: Improve tab handling for both fixed and free source
3667 forms, and ignore carriage-returns on input, while generally
3668 improving the code. ffelex_handle_tab_ has been renamed and
3669 reinvented as ffelex_image_char_, among other things.
3671 * malloc.c, malloc.h: Switch to valid ANSI C replacement for
3672 ARRAY_ZERO, and kill the full number of bytes in pools and
3673 areas.
3675 * proj.h.in (ARRAY_ZERO, ARRAY_ZERO_SIZE): Remove.
3677 * ste.c (ffeste_io_cilist_, ffeste_io_icilist_, ffeste_R838,
3678 ffeste_R839): Issue diagnostic if a too-narrow variable used in an
3679 ASSIGN context despite changes to this code and code in com.c.
3681 * where.c, where.h: Switch to valid ANSI C replacement for
3682 ARRAY_ZERO.
3684 Fri Feb 17 03:35:19 1995 Craig Burley (burley@gnu.ai.mit.edu)
3686 * Version 0.5.8 released.
3688 * INSTALL: In quick-build case, list g77 target first so g77
3689 gets installed. Also, explain that gcc gets built and installed
3690 as well, even though this isn't really what we want (and maybe
3691 we'll find a way around this someday).
3693 Fri Feb 17 02:35:41 1995 Craig Burley (burley@gnu.ai.mit.edu)
3695 * Version 0.5.7 released.
3697 * Makefile.in (CONFIG_H, HCONFIG_H, TCONFIG_H, TM_H): Remove
3698 ../ prefix in front of .h files, since they're in the cd.
3700 Fri Feb 17 01:50:48 1995 Craig Burley (burley@gnu.ai.mit.edu)
3702 * Version 0.5.6 released.
3704 Thu Feb 16 20:26:54 1995 Craig Burley (burley@gnu.ai.mit.edu)
3706 * ../README.g77: Remove description of g77 as "not-yet-published".
3708 * CREDITS: More changes.
3710 * Make-lang.in (G77STAGESTUFF): Remove cktyps stuff.
3712 * Makefile.in (CONFIG_H, HCONFIG_H, TCONFIG_H, TM_H): Don't
3713 prefix gcc dir with $(srcdir) since these don't live there,
3714 they are created in the build dir by gcc's configure. Add
3715 a note explaining what these macros are about.
3716 Update dependencies via deps-kinda.
3718 * README.NEXTSTEP: Credit Toon, and per his request, add his
3719 email address.
3721 * com.h (FFECOM_DETERMINE_TYPES): #include "config.j".
3723 * config.j, convert.j, flags.j, hconfig.j, rtl.j, tconfig.j,
3724 tm.j, tree.j: Don't #include if already done.
3726 * convert.j: #include "tree.j" first, as convert.h clearly depends
3727 on trees being defined.
3729 * rtl.j: #include "config.j" first, since there's some stuff
3730 in rtl.h that assumes it has been #included.
3732 * tree.j: #include "config.j" first, or real.h makes inconsistent
3733 decision about return type of ereal_atof, leading to bugs, and
3734 because tree.h/real.h assume config.h already included.
3736 Wed Feb 15 14:40:20 1995 Craig Burley (burley@gnu.ai.mit.edu)
3738 * Version 0.5.5 released.
3740 * Copyright notices updated to be FSF-style.
3742 * INSTALL: Some more clarification regarding building just f77.
3744 * Make-lang.in (F77_SRCS): Update wrt changing some .h to .j.
3745 (install-libf77): Fix typo in new parenthetical note.
3747 * Makefile.in (f/*.o): Update.
3748 (CONFIG_H, CONVERT_H, FLAGS_H, GLIMITS_H, HCONFIG_H, RTL_H,
3749 TCONFIG_H, TM_H, TREE_H): Update/new symbols.
3750 (deps-kinda): More fixes wrt changing some .h to .j.
3751 Document and explain this rule a bit better.
3752 Accommodate changes in output of gcc -MM.
3754 * *.h, *.c: Change #include's so proj.h not assumed to #include
3755 malloc.h or config.h (now config.j), and so new .j files are
3756 used instead of old .h ones.
3758 * com.c (ffecom_init_0): Use FLOAT_TYPE_SIZE for f2c's
3759 TYLONG/TYLOGICAL type codes, to get g77 working on Alpha.
3761 * com.h: Make all f2c-related integral types "int", not "long
3762 int".
3764 * config.j, convert.j, flags.j, glimits.j, hconfig.j, rtl.j,
3765 tconfig.j, tm.j, tree.j: New files wrapping around gbe
3766 .h files.
3768 * config.h, convert.h, flags.h, glimits.h, hconfig.h, rtl.h,
3769 tconfig.h, tm.h, tree.h: Deleted so new .j files
3770 can #include the gbe files directly, instead of using "../",
3771 and thus do better with various kinds of builds.
3773 * proj.h: Delete unused NO_STDDEF and related stuff.
3775 Tue Feb 14 08:28:08 1995 Craig Burley (burley@gnu.ai.mit.edu)
3777 * BUGS: Remove item #12, cross-compiling & autoconf scripts
3778 reportedly expected to work properly (according to d.love).
3780 * INSTALL: Add explanation of d.love's patch to config-lang.in.
3781 Add explanation of how to install just g77 when gcc already installed.
3782 Add note about usability of "-Wall". Add note about bug-
3783 reporting.
3785 * Make-lang.in ($(srcdir)/f/conf-proj): Add comment about why
3786 conf-proj.out.
3787 (install-libf77): Echo parenthetical note to user about how to do
3788 just the (aborted) libf2c installation.
3789 (deps-kinda): Update to work with new configuration/build stuff.
3791 * bad.c (ffebad_finish): Put capitalized "warning:" &c message
3792 as prefix on any diagnostic without pointers into source.
3794 * bad.def (FFEBAD_TOO_BIG_INIT): Add this warning message.
3796 * config-lang.in: Add Dave Love's patch to catch case where
3797 back-end patches not applied and abort configuration.
3799 * data.c (ffedata_gather_, ffedata_value_): Warn when about
3800 to initialize a large aggregate area, due to design flaw resulting
3801 in too much time/space used to handle such cases.
3802 Use COMMON area name, and first notice of symbol, for multiple-
3803 initialization diagnostic, instead of member symbol and unknown
3804 location.
3805 (FFEDATA_sizeTOO_BIG_INIT_): New macro per above.
3807 Mon Feb 13 13:54:26 1995 Dave Love <d.love@dl.ac.uk>
3809 * Make-lang.in (F77_SRCS): Use $(srcdir)/f/proj.h.in, not
3810 $(srcdir)/f/proj.h for build outside srcdir.
3812 Sun Feb 12 13:37:11 1995 Craig Burley (burley@gnu.ai.mit.edu)
3814 * ../README.g77: Clarify procedures for unpacking, add asterisks
3815 to mark important things the user must do.
3817 * Fix dates in/add dates to ../README.g77, BUGS, CREDITS, DOC,
3818 INSTALL, PROJECTS, README.
3820 Sun Feb 12 00:26:10 1995 Craig Burley (burley@gnu.ai.mit.edu)
3822 * Version 0.5.4 released.
3824 * Make-lang.in (f/proj.h): Reproduce this rule here from
3825 Makefile.in.
3826 ($(srcdir)/f/conf-proj): Put autoconf's stdout in temp file
3827 conf-proj.out, then mv to conf-proj only if successful, so
3828 conf-proj not touched if autoconf not installed.
3830 * Makefile.in ($(srcdir)/conf-proj): See Make-lang.in's similar
3831 rule.
3833 Sat Feb 11 20:56:02 1995 Craig Burley (burley@gnu.ai.mit.edu)
3835 * BUGS: Clarify some bugs.
3837 * DOC: Many improvements and fixes.
3839 * README: Move bulk of text, edited, to ../README.g77, and
3840 replace with pointer to that file.
3842 * com.c (ffecom_init_0): Comment out warning about sizeof(ftnlen)
3843 as per ste.c change. Add text about ASSIGN to help user understand
3844 what is being warned about.
3846 * conf-proj.in: Fix typos in comments.
3848 * proj.h.in: Add ARRAY_ZERO_SIZE to parallel malloc.h's version,
3849 in case it proves to be needed.
3851 * ste.c: Comment out assertions requiring sizeof(ftnlen) >=
3852 sizeof(char *), in the hopes that overflow will never happen.
3853 (ffeste_R838): Change assertion to fatal() with at least
3854 partially helpful message.
3856 Sat Feb 11 12:38:00 1995 Craig Burley (burley@gnu.ai.mit.edu)
3858 * com.c (ffecom_vardesc_): Crash if typecode is -1.
3860 * ste.c (ffeste_io_dolio_): Crash if typecode is -1.
3862 Sat Feb 11 09:51:57 1995 Craig Burley (burley@gnu.ai.mit.edu)
3864 * ste.c: In I/O code tests for item arrayness, sort of revert
3865 to much earlier code that tests original exp, but also check
3866 in newer way just in case. Newer way alone treated FOO(1:40)
3867 as an array, not sure why older way alone didn't work, but I
3868 think maybe it was when diagnosed code was involved, and
3869 since there are now checks for error_mark_node, maybe the old
3870 way alone would work. But better to be safe; both original
3871 ffebld exp _and_ the transformed tree must indicate an array
3872 for the size-determination code to be used, else just 1/2 elements
3873 assumed. And this text is for EMACS: (foo at bar).
3875 Fri Feb 10 11:05:50 1995 Craig Burley (burley@gnu.ai.mit.edu)
3877 * ste.c: In many cases, surround statement-expansion code
3878 with ffecom_push_calltemps () and ffecom_pop_calltemps ()
3879 so COMPLEX-returning functions can have temporaries pushed
3880 in "auto-pop" mode and have them auto-popped at the end of
3881 the statement.
3883 Wed Feb 8 14:35:10 1995 Dave Love <d.love@dl.ac.uk>
3885 * runtime/f2c.h.in (ftnlen, ftnint): Make same size as integer.
3887 * runtime/libI77/err.c (f_init): Thinko in MISSING_FILE_ELEMS
3888 conditional.
3889 * runtime/libI77/wrtfmt.c (mv_cur): Likewise.
3890 * runtime/libI77/wsfe.c (x_putc): Likewise.
3892 * runtime/libF77/signal_.c (signal_): Return 0 (this is a
3893 subroutine).
3895 * Makefile.in (f/proj.h): Depend on com.h.
3896 * Make-lang.in (include/f2c.h): Likewise (and proj.h).
3897 (install-libf77): Also install f2c.h.
3899 * runtime/libI77/Makefile.in (*.o): Add f2c.h dependency.
3900 * runtime/libF77/Makefile.in: Likewise.
3902 Wed Feb 8 13:56:47 1995 Craig Burley (burley@gnu.ai.mit.edu)
3904 * stc.c (ffestc_R501_item): Don't reset kind/where to NONE when
3905 setting basictype/kindtype info for symbol, or especially
3906 its function/result twin, because kind/where might not be NONE.
3908 Tue Feb 7 14:47:26 1995 Dave Love <d.love@dl.ac.uk>
3910 * Make-lang.in (include/f2c.h:): Set shell variable src more
3911 robustly (independent of whether srcdir is relative or absolute).
3912 * Makefile.in (f/proj.h:): Likewise.
3914 * conf-proj.in: Check need for LAME_ASSERT. Fix indentation in
3915 check for LAME_STDIO (cosmetic only with ANSI C).
3917 * com.h: Extra ...SIZE stuff taken from com.c.
3919 * com.c (FFECOM_DETERMINE_TYPES): Define before including com.h.
3920 (BITS_PER_WORD etc.) Remove and use conditional definitions to com.h.
3922 * runtime/configure.in: #define FFECOM_DETERMINE_TYPES for com.h in
3923 f2c type determination.
3925 * tm.h: Remove (at least pro tem) because of relative path and use
3926 top-level one.
3928 * Make-lang.in (include/f2c.h:): Set shell variable src more
3929 robustly (independent of whether srcdir is relative or absolute).
3930 * Makefile.in (f/proj.h:): Likewise.
3932 Mon Feb 6 19:58:32 1995 Dave Love <d.love@dl.ac.uk>
3934 * g77.c (append_arg): Use K&R declaration for, e.g. SunOS4 build.
3936 Fri Feb 3 20:33:14 1995 Craig Burley (burley@gnu.ai.mit.edu)
3938 * g77.c (main): Treat -l like filename in terms of -x handling.
3939 Rewrite arglist mechanism for ease of maintenance.
3940 Make sure every -lf2c is followed by -lm and vice versa.
3942 * Make-lang.in: Put complete list of sources in F77_SRCS def
3943 so changing a .h file, for example, causes rebuild.
3945 * Makefile.in: Change test for nextstep to m68k-next-nextstep* so
3946 all versions of nextstep on m68k get the necessary flag.
3948 Fri Feb 3 19:10:32 1995 Dave Love <d.love@dl.ac.uk>
3950 * INSTALL: Note about possible conflict with existing libf2c.a and
3951 f2c.h.
3953 * Make-lang.in (f77.distclean): Tidy and move deletion of
3954 f/config.cache to mostlyclean.
3955 (install-libf77): Test for $(libdir)/libf2c.* and barf if found
3956 unless F2CLIBOK defined.
3958 * runtime/Makefile.in (all): Change path to include directory (and
3959 elsewhere).
3960 (INCLUDES): Remove (unused/misleading).
3961 (distclean): Include f2c.h.
3962 (clean): Include config.cache.
3964 * runtime/libF77/Makefile.in (.SUFFIXES): Correct typo.
3965 (ALL_CFLAGS) Fix up include search path to find f2c.h in top level
3966 includes always.
3967 (all): Depend on f2c.h.
3968 * runtime/libI77/Makefile.in (.SUFFIXES): Likewise.
3970 Thu Feb 2 17:17:06 1995 Dave Love <d.love@dl.ac.uk>
3972 * INSTALL: Note about --srcdir and GNU make.
3974 * runtime/f2c.h.in (Pad_UDread, ALWAYS_FLUSH): Reomve the #defines
3975 per below.
3977 * runtime/configure.in (Pad_UDread, ALWAYS_FLUSH): Define these
3978 here, not in f2c.h as they'r eonly relevant for building.
3979 * runtime/configure: Regenerated.
3981 * config-lang.in: Warn about using GNU make outside source tree
3982 since I can't get Irix5 or SunOS4 makes to work in this case.
3984 * Makefile.in (VPATH): Don't set it here.
3985 (srcdir): Make it the normal `.' (overridden) at top level.
3986 (all.indirect): New dependency `dircheck'.
3987 (f771): Likewise
3988 (dircheck): New target for foolproofing.
3989 (f/proj.h:): Change finding source.
3990 (CONFIG_H): Don't use this as the relative path in the include loses
3991 f builddir != srcdir.
3993 * config.h: Remove per CONFIG_H change above.
3995 * Make-lang.in (F77_FLAGS_TO_PASS): Remove GCC_FOR_TARGET.
3996 (f771:): Pass VPATH, srcdir to sub-make.
3997 (f/Makefile:): New target.
3998 (stmp-int-hdrs): new variable for cheating build.
3999 (f77-runtime:): Alter GCC_FOR_TARGET treatment.
4000 (include/f2c.h f/runtime/Makefile:) Likewise.
4001 (f77-runtime-unsafe:): New (cheating) target.
4003 Thu Feb 2 12:09:51 1995 Craig Burley (burley@gnu.ai.mit.edu)
4005 * BUGS: Update regarding losing EQUIVALENCE members in -g, and
4006 regarding RS/6000 problems in the back end.
4008 * CREDITS: Make some changes as requested.
4010 * com.c (ffecom_member_trunk_): Remove unused static variable.
4011 (ffecom_finish_symbol_transform_): Improve comments.
4012 (ffecom_let_char_): Fix size of temp address-type var.
4013 (ffecom_member_phase2_): Try fixing problem fixed by change
4014 to ffecom_transform_equiv_ (f_m_p2_ function currently not used).
4015 (ffecom_transform_equiv_): Remove def of unused static variable.
4016 Comment-out use of ffecom_member_phase2_, until problems with
4017 back end fixed.
4018 (ffecom_push_tempvar): Fix assertion to not crash okay code.
4020 * com.h: Remove old, commented-out code.
4021 Add prototype for warning() in back end.
4023 * ste.c (ffeste_io_dofio_, ffeste_io_dolio_, ffeste_io_douio_,
4024 ffeste_io_icilist_): Check correct type of variable for arrayness.
4026 Sun Jan 29 14:41:42 1995 Dave Love <d.love@dl.ac.uk>
4028 * BUGS: Remove references to my configure bugs; add another.
4030 * runtime/Makefile.in (AR_FLAGS): Provide default value.
4032 * runtime/f2c.h.in (integer, logical): Take typedefs from
4033 F2C_INTEGER configuration parameter again.
4034 (NON_UNIX_STDIO): don't define it.
4036 * runtime/configure.in: Bring type checks for f2c.h in line with
4037 com.h.
4038 (MISSING_FILE_ELEMS): New variable to determine whether the relevant
4039 elements of the FILE struct exist, independent of NON_UNIX_STDIO.
4040 * runtime/libI77/{err,wrtfmt,wsfe}.c (MISSING_FILE_ELEMS): Use new
4041 parameter.
4043 * config-lang.in: Comment out more of f2c rules (c.f. Make-lang.in).
4044 (This stuff is relevant iff you gave configure --enable-f2c.)
4045 Create f/runtime directory tree iff not building in source
4046 directory.
4048 * Makefile.in (srcdir): Append slash so we get the right value when
4049 not building in the source directory. This is a consequence of not
4050 building the `f' sources in `f'.
4051 (VPATH): Override configure's value for reasons above.
4052 (f/proj.h f/conf-proj): New rules to build proj.h by
4053 autoconfiguration.
4055 * proj.h: Rename to proj.h.in for autoconfiguration.
4056 * proj.h.in: New as above.
4057 * conf-proj conf-proj.in: New files for autoconfiguration.
4059 * Make-lang.in (include/f2c.h f/runtime/Makefile:): Change the order
4060 of setting the sh variables so that the right GCC_FOR_TARGET is
4061 used.
4062 (f77.*clean:) Add products of new configuration files and make sure
4063 all the *clean targets do something (unlike the ones in
4064 cp/Make-lange.in).
4066 * com.h (FFECOM_f2cINTEGER, FFECOM_f2cLOGICAL): Define as long or
4067 int appropriately to ensure sizeof(real) == sizeof(integer).
4069 * PROJECTS: Library section.
4071 * runtime/libI77/endfile.c: Don't #include sys/types.h conditional
4072 on NON_UNIX_STDIO since rawio.h needs size_t.
4073 * runtime/libI77/uio.c: #include <sys/types.h> for size_t if not
4074 KR_headers.
4076 Wed Jan 25 03:31:51 1995 Craig Burley (burley@gnu.ai.mit.edu)
4078 * Version 0.5.3 released.
4080 * INSTALL: Revise.
4082 * Make-lang.in: Comment out rules for building f2c itself (f/f2c/).
4084 * README: Revise.
4086 * com.c (ffecom_init_0): Warn if ftnlen or INTEGER not big enough
4087 to hold a char *.
4089 * gbe/2.6.2.diff: Update.
4091 Mon Jan 23 17:10:49 1995 Craig Burley (burley@gnu.ai.mit.edu)
4093 * TODO: Remove.
4094 BUGS: New file.
4095 PROJECTS: New file.
4096 CREDITS: New file.
4098 * cktyps*: Remove.
4099 Make-lang.in: Remove cktyps stuff.
4100 Makefile.in: Remove cktyps stuff.
4102 * DOC: Add info on changes for 0.5.3.
4104 * bad.c: Put "warning:" &c on diagnostic messages.
4105 Don't output informational messages if warnings disabled.
4107 Thu Jan 19 12:38:13 1995 Craig Burley (burley@gnu.ai.mit.edu)
4109 * g77.c: Avoid putting out useless "-xnone -xf77" pairs so
4110 larger command lines can be accommodated.
4111 Recognize both `-xlang' and `-x lang'.
4112 Recognize `-xnone' and `-x none' to mean what it does, instead
4113 of treating "none" as any other language.
4114 Some minor, slight improvements in the way args are handled
4115 (hopefully for clearer, more maintainable code), including
4116 consistency checks on arg count just in case.
4118 Wed Jan 18 16:41:57 1995 Craig Burley (burley@gnu.ai.mit.edu)
4120 * DOC: Explain -fautomatic better.
4122 * INSTALL: Describe libf2c.a better.
4124 * Make-lang.in, Makefile.in: Build f771 &c with gcc/ as cd instead
4125 of gcc/f/ so debugging info is better (source file tracking).
4126 Add new source file type.c.
4128 * Makefile.in: For nextstep3, link f771 with -segaddr __DATA
4129 6000000. Fix typo. Change deps-kinda target to handle building
4130 from gcc/. Update dependencies.
4132 * bld-op.def, bld.h, bld.c: Remove opBACKEND and all related
4133 stuff.
4134 Remove consistency tests that cause compiler warnings.
4136 * cktyps.c: Remove all typing checking.
4138 * com-rt.def: Change all rttypeFLOAT_ intrinsics to rttypeDOUBLE_,
4139 to precisely match how they're declared in libf2c.
4141 * com.h, com.c: Revise to more elegantly track related stuff
4142 in the version of f2c.h used to build libf2c.
4144 * com.c: Increase FFECOM_sizeMAXSTACKITEM, and if 0 or undefined
4145 when checked to determine where to put entity, treat as infinite.
4146 Rewrite temporary mechanism to be based on trees instead of
4147 ffeinfo stuff, and make it much simpler. Change interface
4148 accordingly.
4149 Fixes to better track types of things, make appropriate
4150 conversions, etc. E.g. when making an arg for a libf2c
4151 function, make sure it's of the right type (such as ftnlen).
4152 Delete opBACKEND transformation code.
4153 (ffecom_init_0): Smoother initialization of types, especially
4154 paying attention to using consistent rules for making INTEGER,
4155 REAL, DOUBLE PRECISION, etc., and for deciding their "*N"
4156 and kind values that will work across all g77 platforms.
4157 No longer require per-target configuration info in target.h
4158 or config/*/*; use new type module to store size, alignment.
4159 (ffecom_member_phase2): Declare COMMON/EQUIVALENCE group members
4160 so debugger sees them.
4161 (ffecom_finish_progunit): Transform all symbols in program unit,
4162 so -g will show they all exist.
4164 * expr.c (ffeexpr_collapse_substr): Handle strange substring
4165 range values.
4167 * info.h, info.c: Provide connection to new type module.
4168 Remove tests that yield compiler warnings.
4170 * intrin.c (ffeintrin_is_intrinsic): Properly handle deleted
4171 intrinsic.
4173 * lex.c (ffelex_file_fixed): Remove redundant/buggy code.
4175 * stc.c (ffestc_kindtype_kind_, ffestc_kindtype_star_): Replace
4176 boring switch stmt with simple call to new type module. This
4177 sort of thing is a reason to get up in the morning.
4179 * ste.c: Update to handle new interface for
4180 ffecom_push/pop_tempvar.
4181 Fixes to better track types of things.
4182 Fixes to not crash for certain diagnosed constructs.
4183 (ffeste_begin_iterdo_): Check only constants for overflow to avoid
4184 spurious diagnostics.
4185 Don't convert larger integer (say, INTEGER*8) to canonical integer
4186 for iteration count.
4188 * stw.h: Track DO iteration count temporary variable.
4190 * symbol.c: Remove consistency tests that cause compiler warnings.
4192 * target.c (ffetarget_aggregate_info): Replace big switch with
4193 little call to new type module.
4194 (ffetarget_layout): Remove consistency tests that cause
4195 compiler warnings.
4196 (ffetarget_convert_character1_typeless): Pick up length of
4197 typeless type from new type module.
4199 * target.h: Crash build if target float bit pattern cannot be
4200 precisely determined.
4201 Remove all the type cruft now determined by ffecom_init_0
4202 at invocation time and maintained in new type module.
4203 Put casts on second arg of all REAL_VALUE_TO_TARGET_DOUBLE
4204 uses so compiler warnings avoided (requires target float bit
4205 pattern to be precisely determined, hence code to crash build).
4207 * top.c: Add inits/terminates for new type module.
4209 * type.h, type.c: New module.
4211 * gbe/2.6.2.diff: Remove all patches to files in gcc/config/
4212 directory and its subdirectories.
4214 Mon Jan 9 19:23:25 1995 Dave Love <d.love@dl.ac.uk>
4216 * com.h (FFECOM_F2C_INTEGER_TYPE_NODE): Define and use instead of
4217 long_integer_type_node where appropriate.
4219 Tue Jan 3 14:56:18 1995 Dave Love <d.love@dl.ac.uk>
4221 * com.h: Make ffecom_f2c_logical_type_node long, consistent with
4222 integer.
4224 Fri Dec 2 20:07:37 1994 Dave Love <d.love@dl.ac.uk>
4226 * config-lang.in (stagestuff): Add f2c conditionally.
4227 * Make-lang.in: Add f2c and related targets.
4228 * f2c: Add the directory.
4230 Fri Nov 25 22:17:26 1994 Dave Love <d.love@dl.ac.uk>
4232 * Makefile.in (FLAGS_TO_PASS): pass $(CROSS)
4233 * Make-lang.in: more changes to runtime targets
4235 Thu Nov 24 18:03:21 1994 Dave Love <d.love@dl.ac.uk>
4237 * Makefile.in (FLAGS_TO_PASS): define for sub-makes
4239 * g77.c (main): change f77-cpp-output to f77-cpp-input (.F files)
4241 Wed Nov 23 15:22:53 1994 Dave Love <d.love@dl.ac.uk>
4243 * bad.c (ffebad_finish): kluge to fool emacs19 into finding errors:
4244 add trailing space to <file>:<line>:
4246 Tue Nov 22 11:30:50 1994 Dave Love <d.love@dl.ac.uk>
4248 * runtime/libF77/signal_.c (RETSIGTYPE): added
4250 Mon Nov 21 13:04:13 1994 Dave Love <d.love@dl.ac.uk>
4252 * Makefile.in (compiler): add runtime
4254 * config-lang.in (stagestuff): add libf2c.a to stagestuff
4256 * Make-lang.in:
4257 G77STAGESTUFF <- MORESTAGESTUFF
4258 f77-runtime: new target, plus supporting ones
4260 * runtime: add the directory, containing libI77, libF77 and autoconf
4261 stuff
4263 * g++.1: remove
4265 * g77.1: minor fixes
4267 Thu Nov 17 15:18:05 1994 Craig Burley (burley@gnu.ai.mit.edu)
4269 * Version 0.5.2 released.
4271 * bad.def: Modify wording of FFEBAD_UNIMPL_STMT to indicate
4272 that it covers a wide array of possible problems (that, someday,
4273 should be handled via separate diagnostics).
4275 * lex.c: Allow $ in identifiers if -fdollar-ok.
4276 * top.c: Support -fdollar-ok.
4277 * top.h: Support -fdollar-ok.
4278 * target.h: Support -fdollar-ok.
4279 * DOC: Describe -fdollar-ok.
4281 * std.c (ffestd_R1229_finish): Fix bug so stand-alone build works.
4282 * ste.c (ffeste_R819A): Fix bug so stand-alone build works.
4284 * Make: Improvements for stand-alone build.
4286 * Makefile.in: Fix copyright text at top of file.
4288 * LINK, SRCS, UNLINK: Removed. Not particularly useful now that
4289 g77 sources live in their own subdirectory.
4291 * g77.c (main): Cast arg to bzero to avoid warning. (This is
4292 identical to Kenner's fix to cp/g++.c.)
4294 * gbe/: New subdirectory, to contain .diff files for various
4295 versions of the GNU CC back end.
4297 * gbe/README: New file.
4298 * gbe/2.6.2.diff: New file.
4300 Tue Nov 8 10:23:10 1994 Dave Love <d.love@dl.ac.uk>
4302 * Make-lang.in: don't install as f77 as well as g77 to avoid
4303 confusion with system's compiler (especially while testing)
4305 * g77.c (main): use -lf2c and -lm; fix sense of test for .f/.F files
4307 Fri Oct 28 09:45:00 1994 Craig Burley (burley@gnu.ai.mit.edu)
4309 * Version 0.5.1 released.
4311 * gcc.c: Invoke f771 instead of f-771.
4313 Fri Oct 28 02:00:44 1994 Craig Burley (burley@gnu.ai.mit.edu)
4315 * Version 0.5.0 released.
4317 Fri Oct 14 15:03:35 1994 Craig Burley (burley@gnu.ai.mit.edu)
4319 * Makefile.in: Handle the Fortran-77 front-end in a subdirectory.
4320 * f-*: Move Fortran-77 front-end to f/*.