1 Fri Feb 20 10:11:20 1998 Craig Burley <burley@gnu.org>
3 * Make-lang.in (f77.install-common): Don't install, and
4 don't uninstall existing, Info files if f/g77.info
5 doesn't exit. (This is a somewhat modified version
6 of an egcs patch on 1998-01-07 12:05:51 by Bruno Haible
7 <bruno@linuix.mathematik.uni-karlsruhe.de>.
9 Sun Jan 11 02:14:47 1998 Craig Burley <burley@gnu.org>
11 Fix 970626-2.f by not doing most back-end processing
12 when current_function_decl is an ERROR_MARK, and by
13 making that the case when its type would be an ERROR_MARK:
14 * com.c (ffecom_start_progunit_, finish_function,
15 lang_printable_name, start_function,
16 ffecom_finish_symbol_transform_): Test for ERROR_MARK.
17 * std.c (ffestd_stmt_pass_): Don't do any downstream
18 processing if ERROR_MARK.
20 Support FORMAT(I<1+2>) (constant variable-FORMAT
22 * bad.def (FFEBAD_FORMAT_VARIABLE): New diagnostic.
23 * std.c (ffestd_R1001rtexpr_): New function.
24 (ffestd_R1001dump_, ffestd_R1001dump_1005_1_,
25 ffestd_R1001dump_1005_2_, ffestd_R1001dump_1005_3_,
26 ffestd_R1001dump_1005_4_, ffestd_R1001dump_1005_5_,
27 ffestd_R1001dump_1010_2_, ffestd_R1001dump_1010_3_,
28 ffestd_R1001dump_1010_4_, ffestd_R1001dump_1010_5_):
29 Use new function instead of ffestd_R1001error_.
31 * stb.c (ffestb_R10014_, ffestb_R10016_, ffestb_R10018_,
32 ffestb_R100110_): Restructure `for' loop for style.
34 Fri Jan 9 19:09:07 1998 Craig Burley <burley@gnu.org>
36 * intrin.c (ffeintrin_check_): No longer ignore explicit
37 omitted trailing args.
39 Tue Dec 23 14:58:04 1997 Craig Burley <burley@gnu.org>
41 Improve run-time diagnostic for "PRINT '(I1', 42":
42 * com.c (ffecom_char_args_x_): Renamed from ffecom_char_args_,
43 which is now a macro (to avoid lots of changes to other code)
44 with new arg, ffecom_char_args_with_null_ being another new
45 macro to call same function with different value for new arg.
46 This function now appends a null byte to opCONTER expression
47 if the new arg is TRUE.
48 (ffecom_arg_ptr_to_expr): Support NULL length pointer.
49 * ste.c (ffeste_io_cilist_):
50 (ffeste_io_icilist_): Pass NULL length ptr for
51 FORMAT expression, so null byte gets appended where
53 * target.c (ffetarget_character1):
54 (ffetarget_concatenate_character1):
55 (ffetarget_substr_character1):
56 (ffetarget_convert_character1_character1):
57 (ffetarget_convert_character1_hollerith):
58 (ffetarget_convert_character1_integer4):
59 (ffetarget_convert_character1_logical4):
60 (ffetarget_convert_character1_typeless):
61 (ffetarget_hollerith): Append extra phantom null byte as
62 part of FFETARGET-NULL-BYTE kludge.
64 * intrin.c (ffeintrin_fulfill_generic): Don't generate
65 FFEBAD_INTRINSIC_TYPE for CHARACTER*(*) intrinsic.
67 * com.c (ffecom_gfrt_basictype):
68 (ffecom_gfrt_kindtype):
70 (FFECOM_rttypeVOIDSTAR_): New return type `void *', for
72 * com-rt.def (FFECOM_rttypeSIGNAL): Now returns `void *'.
73 * intdoc.c: Replace `p' kind specifier with `7'.
74 * intrin.c (ffeintrin_check_, ffeintrin_init_0): Replace
75 `p' kind specifier with `7'.
76 * intrin.def (FFEINTRIN_impLOC, FFEINTRIN_impSIGNAL_func,
77 FFEINTRIN_impSIGNAL_subr): Replace `p' specifier with `7'.
78 Also, SIGNAL now returns a `void *' status, not `int'.
80 Mon Dec 22 12:41:07 1997 Craig Burley <burley@gnu.org>
82 * intrin.c (ffeintrin_init_0): Remove duplicate
85 Sun Dec 14 02:49:58 1997 Craig Burley <burley@gnu.org>
87 * intrin.c (ffeintrin_init_0): Fix up indentation a bit.
88 Fix bug that prevented checking of arguments other
91 * intdoc.c: Fix up indentation a bit.
93 Mon Dec 1 19:12:36 1997 Craig Burley <burley@gnu.org>
95 * intrin.c (ffeintrin_check_): Fix up indentation a bit more.
97 Mon Dec 1 16:21:08 1997 Craig Burley <burley@gnu.org>
99 * com.c (ffecom_arglist_expr_): Crash if non-supplied
100 optional arg isn't passed as an address.
101 Pass null pointer explicitly, instead of via ffecom routine.
102 If incoming argstring is NULL, substitute pointer to "0".
103 Recognize '0' as ending the usual arg stuff, just like '\0'.
105 Sun Nov 30 22:22:22 1997 Craig Burley <burley@gnu.org>
107 * intrin.def: Fix up spacing a bit.
109 * intdoc.c: Minor fix-ups.
111 * intrin.c (ffeintrin_check_): Fix up indentation a bit.
113 1997-11-17 Dave Love <d.love@dl.ac.uk>
115 * com.c (ffecom_arglist_expr_): Pass null pointers for optional
116 args which aren't supplied.
118 Sun Oct 26 02:36:21 1997 Craig Burley <burley@gnu.ai.mit.edu>
120 * com.c (lang_print_error_function): Fix to more
121 reliably notice when the diagnosed region changes.
123 Sat Oct 25 23:43:36 1997 Craig Burley <burley@gnu.ai.mit.edu>
126 * sta.c, sta.h (ffesta_outpooldisp): New function.
127 * std.c (ffestd_stmt_pass_): Don't kill NULL pool.
128 (ffestd_R842): If pool already preserved, save NULL
129 for pool, because it should be killed only once.
131 * malloc.c [MALLOC_DEBUG]: Put initializer for `name'
132 component in braces, to avoid compiler warning.
134 Fri Oct 10 13:00:48 1997 Craig Burley <burley@gnu.ai.mit.edu>
136 * ste.c (ffeste_begin_iterdo_): Fix loop setup so iteration
137 variable is modified only after the #iterations is calculated;
138 otherwise if the iteration variable is aliased to any of the
139 operands in the start, end, or increment expressions, the
140 wrong #iterations might be calculated.
142 * com.c (ffecom_save_tree): Fix indentation.
144 Mon Sep 29 16:18:21 1997 Craig Burley <burley@gnu.ai.mit.edu>
146 * stu.c (ffestu_list_exec_transition_,
147 ffestu_dummies_transition_): Specify `bool' type for
148 `in_progress' variables.
150 * com.h (assemble_string): Declare this routine (instead
151 of #include'ing "output.h" from gcc) to eliminate warnings
154 Fri Sep 19 01:12:27 1997 Craig Burley <burley@gnu.ai.mit.edu>
156 * expr.c (ffeexpr_reduced_eqop2_):
157 (ffeexpr_reduced_relop2_): Minor fixes to diagnostic code.
159 * fini.c (main): Change return type to `int'.
161 Wed Sep 17 10:47:08 1997 Craig Burley <burley@gnu.ai.mit.edu>
163 * com-rt.def (FFECOM_gfrtDSIGN, FFECOM_gfrtISIGN,
164 FFECOM_gfrtSIGN): Add second argument.
166 * expr.c (ffeexpr_cb_comma_c_): Trivial fixes.
168 Tue Sep 9 01:59:35 1997 Craig Burley <burley@gnu.ai.mit.edu>
170 * Version 0.5.21 released.
172 Tue Sep 9 00:31:01 1997 Craig Burley <burley@gnu.ai.mit.edu>
174 * intdoc.c (dumpem): Put appropriate commentary in
175 output file, so readers know it isn't source.
177 Wed Aug 27 08:08:25 1997 Craig Burley <burley@gnu.ai.mit.edu>
179 * proj.h: Always #include "config.j" first, to pick up
181 * com.c: Change bcopy() and bzero() calls to memcpy()
182 and memset() calls, to make more of g77 ANSI C.
184 1997-08-26 Dave Love <d.love@dl.ac.uk>
186 * Make-lang.in ($(srcdir)/f/runtime/configure,
187 $(srcdir)/f/runtime/libU77/configure): Fix for when srcdir isn't
190 Tue Aug 26 05:59:21 1997 Craig Burley <burley@gnu.ai.mit.edu>
192 * ansify.c (main): Make sure readers of stdout know
193 it's derived from stdin; omit comment text; get source
194 line numbers in future stderr output to be correct.
196 Tue Aug 26 01:36:01 1997 Craig Burley <burley@gnu.ai.mit.edu>
199 * stb.c (ffestb_R5284_): Allow OPEN_PAREN after closing
200 SLASH as well as NAME.
202 Mon Aug 25 23:48:17 1997 Craig Burley <burley@gnu.ai.mit.edu>
204 Changes to allow g77 docs to be built entirely from scratch
205 using any ANSI C compiler, not requiring GNU C:
206 * Make-lang.in ($(srcdir)/f/intdoc.texi): "Pipe" new
207 location of intrinsic documentation data base, f/intdoc.in,
208 through new `ansify' program to append `\n\' to quoted
209 newlines, into f/intdoc.h0. Do appropriate cleanups. Explain.
210 (f77.mostlyclean): Add f/ansify and f/intdoc.h0 to cleanups.
211 * f/ansify.c: New program.
212 * f/intdoc.c: Fix so it conforms to ANSI C.
213 #include f/intdoc.h0 instead of f/intdoc.h.
215 * f/intdoc.h, f/intdoc.in: Rename the former to the latter; no
216 changes made to the content in this patch!
217 * f/intrin.h (ffeintrinFamily): Fix to conform to ANSI C.
219 Sun Aug 24 06:52:48 1997 Craig Burley <burley@gnu.ai.mit.edu>
221 Fix up g77 compiler data base for libf2c routines:
222 * com-rt.def (FFECOM_gfrtSIGNAL): Change return type to
223 FTNINT to match actual code.
225 * com.c (ffecomRttype_): Replace FFECOM_rttypeINT_ with
226 FFECOM_rttypeFTNINT_.
227 Add and fix up comments.
228 (ffecom_make_gfrt_, ffecom_gfrt_basictype,
229 ffecom_gfrt_kindtype): Replace FFECOM_rttypeINT_ with
230 FFECOM_rttypeFTNINT_; add FFECOM_rttypeDOUBLEREAL_.
232 Wed Aug 20 17:18:40 1997 Craig Burley <burley@gnu.ai.mit.edu>
234 * global.c (ffeglobal_ref_progunit_): It's okay to have
235 a different CHARACTER*n length for a reference if the
236 existing length is for another reference, not a definition.
238 Mon Aug 18 14:27:18 1997 Craig Burley <burley@gnu.ai.mit.edu>
241 * global.c (ffeglobal_new_progunit_): Distinguish
242 between previously defined, versus inferred, filewide
243 when it comes to diagnostics.
246 * global.c (ffeglobal_ref_progunit_): Change BDATA into EXT
247 right at the beginning, so EXTERNAL FOO followed later
248 by SUBROUTINE FOO is not diagnosed.
251 * com-rt.def (FFECOM_gfrtALARM): Returns `integer', not
254 Sun Aug 17 03:32:44 1997 Craig Burley <burley@gnu.ai.mit.edu>
256 Fix up problems when virtual memory exhausted:
257 * malloc.c (malloc_new_): Use gcc's xmalloc(), so we
258 print a nicer message when malloc returns no memory.
259 (malloc_resize_): Ditto for xrealloc().
261 * Make-lang.in, Makefile.in: Comment out lines containing
264 Sat Aug 16 19:41:33 1997 Craig Burley <burley@gnu.ai.mit.edu>
266 * com.c (ffecom_make_gfrt_): For rttypeREAL_F2C_, return
267 double_type_node; for rttypeREAL_GNU_, return
270 1997-08-13 Dave Love <d.love@dl.ac.uk>
272 * config-lang.in (diff_excludes): Add some hints about known
273 problematic platforms.
275 1997-08-13 Dave Love <d.love@dl.ac.uk>
277 * intdoc.h: Document `alarm'.
279 Mon Aug 11 21:19:22 1997 Craig Burley <burley@gnu.ai.mit.edu>
281 * Make-lang.in ($(RUNTIMESTAGESTUFF)): Add
284 Mon Aug 11 01:52:03 1997 Craig Burley <burley@gnu.ai.mit.edu>
286 * com.c (ffecom_build_complex_constant_): Go with the
287 new build_complex() approach used in gcc-2.8.
289 * com.c (ffecom_sym_transform_): Don't set
290 DECL_IN_SYSTEM_HEADER for a tree node that isn't
291 a VAR_DECL, which happens when var is in common!
293 * com.c (ffecom_expr_intrinsic_) (case FFEINTRIN_impALARM):
294 No need to test codegen_imp -- there's only one valid here.
296 * intrin.def (FFEINTRIN_impALARM): Specify `Status' argument
299 Fri Aug 8 05:40:23 1997 Craig Burley <burley@gnu.ai.mit.edu>
301 Substantial changes to accommodate distinctions among
302 run-time routines that support intrinsics, and between
303 routines that compute and return the same type vs. those
304 that compute one type and return another (or `void'):
305 * com-rt.def: Specify new return type REAL_F2C_ instead
306 of many DOUBLE_, COMPLEX_F2C_ instead of COMPLEX_, and
308 Clear up the *BES* routines "once and for all".
309 * com.c: New return types.
310 (ffecom_convert_narrow_, ffecom_convert_widen_):
311 New functions that are "safe" variants of convert(),
312 to catch errors that ffecom_expr_intrinsic_() now
314 (ffecom_arglist_expr_): Ensure arguments are not
315 converted to narrower types.
316 (ffecom_call_): Ensure return value is not converted
318 (ffecom_char_args_): Use new ffeintrin_gfrt_direct()
320 (ffecom_expr_intrinsic_): Simplify how run-time
321 routine is selected (via `gfrt' only now; lose the
322 redundant `ix' variable).
323 Eliminate the `library' label; any code that doesn't
324 return directly just `break's out now with `gfrt'
326 Set `gfrt' to default choice initially, either a
327 fast direct form or, if not available, a slower
328 indirect-callable form.
329 (ffecom_make_gfrt_): No longer need to do special
330 check for complex; it's built into the new return-type
332 (ffecom_ptr_to_expr): Use new ffeintrin_gfrt_indirect()
334 * intrin.c, intrin.h: `gfrt' field replaced with three fields,
335 so it is easier to provide faster direct-callable and
336 GNU-convention indirect-callable routines in the future.
337 DEFIMP macro adjusted accordingly, along with all its uses.
338 (ffeintrin_gfrt_direct): New function.
339 (ffeintrin_gfrt_indirect): Ditto.
340 (ffeintrin_is_actualarg): If `-fno-f2c' is in effect,
341 require a GNU-callable version of intrinsic instead of
342 an f2c-callable version, so indirect calling is still checked.
343 * intrin.def: Replace one GFRT field with the three new fields,
344 as appropriate for each DEFIMP intrinsic.
346 * com.c (ffecom_stabilize_aggregate_,
347 ffecom_convert_to_complex_): Make these `static'.
349 Thu Aug 7 11:24:34 1997 Craig Burley <burley@gnu.ai.mit.edu>
351 Provide means for front end to determine actual
352 "standard" return type for an intrinsic if it is
353 passed as an actual argument:
354 * com.h, com.c (ffecom_gfrt_basictype,
355 ffecom_gfrt_kindtype): New functions.
356 (ffecom_gfrt_kind_type_): Replaced with new function.
358 (ffecom_make_gfrt_): No longer need do anything
361 * intrin.c (ffeintrin_basictype, ffeintrin_kindtype):
362 Now returns correct type info for specific intrinsic
363 (based on type of run-time-library implementation).
365 Wed Aug 6 23:08:46 1997 Craig Burley <burley@gnu.ai.mit.edu>
367 * global.c (ffeglobal_ref_progunit_): Don't reset
368 number of arguments just due to new type info,
369 so useful warnings can be issued.
371 1997-08-06 Dave Love <d.love@dl.ac.uk>
373 * intrin.def: Fix IDATE_vxt argument order.
374 * intdoc.h: Likewise.
376 Thu Jul 31 22:22:03 1997 Craig Burley <burley@gnu.ai.mit.edu>
378 * global.c (ffeglobal_proc_ref_arg): If REF/DESCR
379 disagreement, DESCR is CHARACTER, and types disagree,
380 pretend the argsummary agrees so the message ends up
381 being about type disagreement.
382 (ffeglobal_proc_def_arg): Ditto.
384 * expr.c (ffeexpr_token_first_rhs_3_): Set info for LABTOK
385 to NONE of everything, to avoid misdiagnosing filewide
386 usage of alternate returns.
388 Sun Jul 20 23:07:47 1997 Craig Burley <burley@gnu.ai.mit.edu>
390 * com.c (ffecom_sym_transform_): If type gets set
391 to error_mark_node, just return that for transformed symbol.
392 (ffecom_member_phase2_): If type gets set to error_mark_node,
394 (ffecom_check_size_overflow_): Add `dummy' argument to
395 flag that type is for a dummy, update all callers.
397 Sun Jul 13 17:40:53 1997 Craig Burley <burley@gnu.ai.mit.edu>
400 * where.c (ffewhere_set_from_track): If start point
401 is too large, just use initial start point. 0.6 should
402 fix all this properly.
405 * com.c (ffecom_sym_transform_): Preserve error_mark_node for type.
406 (ffecom_type_localvar_): Ditto.
407 (ffecom_sym_transform_): If type is error_mark_node,
408 don't error-check decl size, because back end responds by
409 setting that to an integer 0 instead of error_mark_node.
410 (ffecom_transform_common_): Same as earlier fix to _transform_
411 in that size is checked by dividing BITS_PER_UNIT instead of
413 (ffecom_transform_equiv_): Ditto.
416 * stb.c (ffestb_R10014_): Fix flaky fall-through in error
417 test for FFELEX_typeCONCAT by just replicating the code,
418 and do FFELEX_typeCOLONCOLON while at it.
420 1997-07-07 Dave Love <d.love@dl.ac.uk>
422 * intdoc.h: Add various missing pieces; correct GMTIME, LTIME
425 * intrin.def, com-rt.def: Add alarm.
427 * com.c (ffecom_expr_intrinsic_): Add case for alarm.
429 Thu Jun 26 04:19:40 1997 Craig Burley <burley@gnu.ai.mit.edu>
432 * com.c (ffecom_sym_transform_): For sanity-check compare
433 of gbe size of local variable to g77 expectation,
434 use varasm.c/assemble_variable technique of dividing
435 BITS_PER_UNIT out of gbe info instead of multiplying
436 g77 info up, to avoid crash when size in bytes is very
437 large, and overflows an `int' or similar when multiplied.
440 * com.c (ffecom_finish_symbol_transform_): Don't bother
441 transforming a dummy argument, to avoid a crash.
442 * ste.c (ffeste_R1227): Don't return a value if the
443 result decl, or its type, is error_mark_node.
446 * lex.c (ffelex_splice_tokens): `-fdollar-ok' is
447 irrelevant to whether a DOLLAR token should be made
448 from an initial character of `$'.
451 * stb.c (ffestb_do3_): DO iteration variable is an
452 lhs, not rhs, expression.
454 Fix 970626-7.f and 970626-8.f:
455 * expr.c (ffeexpr_cb_comma_i_1_): Set IMPDO expression
456 to have clean info, because undefined rank, for example,
457 caused crash on mangled source on UltraSPARC but not
458 on Alpha for a series of weird reasons.
459 (ffeexpr_cb_close_paren_): If not CLOSE_PAREN, push
460 opANY expression onto stack instead of attempting
461 to mimic what program might have wanted.
462 (ffeexpr_cb_close_paren_): Don't wrap opPAREN around
463 opIMPDO, just warn that it's gratuitous.
464 * bad.def (FFEBAD_IMPDO_PAREN): New warning.
467 * expr.c (ffeexpr_declare_parenthesized_): Must shut down
468 parsing in kindANY case, otherwise the parsing engine might
469 decide there's an ambiguity.
470 (ffeexpr_token_name_rhs_): Eliminate parentypeSUBROUTINE_
471 case, so we crash right away if it comes through.
472 * st.c, st.h, sta.c, sta.h (ffest_shutdown, ffesta_shutdown):
475 Tue Jun 24 19:47:29 1997 Craig Burley <burley@gnu.ai.mit.edu>
477 * com.c (ffecom_check_size_overflow_): New function
478 catches some cases of the size of a type getting
479 too large. varasm.c must catch the rest.
480 (ffecom_sym_transform_): Use new function.
481 (ffecom_type_localvar_): Ditto.
483 Mon Jun 23 01:09:28 1997 Craig Burley <burley@gnu.ai.mit.edu>
485 * global.c (ffeglobal_proc_def_arg): Fix comparison
487 (ffeglobal_proc_ref_arg): Ditto.
489 * lang-options.h, top.c: Rename `-fdebug' to `-fxyzzy',
490 since it's an unsupported internals option and some
491 poor user might guess that it does something.
493 * bad.def: Make a warning for each filewide diagnostic.
494 Put all filewides together.
495 * com.c (ffecom_sym_transform_): Don't substitute
496 known global tree for global entities when `-fno-globals'.
497 * global.c (ffeglobal_new_progunit_): Don't produce
498 fatal diagnostics about globals when `-fno-globals'.
499 Instead, produce equivalent warning when `-Wglobals'.
500 (ffeglobal_proc_ref_arg): Ditto.
501 (ffeglobal_proc_ref_nargs): Ditto.
502 (ffeglobal_ref_progunit_): Ditto.
503 * lang-options.h, top.c, top.h: New `-fno-globals' option.
505 Sat Jun 21 12:32:54 1997 Craig Burley <burley@gnu.ai.mit.edu>
507 * expr.c (ffeexpr_fulfill_call_): Set array variable
508 to avoid warning about uninitialized variable.
510 * Make-lang.in: Get rid of any setting of HOST_* macros,
511 since these will break gcc's build!
512 * makefile: New file to make building derived files
515 Thu Jun 19 18:19:28 1997 Craig Burley <burley@gnu.ai.mit.edu>
517 * g77.c (main): Install Emilio Lopes' patch to support
518 Ratfor, and to fix the printing of the version string
519 to go to stderr, not stdout.
520 * lang-specs.h: Install Emilio Lopes' patch to support
521 Ratfor, and patch the result to support picking up
522 `*f771' from the `specs' file.
524 Thu Jun 12 14:36:25 1997 Craig Burley <burley@gnu.ai.mit.edu>
526 * storag.c (ffestorag_update_init, ffestorag_update_save):
527 Also update parent, in case equivalence processing
528 has already eliminated pointers to it via the
529 local equivalence info.
531 Tue Jun 10 14:08:26 1997 Craig Burley <burley@gnu.ai.mit.edu>
533 * intdoc.c: Add cross-reference to end of description
534 of any generic intrinsic pointing to other intrinsics
537 Warn about explicit type declaration for intrinsic
538 that disagrees with invocation:
539 * expr.c (ffeexpr_paren_rhs_let_): Preserve type info
540 for intrinsic functions.
541 (ffeexpr_token_funsubstr_): Ditto.
542 * intrin.c (ffeintrin_fulfill_generic): Warn if type
543 info of fulfilled intrinsic invocation disagrees with
544 explicit type info given symbol.
545 (ffeintrin_fulfill_specific): Ditto.
546 * stc.c (ffestc_R1208_item): Preserve type info
548 (ffestc_R501_item): Ditto.
550 Mon Jun 9 17:45:44 1997 Craig Burley <burley@gnu.ai.mit.edu>
552 * com.c (ffecom_expr_intrinsic_): Fix several of the
553 libU77/libF77-unix handlers to properly convert their
556 * com-rt.def (FFECOM_gfrtFSTAT): Append missing "i" to
559 Fri Jun 6 14:37:30 1997 Craig Burley <burley@gnu.ai.mit.edu>
561 * com.c (ffecom_expr_intrinsic_): Have a case statement
562 for every intrinsic implementation, so missing ones
563 are caught via gcc warnings.
564 Don't call ffeintrin_codegen_imp anymore.
565 * intrin.c (ffeintrin_fulfill_generic): Remove cg_imp
567 (ffeintrin_codegen_imp): Delete this function.
568 * intrin.def, intrin.h: Remove DEFIMQ stuff from here
571 Thu Jun 5 13:03:07 1997 Craig Burley <burley@gnu.ai.mit.edu>
573 * top.c (ffe_decode_option): New -fbadu77-intrinsics-*
576 * intrin.h: New BADU77 family.
577 * intrin.c (ffeintrin_state_family): Ditto.
579 Implement new scheme to track intrinsic names vs. forms:
580 * intrin.c (ffeintrin_fulfill_generic),
581 (ffeintrin_fulfill_specific), (ffeintrin_is_intrinsic),
582 intrin.def: The documented name is now either in the
583 generic info or, if no generic, in the specific info.
584 For a generic, the specific info contains merely the
585 distinguishing form (usually "function" or "subroutine"),
586 used for diagnostics about ambiguous references and
587 in the documentation.
589 * intrin.def: Clean up formatting of DEFNAME block.
590 Convert many libU77 intrinsics into generics that
591 support both subroutine and function forms.
592 Put the function forms of side-effect routines into
593 the new BADU77 family.
594 Make MCLOCK and TIME return INTEGER*4 again, and add
595 INTEGER*8 equivalents called MCLOCK8 and TIME8.
596 Fix up more status return values to be written and
597 insist on them being I1 as well.
598 * com.c (ffecom_expr_intrinsic_): Lots of changes to
599 support new libU77 intrinsic interfaces.
601 Mon Jun 2 00:37:53 1997 Craig Burley <burley@gnu.ai.mit.edu>
603 * com.c (ffecom_init_0): Pointer type is now INTEGER(KIND=7),
604 not INTEGER(KIND=0), since we want to reserve KIND=0 for
607 Thu May 29 14:30:33 1997 Craig Burley <burley@gnu.ai.mit.edu>
609 Fix bugs preventing CTIME(I*4) from working correctly:
610 * com.c (ffecom_char_args_): For FUNCREF case, process
611 args to intrinsic just as they would be in
612 ffecom_expr_intrinsic_.
613 * com-rt.def (FFECOM_gfrtCTIME, FFECOM_gfrtTTYNAM): Fix
614 argument decls to specify `&'.
616 Wed May 28 22:19:49 1997 Craig Burley <burley@gnu.ai.mit.edu>
618 Fix gratuitous warnings exposed by dophot aka 970528-1:
619 * global.c (ffeglobal_proc_def_arg, ffeglobal_proc_ref_arg):
620 Support distinct function/subroutine arguments instead of
623 * expr.c (ffeexpr_fulfill_call_): A SYMTER with kindNONE
624 also is a procedure (either function or subroutine).
626 Mon May 26 20:25:31 1997 Craig Burley <burley@gnu.ai.mit.edu>
628 * bad.def: Have several lexer diagnostics refer to
629 documentation for people who need more info on what Fortran
630 source code is supposed to look like.
632 * expr.c (ffeexpr_reduced_bool1_), bad.def: New diagnostics
633 specific to .NOT. now mention only one operand instead
636 * g77.c: Recognize -fsyntax-only, similar to -c etc.
637 (lookup_option): Fix bug that prevented non-`--' options
638 from being recognized.
640 Sun May 25 04:29:04 1997 Craig Burley <burley@gnu.ai.mit.edu>
642 * intrin.def (FFEINTRIN_impCTIME): Accept `I*' expression
643 for STime instead of requiring `I2'.
645 Tue May 20 16:14:40 1997 Craig Burley <burley@gnu.ai.mit.edu>
647 * symbol.c (ffesymbol_reference): All references to
648 standard intrinsics are considered explicit, so as
649 to avoid generating basically useless warnings.
650 * intrin.c, intrin.h (ffeintrin_is_standard): Returns TRUE
651 if intrinsic is standard.
653 Sun May 18 21:14:59 1997 Craig Burley <burley@gnu.ai.mit.edu>
655 * com-rt.def: Changed all external names of the
656 form `"\([a-z0-9]*\)_' to `"G77_\1_0"' so as to
657 allow any name valid as an intrinsic to be used
658 as such and as a user-defined external procedure
659 name or common block as well.
661 Thu May 8 13:07:10 1997 Craig Burley <burley@gnu.ai.mit.edu>
663 * expr.c (ffeexpr_cb_end_notloc_): For %VAL, %REF, and
664 %DESCR, copy arg info into new node.
666 Mon May 5 14:42:17 1997 Craig Burley <burley@gnu.ai.mit.edu>
668 From Uwe F. Mayer <mayer@math.Vanderbilt.Edu>:
669 * Make-lang.in (g77-cross): Fix typo in g77.c path.
671 From Brian McIlwrath <bkm@star.rl.ac.uk>:
672 * lang-specs.h: Have g77 pick up options from a section
673 labeled `*f771' of the `specs' file.
675 Sat May 3 02:46:08 1997 Craig Burley <burley@gnu.ai.mit.edu>
677 * intrin.def (FFEINTRIN_defSIGNAL): Add optional `Status'
678 argument that com.c already expects (per Dave Love).
680 More changes to support better tracking of (filewide)
681 globals, in particular, the arguments to procedures:
682 * bad.def (FFEBAD_FILEWIDE_NARGS, FFEBAD_FILEWIDE_NARGS_W,
683 FFEBAD_FILEWIDE_ARG, FFEBAD_FILEWIDE_ARG_W): New diagnostics.
684 * expr.c (ffebad_fulfill_call_): Provide info on each
685 argument to ffeglobal.
686 * global.c, global.h (ffeglobal_proc_def_arg,
687 ffeglobal_proc_def_nargs, ffeglobal_proc_ref_arg,
688 ffeglobal_proc_ref_args): New functions.
689 (ffeglobalArgSummary, ffeglobalArgInfo_): New types.
691 Tue Apr 29 18:35:41 1997 Craig Burley <burley@gnu.ai.mit.edu>
693 More changes to support better tracking of (filewide)
695 * expr.c (ffeexpr_fulfill_call_): New function.
696 (ffeexpr_token_name_lhs_): Call after building procedure
697 reference expression. Also leave info field for ANY-ized
699 (ffeexpr_token_arguments_): Ditto.
701 Mon Apr 28 20:04:18 1997 Craig Burley <burley@gnu.ai.mit.edu>
703 Changes to support better tracking of (filewide)
704 globals, mainly to avoid crashes due to inlining:
705 * bad.def: Go back to quoting intrinsic names,
706 (FFEBAD_FILEWIDE_DISAGREEMENT, FFEBAD_FILEWIDE_TIFF,
707 FFEBAD_FILEWIDE_TYPE_MISMATCH): New diagnostics.
708 (FFEBAD_INTRINSIC_EXPIMP, FFEBAD_INTRINSIC_GLOBAL): Reword
710 * com.c (ffecom_do_entry_, ffecom_start_progunit_,
711 ffecom_sym_transform_): Accommodate new FFEGLOBAL_typeEXT
713 * expr.c (ffeexpr_sym_lhs_call_, ffeexpr_sym_lhs_extfunc_,
714 ffeexpr_sym_rhs_actualarg_, ffeexpr_declare_parenthesized_,
715 ffeexpr_paren_rhs_let_, ffeexpr_token_funsubstr_):
716 Fill in real kind info instead of leaving NONE where
718 Register references to intrinsics and globals with ffesymbol
719 using new ffesymbol_reference function instead of
721 * global.c (ffeglobal_type_string_): New array for
723 * global.h, global.c:
724 Replace ->init mechanism with ->tick mechanism.
725 Move other common-related members into a substructure of
726 a union, so the proc substructure can be introduced
727 to include members related to externals other than commons.
728 Don't complain about ANY-ized globals; ANY-ize globals
729 once they're complained about, in any case where code
730 generation could become a problem.
731 Handle global entries that have NONE type (seen as
732 intrinsics), EXT type (seen as EXTERNAL), and so on.
733 Keep track of kind and type of externals, both via
734 definition and via reference.
735 Diagnose disagreements about kind or type of externals
737 (ffeglobal_ref_intrinsic, ffeglobal_ref_progunit_): New
739 * stc.c (ffestc_R1207_item, ffestc_R1208_item,
740 ffestc_R1219, ffestc_R1226):
741 Call ffesymbol_reference, not ffesymbol_globalize.
742 * stu.c (ffestu_sym_end_transition,
743 ffestu_sym_exec_transition):
744 Call ffesymbol_reference, not ffesymbol_globalize.
745 * symbol.c (ffesymbol_globalize): Removed...
746 (ffesymbol_reference): ...to this new function,
747 which more generally registers references to symbols,
748 globalizes globals, and calls on the ffeglobal module
749 to check globals filewide.
751 * global.h, global.c: Rename some macros and functions
752 to more clearly distinguish common from other globals.
755 * com.c (ffecom_sym_transform_): Trees describing
756 filewide globals must be allocated on permanent obstack.
758 * expr.c (ffeexpr_token_name_lhs_): Don't generate
759 gratuitous diagnostics for FFEINFO_whereANY case.
761 Thu Apr 17 03:27:18 1997 Craig Burley <burley@gnu.ai.mit.edu>
763 * global.c: Add support for flagging intrinsic/global
764 confusion via warnings.
765 * bad.def (FFEBAD_INTRINSIC_EXPIMP,
766 FFEBAD_INTRINSIC_GLOBAL): New diagnostics.
767 * expr.c (ffeexpr_token_funsubstr_): Ditto.
768 (ffeexpr_sym_lhs_call_): Ditto.
769 (ffeexpr_paren_rhs_let_): Ditto.
770 * stc.c (ffestc_R1208_item): Ditto.
772 Wed Apr 16 22:40:56 1997 Craig Burley <burley@gnu.ai.mit.edu>
774 * expr.c (ffeexpr_declare_parenthesized_): INCLUDE
775 context can't be an intrinsic invocation either.
777 Fri Mar 28 10:43:28 1997 Craig Burley <burley@gnu.ai.mit.edu>
779 * expr.c (ffeexpr_token_arguments_): Make sure top of
780 exprstack is operand before dereferencing operand field.
782 * lex.c (ffelex_prepare_eos_): Fill up truncated
783 hollerith token, so crash on null ->text field doesn't
786 * stb.c (ffestb_R10014_): If NAMES isn't recognized (or
787 the recognized part is followed in the token by a
788 non-digit), don't try and collect digits, as there
789 might be more than FFEWHERE_indexMAX letters to skip
790 past to do so -- and the code is diagnosed anyway.
792 Thu Mar 27 00:02:48 1997 Craig Burley <burley@gnu.ai.mit.edu>
794 * com.c (ffecom_sym_transform_): Force local
795 adjustable array onto stack.
797 * stc.c (ffestc_R547_item_object): Don't actually put
798 the symbol in COMMON if the symbol has already been
799 EQUIVALENCE'd to a different COMMON area.
801 * equiv.c (ffeequiv_add): Don't actually do anything
802 if there's a disagreement over which COMMON area is
805 Tue Mar 25 03:35:19 1997 Craig Burley <burley@gnu.ai.mit.edu>
807 * com.c (ffecom_transform_common_): If no explicit init
808 of COMMON area, don't actually init it even though
809 storage area suggests it.
811 Mon Mar 24 12:10:08 1997 Craig Burley <burley@gnu.ai.mit.edu>
813 * lex.c (ffelex_image_char_): Avoid overflowing the
814 column counter itself, as well as the card image.
816 * where.c (ffewhere_line_new): Cast ffelex_line_length()
817 to (size_t) so 255 doesn't overflow to 0!
819 * stc.c (ffestc_labeldef_notloop_begin_): Don't gratuitously
820 terminate loop before processing statement, so block
821 doesn't disappear out from under EXIT/CYCLE processing.
822 (ffestc_labeldef_notloop_): Has old code from above
823 function, instead of just calling it.
825 * expr.c (ffeexpr_cb_comma_i_4_): Don't skip over
826 arbitrary token (such as EOS).
828 * com.c (ffecom_init_zero_): Handle RECORD_TYPE and
829 UNION_TYPE so -fno-zeros works with -femulated-complex.
831 1997-03-12 Dave Love <d.love@dl.ac.uk>
833 * intrin.def: New intrinsics INT2, INT8, CPU_TIME. Fix AND, OR,
834 XOR. [Integrated by burley, AND/OR/XOR already fixed, INT8
835 implementation changed/fixed.]
837 Wed Mar 12 10:40:08 1997 Craig Burley <burley@gnu.ai.mit.edu>
839 * Make-lang.in ($(srcdir)/f/intdoc.texi): Simplify rules
840 so building f/intdoc is not always necessary; remove
841 f/intdoc after running it if it is built.
843 Tue Mar 11 23:42:00 1997 Craig Burley <burley@gnu.ai.mit.edu>
845 * intrin.def (FFEINTRIN_impAND, FFEINTRIN_impOR,
846 FFEINTRIN_impXOR): Use the IAND, IOR, and IEOR implementations
847 of these, instead of crashing in ffecom_expr_intrinsic_
848 or adding case labels there.
850 Mon Mar 10 22:51:23 1997 Craig Burley <burley@gnu.ai.mit.edu>
852 * intdoc.c: Fix so any C compiler can compile this.
854 Fri Feb 28 13:16:50 1997 Craig Burley <burley@gnu.ai.mit.edu>
856 * Version 0.5.20 released.
858 Fri Feb 28 01:45:25 1997 Craig Burley <burley@gnu.ai.mit.edu>
860 * Make-lang.in (RUNTIMESTAGESTUFF, LIBU77STAGESTUFF):
861 Move some files incorrectly in the former to the latter,
862 and add another file or two to the latter.
864 New meanings for (KIND=n), and new denotations in the
865 little language describing intrinsics:
866 * com.c (ffecom_init_0): Assign new meanings.
867 * intdoc.c: Document new meanings.
868 Support the new denotations.
869 * intrin.c: Employ new meanings, mapping them to internal
870 values (which are the same as they ever were for now).
871 Support the new denotations.
872 * intrin.def: Switch DEFIMP table to the new denotations.
874 * intrin.c (ffeintrin_check_): Fix bug that was leaving
875 LOC() and %LOC() returning INTEGER*4 on systems where
876 it should return INTEGER*8.
878 * type.c: Canonicalize function definitions, for etags
881 Wed Feb 26 20:43:03 1997 Craig Burley <burley@gnu.ai.mit.edu>
883 * com.c (ffecom_init_0): Choose INTEGER(KIND=n) types,
884 where n is 2, 3, and 4, according to the new docs
885 instead of according to the old C correspondences
886 (which seem less useful at this point).
888 * equiv.c (ffeequiv_destroy_): New function.
889 (ffeequiv_layout_local_): Use this new function
890 whenever the laying out of a local equivalence chain
891 is aborted for any reason.
892 Otherwise ensure that symbols no longer reference
893 the stale ffeequiv entries that result when they
894 are killed off in this procedure.
895 Also, the rooted symbol is one that has storage,
896 it really is irrelevant whether it has an equiv entry
897 at this point (though the code to remove the equiv
898 entry was put in at the end, just in case).
899 (ffeequiv_kill): When doing internal checks, make
900 sure the victim isn't named by any symbols it points
901 to. Not as complete a check as looking through the
902 entire symbol table (which does matter, since some
903 code in equiv.c used to remove symbols from the lists
904 for an ffeequiv victim but not remove that victim as the
905 symbol's equiv info), but this check did find some
906 real bugs in the code (that were fixed).
908 Mon Feb 24 16:42:13 1997 Craig Burley <burley@gnu.ai.mit.edu>
910 * com.c (ffecom_expr_intrinsic_): Fix a couple of
911 warnings about uninitialized variables.
912 * intrin.c (ffeintrin_check_): Ditto, but there were
913 a couple of _real_ uninitialized-variable _bugs_ here!
914 (ffeintrin_fulfill_specific): Ditto, no real bug here.
916 Sun Feb 23 15:01:20 1997 Craig Burley <burley@gnu.ai.mit.edu>
918 Clean up diagnostics (especially about intrinsics):
919 * bad.def (FFEBAD_UNIMPL_STMT): Remove.
920 (FFEBAD_INTRINSIC_*, FFEBAD_NEED_INTRINSIC): Clean these
921 up so they're friendlier.
922 (FFEBAD_INTRINSIC_CMPAMBIG): New.
923 * intrin.c (ffeintrin_fulfill_generic,
924 ffeintrin_fulfill_specific, ffeintrin_is_intrinsic):
926 generic or specific name text (which is for doc purposes
927 anyway) over implementation name text (which is for
929 * intrin.def: Use more descriptive name texts for generics
930 and specifics in cases where the names themselves are not
931 enough (e.g. IDATE, which has two forms).
933 Fix some intrinsic mappings:
934 * intrin.def (FFEINTRIN_specIDINT, FFEINTRIN_specAND,
935 FFEINTRIN_specDFLOAT, FFEINTRIN_specDREAL, FFEINTRIN_specOR,
936 FFEINTRIN_specXOR): Now have their own implementations,
937 instead of borrowing from others.
938 (FFEINTRIN_specAJMAX0, FFEINTRIN_specAJMIN0, FFEINTRIN_specBJTEST,
939 FFEINTRIN_specDFLOTJ, FFEINTRIN_specFLOATJ, FFEINTRIN_specJIABS,
940 FFEINTRIN_specJIAND, FFEINTRIN_specJIBCLR, FFEINTRIN_specJIBITS,
941 FFEINTRIN_specJIBSET, FFEINTRIN_specJIDIM, FFEINTRIN_specJIDINT,
942 FFEINTRIN_specJIDNNT, FFEINTRIN_specJIEOR, FFEINTRIN_specJIFIX,
943 FFEINTRIN_specJINT, FFEINTRIN_specJIOR, FFEINTRIN_specJISHFT,
944 FFEINTRIN_specJISHFTC, FFEINTRIN_specJISIN, FFEINTRIN_specJMAX0,
945 FFEINTRIN_specJMAX1, FFEINTRIN_specJMIN0, FFEINTRIN_specJMIN1,
946 FFEINTRIN_specJMOD, FFEINTRIN_specJNINT, FFEINTRIN_specJNOT,):
947 Turn these implementations off, since it's not clear
948 just what types they expect in the context of portable Fortran.
949 (DFLOAT): Now in FVZ family, since f2c supports them
951 Support intrinsic inquiry functions (BIT_SIZE, LEN):
952 * intrin.c: Allow `i' in <arg_extra>.
953 * intrin.def (FFEINTRIN_impBIT_SIZE, FFEINTRIN_impLEN):
956 Sat Feb 22 13:34:09 1997 Craig Burley <burley@gnu.ai.mit.edu>
958 Only warn, don't error, for reference to unimplemented
960 * bad.def (FFEBAD_INTRINSIC_UNIMPLW): Warning version
962 * intrin.c (ffeintrin_is_intrinsic): Use new warning
963 version of _UNIMPL (FFEBAD_INTRINSIC_UNIMPLW).
965 Complain about REAL(Z) and AIMAG(Z) (Z is DOUBLE COMPLEX):
966 * bad.def (FFEBAD_INTRINSIC_CMPAMBIG): New diagnostic.
967 * expr.c: Needed #include "intrin.h" anyway.
968 (ffeexpr_token_intrincheck_): New function handles delayed
969 diagnostic for "REAL(REAL(expr)" if next token isn't ")".
970 (ffeexpr_token_arguments_): Do most of the actual checking here.
971 * intrin.h, intrin.c (ffeintrin_fulfill_specific): New
972 argument, check_intrin, to tell caller that intrin is REAL(Z)
973 or AIMAG(Z). All callers updated, mostly to pass NULL in
975 (ffeintrin_check_): Also has new arg check_intrin for same
976 purpose. All callers updated the same way.
977 * intrin.def (FFEINTRIN_impAIMAG): Change return type
978 from "R0" to "RC", to accommodate f2c (and perhaps other
979 non-F90 F77 compilers).
980 * top.h, top.c: New option -fugly-complex.
982 New GNU intrinsics REALPART, IMAGPART, and COMPLEX:
983 * com.c (ffecom_expr_intrinsic_): Implement impCOMPLEX
984 and impREALPART here. (specIMAGPART => specAIMAG.)
985 * intrin.def: Add the intrinsics here.
987 Rename implementations of VXTIDATE and VXTTIME to IDATEVXT
988 and TIMEVXT, so they sort more consistently:
989 * com.c (ffecom_expr_intrinsic_):
992 Delete intrinsic group `dcp', add `gnu', etc.:
993 * intrin.c (ffeintrin_state_family): FFEINTRIN_familyGNU
994 replaces FFEINTRIN_familyDCP, and gets state from `gnu'
996 Get rid of FFEINTRIN_familyF2Z, nobody needs it.
997 Move FFEINTRIN_specDCMPLX from DCP family to FVZ family,
999 Move FFEINTRIN_specDFLOAT from F2C family to FVZ family.
1000 (FFEINTRIN_specZABS, FFEINTRIN_specZCOS, FFEINTRIN_specZEXP,
1001 FFEINTRIN_specZLOG, FFEINTRIN_specZSIN, FFEINTRIN_specZSQRT):
1002 Move these from F2Z family to F2C family.
1003 * intrin.h (FFEINTRIN_familyF2Z, FFEINTRIN_familyDCP): Remove.
1004 (FFEINTRIN_familyGNU): Add.
1005 * top.h, top.c: Replace `dcp' with `gnu'.
1007 * com.c (ffecom_expr_intrinsic_): Clean up by collecting
1008 simple conversions into one nice, conceptual place.
1009 Fix up some intrinsic subroutines (MVBITS, KILL, UMASK) to
1010 properly push and pop call temps, to avoid wasting temp
1013 * g77.c (doit): Toon says variables should be defined
1014 before being referenced. Spoilsport.
1016 * intrin.c (ffeintrin_check_): Now Dave's worried about
1017 warnings about uninitialized variables. Okay, so for
1018 basic return values 'g' and 's', they _were_
1019 uninitialized -- is determinism really _that_ useful?
1021 * intrin.def (FFEINTRIN_impFGETC): Fix STATUS argument
1022 so that it is INTENT(OUT) instead of INTENT(IN).
1024 1997-02-21 Dave Love <d.love@dl.ac.uk>
1026 * intrin.def, com.c: Support Sun-type `short' and `long'
1027 intrinsics. Perhaps should also do Microcruft-style `int2'.
1029 Thu Feb 20 15:16:53 1997 Craig Burley <burley@gnu.ai.mit.edu>
1031 * com.c (ffecom_expr_intrinsic_): Clean up indentation.
1032 Support SECONDSUBR intrinsic implementation.
1033 Rename SECOND to SECONDFUNC for direct support via library.
1035 * g77.c: Fix to return proper status value to shell,
1036 by obtaining it from processes it spawns.
1038 * intdoc.c: Fix minor typo.
1040 * intrin.def: Turn SECOND into generic that maps into
1041 function and subroutine forms.
1043 * intrin.def: Make FLOAT and SNGL into specific intrinsics.
1045 * intrin.def, intrin.h: Change the way DEFGEN and DEFSPEC
1046 macros work, to save on verbage.
1048 Mon Feb 17 02:08:04 1997 Craig Burley <burley@gnu.ai.mit.edu>
1050 New subsystem to automatically generate documentation
1052 * Make-lang.in ($(srcdir)/f/g77.info,
1053 $(srcdir)/f/g77.dvi): Move g77 doc rules around.
1054 Add to g77 doc rules the new subsystem.
1055 (f77.mostlyclean, f77.maintainer-clean): Also clean up
1056 after new doc subsystem.
1057 * intdoc.c, intdoc.h: New doc subsystem code.
1058 * intrin.h [FFEINTRIN_DOC]: When 1, don't pull in
1059 stuff not needed by doc subsystem.
1061 Improve on intrinsics mechanism to both be more
1062 self-documenting and to catch more user errors:
1063 * intrin.c (ffeintrin_check_): Recognize new arg-len
1064 and arg-rank information, and check it.
1065 Move goto and signal indicators to the basic type.
1066 Permit reference to arbitrary argument number, not
1067 just first argument (for BESJN and BESYN).
1068 (ffeintrin_init_0): Check and accept new notations.
1069 * intrin.c, intrin.def: Value in COL now identifies
1070 arguments starting with number 0 being the first.
1072 Some minor intrinsics cleanups (resulting from doc work):
1073 * com.c (ffecom_expr_intrinsic_): Implement FLUSH
1074 directly once again, handle its optional argument,
1075 so it need not be a generic (awkward to handle in docs).
1076 * intrin.def (BESJ0, BESJ1, BESJN, BESY0, BESY1, BESYN,
1077 CHDIR, CHMOD, CTIME, DBESJ0, DBESJ1, DBESJN, DBESY0,
1078 DBESY1, DBESYN, DDIM, ETIME, FGETC, FNUM, FPUTC, FSTAT,
1079 GERROR, GETCWD, GETGID, GETLOG, GETPID, GETUID, GMTIME,
1080 HOSTNM, IDATE, IERRNO, IIDINT, IRAND, ISATTY, ITIME, JIDINT,
1081 LNBLNK, LSTAT, LTIME, MCLOCK, PERROR, SRAND, SYMLNK, TTYNAM,
1082 UMASK): Change capitalization of initcaps (official) name
1083 to be consistent with Burley's somewhat arbitrary rules.
1084 (BESJN, BESYN): These have return arguments of same type
1085 as their _second_ argument.
1086 (FLUSH): Now a specific, not generic, intrinsic, with one
1088 (FLUSH1): Eliminated.
1089 Add arg-len and arg-rank info to several intrinsics.
1090 (ITIME): Change argument type from REAL to INTEGER.
1092 Tue Feb 11 14:04:42 1997 Craig Burley <burley@gnu.ai.mit.edu>
1094 * Make-lang.in (f771): Invocation of Makefile now done
1095 with $(srcdir)=gcc to go along with $(VPATH)=gcc.
1096 ($(srcdir)/f/runtime/configure,
1097 $(srcdir)/f/runtime/libU77/configure): Break these out
1098 so spurious triggers of this rule don't happen (as when
1099 configure.in is more recent than libU77/configure).
1100 (f77.rebuilt): Distinguish source versus build files,
1101 so this target can be invoked from build directory and
1103 * Makefile.in: This now expects $(srcdir) to be the gcc
1104 source directory, not gcc/f, to agree with $(VPATH).
1105 Accordingly, $(INCLUDES) has been fixed, various cruft
1106 removed, the removal of f771 has been fixed to remove
1107 the _real_ f771 (not the one in gcc's parent directory),
1110 * lex.c: Part of ffelex_finish_statement_() now done
1111 by new function ffelex_prepare_eos_(), so that, in one
1112 popular case, the EOS can be prepared while the pointer
1113 is at the end of the non-continued line instead of the
1114 end of the line that marks no continuation. This improves
1115 the appearance of diagnostics substantially.
1117 Mon Feb 10 12:44:06 1997 Craig Burley <burley@gnu.ai.mit.edu>
1119 * Make-lang.in: runtime Makefile's, and include/f2c.h,
1120 also depend on f/runtime/configure and f/runtime/libU77/configure.
1122 Fix various libU77 routines:
1123 * com-rt.def (FFECOM_gfrtCTIME, FFECOM_gfrtMCLOCK,
1124 FFECOM_gfrtTIME): These now use INTEGER*8 for time values,
1125 for compatibility with systems like Alpha.
1126 (FFECOM_gfrtSYSTEM_CLOCK, FFECOM_gfrtTTYNAM): Delete incorrect
1127 trailing underscore in routine names.
1128 * intrin.c, intrin.def: Support INTEGER*8 return values and
1129 arguments ('4'). Change FFEINTRIN_impCTIME, FFEINTRIN_impMCLOCK,
1130 and FFEINTRIN_impTIME accordingly.
1131 (ffeintrin_is_intrinsic): Don't give caller a clue about
1132 form of intrinsic -- shouldn't be needed at this point.
1134 Cope with generic intrinsics that are subroutines and functions:
1135 * com.c (ffecom_finish_symbol_transform_, ffecom_expr_transform_):
1136 Don't transform an intrinsic that is not known to be a subroutine
1137 or a function. (Maybe someday have to avoid transforming
1138 any intrinsic with an undecided or unknown implementation.)
1139 * expr.c (ffeexpr_declare_unadorned_,
1140 ffeexpr_declare_parenthesized_): Ok to invoke generic
1141 intrinsic that has at least one subroutine form as a
1143 Ok to pass intrinsic as actual arg if it has a known specific
1144 intrinsic form that is valid as actual arg.
1145 (ffeexpr_declare_parenthesized_): An unknown kind of
1146 intrinsic has a paren_type chosen based on context.
1147 (ffeexpr_token_arguments_): Build funcref/subrref based
1148 on context, not on kind of procedure being called.
1149 * intrin.h, intrin.c (ffeintrin_is_intrinsic): Undo changes of
1150 Tue Feb 4 23:12:04 1997 by me, change all callers to leave
1151 intrinsics as FFEINFO_kindNONE at this point. (Some callers
1152 also had unused variables deleted as a result.)
1154 Enable all intrinsic groups (especially f90 and vxt):
1155 * target.h (FFETARGET_defaultSTATE_DCP, FFETARGET_defaultSTATE_F2C,
1156 FFETARGET_defaultSTATE_F90, FFETARGET_defaultSTATE_MIL,
1157 FFETARGET_defaultSTATE_UNIX, FFETARGET_defaultSTATE_VXT):
1158 Delete these macros, let top.c set them directly.
1159 * top.c (ffeintrinsic_state_dcp_, ffe_intrinsic_state_f2c_,
1160 ffe_intrinsic_state_f90_, ffe_intrinsic_state_mil_,
1161 ffe_intrinsic_state_unix_, ffe_intrinsic_state_vxt_):
1162 Enable all these directly.
1164 Sat Feb 8 03:21:50 1997 Craig Burley <burley@gnu.ai.mit.edu>
1166 * g77.c: Incorporate recent changes to ../gcc.c.
1167 For version magic (e.g. `g77 -v'), instead of compiling
1168 /dev/null, write, compile, run, and then delete a small
1169 program that prints the version numbers of the three
1170 components of libf2c (libF77, libI77, and libU77),
1171 so we get this info with bug reports.
1172 Also, this change reduces the chances of accidentally
1173 linking to an old (complex-alias-problem) libf2c.
1174 Fix `-L' so the argument is expected in `-Larg'.
1176 * com.h (FFECOM_f2cLONGINT): For INTEGER*8 support in f2c.h,
1177 dynamically determine proper type here, instead of
1178 assuming `long long int' is correct.
1180 Tue Feb 4 23:12:04 1997 Craig Burley <burley@gnu.ai.mit.edu>
1182 Add libU77 library from Dave Love <d.love@dl.ac.uk>:
1183 * Make-lang.in (f77-runtime): Depend on new Makefile.
1184 (f/runtime/libU77/Makefile): New rule.
1185 Also configure libU77.
1186 ($(srcdir)/f/runtime/configure: Use Makefile.in,
1187 so configuration doesn't have to have happened.
1188 (f77.mostlyclean, f77.clean, f77.distclean,
1189 f77.maintainer-clean): Some fixups here, but more work
1191 (RUNTIMESTAGESTUFF): Add libU77's config.status.
1192 (LIBU77STAGESTUFF, f77.stage1, f77.stage2, f77.stage3,
1193 f77.stage4): New macro, appropriate uses added.
1194 * com-rt.def: Add libU77 procedures.
1195 * com.c (ffecom_f2c_ptr_to_integer_type_node,
1196 ffecom_f2c_ptr_to_real_type_node): New type nodes.
1197 (FFECOM_rttypeCHARACTER_): New type of run-time function.
1198 (ffecom_char_args_): Handle CHARACTER*n intrinsics
1199 where n != 1 here, instead of in ffecom_expr_intrinsic_.
1200 (ffecom_expr_intrinsic_): New code to handle new
1202 In particular, change how FFEINTRIN_impFLUSH is handled.
1203 (ffecom_make_gfrt_): Handle new type of run-time function.
1204 (ffecom_init_0): Initialize new type nodes.
1205 * config-lang.in: New libU77 directory.
1206 * intrin.h, intrin.c (ffeintrin_is_intrinsic): Handle
1207 potential generic for subroutine _and_ function
1208 specifics via two new arguments. All callers changed.
1209 Properly ignore deleted/disabled intrinsics in resolving
1211 (ffeintrin_check_, ffeintrin_init_0): Handle CHARACTER intrinsics of (*)
1213 * intrin.def: Permission granted by FSF to place this in
1214 public domain, which will allow it to serve as source
1215 for both g77 program and its documentation.
1216 Add libU77 intrinsics.
1217 (FLUSH): Now a generic, not specific, intrinsic.
1218 (DEFIMP): Now support return modifier for CHARACTER intrinsics.
1220 * com-rt.def (FFECOM_gfrtDIM, FFECOM_gfrtERF,
1221 FFECOM_gfrtERFC, FFECOM_gfrtEXP, FFECOM_gfrtSIGN,
1222 FFECOM_gfrtSIN, FFECOM_gfrtSINH, FFECOM_gfrtTAN,
1223 FFECOM_gfrtTANH, FFECOM_gfrtPOW_RI): Change "&r" to "&f".
1225 Sat Feb 1 12:15:09 1997 Craig Burley <burley@gnu.ai.mit.edu>
1227 * Version 0.5.19.1 released.
1229 * com.c (ffecom_expr_, ffecom_expr_intrinsic_,
1230 ffecom_tree_divide_): FFECOM_gfrtPOW_ZI,
1231 FFECOM_gfrtCONJG, FFECOM_gfrtDCONJG,
1232 FFECOM_gfrtCCOS, FFECOM_gfrtCDCOS,
1233 FFECOM_gfrtCLOG, FFECOM_gfrtCDLOG,
1234 FFECOM_gfrtCSIN, FFECOM_gfrtCDSIN,
1235 FFECOM_gfrtCSQRT, FFECOM_gfrtCDSQRT,
1236 FFECOM_gfrtDIV_CC, FFECOM_gfrtDIV_ZZ: These all require
1237 result to _not_ overlap one or more inputs.
1239 Sat Feb 1 00:25:55 1997 Craig Burley <burley@gnu.ai.mit.edu>
1241 * com.c (ffecom_init_0): Do internal checks only if
1242 -fset-g77-defaults not specified.
1244 Fix %LOC(), LOC() to return sufficiently wide type:
1245 * com.h, com.c (ffecom_pointer_kind_, ffecom_label_kind_,
1246 ffecom_pointer_kind(), ffecom_label_kind()): New globals
1247 and accessor macros hold kind for integer pointers on target
1249 (ffecom_init_0): Determine narrowest INTEGER type that
1250 can hold a pointer (usually INTEGER*4 or INTEGER*8),
1251 store it in ffecom_pointer_kind_, etc.
1252 * expr.c (ffeexpr_cb_end_loc_): Use right type for %LOC().
1253 * intrin.c (ffeintrin_check_, ffeintrin_init_0): Support
1254 new 'p' kind for type of intrinsic.
1255 * intrin.def (FFEINTRIN_impLOC): Returns "Ip" instead of "I1",
1256 so LOC() type is correct for target machine.
1258 Support -fugly-assign:
1259 * lang-options.h, top.h, top.c (ffe_decode_option):
1260 Accept -fugly-assign and -fno-ugly-assign.
1261 * com.c (ffecom_expr_): Handle -fugly-assign.
1262 * expr.c (ffeexpr_finished_): Check right type for ASSIGN
1265 Fri Jan 31 14:30:00 1997 Craig Burley <burley@gnu.ai.mit.edu>
1267 Remove last vestiges of -fvxt-not-f90:
1268 * stb.c (ffestb_R10012_, ffestb_R10014_, ffestb_V0201_):
1271 Fri Jan 31 02:13:54 1997 Craig Burley <burley@gnu.ai.mit.edu>
1273 * top.c (ffe_decode_option): Warn if -fugly is specified,
1276 * symbol.h: No need to #include "bad.h".
1278 Reorganize features from -fvxt-not-f90 to -fvxt:
1279 * lang-options.h, top.h, top.c:
1280 Accept -fvxt and -fno-vxt, but not -fvxt-not-f90 or -ff90-not-vxt.
1281 Warn if the latter two are used.
1282 * expr.c (ffeexpr_nil_rhs_): Double-quote means octal constant.
1283 (ffeexpr_token_rhs_): Double-quote means octal constant.
1284 * target.h (FFETARGET_defaultIS_VXT_NOT_90): Delete macro
1285 definition, no longer needed.
1287 Make some -ff90 features the default:
1288 * data.c (ffedata_value): DATA implies SAVE.
1289 * src.h (ffesrc_is_name_noninit): Underscores always okay.
1291 Fix up some more #error directives by quoting their text:
1292 * bld.c (ffebld_constant_is_zero):
1295 Sat Jan 18 18:22:09 1997 Craig Burley <burley@gnu.ai.mit.edu>
1297 * g77.c (lookup_option, main): Recognize `-Xlinker',
1298 `-Wl,', `-l', `-L', `--library-directory', `-o',
1300 (lookup_option): Don't depend on SWITCH_TAKES_ARG
1301 being correct, it might or might not have `-x' in
1302 it depending on host.
1303 Return NULL argument if it would be an empty string.
1304 (main): If no input files (by gcc.c's definition)
1305 but `-o' or `--output' specified, produce diagnostic
1306 to avoid overwriting output via gcc.
1307 Recognize C++ `+e' options.
1308 Treat -L as another non-magical option (like -B).
1309 Don't append_arg `-x' twice.
1311 Fri Jan 10 23:36:00 1997 Craig Burley <burley@gnu.ai.mit.edu>
1313 * top.c [BUILT_FOR_270] (ffe_decode_option): Make
1314 -fargument-noalias-global the default.
1316 Fri Jan 10 07:42:27 1997 Craig Burley <burley@gnu.ai.mit.edu>
1318 Enable inlining of previously-compiled program units:
1319 * com.c (ffecom_do_entry_, ffecom_start_progunit_):
1320 Register new public function in ffeglobal database.
1321 (ffecom_sym_transform_): Any GLOBAL or potentially GLOBAL
1322 symbol should be looked up in ffeglobal database and
1323 that tree node used, if found. That way, gcc knows
1324 the references are to those earlier definitions, so it
1325 can emit shorter branches/calls, inline, etc.
1326 (ffecom_transform_common_): Minor change for clarity.
1327 * expr.c (ffeexpr_sym-lhs_call_, ffeexpr_sym_lhs_extfunc_,
1328 ffeexpr_sym_rhs_actualarg_, ffeexpr_paren_rhs_let_,
1329 ffeexpr_token_funsubstr_): Globalize symbol as needed.
1330 * global.c (ffeglobal_promoted): New function to look up
1331 existing local symbol in ffeglobal database.
1332 * global.h: Declare new function.
1333 * name.h (ffename_token): New macro, plus alphabetize.
1334 * stc.c (ffestc_R1207_item): Globalize EXTERNAL symbol.
1335 * stu.c (ffestu_sym_end_transition, ffestu_sym_exec_transition):
1336 Globalize symbol as needed.
1337 * symbol.h, symbol.c (ffesymbol_globalize): New function.
1339 Thu Jan 9 14:20:00 1997 Craig Burley <burley@gnu.ai.mit.edu>
1341 * ste.c (ffeste_R809): Produce a diagnostic for SELECT CASE
1342 on CHARACTER type, instead of crashing.
1344 Thu Jan 9 00:52:45 1997 Craig Burley <burley@gnu.ai.mit.edu>
1346 * stc.c (ffestc_order_entry_, ffestc_order_format_,
1347 ffestc_R1226): Allow ENTRY and FORMAT before IMPLICIT
1348 NONE, by having them transition only to state 1 instead
1349 of state 2 (which is disallowed by IMPLICIT NONE).
1351 Mon Jan 6 22:44:53 1997 Craig Burley <burley@gnu.ai.mit.edu>
1353 Fix AXP bug found by Rick Niles (961201-1.f):
1354 * com.c (ffecom_init_0): Undo my 1996-05-14 change, as
1355 it is incorrect and prevented easily finding this bug.
1356 * target.h [__alpha__] (ffetargetReal1, ffetargetReal2):
1357 Use int instead of long.
1358 (ffetarget_cvt_r1_to_rv_, ffetarget_cvt_rv_to_r1_,
1359 ffetarget_cvt_r2_to_rv_, ffetarget_cvt_rv_to_r2_):
1360 New functions that intercede for callers of
1361 REAL_VALUE_(TO|UNTO)_TARGET_(SINGLE|DOUBLE).
1362 All callers changed, and damaging casts to (long *) removed.
1364 Sun Jan 5 03:26:11 1997 Craig Burley <burley@gnu.ai.mit.edu>
1366 * Make-lang.in (g77, g77-cross): Depend on both g77.c and
1367 zzz.c, in $(srcdir)/f/.
1369 Better design for -fugly-assumed:
1370 * stc.c (ffestc_R501_item, ffestc_R524_item,
1371 ffestc_R547_item_object): Pass new is_ugly_assumed flag.
1372 * stt.c, stt.h (ffestt_dimlist_as_expr,
1373 ffestt_dimlist_type): New is_ugly_assumed flag now
1374 controls whether "1" is treated as "*".
1375 Don't treat "2-1" or other collapsed constants as "*".
1377 Sat Jan 4 15:26:22 1997 Craig Burley <burley@gnu.ai.mit.edu>
1379 * stb.c (ffestb_R10012_): Don't confirm on FORMAT(A,)
1380 or even FORMAT(A,,B), as R1229 only warns about the
1381 former currently, and this seems reasonable.
1383 Improvements to diagnostics:
1384 * sta.c (ffesta_second_): Don't add any ffestb parsers
1385 unless they're specifically called for.
1386 Set up ffesta_tokens[0] before calling ffestc_exec_transition,
1387 else stale info might get used.
1388 (ffesta_save_): Do a better job picking which parser to run
1389 after running all parsers with no confirmed possibles.
1390 (FFESTA_maxPOSSIBLES_): Decrease from 100 now that so few
1391 possibles are ever on the list at a given time.
1392 (struct _ffesta_possible): Add named attribute.
1393 (ffesta_add_possible_exec_, ffesta_add_possible_nonexec_):
1394 Make these into macros that call a single function that now
1395 sets the named attribute.
1396 (ffesta_add_possible_unnamed_exec_,
1397 ffeseta_add_possible_unnamed_nonexec_): New macros.
1398 (ffesta_second_): Designate unnamed possibles as
1400 * stb.c (ffestb_R1229, ffestb_R12291_): Use more general
1401 diagnostic, so things like "POINTER (FOO, BAR)" are
1402 diagnosed as unrecognized statements, not invalid statement
1404 * stb.h, stb.c (ffestb_unimplemented): Remove function.
1406 1996-12-30 Dave Love <d.love@dl.ac.uk>
1408 * com.c: #include libU77/config.h
1409 (ffecom_f2c_ptr_to_integer_type_node,
1410 ffecom_f2c_ptr_to_integer_type_node): New variables.
1411 (ffecom_init_0): Use them.
1412 (ffecom_expr_intrinsic_): Many news cases for libU77 intrinsics.
1414 * com-rt.def: New definitions for libU77.
1415 * intrin.def: Likewise. Also correct ftell arg spec.
1417 * Makefile.in (f/runtime/libU77/config.h): New target for com.c
1419 * Make-lang.in (f771): Depend on f/runtime/Makefile for the above.
1421 Sat Dec 28 12:28:29 1996 Craig Burley <burley@gnu.ai.mit.edu>
1423 * stt.c (ffestt_dimlist_type): Treat ([...,]1) in dimlist
1424 as ([...,]*) if -fugly-assumed, so assumed-size array
1425 detected early enough.
1427 Thu Dec 19 14:01:57 1996 Craig Burley <burley@gnu.ai.mit.edu>
1429 * target.h (FFETARGET_REAL_VALUE_FROM_INT_): Conditionalize
1430 definition on BUILT_FOR_280, not BUILT_WITH_280, since
1431 the name of the macro was (properly) changed since 0.5.19.
1433 Fix warnings/errors resulting from ffetargetOffset becoming
1434 `long long int' instead of `unsigned long' as of 0.5.19,
1435 while ffebitCount remains `unsigned long':
1436 * bld.c (ffebld_constantarray_dump): Avoid warnings by
1437 using loop var of appropriate type, and using casts.
1438 * com.c (ffecom_expr_): Use right type for loop var.
1439 (ffecom_sym_transform_, ffecom_transform_equiv_):
1440 Cast to right type in assertions.
1441 * data.c (ffedata_gather_, ffedata_value_): Cast to right
1442 type in assertions and comparisons.
1444 Wed Dec 18 12:07:11 1996 Craig Burley <burley@gnu.ai.mit.edu>
1446 Patch from Alexandre Oliva <oliva@dcc.unicamp.br>:
1447 * Makefile.in (all.indirect): Don't pass -bbigtoc option
1450 Cope with new versions of gcc:
1451 * com.h (BUILT_FOR_280): New macro.
1452 * com.c (ffecom_ptr_to_expr): Conditionalize test of
1454 (ffecom_build_complex_constant_): Conditionalize calling
1455 sequence for build_complex.
1457 Sat Dec 7 07:15:17 1996 Craig Burley <burley@gnu.ai.mit.edu>
1459 * Version 0.5.19 released.
1461 Fri Dec 6 12:23:55 1996 Craig Burley <burley@gnu.ai.mit.edu>
1463 * g77.c: Default to assuming "f77" is in $LANGUAGES, since
1464 the LANGUAGE_F77 macro isn't defined by anyone anymore (but
1465 might as well leave the no-f77 code in just in case).
1466 * Make-lang.in (g77, g77-cross): Don't define LANGUAGE_F77
1469 1996-12-06 Dave Love <d.love@dl.ac.uk>
1471 * Make-lang.in (g77, g77-cross): Revert to building `g77' or not
1472 conditional on `f77' in LANGUAGES.
1474 Wed Dec 4 13:08:44 1996 Craig Burley <burley@gnu.ai.mit.edu>
1476 * Make-lang.in (g77, g77-cross): No libs or lib dependencies
1477 in case where "f77" is not in $LANGUAGES.
1479 * lex.c (ffelex_image_char_, ffelex_file_fixed,
1480 ffelex_file_free): Fixes to properly handle lines with
1481 null character, and too-long lines as well.
1483 * lex.c: Call ffebad_start_msg_lex instead of
1484 ffebad_start_msg throughout.
1486 Sun Dec 1 21:19:55 1996 Craig Burley <burley@gnu.ai.mit.edu>
1488 Fix-up for 1996-11-25 changes:
1489 * com.c (ffecom_member_phase2_): Subtract out 0 offset for
1490 elegance and consistency with EQUIVALENCE aggregates.
1491 (ffecom_sym_transform_): Ditto for LOCAL/COMMON, and
1492 ensure we get the same parent storage area.
1493 * data.c (ffedata_gather_, ffedata_value_): Subtract out
1496 Wed Nov 27 13:55:57 1996 Craig Burley <burley@gnu.ai.mit.edu>
1498 * proj.h: Quote the text of the #error message, to avoid
1499 strange-looking diagnostics from non-gcc ANSI compilers.
1501 * top.c: Make -fno-debug-kludge the default.
1503 Mon Nov 25 20:13:45 1996 Craig Burley <burley@gnu.ai.mit.edu>
1505 Provide more info on EQUIVALENCE mismatches:
1506 * bad.def (FFEBAD_EQUIV_MISMATCH): More detailed message.
1507 * equiv.c (ffeequiv_layout_local_, ffeequiv_layout_cblock):
1508 More details for FFEBAD_EQUIV_MISMATCH.
1510 Fix problem with EQUIVALENCE handling:
1511 * equiv.c (ffeequiv_layout_local_): Redesign algorithm --
1512 old one was broken, resulting in rejection of good code.
1513 (ffeequiv_offset_): Add argument, change callers.
1514 Clean up the code, fix up the (probably unused) negative-value
1516 * com.c (ffecom_sym_transform_): For local EQUIVALENCE
1517 member, subtract out aggregate offset (which is <= 0).
1519 Thu Nov 21 12:44:56 1996 Craig Burley <burley@gnu.ai.mit.edu>
1521 Change type of ffetargetOffset from `unsigned long' to `long long':
1522 * bld.c (ffebld_constantarray_dump): Change printf formats.
1523 * storag.c (ffestorag_dump): Ditto.
1524 * symbol.c (ffesymbol_report): Ditto.
1525 * target.h (ffetargetOffset_f): Ditto and change type itself.
1527 Handle situation where list of languages does not include f77:
1528 * Make-lang.in: Define LANGUAGE_F77 to 1 only if `f77' is in
1529 the $LANGUAGES macro for the build.
1530 * g77.c: Compile to a (nearly) no-op program if LANGUAGE_F77
1531 is not defined to 1.
1533 Fixes to delay confirmation of READ, WRITE, and GOTO statements
1534 so the corresponding assignments to same-named CHAR*(*) arrays
1536 * stb.c (ffestb_R90915_, ffestb_91014_): New functions.
1537 (ffestb_goto3_, ffestb_goto5_): Move confirmation from 3 to 5
1538 for the OPEN_PAREN case.
1539 (ffestb_R9091_, ffestb_R9094_, ffestb_R90913_, ffestb_R90914_,
1540 ffestb_R91012_, ffestb_R91013_): Use new functions, and confirm
1541 except for the OPEN_PAREN case.
1543 Fixes to not confirm declarations with an open paren where
1544 an equal sign or other assignment-like token might be, so the
1545 corresponding assignments to same-named CHAR*(*) arrays work:
1546 (ffestb_decl_entsp_5_): Move assertion so we crash on that first,
1547 if it turns out to be wrong, before the less-debuggable crash
1548 on mistaken confirmation.
1549 (ffestb_decl_entsp_6_, ffestb_decl_entsp_7_, ffestb_decl_entsp_8_):
1550 Include OPEN_PAREN in list of assignment-only tokens.
1552 Fix more diagnosed-crash bugs:
1553 * stu.c (ffestu_sym_end_transition): ANY-ize an adjustable array
1554 with bad dimension expressions even if still stateUNCERTAIN.
1555 (ffestu_symter_end_transition_, ffestu_symter_exec_transition_):
1556 Return TRUE for opANY as well.
1557 For code elegance, move opSYMTER case into first switch.
1559 1996-11-17 Dave Love <d.love@dl.ac.uk>
1561 * lex.c: Fix last change.
1563 1996-11-14 Dave Love <d.love@dl.ac.uk>
1565 * Make-lang.in, config-lang.in: Remove the (broken) libU77 stuff,
1568 Thu Nov 14 15:40:59 1996 Craig Burley <burley@gnu.ai.mit.edu>
1570 * bad.def (FFEBAD_UNIMPL_STMT): Explain that invalid
1571 intrinsic references can trigger this message, too.
1573 1996-11-12 Dave Love <d.love@dl.ac.uk>
1575 * lex.c: Declare dwarfout routines.
1577 * config-lang.in: Sink grep o/p.
1579 Mon Nov 11 14:21:13 1996 Craig Burley <burley@gnu.ai.mit.edu>
1581 * g77.c (main): Might as well print version number
1582 for --verbose as well.
1584 Thu Nov 7 18:41:41 1996 Craig Burley <burley@gnu.ai.mit.edu>
1586 * expr.c, lang-options.h, target.h, top.c, top.h: Split out
1587 remaining -fugly stuff into -fugly-logint and -fugly-comma,
1588 leaving -fugly as simply a `macro' that expands into other
1589 options, and eliminate defaults for some of the ugly stuff
1592 * Make-lang.in (gcc-cross): Compile zzz.c, not version.o (!),
1593 in to get version info for this target.
1595 * config-lang.in: Test for GBE patch application based
1596 on whether 2.6.x or 2.7.x GBE is detected.
1598 Wed Nov 6 14:19:45 1996 Craig Burley <burley@gnu.ai.mit.edu>
1600 * Make-lang.in (g77): Compile zzz.c in to get version info.
1601 * g77.c: Add support for --help and --version.
1603 * g77.c (lookup_option): Short-circuit long-winded tests
1604 when second char is not hyphen, just to save a spot of time.
1606 Sat Nov 2 13:50:31 1996 Craig Burley <burley@gnu.ai.mit.edu>
1608 * intrin.def: Add FTELL and FSEEK intrinsics, plus new
1609 `g' codes for alternate-return (GOTO) arguments.
1610 * intrin.c (ffeintrin_check_): Support `g' codes.
1611 * com-rt.def: Add ftell_() and fseek_() to database.
1612 * com.c (ffecom_expr_intrinsic_): Ditto. Also, let each
1613 subroutine intrinsic decide for itself what to do with
1614 tree_type, the default being NULL_TREE once again (so
1615 ffecom_call_ doesn't think it's supposed to cast the
1616 function call to the type in the fall-through case).
1618 * ste.c (ffeste_R909_finish): Don't special-case list-directed
1619 I/O, now that libf2c can return non-zero status codes.
1620 (ffeste_R910_finish): Ditto.
1621 (ffeste_io_call_): Simplify logic.
1623 (ffeste_subr_beru_):
1626 (ffeste_R909_start):
1628 (ffeste_R909_finish):
1629 (ffeste_R910_start):
1631 (ffeste_R910_finish):
1632 (ffeste_R911_start):
1633 (ffeste_R923A): Ditto all the above.
1635 Thu Oct 31 20:56:28 1996 Craig Burley <burley@gnu.ai.mit.edu>
1637 * config-lang.in, Make-lang.in: Rename flag file
1638 build-u77 to build-libu77, for consistency with
1639 install-libf2c and such.
1641 * config-lang.in: Don't complain about failure to patch
1642 if pre-2.7.0 gcc is involved (since our patch for that
1643 doesn't add support for tooning).
1645 Sat Oct 26 05:56:51 1996 Craig Burley <burley@gnu.ai.mit.edu>
1647 * bad.def (FFEBAD_TYPELESS_TOO_LARGE): Remove this
1648 unused and redundant diagnostic.
1650 Sat Oct 26 00:45:42 1996 Craig Burley <burley@gnu.ai.mit.edu>
1652 * target.c (ffetarget_integerhex): Fix dumb bug.
1654 1996-10-20 Dave Love <d.love@dl.ac.uk>
1656 * gbe/2.7.2.1.diff: New file.
1658 * Makefile.in (F771_LDFLAGS): Add -bbigtoc for AIX4.1 up, suggested by
1659 endo@material.tohoku.ac.jp [among others!].
1661 Sat Oct 19 03:11:14 1996 Craig Burley <burley@gnu.ai.mit.edu>
1663 * bad.def, bld.c, bld.h, expr.c, lang-options.h, target.c,
1664 target.h, top.c, top.h (ffebld_constant_new_integerbinary,
1665 ffebld_constant_new_integerhex, ffebld_constant_new_integeroctal,
1666 ffeexpr_token_name_apos_name_, ffetarget_integerbinary,
1667 ffetarget_integerhex, ffetarget_integeroctal): Support
1668 new -fno-typeless-boz option with new functions, mods to
1669 existing octal-handling functions, new macros, new error
1670 messages, and so on.
1672 * com.c, lang-options.h, top.c, top.h (ffecom_notify_primary_entry):
1673 Print program unit name on stderr if -fno-silent (new option).
1675 * lang-options.h, top.c, top.h, stt.c (ffestt_dimlist_as_expr):
1676 Treat ([...,]1) in dimlist as ([...,]*) if -fugly-assumed
1679 * lang-options.h: Comment out options duplicated in gcc/toplev.c,
1680 because, somehow, having them commented in and building on my
1681 DEC Alpha results in a cc1 that always segfaults, and gdb that
1682 also segfaults whenever it debugs it up to init_lex() calling
1685 Thu Oct 17 00:39:27 1996 Craig Burley <burley@gnu.ai.mit.edu>
1687 * stb.c (ffestb_R10013_): Don't change meaning of .sign until
1688 after previous meaning/value used to set sign of value
1691 Sun Oct 13 22:15:23 1996 Craig Burley <burley@gnu.ai.mit.edu>
1693 * top.c (ffe_decode_option): Don't set back-end flags
1694 that are nonexistent prior to gcc 2.7.0.
1696 Sun Oct 13 12:48:45 1996 Craig Burley <burley@gnu.ai.mit.edu>
1698 * com.c (convert): Don't convert emulated complex expr to
1699 real (via REALPART_EXPR) if the target type is (emulated)
1702 Wed Oct 2 21:57:12 1996 Craig Burley <burley@gnu.ai.mit.edu>
1704 * com.c (ffecom_debug_kludge_): Set DECL_IN_SYSTEM_HEADER so
1705 -Wunused doesn't complain about these manufactured decls.
1706 (ffecom_expr_): Ditto, for original (non-ASSIGN'ed) variable.
1707 (ffecom_transform_equiv_): Clear DECL_IGNORED_P for aggregate
1708 area so it shows up as a debug-accessible symbol.
1709 (pushdecl): Default for "invented" identifiers (a g77-specific
1710 concept for now) is that they are artificial, in system header,
1711 ignored for debugging purposes, used, and (for types) suppressed.
1712 This ought to be overkill.
1714 Fri Sep 27 23:13:07 1996 Craig Burley <burley@gnu.ai.mit.edu>
1716 * ste.c (ffeste_begin_iterdo_, ffeste_end_iterdo_): Support
1717 one-trip DO loops (F66-style).
1718 * lang-options.h, top.c, top.h (-fonetrip): New option.
1720 Thu Sep 26 00:18:40 1996 Craig Burley <burley@gnu.ai.mit.edu>
1722 * com.c (ffecom_debug_kludge_): New function.
1723 (ffecom_sym_transform_): Use new function for COMMON and EQUIVALENCE
1726 * lang-options.h, top.c, top.h (-fno-debug-kludge):
1729 1996-09-24 Dave Love <d.love@dl.ac.uk>
1731 * Make-lang.in (include/f2c.h):
1732 Remove dependencies on xmake_file and tmake_file.
1733 They expand inconsistently in 2.8 c.f. 2.7; $(GCC_PARTS) depends on
1736 1996-09-22 Dave Love <d.love@dl.ac.uk>
1738 * config-lang.in: Add --enable-libu77 option handling.
1741 Conditionally add --enable-libu77 when running runtime configure.
1742 Define LIBU77STAGESTUFF and use it in relevant rules.
1744 1996-08-21 Dave Love <d.love@dl.ac.uk>
1746 * Make-lang.in (f77-runtime):
1747 `stmp-hdrs' should have been `stmp-headers'.
1749 1996-08-20 Dave Love <d.love@dl.ac.uk>
1751 * Make-lang.in (f77-runtime):
1752 Depend on stmp-hdrs, not stmp-int-hdrs, since libF77
1755 Sat Jun 22 18:17:11 1996 Craig Burley <burley@gnu.ai.mit.edu>
1757 * com.c (ffecom_tree_divide_): Fix RECORD_TYPE case to
1758 look at type of first field, properly, to determine
1759 whether to call c_div or z_div.
1761 Tue Jun 4 04:27:18 1996 Craig Burley <burley@gnu.ai.mit.edu>
1763 * com.c (ffecom_build_complex_constant_): Explicitly specify
1765 (ffecom_expr_): Fix thinko.
1766 (ffecom_2): For COMPLEX_EXPR, explicitly specify TREE_PURPOSE.
1768 Mon May 27 16:23:43 1996 Craig Burley <burley@gnu.ai.mit.edu>
1770 Changes to optionally avoid gcc's back-end complex support:
1771 * com.c (ffecom_stabilize_aggregate_): New function.
1772 (ffecom_convert_to_complex_): New function.
1773 (ffecom_make_complex_type_): New function.
1774 (ffecom_build_complex_constant_): New function.
1775 (ffecom_expr_): For opCONVERT of non-COMPLEX to COMPLEX,
1776 don't bother explicitly converting to the subtype first,
1777 because gcc does that anyway, and more code would have
1778 to be added to find the subtype for the emulated-complex
1780 (ffecom_f2c_make_type_): Use ffecom_make_complex_type_
1781 instead of make_node etc. to make a complex type.
1782 (ffecom_1, ffecom_2): Translate operations on COMPLEX operands
1783 to appropriate operations when emulating complex.
1784 (ffecom_constantunion): Use ffecom_build_complex_constant_
1785 instead of build_complex to build a complex constant.
1786 (ffecom_init_0): Change point at which types are laid out
1787 for improved consistency.
1788 Use ffecom_make_complex_type_ instead of make_node etc.
1789 to make a complex type.
1790 Always calculate storage sizes from TYPE_SIZE, never TYPE_PRECISION.
1791 (convert): Use e, not expr, since we've copied into that anyway.
1792 For RECORD_TYPE cases, do emulated-complex conversions.
1793 (ffecom_f2c_set_lio_code_): Always calculate storage sizes
1794 from TYPE_SIZE, never TYPE_PRECISION.
1795 (ffecom_tree_divide_): Allow RECORD_TYPE to also be handled
1796 by run-time library.
1797 (ffecom_expr_intrinsic_): Handle possible RECORD_TYPE as argument
1800 * top.h, top.c, lang-options.h: Support new -f(no-)emulate-complex option.
1802 * com.c (ffecom_sym_transform_): Clarify and fix typos in comments.
1804 Mon May 20 02:06:27 1996 Craig Burley <burley@gnu.ai.mit.edu>
1806 * target.h: Use new REAL_VALUE_UNTO_TARGET_* macros instead
1807 of REAL_VALUE_FROM_TARGET_DOUBLE and _SINGLE.
1808 Explicitly use long instead of HOST_WIDE_INT for emulation
1809 of ffetargetReal1 and ffetargetReal2.
1811 1996-05-20 Dave Love <d.love@dl.ac.uk>
1814 Test for patch being applied with flag_move_all_movables in toplev.c.
1816 * install.texi (Patching GNU Fortran):
1817 Mention overriding X_CFLAGS rather than
1818 editing proj.h on SunOS4.
1820 * Make-lang.in (F77_FLAGS_TO_PASS):
1821 Add X_CFLAGS (convenient for SunOS4 kluge, in
1823 (f77.{,mostly,dist}clean): Reorder things, in particular not to delete
1824 Makefiles too early.
1826 * g77.c (DEFAULT_SWITCH_TAKES_ARG): Define a la gcc.c in the
1827 current GCC snapshot.
1829 Tue May 14 00:24:07 1996 Craig Burley <burley@gnu.ai.mit.edu>
1831 Changes for DEC Alpha AXP support:
1832 * com.c (ffecom_init_0): REAL_ARITHMETIC means internal
1833 REAL/DOUBLE PRECISION might well have a different size
1834 than the compiled type, so don't crash if this is the
1836 * target.h: Use `int' for ffetargetInteger1,
1837 ffetargetLogical1, and magical tests. Set _f format
1838 strings accordingly.
1840 Tue Apr 16 14:08:28 1996 Craig Burley <burley@gnu.ai.mit.edu>
1842 * top.c (ffe_decode_option): -Wall no longer implies
1845 Sat Apr 13 14:50:06 1996 Craig Burley <burley@gnu.ai.mit.edu>
1847 * com.c (ffecom_char_args_): If item is error_mark_node,
1848 set *length that way, too.
1850 * com.c (ffecom_expr_power_integer_): If either operand
1851 is error_mark_node, return that.
1853 * com.c (ffecom_intrinsic_len_): If item is error_mark_node,
1854 return that for length.
1856 * expr.c (ffeexpr_declare_unadorned_,
1857 ffeexpr_declare_parenthesized_): Instead of crashing
1858 on unexpected contexts, produce a diagnostic.
1860 * intrin.c (ffeintrin_check_), intrin.def (impSIGNAL):
1861 Allow procedure as second arg to SIGNAL intrinsic.
1863 * stu.c (ffestu_symter_end_transition_): New function.
1864 (ffestu_symter_exec_transition_): Return bool arg.
1865 Always transition symbol (don't inhibit when !whereNONE).
1866 (ffestu_sym_end_transition): If DUMMY/LOCAL arg has any
1867 opANY exprs in its dimlist, diagnose it so it doesn't
1868 make it through to later stages that try to deal with
1870 (ffestu_sym_exec_transition): If sym has any opANY exprs
1871 in its dimlist, diagnose it so it becomes opANY itself.
1873 * symbol.c (ffesymbol_error): If token arg is NULL,
1874 just ANY-ize the symbol -- don't produce diagnostic.
1876 Mon Apr 1 10:14:02 1996 Craig Burley <burley@gnu.ai.mit.edu>
1878 * Version 0.5.18 released.
1880 Mon Mar 25 20:52:24 1996 Craig Burley <burley@gnu.ai.mit.edu>
1882 * com.c (ffecom_expr_power_integer_): Don't generate code
1883 that compares COMPLEX (or, as it happens, REAL) via "LT_EXPR",
1884 since the back end crashes on that. (This code would never
1885 be executed anyway, but the test that avoids it has now been
1886 translated to control whether the code gets generated at all.)
1889 * com.c (ffecom_type_localvar_): Handle variable-sized
1890 dimension bounds expressions here, so they get calculated
1891 and saved on procedure entry. Fixes 960323-4.f.
1893 * com.c (ffecom_notify_init_symbol): Symbol has no init
1894 info at all if only zeros have been used to initialize it.
1897 * expr.c, expr.h (ffeexpr_type_combine): Renamed from
1898 ffeexpr_type_combine_ and now a public procedure; last arg now
1899 a token, instead of an internal structure used to extract a token.
1900 Now allows the outputs to be aliased with the inputs.
1901 Now allows a NULL token to mean "don't report error".
1902 (ffeexpr_reduced_bool2_, ffeexpr_reduced_eqop2_,
1903 ffeexpr_reduced_math2_, ffeexpr_reduced_power_,
1904 ffeexpr_reduced_relop2_): Handle new calling sequence for
1905 ffeexpr_type_combine.
1906 * (ffeexpr_convert): Don't put an opCONVERT node
1907 in just because the size is unknown; all downstream code
1908 should be able to deal without it being there anyway, and
1909 getting rid of it allows new intrinsic code to more easily
1910 combine types and such without generating bad code.
1911 * info.c, info.h (ffeinfo_kindtype_max): Rewrite to do
1912 proper comparison of size of types, not just comparison
1913 of their internal kind numbers (so I2.eq.I1 doesn't promote
1914 I1 to I2, rather the other way around).
1915 * intrin.c (ffeintrin_check_): Combine types of arguments
1916 in COL a la expression handling, for greater flexibility
1917 and permissiveness (though, someday, -fpedantic should
1918 report use of this kind of thing).
1919 Make sure Hollerith/typeless where CHARACTER expected is
1920 rejected. This all fixes 960323-2.f.
1922 * ste.c (ffeste_begin_iterdo_): Fix some more type conversions
1923 so INTEGER*2-laden DO loops don't crash at compile time on
1924 certain machines. Believed to fix 960323-1.f.
1926 * stu.c (ffestu_sym_end_transition): Certainly reject
1927 whereDUMMY not in any dummy list, whether stateUNCERTAIN
1928 or stateUNDERSTOOD. Fixes 960323-0.f.
1930 Tue Mar 19 13:12:40 1996 Craig Burley <burley@gnu.ai.mit.edu>
1932 * data.c (ffedata_value): Fix crash on opANY, and simplify
1933 the code at the same time.
1935 * Make-lang.in (f77-runtime): Also depends on lib[FI]77/Makefile...
1936 (include/f2c.h...): ...which in turn depend on */Makefile.in.
1937 (f77.rebuilt): Rebuild runtime stuff too.
1939 * intrin.c (ffeintrin_check_): Accommodate TYPELESS/HOLLERITH
1940 types, convert args as necessary, etc.
1942 * expr.c (ffeexpr_convert): Fix test for TYPELESS/HOLLERITH
1943 to obey the docs; crash if no source token when error.
1944 (ffeexpr_collapse_convert): Crash if no token when error.
1946 Mon Mar 18 15:51:30 1996 Craig Burley <burley@gnu.ai.mit.edu>
1948 * com.c (ffecom_init_zero_): Renamed from
1949 ffecom_init_local_zero_; now handles top-level
1950 (COMMON) initializations too.
1952 * bld.c (ffebld_constant_is_zero):
1953 * com.c (ffecom_symbol_transform_, ffecom_sym_transform_assign_,
1954 ffecom_transform_common_, ffecom_transform_equiv_):
1967 * top.h: All of this is mostly housekeeping-type changes
1968 to support -f(no-)zeros, i.e. not always stuff zero
1969 values into the initializer fields of symbol/storage objects,
1970 but still track that they have been given initial values.
1972 * bad.def: Fix wording for DATA-related diagnostics.
1974 * com.c (ffecom_sym_transform_assign_): Don't check
1975 any EQUIVALENCE stuff for local ASSIGN, the check was
1976 bad (crashing), and it's not necessary, anyway.
1978 * com.c (ffecom_expr_intrinsic_): For MAX and MIN,
1979 ignore null arguments as far arg[123], and fix handling
1980 of ANY arguments. (New intrinsic support now allows
1981 spurious trailing null arguments.)
1983 * com.c (ffecom_init_0): Add HOLLERITH (unsigned)
1984 equivalents for INTEGER*2, *4, and *8, so shift intrinsics
1985 and other things that need unsigned versions of signed
1988 Sat Mar 16 12:11:40 1996 Craig Burley <burley@gnu.ai.mit.edu>
1990 * storag.c (ffestorag_exec_layout): Treat adjustable
1991 local array like dummy -- don't create storage object.
1992 * com.c (ffecom_sym_transform_): Allow for NULL storage
1993 object in LOCAL case (adjustable array).
1995 Fri Mar 15 13:09:41 1996 Craig Burley <burley@gnu.ai.mit.edu>
1997 * com.c (ffecom_sym_transform_): Allow local symbols
1998 with nonconstant sizes (adjustable local arrays).
1999 (ffecom_type_localvar_): Allow dimensions with nonconstant
2000 component (adjustable local arrays).
2001 * expr.c: Various minor changes to handle adjustable
2002 local arrays (a new case of stateUNCERTAIN).
2003 * stu.c (ffestu_sym_end_transition,
2004 ffestu_sym_exec_transition): Ditto.
2005 * symbol.def: Update docs to reflect these changes.
2007 * com.c (ffecom_expr_): Reduce space/time needed for
2008 opACCTER case by handling it here instead of converting
2009 it to opARRTER earlier on.
2010 (ffecom_notify_init_storage): Don't convert ACCTER to ARRTER.
2011 (ffecom_notify_init_symbol): Ditto.
2013 * com.c (ffecom_init_0): Crash and burn if any of the types'
2014 sizes, according to the GBE, disagrees with the sizes of
2015 the FFE's internal implementation. This might catch
2016 Alpha/SGI bugs earlier.
2018 Fri Mar 15 01:09:41 1996 Craig Burley <burley@gnu.ai.mit.edu>
2020 * com-rt.def, com.c, com.h: Changes for rewrite of intrinsic
2022 * com.c (ffecom_arglist_expr_): New function.
2023 (ffecom_widest_expr_type_): New function.
2024 (ffecom_expr_intrinsic_): Reorganize, some rewriting.
2025 (ffecom_f2c_make_type_): Layout complex types.
2026 (ffecom_gfrt_args_): New function.
2027 (ffecom_list_expr): Trivial change for consistency.
2029 * expr.c (ffeexpr_token_name_rhs_): Go back to getting
2030 type from specific, not implementation, info.
2031 (ffeexpr_token_funsubstr_): Set intrinsic implementation too!
2032 * intrin.c: Major rewrite of most portions.
2033 * intrin.def: Major rearchitecting of tables.
2034 * intrin.h (ffeintrin_basictype, ffeintrin_kindtype):
2035 Now (once again) take ffeintrinSpec as arg, not ffeintrinImp;
2036 for now, these return NONE, since they're not really needed
2037 and adding the necessary info to the tables is not trivial.
2038 (ffeintrin_codegen_imp): New function.
2039 * stc.c (ffestc_R1208_item): Change way ffeintrin funcs called,
2040 back to original per above; but comment out the code anyway.
2042 * intrin.c (ffe_init_0): Do internal checks only if
2043 -fset-g77-defaults not specified.
2045 * lang-options.h: Add -fset-g77-defaults option.
2046 * lang-specs.h: Always pass -fset-g77-defaults.
2047 * top.c, top.h: New option.
2049 Sat Mar 9 17:49:50 1996 Craig Burley <burley@gnu.ai.mit.edu>
2051 * Make-lang.in (stmp-int-hdrs): Use --no-validate when
2052 generating the f77.rebuilt files (BUGS, INSTALL, NEWS)
2053 so cross-references can work properly in g77.info
2054 without a lot of hassle. Users can probably deal with
2055 the way they end up looking in the f77.rebuilt files.
2057 * bld.c (ffebld_constant_new_integer4_val): INTEGER*8
2058 support -- new function.
2059 (ffebld_constant_new_logical4_val): New function.
2060 * com.c (ffecom_f2c_longint_type_node): New type.
2061 (FFECOM_rttypeLONGINT_): New return type code.
2062 (ffecom_expr_): Add code to invoke pow_qq instead
2063 of pow_ii for INTEGER4 (INTEGER*8) case.
2064 If ffecom_expr_power_integer_ returns NULL_TREE, just do
2066 (ffecom_make_gfrt_): Handle new type.
2067 (ffecom_expr_power_integer_): Let caller do the work if in
2068 dummy-transforming case, since
2069 caller now knows about INTEGER*8 and such, by returning
2071 * expr.c (ffeexpr_reduced_power_): Complain about non-INTEGER
2072 raised to INTEGER4 (INTEGER*8) power.
2074 * target.c (ffetarget_power_integerdefault_integerdefault):
2076 * com.c (ffecom_expr_power_integer_): Fix (-1)**(-8) and similar
2077 to ABS() the integral result if the exponent is negative
2080 * ste.c (ffeste_begin_iterdo_): Clean up a type ref.
2081 Always convert iteration count to _default_ INTEGER.
2083 * sta.c (ffesta_second_): Add BYTE and WORD type/stmts;
2084 changes by Scott Snyder <snyder@d0sgif.fnal.gov>.
2085 * stb.c (ffestb_decl_recursive): Ditto.
2086 (ffestb_decl_recursive): Ditto.
2087 (ffestb_decl_entsp_2_): Ditto.
2088 (ffestb_decl_entsp_3_): Ditto.
2089 (ffestb_decl_funcname_2_): Ditto.
2090 (ffestb_decl_R539): Ditto.
2091 (ffestb_decl_R5395_): Ditto.
2092 * stc.c (ffestc_establish_declstmt_): Ditto.
2093 * std.c (ffestd_R539item): Ditto.
2094 (ffestd_R1219): Ditto.
2096 * str-1t.fin: Ditto.
2097 * str-2t.fin: Ditto.
2099 * expr.c (ffeexpr_finished_): For DO loops, allow
2100 any INTEGER type; convert LOGICAL (assuming -fugly)
2101 to corresponding INTEGER type instead of always default
2102 INTEGER; let later phases do conversion of DO start,
2103 end, incr vars for implied-DO; change checks for non-integral
2104 DO vars to be -Wsurprising warnings.
2105 * ste.c (ffeste_io_impdo_): Convert start, end, and incr
2106 to type of DO variable.
2108 * com.c (ffecom_init_0): Add new types for [IL][234],
2109 much of which was done by Scott Snyder <snyder@d0sgif.fnal.gov>.
2113 Wed Mar 6 14:08:45 1996 Craig Burley <burley@gnu.ai.mit.edu>
2115 * top.c (ffe_init_gbe_): Make -frerun-loop-opt the default.
2117 Mon Mar 4 12:27:00 1996 Craig Burley <burley@gnu.ai.mit.edu>
2119 * expr.c (ffeexpr_exprstack_push_unary_): Really warn only
2120 about two successive _arithmetic_ operators.
2122 * stc.c (ffestc_R522item_object): Allow SAVE of (understood)
2125 * top.c (ffe_decode_option): New -f(no-)second-underscore options.
2126 * top.h: New options.
2127 * com.c (ffecom_get_external_identifier_, ffecom_get_identifier_):
2130 * Make-lang.in (f77.maintainer-clean): Clean f/BUGS, f/INSTALL,
2132 ($(srcdir)/f/BUGS, $(srcdir)/f/INSTALL, $(srcdir)/f/NEWS):
2134 ($(srcdir)/f/g77.info, $(srcdir)/f/g77.dvi): Depend on
2135 f/bugs.texi and f/news.texi.
2136 (f77.install-man): Install f77 man pages (if enabled).
2137 (f77.uninstall): Uninstall info docs, f77 man pages (if enabled).
2139 * top.c (ffe_init_gbe_): New function.
2140 (ffe_decode_option, ffe_file): Call ffe_init_gbe_ to
2141 set defaults for gcc options.
2143 Sat Jan 20 13:57:19 1996 Craig Burley <burley@gnu.ai.mit.edu>
2145 * com.c (ffecom_get_identifier_): Eliminate needless
2146 comparison of results of strchr.
2148 Tue Dec 26 11:41:56 1995 Craig Burley <burley@gnu.ai.mit.edu>
2150 * Make-lang.in: Add rules for new files g77.texi, g77.info,
2152 Reorganize the *clean rules to more closely parallel gcc's.
2154 * config-lang.in: Exclude g77.info from diffs.
2156 Sun Dec 10 02:29:13 1995 Craig Burley <burley@gnu.ai.mit.edu>
2158 * expr.c (ffeexpr_declare_unadorned_,
2159 ffeexpr_declare_parenthesized_): Break out handling of
2160 contextDATAIMPDO[INDEX,CTRL] so it's independent of symbol state.
2161 Don't exec-transition these here (let ffeexpr_sym_impdoitem_
2162 handle that when appropriate). Don't "declare" them twice.
2164 Tue Dec 5 06:48:26 1995 Craig Burley <burley@gnu.ai.mit.edu>
2166 * stc.c (ffestc_promote_sfdummy_): Allow whereNONE parent
2167 symbol, since it is not necessarily known whether it will
2168 become LOCAL or DUMMY.
2170 Mon Dec 4 03:46:55 1995 Craig Burley <burley@gnu.ai.mit.edu>
2172 * lex.c (ffelex_display_token, ffelex_type_string_): Resurrect
2173 these from their old versions and update them for possible invocation
2175 * lex.h (ffelex_display_token): Declare this in case anyone
2176 else wants to call it.
2178 * lex.c (ffelex_total_tokens_): Have this reflect actual allocated
2179 tokens, no longer include outstanding "uses" of tokens.
2181 * malloc.c, malloc.h (MALLOC_DEBUG): New macro to control
2182 checking of whether callers follow rules, now defaults to 0
2183 for "no checking" to improve compile times.
2185 * malloc.c (malloc_pool_kill): Fix bug that could prevent
2186 subpool from actually being killed (wasn't setting its use
2189 * proj.h, *.c (dmpout): Replace all occurrences of `stdout'
2190 and some of `stderr' with `dmpout', so where to dump debugging
2191 output can be easily controlled during build; add default
2192 for `dmpout' of `stderr' to proj.h.
2194 Sun Dec 3 00:56:29 1995 Craig Burley <burley@gnu.ai.mit.edu>
2196 * com.c (ffecom_return_expr): Eliminate attempt at warning
2197 about unset return values, since the back end does this better,
2198 with better wording, and is not triggered by clearly working
2199 (but spaghetti) code as easily as this test.
2201 Sat Dec 2 08:28:56 1995 Craig Burley <burley@gnu.ai.mit.edu>
2203 * target.c (ffetarget_power_*_integerdefault): Raising 0 to
2204 integer constant power should not be an error condition;
2205 if so, other code should catch 0 to any power, etc.
2207 * bad.def (FFEBAD_BAD_POWER): 0**integer now a warning instead
2210 Fri Dec 1 00:12:03 1995 Craig Burley (burley@gnu.ai.mit.edu)
2212 * bad.def: Clarify diagnostic regarding complex constant elements.
2213 * expr.c (ffeexpr_cb_comma_c_): Capitalize real/imaginary
2214 for clarified diagnostic.
2216 * com.c (ffecom_close_include_): Close the file!
2218 * lex.c (ffelex_file_fixed): Update line info if the line
2219 has any content, not just if it finishes a previous line
2221 (ffelex_file_free): Clarify switch statement code.
2223 Sat Nov 18 19:37:22 1995 Craig Burley (burley@gnu.ai.mit.edu)
2225 * Version 0.5.17 released.
2227 Fri Nov 17 14:27:24 1995 Craig Burley (burley@gnu.ai.mit.edu)
2229 * Make-lang.in: Fix typo in comment.
2231 * Makefile.in (f/fini.o, f/proj-h.o): Don't use `$<' since
2232 not all makes support it (e.g. NeXT make), use explicit
2233 source name instead (with $(srcdir) and munging).
2234 (ASSERT_H): assert.h lives in source dir, not build dir.
2236 Thu Nov 16 12:47:50 1995 Craig Burley (burley@gnu.ai.mit.edu)
2238 * com.c (ffecom_init_0): Fix dumb bug in code to produce
2239 warning message about non-32-bit-systems.
2241 * stc.c (ffestc_R501_item): Parenthesize test to make
2242 warning go away (and perhaps fix bug).
2244 Thu Nov 16 03:43:33 1995 Craig Burley (burley@gnu.ai.mit.edu)
2246 * g77.c: Upgrade to 2.7.0's gcc.c.
2247 Fix -v to pass a temp name instead of "/dev/null" for "-o".
2249 Fri Nov 10 19:16:05 1995 Craig Burley (burley@gnu.ai.mit.edu)
2251 * ste.c (ffeste_begin_iterdo_): Add Toon's change to
2252 make loops faster on some machines (implement termination
2253 condition as "--i >= 0" instead of "i-- > 0").
2255 Thu Nov 2 03:58:17 1995 Craig Burley (burley@gnu.ai.mit.edu)
2257 * Make-lang.in: Remove unnecessary $(exeext) a la cp/Make-lang.in.
2259 * com.c (ffecom_expr_): Restore old strategy for assignp variant
2260 of opSYMTER case...always return the ASSIGN version of var.
2261 That way, `-O -Wuninitialized' will catch "I=3;GOTO I;END"
2262 (though the diagnostic will refer to `__g77_ASSIGN_i').
2264 * com.c (ffecom_expr_power_integer_): For constant rhs case,
2265 wrap every new eval of lhs in save_expr() so it is clear to
2266 back end that MULT_EXPR(lhs,lhs) has identical operands,
2267 otherwise for an rhs like 32767 it generates around 65K pseudo
2268 registers, which which stupid_life_analysis cannot cope
2269 (due to reg_renumber in regs.h being `short *' instead of
2272 * com.c (ffecom_expr_): Speed up implementation of LOGICAL
2273 versions of opNOT, opAND, opOR, opXOR/opNEQV, and opEQV by
2274 assuming the values actually are kosher LOGICAL bit patterns.
2275 Also simplify code that implements some of the INTEGER versions
2278 * com.c (skip_redundant_dir_prefix, read_name_map,
2279 ffecom_open_include_, signed_type, unsigned_type): Fold in
2280 changes to cccp.c made from 2.7.0 through ss-950826.
2282 * equiv.c (ffeequiv_layout_local_): Kill the equiv list
2285 * expr.c (ffeexpr_reduced_eqop2_): Issue specific diagnostic
2286 regarding usage of .EQV./.NEQV. in preference to .EQ./.NE..
2288 * intrin.c: Add ERF and ERFC as generic intrinsics.
2291 * sta.c (ffesta_save_, ffesta_second_): Whoever calls
2292 ffestd_exec_begin must also set ffesta_seen_first_exec = TRUE,
2293 and anytime stc sees an exec transition, it must do both.
2294 stc.c (ffestc_eof): Same.
2296 * stc.c (ffestc_promote_sfdummy_): If failed implicit typing
2297 or CHARACTER*(*) arg, after calling ffesymbol_error, don't
2298 reset info to ENTITY/DUMMY, because ffecom_sym_transform_
2299 doesn't expect such a thing with ANY/ANY type.
2301 * target.h (*logical*): Change some of these so they parallel
2302 changes in com.c, e.g. for _eqv_, use (l)==(r) instead of
2303 !!(l)==!!(r), to get a more faithful result.
2305 Fri Oct 27 07:06:59 1995 Craig Burley (burley@gnu.ai.mit.edu)
2307 * com.c (ffecom_sym_transform_): Simplify code for local
2310 * expr.c (ffeexpr_exprstack_push_unary_): Warn about two
2311 successive operators.
2312 (ffeexpr_exprstack_push_binary_): Warn about "surprising"
2313 operator precedence, as in "-2**2".
2315 * lang-options.h: Add -W(no-)surprising options.
2317 * parse.c (yyparse): Don't reset -fpedantic if not -pedantic.
2319 * top.c (ffe_decode_option): Support new -Wsurprising option.
2322 Mon Oct 23 09:14:15 1995 Craig Burley (burley@gnu.ai.mit.edu)
2324 * com.c (ffecom_finish_symbol_transform_): Don't transform
2325 NONE/NONE (CHARACTER*(*)) vars, as these don't mean anything
2326 in debugging terms, and can't be turned into anything
2327 in the back end (so ffecom_sym_transform_ crashes on them).
2329 * com.c (ffecom_expr_): Change strategy for assignp variant
2330 of opSYMTER case...always return the original var unless
2331 it is not wide enough.
2333 * ste.c (ffeste_io_cilist_): Clarify diagnostic for ASSIGN
2334 involving too-narrow variable. This shouldn't happen, though.
2335 (ffeste_io_icilist_): Ditto.
2336 (ffeste_R838): Ditto.
2337 (ffeste_R839): Ditto.
2339 Thu Oct 19 03:21:20 1995 Craig Burley (burley@gnu.ai.mit.edu)
2341 * com.c (ffecom_sym_transform_assign_): Set TREE_STATIC
2342 using the same decision-making process as used for their twin
2343 variables, so ASSIGN can last across RETURN/CALL as appropriate.
2345 Fri Sep 22 20:21:18 1995 Craig Burley (burley@gnu.ai.mit.edu)
2347 * Makefile.in: fini is a host program, so it needs a host-compiled
2348 version of proj.o, named proj-h.o. f/fini, f/fini.o, and
2349 f/proj-h.o targets updated accordingly.
2351 * com.c (__eprintf): New function.
2353 Wed Sep 20 02:26:36 1995 Craig Burley (burley@gnu.ai.mit.edu)
2355 * lang-options.h: Add omitted -funix-intrinsics-* options.
2357 * malloc.c (malloc_find_inpool_): Check for infinite
2358 loop, crash if detected (user reports encountering
2359 them in some large programs, this might help track
2362 Thu Sep 7 13:00:32 1995 Craig Burley (burley@gnu.ai.mit.edu)
2364 * com.c (lang_print_error_function): Don't dereference null
2365 pointer when outside any program unit.
2366 (ffecom_let_char_, ffecom_arg_ptr_to_expr): If catlist
2367 item or length ever error_mark_node, don't continue processing,
2368 since back-end functions like build_pointer_type crash on
2369 error_mark_node's (due to pushing bad obstacks, etc.).
2371 Wed Aug 30 15:58:35 1995 Craig Burley (burley@gnu.ai.mit.edu)
2373 * Version 0.5.16 released.
2375 Mon Aug 28 12:24:20 1995 Craig Burley (burley@gnu.ai.mit.edu)
2377 * bad.c (ffebad_finish): Fix botched message when no places
2378 are printed (due to unknown line info, etc.).
2380 * std.c (ffestd_subr_labels_): Do a better job finding
2381 line info in the case of typeANY and diagnostics.
2383 Fri Aug 25 15:19:29 1995 Craig Burley (burley@gnu.ai.mit.edu)
2385 * com.c (DECL_ARTIFICIAL): Surround all references to this
2386 macro with #if !BUILT_FOR_270 and #endif.
2387 (init_lex): Surround print_error_function decl with
2388 #if !BUILT_FOR_270 and #endif.
2389 (lang_init): Call new ffelex_hash_kludge function to solve
2390 problem with preprocessed files that have INCLUDE statements.
2392 * lex.c (ffelex_getc_): New function.
2393 (ffelex_cfelex_): Use ffelex_getc_ instead of getc in any
2394 paths of code that can be affected by ffelex_hash_kludge.
2395 Don't make an EOF token for unrecognized token; set token
2396 to NULL instead, to avoid problems when not initialized.
2397 (ffelex_hash_): Use ffelex_getc_ instead of getc in any
2398 paths of code that can be affected by ffelex_hash_kludge.
2399 Test token returned by ffelex_cfelex_ for NULL, meaning
2401 Get rid of useless used_up variable.
2402 Don't do ffewhere stuff or kill any tokens if in
2404 (ffelex_file_fixed, ffelex_file_free): Use ffelex_getc_
2405 instead of getc in any paths of code that can be affected
2406 by ffelex_hash_kludge.
2407 (ffelex_hash_kludge): New function.
2409 * lex.h (ffelex_hash_kludge): New function.
2411 Wed Aug 23 15:17:40 1995 Craig Burley (burley@gnu.ai.mit.edu)
2413 * com.c: Implement -f(no-)underscoring options by always
2414 compiling in code to do it, and having that code inhibit
2415 itself when -fno-underscoring is in effect. This option
2416 overrides -f(no-)f2c for this purpose; -f(no-)f2c returns
2417 to it's <=0.5.15 behavior of affecting only how code
2418 is generated, not how/whether names are mangled.
2420 * target.h: Redo specification of appending underscores so
2421 the macros are named "_default" instead of "_is" and the
2422 two-underscore macro defaults to 1.
2424 * top.c, top.h (underscoring): Add appropriate stuff
2425 for the -f(no-)underscoring options.
2427 Tue Aug 22 10:25:01 1995 Craig Burley (burley@gnu.ai.mit.edu)
2429 * bad.c (ffebad_finish): Call report_error_function (in toplev.c)
2430 to better identify location of problem.
2431 Say "(continued):" instead of "(continued:)" for consistency.
2433 * com.c (ffecom_gen_sfuncdef_): Set and reset new
2434 ffecom_nested_entry_ variable to hold ffesymbol being compiled.
2435 (lang_print_error_function): New function from toplev.c.
2436 Use ffecom_nested_entry_ to help determine which name
2437 and kind-string to print.
2438 (ffecom_expr_intrinsic_): Handle EXIT and FLUSH invocations
2439 with different calling sequences than library functions.
2440 Have SIGNAL and SYSTEM push and pop calltemps, and convert
2441 their return values to the destination type (just in case).
2442 (FFECOM_rttypeINT_): New return type for `int', in case
2443 gcc/f/runtime/libF77/system_.c(system_) is really supposed
2444 to return `int' instead of `ftnint'.
2446 * com.h (report_error_function): Declare this.
2448 * equiv.c (ffeequiv_layout_local_): Don't forget to consider
2449 root variable itself as possible "first rooted variable",
2450 else might never set symbol and then crash later.
2452 * intrin.c (ffeintrin_check_exit_): Change to allow no args
2453 and rename to ffeintrin_check_int_1_o_ for `optional'.
2454 #define ffeintrin_check_exit_ and _flush_ to this new
2455 function, so intrin.def can refer to the appropriate names.
2457 * intrin.def (FFEINTRIN_impFLUSH): Validate using
2458 ffeintrin_check_flush_ so passing an INTEGER arg is allowed.
2460 * lex.c (ffelex_file_push_, ffelex_file_pop_): New functions
2461 to manage input_file_stack in gbe.
2462 (ffelex_hash_): Call new functions (instead of doing code).
2463 (ffelex_include_): Call new functions to update stack for
2464 INCLUDE (_hash_ handles cpp output of #include).
2466 Mon Aug 21 08:09:04 1995 Craig Burley (burley@gnu.ai.mit.edu)
2468 * Makefile.in: Put `-W' in front of every `-Wall', since
2469 2.7.0 requires that to engage `-Wunused' for parameters.
2471 * com.c: Mark all parameters as artificial, so
2472 `-W -Wunused' doesn't complain about unused ones (since
2473 there's no way right not to individually specify attributes
2476 * proj.h: Don't #define UNUSED if already defined, regardless
2479 Sun Aug 20 16:03:56 1995 Craig Burley (burley@gnu.ai.mit.edu)
2481 * gbe/2.7.0.diff: Regenerate.
2483 * lang-options.h, lang-specs.h: If not __STDC__ (ANSI C),
2484 avoid doing anything, especially the stringizing in -specs.h.
2486 Thu Aug 17 03:36:12 1995 Craig Burley (burley@gnu.ai.mit.edu)
2488 * lang-specs.h: Remove useless optional settings of -traditional,
2489 since -traditional is always set anyway.
2491 Wed Aug 16 16:56:46 1995 Craig Burley (burley@gnu.ai.mit.edu)
2493 * Make-lang.in (F2C_INSTALL_FLAG, F2CLIBOK): More
2494 control over whether to install f2c-related stuff.
2495 (install-f2c-*): New targets to install f2c-related
2496 stuff in system, not just gcc, directories.
2498 * com.c: Change calls to ffecom_get_invented_identifier
2499 to use generally more predictable names.
2500 Change calls to build_range_type to ensure consistency
2501 of types of operands.
2502 (ffecom_get_external_identifier_): Change to accept
2503 symbol info, not just text, so it can use f2c flag for
2504 symbol to decide whether to append underscore(s).
2505 (ffecom_get_identifier_): Don't change names if f2c flag
2506 off for compilation.
2507 (ffecom_type_permanent_copy_): Use same type for new max as
2509 (ffecom_notify_init_storage): Offline fixups for stand-alone.
2511 * data.c (ffedata_gather): Explicitly test for common block,
2512 since it's no longer always the case that a local EQUIVALENCE
2513 group has no symbol ptr (it now can, if a user-predictable
2514 "rooted" symbol has been identified).
2516 * equiv.c: Add some debugging stuff.
2517 (ffeequiv_layout_local_): Set symbol ptr with user-predictable
2518 "rooted" symbol, for giving the invented aggregate a
2521 * g77.c (append_arg): Allow for 20 extra args instead of 10.
2522 (main): For version-only case, add `-fnull-version' and, unless
2523 explicitly omitted, `-lf2c -lm'.
2525 * lang-options.h: New "-fnull-version" option.
2527 * lang-specs.h: Support ".fpp" suffix for preprocessed source
2528 (useful for OS/2, MS-DOS, other case-insensitive systems).
2530 * stc.c (ffestc_R544_equiv_): Swap way lists are merged so this
2531 is consistent with the order in which lists are built, making
2532 user predictability of invented aggregate name much higher.
2534 * storag.c, storag.h (FFESTORAG_typeDUMMY): Delete this enum.
2536 * top.c: Accept, but otherwise ignore, `-fnull-version'.
2538 Tue Aug 15 07:01:07 1995 Craig Burley (burley@gnu.ai.mit.edu)
2540 * DOC, INSTALL, PROJECTS: Extensive improvements to documentation.
2542 Sun Aug 13 01:55:18 1995 Craig Burley (burley@gnu.ai.mit.edu)
2544 * INSTALL (f77-install-ok): Document the use of this file.
2546 * Make-lang.in (F77_INSTALL_FLAG): New flag to control
2547 whether to install an `f77' command (based on whether
2548 a file named `f77-install-ok' exists in the source or
2549 build directory) to replace the broken attempt to use
2550 comment lines to avoid installing `f77' (broken in the
2551 sense that it prevented installation of `g77').
2553 Mon Aug 7 06:14:26 1995 Craig Burley (burley@gnu.ai.mit.edu)
2555 * DOC: Add new sections for g77 & gcc compiler options,
2556 source code form, and types, sizes and precisions.
2557 Remove lots of old "delta-version" info, or at least
2560 * INSTALL: Add info here that used to be in DOC.
2563 * g77.c (lookup_option, main): Check for --print-* options,
2564 so we avoid adding version-determining stuff.
2566 Wed Jul 26 15:51:03 1995 Craig Burley (burley@gnu.ai.mit.edu)
2568 * Make-lang.in, Makefile.in (input.j, INPUT_H): New file.
2569 Update dependencies accordingly.
2571 * bad.c (ffebad_here): Okay to use unknown line/col.
2573 * compilers.h (@f77-cpp-input): Remove -P option now that
2574 # directives are handled by f771. Update other options
2575 to be more consistent with @c in gcc/gcc.c. Don't run f771
2576 if -E specified, etc., a la @c.
2577 (@f77): Don't run f771 if -E specified, etc., a la @c.
2579 * config-lang.in: Avoid use of word "guaranteed".
2581 * input.j: New file to wrap around gcc/input.h.
2583 * lex.j: Add support for parsing # directives output by cpp.
2584 (ffelex_cfebackslash_): New function.
2585 (ffelex_cfelex_): New function.
2586 (ffelex_get_directive_line_): New function.
2587 (ffelex_hash_): New function.
2588 (ffelex_include_): Change to not use ffewhere_file_(begin|end).
2589 Also fix bug in pointing to next line (for diagnostics, &c)
2590 following successful INCLUDE.
2591 (ffelex_next_line_): New function that does chunk of code
2592 seen in several places elsewhere in the lexers.
2593 (ffelex_file_fixed): Delay finishing statement until source
2594 line is registered with ffewhere, so INCLUDE processing
2595 picks up the info correctly.
2596 Okay to kill or use unknown line/col objects now.
2597 Handle HASH (#) lines.
2598 Reorder tests for insubstantial lines to put most frequent
2599 occurrences at top, for possible minor speedup.
2600 Some general consolidation of code.
2601 (ffelex_file_free): Handle HASH (#) lines.
2602 Okay to kill or use unknown line/col objects now.
2603 Some general consolidation of code.
2604 (ffelex_init_1): Detect HASH (#) lines.
2605 (ffelex_set_expecting_hollerith): Okay to kill or use unknown
2606 line/col objects now.
2608 * lex.h (FFELEX_typeHASH): New enum.
2610 * options-lang.h (-fident, -fno-ident): New options.
2612 * stw.c (ffestw_update): Okay to kill unknown line/col objects
2615 * target.h (FFETARGET_okREALQUAD, FFETARGET_okCOMPLEXDOUBLE,
2616 FFETARGET_okCOMPLEXQUAD): #define these appropriately.
2618 * top.c: Include flag.j wrapper, not flags.h directly.
2619 (ffe_is_ident_): New flag.
2620 (ffe_decode_option): Handle -fident and -fno-ident.
2621 (ffe_file): Replace obsolete ffewhere_file_(begin|end) with
2624 * top.h (ffe_is_ident_, ffe_is_ident, ffe_set_is_ident):
2625 New flag and access functions.
2627 * where.c, where.h: Remove all tracking of parent file.
2628 (ffewhere_file_begin, ffewhere_file_end): Delete these.
2629 (ffewhere_line_use): Make it work with unknown line object.
2631 Mon Jul 17 03:04:09 1995 Craig Burley (burley@gnu.ai.mit.edu)
2633 * com.c (ffecom_sym_transform_): Set DECL_IN_SYSTEM_HEADER
2634 flag for any local vars used as stmtfunc dummies or DATA
2635 implied-DO iter vars, so no -Wunused warnings are produced
2636 for them (a la f2c).
2637 (ffecom_init_0): Do "extern int xargc;" for IARGC() intrinsic.
2638 Warn if target machine not 32 bits, since g77 isn't yet
2639 working on them at all well.
2641 * expr.c (ffeexpr_sym_lhs_call_, ffeexpr_sym_lhs_data_,
2642 ffeexpr_sym_lhs_extfunc_, ffeexpr_sym_rhs_actualarg_,
2643 ffeexpr_sym_rhs_let_, ffeexpr_paren_rhs_let_): Don't
2644 gratuitously set attr bits that don't apply just
2645 to avoid null set meaning error; instead, use explicit
2646 error flag, and allow null attr set, to
2647 fix certain bugs discovered by looking at this code.
2649 * g77.c: Major changes to improve support for gcc long options,
2650 to make `g77 -v' report more useful info, and so on.
2652 Mon Jul 3 14:49:16 1995 Craig Burley (burley@gnu.ai.mit.edu)
2654 * DOC, com.c, intrin.h, intrin.c, intrin.def, target.h, top.c,
2655 top.h: Add new `unix' group of intrinsics, which includes the
2656 newly added ERF, ERFC, EXIT, plus even newer ABORT, DERF, DERFC,
2657 FLUSH, GETARG, GETENV, SIGNAL, and SYSTEM.
2659 Tue Jun 27 23:01:05 1995 Craig Burley (burley@gnu.ai.mit.edu)
2661 * bld.c, bld.h (ffebld_constant_pool,
2662 ffebld_constant_character_pool): Use a single macro (the
2663 former) to access the pool for allocating constants, instead
2664 of latter in public and FFEBLD_CONSTANT_POOL_ internally
2665 in bld.c (which was the only one that was correct before
2666 these changes). Add verification of integrity of certain
2667 heap-allocated areas.
2669 * com.c (ffecom_overlap_, ffecom_args_overlap_,
2670 ffecom_tree_canonize_ptr_, ffecom_tree_canonize_ref_): New
2671 functions to optimize calling COMPLEX and, someday, CHARACTER
2672 functions requiring additional argument to be passed.
2673 (ffecom_call_, ffecom_call_binop_, ffecom_expr_,
2674 ffecom_expr_intrinsic_): Change calling
2675 sequences to include more info on possible destination.
2676 (ffecom_expr_intrinsic_): Add ERF(), ERFC(), and EXIT()
2678 (ffecom_sym_transform_): For assumed-size arrays, set high
2679 bound to highest possible value instead of low bound, to
2680 improve validity of overlap checking.
2681 (duplicate_decls): If olddecl and newdecl are the same,
2682 don't do any munging, just return affirmative.
2684 * expr.c: Change ffecom_constant_character_pool() to
2685 ffecom_constant_pool().
2687 * info.c (ffeinfo_new): Compile this version if not being
2690 * info.h (ffeinfo_new): Don't define macro if not being
2693 * intrin.c, intrin.def: Add ERF(), ERFC(), and EXIT() intrinsics.
2694 (ffeintrin_check_exit_): New for EXIT() subroutine intrinsic.
2696 * malloc.c, malloc.h (malloc_verify_*): New functions to verify
2697 integrity of heap-storage areas.
2699 * stc.c (ffestc_R834, ffestc_R835): Handle possibility that
2700 an enclosing DO won't have a construct name even when the
2701 CYCLE/EXIT does (i.e. without dereferencing NULL).
2703 * target.c, target.h (ffetarget_verify_character1): New function
2704 to verify integrity of heap storage used to hold character constant.
2706 Thu Jun 22 15:36:39 1995 Howard Gordon (flash@super.org)
2708 * stp.h (ffestpVxtcodeIx): Fix typo in typedef for this.
2710 Mon May 29 15:22:31 1995 Craig Burley (burley@gnu.ai.mit.edu)
2712 * *: Make all sorts of changes to accommodate upcoming gcc-2.7.0.
2713 I didn't keep track of them, nor just when I made them, nor
2714 when I (much later, probably in early August 1995) modified
2715 them so they could properly handle both 2.7.0 and 2.6.x.
2717 * com.c (ffecom_expr_power_integer_): Don't expand_start_stmt_expr
2718 if transforming dummy args, because the back end cannot handle
2719 that (it's rejected by the gcc front end), just generate
2720 call to run-time library.
2721 Back out changes in 0.5.15 because more temporaries might be
2722 needed anyway (for COMPLEX**INTEGER).
2723 (ffecom_push_tempvar): Remove inhibitor.
2724 Around start_decl and finish_decl (in particular, arround
2725 expand_decl, which is called by them), push NULL_TREE into
2726 sequence_rtl_expr, an external published by gcc/function.c.
2727 This makes sure the temporary is truly in the function's
2728 context, not the inner context of a statement-valued expression.
2729 (I think the back end is inconsistent here, but am not
2730 interested in convincing the gbe maintainers about this now.)
2731 (pushdecl): Make sure that when pushing PARM_DECLs, nothing
2732 other than them are pushed, as happened for 0.5.15 and which,
2733 if done for other reasons not fixed here, might well indicate
2734 some other problem -- so crash if it happens.
2736 * equiv.c (ffeequiv_layout_local_): If the local equiv group
2737 has a non-nil COMMON field, it should mean that an error has
2738 occurred and been reported, so just trash the local equiv
2739 group and do nothing.
2741 * stc.c (ffestc_promote_sfdummy_): Set sfdummy arg state to
2742 UNDERSTOOD so above checking for duplicate args actually
2743 works, and so we don't crash later in pushdecl.
2745 * ste.c (ffeste_R1001): Set initial value only for VAR_DECLs,
2746 not for, e.g., LABEL_DECLs, which the FORMAT label can be
2747 if it was previously treated as an executable label.
2749 Sat May 20 01:53:53 1995 Craig Burley (burley@gnu.ai.mit.edu)
2751 * com.c (ffecom_sym_transform_): For adjustable arrays,
2752 pass high bound through variable_size in case its primaries
2753 are changed (dumb0.f, and this might also improve
2754 performance so it approaches f2c|gcc).
2756 Fri May 19 11:00:36 1995 Craig Burley (burley@gnu.ai.mit.edu)
2758 * Version 0.5.15 released.
2760 * com.c (ffecom_expr_power_integer_): Push temp vars
2761 before expanding a statement expression, since that seems
2762 to cause temp vars to be "forgotten" after the end of the
2763 expansion in the back end. Disallow more temp-var
2764 pushing during such an expansion, just in case.
2765 (ffecom_push_tempvar): Crash if a new variable needs to be
2766 pushed but cannot be at this point (should never happen).
2768 Wed May 17 12:26:16 1995 Craig Burley (burley@gnu.ai.mit.edu)
2770 * expr.c (ffeexpr_collapse_convert): Add code to convert
2771 LOGICAL to CHARACTER. Reject conversion of REAL or COMPLEX
2772 to CHARACTER entirely, as it cannot be supported with all
2775 * target.h, target.c (ffetarget_convert_character1_logical1):
2778 Sun May 14 00:00:09 1995 Craig Burley (burley@gnu.ai.mit.edu)
2780 * com.c (ffecom_do_entry_, ffecom_gen_sfuncdef_,
2781 ffecom_start_progunit_, ffecom_sym_transform_,
2782 ffecom_init_0, start_function): Changes to have REAL
2783 external functions return same type as DOUBLE PRECISION
2784 external functions when -ff2c is in force; while at it,
2785 some code cleanups done.
2787 * stc.c (ffestc_R547_item_object): Disallow array declarator
2788 if one already exists for symbol.
2790 * ste.c (ffeste_R1227): Convert result variable to type
2791 of function result as seen by back end (e.g. for when REAL
2792 external function actually returns result as double).
2794 * target.h (FFETARGET_defaultFIXED_LINE_LENGTH): New
2795 macro for default for -ffixed-line-length-N option.
2797 * top.c (ffe_fixed_line_length_): Initialize this to new
2798 target.h macro instead of constant 72.
2800 Tue May 9 01:20:03 1995 Craig Burley (burley@gnu.ai.mit.edu)
2802 * lex.c (ffelex_send_token_): If sending CHARACTER token with
2803 null text field, put a single '\0' in it and set length/size
2804 fields to 0 (to fix 950508-0.f).
2805 (ffelex_image_char_): When setting ffelex_bad_line_ to TRUE,
2806 always "close" card image by appending a null char and setting
2807 ffelex_card_length_. As part of this, append useful text
2808 to identify the two kinds of problems that involve this.
2809 (ffelex_file_fixed): Reset ffelex_bad_line_ to FALSE after
2810 seeing a line with invalid first character (fixes 950508-1.f).
2811 If final nontab column is zero, assume tab seen in line.
2812 (ffelex_card_image_): Always make this array 8 characters
2813 longer than reflected by ffelex_card_size_.
2814 (ffelex_init_1): Get final nontab column info from top instead
2817 * options-lang.h: Add -ffixed-line-length- prefix.
2819 * top.h: Add ffe_fixed_line_length() and _set_ version, plus
2820 corresponding extern.
2822 * top.c: Handle -ffixed-line-length- option prefix.
2824 Fri Apr 28 05:40:25 1995 Craig Burley (burley@gnu.ai.mit.edu)
2826 * Version 0.5.14 released.
2828 * Make-lang.in: Add assert.j.
2830 * Makefile.in: Add assert.j.
2832 * assert.j: New file.
2834 Thu Apr 27 16:24:22 1995 Craig Burley (burley@gnu.ai.mit.edu)
2836 * bad.h (ffebad_severity): New function.
2838 * bad.c (ffebad_severity): New function.
2840 * bad.def (FFEBAD_OPEN_INCLUDE): Change severity from SEVERE
2841 to FATAL, since processing continues, and that seems fine.
2843 * com.c: Add facility to handle -I.
2844 (ffecom_file, ffecom_close_include, ffecom_open_include,
2845 ffecom_decode_include_option): New global functions for -I.
2846 (ffecom_file_, ffecom_initialize_char_syntax_,
2847 ffecom_close_include_, ffecom_decode_include_option_,
2848 ffecom_open_include_, append_include_chain, open_include_file,
2849 print_containing_files, read_filename_string, file_name_map,
2850 savestring): New internal functions for -I.
2852 * compilers.h: Pass -I flag(s) to f771 (via "%{I*}").
2854 * lex.c (ffelex_include_): Call ffecom_close_include
2855 to close include file, for its tracking needs for -I,
2856 instead of using fclose.
2858 * options-lang.h: Add -I prefix.
2860 * parse.c (yyparse): Call ffecom_file for main input file,
2861 so -I handling works (diagnostics).
2863 * std.c (ffestd_S3P4): Have ffecom_open_include handle
2864 opening and diagnosing errors with INCLUDE files.
2866 * ste.c (ffeste_begin_iterdo_): Use correct algorithm for
2867 calculating # of iterations -- mathematically similar but
2868 computationally different algorithm was not handling cases
2869 like "DO I=6,5,2" correctly, because (5-6)/2+1 => 1, not 0.
2871 * top.c (ffe_decode_option): Allow -I, restructure a bit
2872 for clarity and, maybe, speed.
2874 Mon Apr 17 13:31:11 1995 Craig Burley (burley@gnu.ai.mit.edu)
2876 * g77.c: Remove -lc, turns out not all systems has it, but
2877 leave other changes in for clarity of code.
2879 Sun Apr 16 21:50:33 1995 Craig Burley (burley@gnu.ai.mit.edu)
2881 * com.c (ffecom_expr_): Implement ARRAY_EXPR as INDIRECT_REF
2882 of appropriate PLUS_EXPRs of ptr_to_expr of array, to see
2883 if this generates better code. (Conditional on
2884 FFECOM_FASTER_ARRAY_REFS.)
2886 Sun Apr 16 00:22:48 1995 Craig Burley (burley@gnu.ai.mit.edu)
2888 * Make-lang.in (F77_SRCS): Remove g77.c, since it doesn't
2889 contribute to building f771.
2891 * Makefile.in (dircheck): Remove/replace with f/Makefile, because
2892 phony targets that are referenced in other real targets get run
2893 when those targets are specified, which is a waste of time (e.g.
2894 when rebuilding and only g77.c has changed, f771 was being linked
2897 * g77.c: Include -lc between -lf2c and -lm throughout.
2899 * implic.c (ffeimplic_establish_symbol): If -Wimplicit, warn if
2900 implicit type given to symbol.
2902 * lex.c (ffelex_include_): Don't gratuitously increment line
2905 * top.h, top.c (ffe_is_warn_implicit_): New global variable and
2906 related access macros.
2907 (ffe_decode_option): Handle -W options, including -Wall and
2910 * where.c (ffewhere_line_new): Don't muck with root line (was
2911 crashing on null input since lexer changes over the past week
2914 Thu Apr 13 16:48:30 1995 Craig Burley (burley@gnu.ai.mit.edu)
2916 * com.c (ffecom_init_0): Register built-in functions for cos,
2918 (ffecom_tree_fun_type_double): New variable.
2919 (ffecom_expr_intrinsic_): Update f2c input and output files
2920 to latest version of f2c (no important g77-related changes
2921 noted, just bug fixes to f2c and such).
2922 (builtin_function): New function from c-decl.c.
2924 * com-rt.def: Refer to built-in functions for cos, sin, and sqrt.
2926 Thu Apr 13 10:25:09 1995 Craig Burley (burley@gnu.ai.mit.edu)
2928 * com.c (ffecom_expr_intrinsic_): Convert 0. to appropriate
2929 type to keep DCMPLX(I) from crashing the compiler.
2930 (ffecom_expr_): Don't convert result from ffecom_tree_divide_.
2931 (ffecom_tree_divide_): Add tree_type argument, have all callers
2932 pass one, and don't convert right-hand operand to it (this is
2933 to make this new function work as much like the old in-line
2934 code used in ffecom_expr_ as possible).
2936 * lex.c: Maintain lineno and input_filename the way the gcc
2939 * std.c (ffestd_exec_end): Save and restore lineno and
2940 input_filename around the second pass, which sets them
2941 appropriately for each saved statement.
2943 Wed Apr 12 09:44:45 1995 Craig Burley (burley@gnu.ai.mit.edu)
2945 * com.c (ffecom_expr_power_integer_): New function.
2946 (ffecom_expr_): Call new function for power op with integer second
2947 argument, for generating better code. Also replace divide
2948 code with call to new ffecom_tree_divide_ function.
2949 Canonicalize calls to ffecom_truth_value(_invert).
2950 (ffecom_tree_divide_): New function.
2952 Wed Apr 5 14:15:44 1995 Craig Burley (burley@gnu.ai.mit.edu)
2954 * lex.c: Change to allocate text for tokens only when actually
2955 needed, which should speed compilation up somewhat.
2956 Change to allow INCLUDE at any point where a statement
2957 can end, i.e. in ffelex_finish_statement_ or when a SEMICOLON
2959 Remove some old, obsolete code.
2960 Clean up layout of entire file to improve formatting,
2962 (ffelex_set_expecting_hollerith): Remove include argument.
2964 Fri Mar 31 23:19:08 1995 Craig Burley (burley@gnu.ai.mit.edu)
2966 * bad.h, bad.c (ffebad_start_msg, ffebad_start_msg_lex):
2967 New functions to generate arbitrary messages.
2968 (FFEBAD_severityPEDANTIC): New severity, to correspond
2969 to toplev's pedwarn() function.
2971 * lex.c (ffelex_backslash_): New function to implement
2972 backslash processing.
2973 (ffelex_file_fixed, ffelex_file_free): Implement new
2974 backslash processing.
2976 * std.c (ffestd_R1001dump_): Don't assume CHARACTER and
2977 HOLLERITH tokens stop at '\0' characters, now that backslash
2978 processing is supported -- use their advertised lengths instead,
2979 and double up the '\002' character for libf2c.
2981 Mon Mar 27 17:10:33 1995 Craig Burley (burley@gnu.ai.mit.edu)
2983 * com.c (ffecom_init_local_zero_): Implement -finit-local-zero.
2984 (ffecom_sym_transform_): Same.
2985 (ffecom_transform_equiv_): Same.
2987 * options-lang.h: Add -f(no-)(init-local-zero,backslash,ugly-init).
2989 * stb.c (ffestb_V020): Reject "TYPEblah(...", which might be
2990 an array assignment.
2992 * target.h, top.h, top.c: Implement -finit-local-zero.
2994 Fri Mar 24 19:56:22 1995 Craig Burley (burley@gnu.ai.mit.edu)
2996 * Make-lang.in, Makefile.in: Remove conf-proj(.in) and
2997 proj.h(.in) rules, plus related config.log, config.cache,
2998 and config.status stuff.
3000 * com.c (ffecom_init_0): Change messages when atof(), bsearch(),
3001 or strtoul() do not work as expected in the start-up test.
3003 * conf-proj, conf-proj.in: Delete.
3005 * lex.c (ffelex_file_fixed): Allow f2c's '&' in column 1
3006 to mean continuation line.
3008 * options-lang.h: New file, #include'd by ../toplev.c.
3010 * proj.h.in: Rename back to proj.h.
3012 * proj.h (LAME_ASSERT): Remove.
3013 (LAME_STDIO): Remove.
3014 (NO_STDDEF): Remove.
3015 (NO_STDLIB): Remove.
3016 (NO_BSEARCH): Remove auto detection, rename to !FFEPROJ_BSEARCH.
3017 (NO_STRTOUL): Remove auto detection, rename to !FFEPROJ_STRTOUL.
3018 (USE_HOST_LIMITS): Remove (maybe still needed by stand-alone?).
3019 (STR, STRX): Do only ANSI C definitions.
3021 Mon Mar 13 10:46:13 1995 Craig Burley (burley@gnu.ai.mit.edu)
3023 * BUGS: Add item about g77 requiring gcc to compile it.
3025 * NEWS: New file listing user-visible changes in the release.
3027 * PROJECTS: Update to include a new item or two, and modify
3028 or delete items that are addressed in this or previous releases.
3030 * bad.c (ffebad_finish): Don't crash if missing string &c,
3031 just substitute obviously distressed string "[REPORT BUG!!]"
3032 for cases where the message/caller are fudgy.
3034 * bad.def: Clean up error messages in a major way, add new ones
3035 for use by changes in target.c.
3037 * com.c (ffecom_expr_): Handle opANY in opCONVERT.
3038 (ffecom_let_char_): Disregard destinations with ERROR_MARK.
3039 (ffecom_1, ffecom_1_fn, ffecom_2, ffecom_2s, ffecom_3,
3040 ffecom_3s, &c): Check all inputs for error_mark_node.
3041 (ffecom_start_progunit_): Don't transform all symbols
3042 in BLOCK DATA, since it never executes, and it is silly
3043 to, e.g., generate all the structures for NAMELIST.
3044 (ffecom_char_length_expr_): Rename to ffecom_intrinsic_len_.
3045 (ffecom_intrinsic_ichar_): New function to handle ICHAR of
3046 arbitrary expression with possible 0-length operands.
3047 (ffecom_expr_intrinsic_): Use ffecom_intrinsic_char_.
3048 For MVBITS, set tree_type to void_type_node.
3049 (ffecom_start_progunit_): Name master function for entry points
3050 after primary entry point so users can easily guess it while
3052 (ffecom_arg_ptr_to_expr): Change treatment of Hollerith,
3053 Typeless, and %DESCR.
3054 (ffecom_expr_): Change treatment of Hollerith.
3056 * data.c (ffedata_gather_): Handle opANY in opCONVERT.
3058 * expr.c (ffeexpr_token_apostrophe_): Issue FFEBAD_NULL_CHAR_CONST
3059 warning as necessary.
3060 (ffeexpr_token_name_rhs_): Set context for args to intrinsic
3061 so that assignment-like concatenation is allowed for ICHAR(),
3062 IACHAR(), and LEN() intrinsics.
3063 (ffeexpr_reduced_*_): Say "an array" instead of "an entity" in
3064 diagnostics, since it's more informative.
3065 (ffeexpr_finished_): For many contexts, check for null expression
3066 and array before trying to do a conversion, to avoid redundant
3069 * g77.1: Fix typo for preprocessed suffix (.F, not .f).
3071 * global.c (ffeglobal_init_common): Warn if initializing
3073 (ffeglobal_pad_common): Enable code to warn if initial
3075 (ffeglobal_size_common): Complain if enlarging already-
3076 initialized common, since it won't work right anyway.
3078 * intrin.c: Add IMAG() intrinsic.
3079 (ffeintrin_check_loc_): Allow opSUBSTR in LOC().
3081 * intrin.def: Add IMAG() intrinsic.
3083 * lex.c: Don't report FFEBAD_NULL_CHAR_CONST errors.
3085 * sta.c, sta.h, stb.c: Changes to clean up error messages (see
3088 * stb.c (ffestb_R100113_): Issue FFEBAD_NULL_CHAR_CONST
3089 warning as necessary.
3091 * stc.c (ffestc_shriek_do_): Don't try to reference doref_line
3092 stuff in ANY case, since it won't be valid.
3093 (ffestc_R1227): Allow RETURN in main program unit, with
3094 appropriate warnings/errors.
3095 (ffestc_subr_format_): Array of any type is a CHAREXPR (F77 C5).
3097 * ste.c (ffeste_begin_doiter_): Couple of fixes to accurately
3098 determine if loop never executes.
3100 * target.c (ffetarget_convert_*_hollerith_): Append spaces,
3101 not zeros, to follow F77 Appendix C, and to warn when
3102 truncation of non-blanks done.
3103 (ffetarget_convert_*_typeless): Rewrite to do typeless
3104 conversions properly, and warn when truncation done.
3105 (ffetarget_print_binary, ffetarget_print_octal,
3106 ffetarget_print_hex): Rewrite to use new implementation of
3108 (ffetarget_typeless_*): Rewrite to use new implementation
3109 of typeless, and to warn about overflow.
3111 * target.h (ffetargetTypeless): New implementation of
3114 * type.h, type.c (ffetype_size_typeless): Remove (incorrect)
3115 implementation of this function and its extern.
3117 Sun Mar 5 18:46:42 1995 Craig Burley (burley@gnu.ai.mit.edu)
3119 * BUGS: Clarify that constant handling would also fix lack of
3120 adequate IEEE-754/854 support to some degree, and typeless
3121 and non-decimal constants.
3123 * com.c (ffecom_type_permanent_copy_): Comment out to avoid
3125 (duplicate_decls): New function a la gcc/c-decl.c.
3126 (pushdecl): Use duplicate_decls to decide whether to return
3127 existing decl or new one, instead of always returning existing
3129 (ffecom_expr_): opPERCENT_LOC now supports CHARACTER arguments.
3130 (ffecom_init_0): Give f2c I/O code 0 for basictypeANY/kindtypeANY.
3131 (ffecom_sym_transform_): For adjustable arrays, pass low bound
3132 through variable_size in case its primaries are changed (950302-1.f).
3134 * com.h: More decls that belong in tree.h &c.
3136 * data.c (ffedata_eval_integer1_): Fix opPAREN case to not
3137 treat value of expression as an error code.
3139 * expr.c (ffeexpr_finished_): Allow opSUBSTR in contextLOC case.
3141 * proj.c: Add "const" as appropriate.
3143 Mon Feb 27 10:04:03 1995 Craig Burley (burley@gnu.ai.mit.edu)
3145 * bad.def (FFEBAD_BAD_SUBSTR): Fix bad grammar in message.
3147 Fri Feb 24 16:21:31 1995 Craig Burley (burley@gnu.ai.mit.edu)
3149 * Version 0.5.13 released.
3151 * INSTALL: Warn that f/zzz.o will compare differently between
3152 stages, since it puts the __TIME__ macro into a string.
3154 * com.c (ffecom_sym_transform_): Transform kindFUNCTION/whereDUMMY
3155 to pointer-to-function, not function.
3156 (ffecom_expr_): Use ffecom_arg_ptr_to_expr instead of
3157 ffecom_char_args_ to handle comparison between CHARACTER
3158 types, so either operand can be a CONCATENATE.
3159 (ffecom_transform_common_): Set size of initialized common area
3160 to global (largest-known) size, even though size of init might
3163 * equiv.c (ffeequiv_offset_): Check symbol info for ANY.
3165 * expr.c (ffeexpr_find_close_paren_, ffeexpr_nil_*): New functions
3166 to handle following the contour of a rejected expression, so
3167 statements like "PRINT(I,I,I)=0" don't cause the PRINT statement
3168 code to get the second passed back to it as if there was a
3169 missing close-paren before it, the comma causing the PRINT code
3170 to confirm the statement, resulting in an ambiguity vis-a-vis
3171 the let statement code.
3172 Use the new ffecom_find_close_paren_ handler when an expected
3173 close-paren is missing.
3174 (ffeexpr_isdigits_): New function, use in all places that
3175 currently use isdigit in repetitive code.
3176 (ffeexpr_collapse_symter): Collapse to ANY if init-expr is ANY,
3177 so as to avoid having symbol get "transformed" if used to
3179 (ffeexpr_token_real_, ffeexpr_token_number_real_): Don't issue
3180 diagnostic about exponent, since it'll be passed along the
3181 handler path, resulting in a diagnostic anyway.
3182 (ffeexpr_token_apos_char_): Use consistent handler path
3183 regardless of whether diagnostics inhibited.
3184 (ffeexpr_token_name_apos_name_): Skip past closing quote/apos
3185 even if not a match or other diagnostic issued.
3186 (ffeexpr_sym_impdoitem_): Exec-transition local SEEN symbol.
3188 * lex.c (ffelex_image_char_): Set ffelex_saw_tab_ if TAB
3189 seen, not if anything other than TAB seen!
3191 * stc.c (ffestc_R537_item): If source is ANY but dest isn't,
3192 set dest symbol's init expr to ANY.
3193 (ffestc_R501_attrib, ffestc_R522, ffestc_R522start): Complain
3194 about conflict between "SAVE" by itself and other uses of
3195 SAVE only in pedantic mode.
3197 * ste.c (ffeste_R1212): Fix loop over labels to always
3198 increment caseno, to avoid pushcase returning 2 for duplicate
3199 values when one of the labels is invalid.
3201 Thu Feb 23 12:42:04 1995 Craig Burley (burley@gnu.ai.mit.edu)
3203 * Version 0.5.12 released.
3205 * Make-lang.in (f77.install-common): Add "else true;" before outer
3206 "fi" per Makefile.in patch.
3208 * Makefile.in (dircheck): Add "else true;" before "fi" per
3209 patch from chs1pm@surrey.ac.uk.
3211 * com.c (ffecom_push_tempvar): If type desired is ERROR_MARK,
3212 return error_mark_node, to avoid crash that results from
3213 making a VAR_DECL with error_mark_node as its type.
3215 * ste.c (ffeste_begin_iterdo_): Convert itercount to INTEGER
3216 anytime calculation of number of iterations ends up with type
3217 other than INTEGER (e.g. DOUBLE PRECISION, REAL).
3219 Thu Feb 23 02:48:38 1995 Craig Burley (burley@gnu.ai.mit.edu)
3221 * Version 0.5.11 released.
3223 * DOC: Explain -fugly-args.
3225 * bad.def (FFEBAD_ACTUALARG): Explain -fugly-args and how to
3226 rewrite code to not require it.
3228 * com.c (ffecom_vardesc_): Handle negative type code, just in
3230 (ffecom_arg_ptr_to_expr): Let ffecom_expr handle hollerith
3231 and typeless constants (move code to ffecom_constantunion).
3232 (ffecom_constantunion): Handle hollerith and typeless constants.
3234 * expr.c (ffecom_finished_): Check -fugly-args in actual-arg
3235 context where hollerith/typeless provided.
3237 * intrin.def (FFEINTRIN_genDFLOAT): Add FFEINTRIN_specDFLOAT.
3238 (FFEINTRIN_specDFLOAT): Add as f2c intrinsic.
3240 * target.h (ffetarget_convert_real[12]_integer,
3241 ffetarget_convert_complex[12]_integer): Pass -1 for high integer
3242 value if low part is negative.
3243 (FFETARGET_defaultIS_UGLY_ARGS): New macro.
3245 * top.c (ffe_is_ugly_args_): New variable.
3246 (ffe_decode_option): Handle -fugly-args and -fno-ugly-args.
3248 * top.h (ffe_is_ugly_args_, ffe_is_ugly_args(),
3249 ffe_set_is_ugly_args()): New variable and macros.
3251 Thu Feb 23 02:48:38 1995 Pedro A M Vazquez (vazquez@iqm.unicamp.br)
3253 * g77.c (sys_errlist): Use const for __FreeBSD__ systems
3256 Wed Feb 22 13:33:43 1995 Craig Burley (burley@gnu.ai.mit.edu)
3258 * Version 0.5.10 released.
3260 * CREDITS: Add Rick Niles.
3262 * INSTALL: Note how to get around lack of makeinfo.
3264 * Make-lang.in (f/proj.h): Remove # comment.
3266 * Makefile.in (f/proj.h): Remove # comment.
3268 * com.c (ffecom_expr_): Simplify opFUNCREF/opSUBRREF conversion.
3269 (ffecom_sym_transform_): For whereGLOBAL and whereDUMMY
3270 kindFUNCTION, use ffecom_tree_fun_type[][] only for non-constant
3271 (non-statement-function) f2c functions.
3272 (ffecom_init_0): ffecom_tree_fun_type[][] and _ptr_to_*_* are
3273 really f2c-interface arrays, so use base type void for COMPLEX
3276 Tue Feb 21 19:01:18 1995 Dave Love <d.love@dl.ac.uk>
3278 * Make-lang.in (f77.install-common): Expurgate the test for and
3279 possible installation of f2c in line with elsewhere. Seems to have
3280 been missing a semicolon anyhow!
3282 Tue Feb 21 11:45:25 1995 Craig Burley (burley@gnu.ai.mit.edu)
3284 * Version 0.5.9 released.
3286 * Make-lang.in (f/proj.h): touch file to register update,
3287 because the previous commands won't necessarily modify it.
3289 * Makefile.in (f/proj.h): touch file to register update,
3290 because the previous commands won't necessarily modify it.
3292 * Makefile.in (f/str-*.h, f/str-*.j): Explicitly specify
3293 output file names, so these targets go in build, not source,
3296 * bits.c, bits.h: Switch to valid ANSI C replacement for
3299 * com.c (ffecom_expr_): Add assignp arg to support ASSIGN better.
3300 If assignp is TRUE, use different tree for FFEBLD_opSYMTER case.
3301 (ffecom_sym_transform_assign_): New function.
3302 (ffecom_expr_assign): New function.
3303 (ffecom_expr_assign_w): New function.
3305 * com.c (ffecom_f2c_make_type_): Do make_signed_type instead
3306 of make_unsigned_type throughout.
3308 * com.c (ffecom_finish_symbol_transform_): Expand scope of
3309 commented-out code to probably produce faster compiler code.
3311 * com.c (ffecom_gen_sfuncdef_): Push/pop calltemps so
3312 COMPLEX works right.
3313 Remove obsolete comment.
3315 * com.c (ffecom_start_progunit_): If non-multi alt-entry
3316 COMPLEX function, primary (static) entry point returns result
3317 directory, not via extra arg -- to agree with ffecom_return_expr
3319 Pretransform all symbols so statement functions are defined
3320 before any code emitted.
3322 * com.c (ffecom_finish_progunit): Don't posttransform all
3323 symbols here -- pretransform them instead.
3325 * com.c (ffecom_init_0): Don't warn about possible ASSIGN
3326 crash, as this shouldn't happen now.
3328 * com.c (ffecom_push_tempvar): Fix to handle temp vars
3329 pushed while context is a statement (nested) function, and
3330 add appropriate commentary.
3332 * com.c (ffecom_return_expr): Check TREE_USED to determine
3333 where return value is unset.
3335 * com.h (struct _ffecom_symbol_): Add note about length_tree
3336 now being used to keep tree for ASSIGN version of symbol.
3338 * com.h (ffecom_expr_assign, ffecom_expr_assign_rw): New decls.
3339 (error): Add this prototype for back-end function.
3341 * fini.c (main): Grab input, output, and include names
3342 directly off the command line instead of making the latter
3343 two out of the first.
3345 * lex.c: Improve tab handling for both fixed and free source
3346 forms, and ignore carriage-returns on input, while generally
3347 improving the code. ffelex_handle_tab_ has been renamed and
3348 reinvented as ffelex_image_char_, among other things.
3350 * malloc.c, malloc.h: Switch to valid ANSI C replacement for
3351 ARRAY_ZERO, and kill the full number of bytes in pools and
3354 * proj.h.in (ARRAY_ZERO, ARRAY_ZERO_SIZE): Remove.
3356 * ste.c (ffeste_io_cilist_, ffeste_io_icilist_, ffeste_R838,
3357 ffeste_R839): Issue diagnostic if a too-narrow variable used in an
3358 ASSIGN context despite changes to this code and code in com.c.
3360 * where.c, where.h: Switch to valid ANSI C replacement for
3363 Fri Feb 17 03:35:19 1995 Craig Burley (burley@gnu.ai.mit.edu)
3365 * Version 0.5.8 released.
3367 * INSTALL: In quick-build case, list g77 target first so g77
3368 gets installed. Also, explain that gcc gets built and installed
3369 as well, even though this isn't really what we want (and maybe
3370 we'll find a way around this someday).
3372 Fri Feb 17 02:35:41 1995 Craig Burley (burley@gnu.ai.mit.edu)
3374 * Version 0.5.7 released.
3376 * Makefile.in (CONFIG_H, HCONFIG_H, TCONFIG_H, TM_H): Remove
3377 ../ prefix in front of .h files, since they're in the cd.
3379 Fri Feb 17 01:50:48 1995 Craig Burley (burley@gnu.ai.mit.edu)
3381 * Version 0.5.6 released.
3383 Thu Feb 16 20:26:54 1995 Craig Burley (burley@gnu.ai.mit.edu)
3385 * ../README.g77: Remove description of g77 as "not-yet-published".
3387 * CREDITS: More changes.
3389 * Make-lang.in (G77STAGESTUFF): Remove cktyps stuff.
3391 * Makefile.in (CONFIG_H, HCONFIG_H, TCONFIG_H, TM_H): Don't
3392 prefix gcc dir with $(srcdir) since these don't live there,
3393 they are created in the build dir by gcc's configure. Add
3394 a note explaining what these macros are about.
3395 Update dependencies via deps-kinda.
3397 * README.NEXTSTEP: Credit Toon, and per his request, add his
3400 * com.h (FFECOM_DETERMINE_TYPES): #include "config.j".
3402 * config.j, convert.j, flags.j, hconfig.j, rtl.j, tconfig.j,
3403 tm.j, tree.j: Don't #include if already done.
3405 * convert.j: #include "tree.j" first, as convert.h clearly depends
3406 on trees being defined.
3408 * rtl.j: #include "config.j" first, since there's some stuff
3409 in rtl.h that assumes it has been #included.
3411 * tree.j: #include "config.j" first, or real.h makes inconsistent
3412 decision about return type of ereal_atof, leading to bugs, and
3413 because tree.h/real.h assume config.h already included.
3415 Wed Feb 15 14:40:20 1995 Craig Burley (burley@gnu.ai.mit.edu)
3417 * Version 0.5.5 released.
3419 * Copyright notices updated to be FSF-style.
3421 * INSTALL: Some more clarification regarding building just f77.
3423 * Make-lang.in (F77_SRCS): Update wrt changing some .h to .j.
3424 (install-libf77): Fix typo in new parenthetical note.
3426 * Makefile.in (f/*.o): Update.
3427 (CONFIG_H, CONVERT_H, FLAGS_H, GLIMITS_H, HCONFIG_H, RTL_H,
3428 TCONFIG_H, TM_H, TREE_H): Update/new symbols.
3429 (deps-kinda): More fixes wrt changing some .h to .j.
3430 Document and explain this rule a bit better.
3431 Accommodate changes in output of gcc -MM.
3433 * *.h, *.c: Change #include's so proj.h not assumed to #include
3434 malloc.h or config.h (now config.j), and so new .j files are
3435 used instead of old .h ones.
3437 * com.c (ffecom_init_0): Use FLOAT_TYPE_SIZE for f2c's
3438 TYLONG/TYLOGICAL type codes, to get g77 working on Alpha.
3440 * com.h: Make all f2c-related integral types "int", not "long
3443 * config.j, convert.j, flags.j, glimits.j, hconfig.j, rtl.j,
3444 tconfig.j, tm.j, tree.j: New files wrapping around gbe
3447 * config.h, convert.h, flags.h, glimits.h, hconfig.h, rtl.h,
3448 tconfig.h, tm.h, tree.h: Deleted so new .j files
3449 can #include the gbe files directly, instead of using "../",
3450 and thus do better with various kinds of builds.
3452 * proj.h: Delete unused NO_STDDEF and related stuff.
3454 Tue Feb 14 08:28:08 1995 Craig Burley (burley@gnu.ai.mit.edu)
3456 * BUGS: Remove item #12, cross-compiling & autoconf scripts
3457 reportedly expected to work properly (according to d.love).
3459 * INSTALL: Add explanation of d.love's patch to config-lang.in.
3460 Add explanation of how to install just g77 when gcc already installed.
3461 Add note about usability of "-Wall". Add note about bug-
3464 * Make-lang.in ($(srcdir)/f/conf-proj): Add comment about why
3466 (install-libf77): Echo parenthetical note to user about how to do
3467 just the (aborted) libf2c installation.
3468 (deps-kinda): Update to work with new configuration/build stuff.
3470 * bad.c (ffebad_finish): Put capitalized "warning:" &c message
3471 as prefix on any diagnostic without pointers into source.
3473 * bad.def (FFEBAD_TOO_BIG_INIT): Add this warning message.
3475 * config-lang.in: Add Dave Love's patch to catch case where
3476 back-end patches not applied and abort configuration.
3478 * data.c (ffedata_gather_, ffedata_value_): Warn when about
3479 to initialize a large aggregate area, due to design flaw resulting
3480 in too much time/space used to handle such cases.
3481 Use COMMON area name, and first notice of symbol, for multiple-
3482 initialization diagnostic, instead of member symbol and unknown
3484 (FFEDATA_sizeTOO_BIG_INIT_): New macro per above.
3486 Mon Feb 13 13:54:26 1995 Dave Love <d.love@dl.ac.uk>
3488 * Make-lang.in (F77_SRCS): Use $(srcdir)/f/proj.h.in, not
3489 $(srcdir)/f/proj.h for build outside srcdir.
3491 Sun Feb 12 13:37:11 1995 Craig Burley (burley@gnu.ai.mit.edu)
3493 * ../README.g77: Clarify procedures for unpacking, add asterisks
3494 to mark important things the user must do.
3496 * Fix dates in/add dates to ../README.g77, BUGS, CREDITS, DOC,
3497 INSTALL, PROJECTS, README.
3499 Sun Feb 12 00:26:10 1995 Craig Burley (burley@gnu.ai.mit.edu)
3501 * Version 0.5.4 released.
3503 * Make-lang.in (f/proj.h): Reproduce this rule here from
3505 ($(srcdir)/f/conf-proj): Put autoconf's stdout in temp file
3506 conf-proj.out, then mv to conf-proj only if successful, so
3507 conf-proj not touched if autoconf not installed.
3509 * Makefile.in ($(srcdir)/conf-proj): See Make-lang.in's similar
3512 Sat Feb 11 20:56:02 1995 Craig Burley (burley@gnu.ai.mit.edu)
3514 * BUGS: Clarify some bugs.
3516 * DOC: Many improvements and fixes.
3518 * README: Move bulk of text, edited, to ../README.g77, and
3519 replace with pointer to that file.
3521 * com.c (ffecom_init_0): Comment out warning about sizeof(ftnlen)
3522 as per ste.c change. Add text about ASSIGN to help user understand
3523 what is being warned about.
3525 * conf-proj.in: Fix typos in comments.
3527 * proj.h.in: Add ARRAY_ZERO_SIZE to parallel malloc.h's version,
3528 in case it proves to be needed.
3530 * ste.c: Comment out assertions requiring sizeof(ftnlen) >=
3531 sizeof(char *), in the hopes that overflow will never happen.
3532 (ffeste_R838): Change assertion to fatal() with at least
3533 partially helpful message.
3535 Sat Feb 11 12:38:00 1995 Craig Burley (burley@gnu.ai.mit.edu)
3537 * com.c (ffecom_vardesc_): Crash if typecode is -1.
3539 * ste.c (ffeste_io_dolio_): Crash if typecode is -1.
3541 Sat Feb 11 09:51:57 1995 Craig Burley (burley@gnu.ai.mit.edu)
3543 * ste.c: In I/O code tests for item arrayness, sort of revert
3544 to much earlier code that tests original exp, but also check
3545 in newer way just in case. Newer way alone treated FOO(1:40)
3546 as an array, not sure why older way alone didn't work, but I
3547 think maybe it was when diagnosed code was involved, and
3548 since there are now checks for error_mark_node, maybe the old
3549 way alone would work. But better to be safe; both original
3550 ffebld exp _and_ the transformed tree must indicate an array
3551 for the size-determination code to be used, else just 1/2 elements
3552 assumed. And this text is for EMACS: (foo at bar).
3554 Fri Feb 10 11:05:50 1995 Craig Burley (burley@gnu.ai.mit.edu)
3556 * ste.c: In many cases, surround statement-expansion code
3557 with ffecom_push_calltemps () and ffecom_pop_calltemps ()
3558 so COMPLEX-returning functions can have temporaries pushed
3559 in "auto-pop" mode and have them auto-popped at the end of
3562 Wed Feb 8 14:35:10 1995 Dave Love <d.love@dl.ac.uk>
3564 * runtime/f2c.h.in (ftnlen, ftnint): Make same size as integer.
3566 * runtime/libI77/err.c (f_init): Thinko in MISSING_FILE_ELEMS
3568 * runtime/libI77/wrtfmt.c (mv_cur): Likewise.
3569 * runtime/libI77/wsfe.c (x_putc): Likewise.
3571 * runtime/libF77/signal_.c (signal_): Return 0 (this is a
3574 * Makefile.in (f/proj.h): Depend on com.h.
3575 * Make-lang.in (include/f2c.h): Likewise (and proj.h).
3576 (install-libf77): Also install f2c.h.
3578 * runtime/libI77/Makefile.in (*.o): Add f2c.h dependency.
3579 * runtime/libF77/Makefile.in: Likewise.
3581 Wed Feb 8 13:56:47 1995 Craig Burley (burley@gnu.ai.mit.edu)
3583 * stc.c (ffestc_R501_item): Don't reset kind/where to NONE when
3584 setting basictype/kindtype info for symbol, or especially
3585 its function/result twin, because kind/where might not be NONE.
3587 Tue Feb 7 14:47:26 1995 Dave Love <d.love@dl.ac.uk>
3589 * Make-lang.in (include/f2c.h:): Set shell variable src more
3590 robustly (independent of whether srcdir is relative or absolute).
3591 * Makefile.in (f/proj.h:): Likewise.
3593 * conf-proj.in: Check need for LAME_ASSERT. Fix indentation in
3594 check for LAME_STDIO (cosmetic only with ANSI C).
3596 * com.h: Extra ...SIZE stuff taken from com.c.
3598 * com.c (FFECOM_DETERMINE_TYPES): Define before including com.h.
3599 (BITS_PER_WORD etc.) Remove and use conditional definitions to com.h.
3601 * runtime/configure.in: #define FFECOM_DETERMINE_TYPES for com.h in
3602 f2c type determination.
3604 * tm.h: Remove (at least pro tem) because of relative path and use
3607 * Make-lang.in (include/f2c.h:): Set shell variable src more
3608 robustly (independent of whether srcdir is relative or absolute).
3609 * Makefile.in (f/proj.h:): Likewise.
3611 Mon Feb 6 19:58:32 1995 Dave Love <d.love@dl.ac.uk>
3613 * g77.c (append_arg): Use K&R declaration for, e.g. SunOS4 build.
3615 Fri Feb 3 20:33:14 1995 Craig Burley (burley@gnu.ai.mit.edu)
3617 * g77.c (main): Treat -l like filename in terms of -x handling.
3618 Rewrite arglist mechanism for ease of maintenance.
3619 Make sure every -lf2c is followed by -lm and vice versa.
3621 * Make-lang.in: Put complete list of sources in F77_SRCS def
3622 so changing a .h file, for example, causes rebuild.
3624 * Makefile.in: Change test for nextstep to m68k-next-nextstep* so
3625 all versions of nextstep on m68k get the necessary flag.
3627 Fri Feb 3 19:10:32 1995 Dave Love <d.love@dl.ac.uk>
3629 * INSTALL: Note about possible conflict with existing libf2c.a and
3632 * Make-lang.in (f77.distclean): Tidy and move deletion of
3633 f/config.cache to mostlyclean.
3634 (install-libf77): Test for $(libdir)/libf2c.* and barf if found
3635 unless F2CLIBOK defined.
3637 * runtime/Makefile.in (all): Change path to include directory (and
3639 (INCLUDES): Remove (unused/misleading).
3640 (distclean): Include f2c.h.
3641 (clean): Include config.cache.
3643 * runtime/libF77/Makefile.in (.SUFFIXES): Correct typo.
3644 (ALL_CFLAGS) Fix up include search path to find f2c.h in top level
3646 (all): Depend on f2c.h.
3647 * runtime/libI77/Makefile.in (.SUFFIXES): Likewise.
3649 Thu Feb 2 17:17:06 1995 Dave Love <d.love@dl.ac.uk>
3651 * INSTALL: Note about --srcdir and GNU make.
3653 * runtime/f2c.h.in (Pad_UDread, ALWAYS_FLUSH): Reomve the #defines
3656 * runtime/configure.in (Pad_UDread, ALWAYS_FLUSH): Define these
3657 here, not in f2c.h as they'r eonly relevant for building.
3658 * runtime/configure: Regenerated.
3660 * config-lang.in: Warn about using GNU make outside source tree
3661 since I can't get Irix5 or SunOS4 makes to work in this case.
3663 * Makefile.in (VPATH): Don't set it here.
3664 (srcdir): Make it the normal `.' (overridden) at top level.
3665 (all.indirect): New dependency `dircheck'.
3667 (dircheck): New target for foolproofing.
3668 (f/proj.h:): Change finding source.
3669 (CONFIG_H): Don't use this as the relative path in the include loses
3670 f builddir != srcdir.
3672 * config.h: Remove per CONFIG_H change above.
3674 * Make-lang.in (F77_FLAGS_TO_PASS): Remove GCC_FOR_TARGET.
3675 (f771:): Pass VPATH, srcdir to sub-make.
3676 (f/Makefile:): New target.
3677 (stmp-int-hdrs): new variable for cheating build.
3678 (f77-runtime:): Alter GCC_FOR_TARGET treatment.
3679 (include/f2c.h f/runtime/Makefile:) Likewise.
3680 (f77-runtime-unsafe:): New (cheating) target.
3682 Thu Feb 2 12:09:51 1995 Craig Burley (burley@gnu.ai.mit.edu)
3684 * BUGS: Update regarding losing EQUIVALENCE members in -g, and
3685 regarding RS/6000 problems in the back end.
3687 * CREDITS: Make some changes as requested.
3689 * com.c (ffecom_member_trunk_): Remove unused static variable.
3690 (ffecom_finish_symbol_transform_): Improve comments.
3691 (ffecom_let_char_): Fix size of temp address-type var.
3692 (ffecom_member_phase2_): Try fixing problem fixed by change
3693 to ffecom_transform_equiv_ (f_m_p2_ function currently not used).
3694 (ffecom_transform_equiv_): Remove def of unused static variable.
3695 Comment-out use of ffecom_member_phase2_, until problems with
3697 (ffecom_push_tempvar): Fix assertion to not crash okay code.
3699 * com.h: Remove old, commented-out code.
3700 Add prototype for warning() in back end.
3702 * ste.c (ffeste_io_dofio_, ffeste_io_dolio_, ffeste_io_douio_,
3703 ffeste_io_icilist_): Check correct type of variable for arrayness.
3705 Sun Jan 29 14:41:42 1995 Dave Love <d.love@dl.ac.uk>
3707 * BUGS: Remove references to my configure bugs; add another.
3709 * runtime/Makefile.in (AR_FLAGS): Provide default value.
3711 * runtime/f2c.h.in (integer, logical): Take typedefs from
3712 F2C_INTEGER configuration parameter again.
3713 (NON_UNIX_STDIO): don't define it.
3715 * runtime/configure.in: Bring type checks for f2c.h in line with
3717 (MISSING_FILE_ELEMS): New variable to determine whether the relevant
3718 elements of the FILE struct exist, independent of NON_UNIX_STDIO.
3719 * runtime/libI77/{err,wrtfmt,wsfe}.c (MISSING_FILE_ELEMS): Use new
3722 * config-lang.in: Comment out more of f2c rules (c.f. Make-lang.in).
3723 (This stuff is relevant iff you gave configure --enable-f2c.)
3724 Create f/runtime directory tree iff not building in source
3727 * Makefile.in (srcdir): Append slash so we get the right value when
3728 not building in the source directory. This is a consequence of not
3729 building the `f' sources in `f'.
3730 (VPATH): Override configure's value for reasons above.
3731 (f/proj.h f/conf-proj): New rules to build proj.h by
3734 * proj.h: Rename to proj.h.in for autoconfiguration.
3735 * proj.h.in: New as above.
3736 * conf-proj conf-proj.in: New files for autoconfiguration.
3738 * Make-lang.in (include/f2c.h f/runtime/Makefile:): Change the order
3739 of setting the sh variables so that the right GCC_FOR_TARGET is
3741 (f77.*clean:) Add products of new configuration files and make sure
3742 all the *clean targets do something (unlike the ones in
3745 * com.h (FFECOM_f2cINTEGER, FFECOM_f2cLOGICAL): Define as long or
3746 int appropriately to ensure sizeof(real) == sizeof(integer).
3748 * PROJECTS: Library section.
3750 * runtime/libI77/endfile.c: Don't #include sys/types.h conditional
3751 on NON_UNIX_STDIO since rawio.h needs size_t.
3752 * runtime/libI77/uio.c: #include <sys/types.h> for size_t if not
3755 Wed Jan 25 03:31:51 1995 Craig Burley (burley@gnu.ai.mit.edu)
3757 * Version 0.5.3 released.
3761 * Make-lang.in: Comment out rules for building f2c itself (f/f2c/).
3765 * com.c (ffecom_init_0): Warn if ftnlen or INTEGER not big enough
3768 * gbe/2.6.2.diff: Update.
3770 Mon Jan 23 17:10:49 1995 Craig Burley (burley@gnu.ai.mit.edu)
3778 Make-lang.in: Remove cktyps stuff.
3779 Makefile.in: Remove cktyps stuff.
3781 * DOC: Add info on changes for 0.5.3.
3783 * bad.c: Put "warning:" &c on diagnostic messages.
3784 Don't output informational messages if warnings disabled.
3786 Thu Jan 19 12:38:13 1995 Craig Burley (burley@gnu.ai.mit.edu)
3788 * g77.c: Avoid putting out useless "-xnone -xf77" pairs so
3789 larger command lines can be accommodated.
3790 Recognize both `-xlang' and `-x lang'.
3791 Recognize `-xnone' and `-x none' to mean what it does, instead
3792 of treating "none" as any other language.
3793 Some minor, slight improvements in the way args are handled
3794 (hopefully for clearer, more maintainable code), including
3795 consistency checks on arg count just in case.
3797 Wed Jan 18 16:41:57 1995 Craig Burley (burley@gnu.ai.mit.edu)
3799 * DOC: Explain -fautomatic better.
3801 * INSTALL: Describe libf2c.a better.
3803 * Make-lang.in, Makefile.in: Build f771 &c with gcc/ as cd instead
3804 of gcc/f/ so debugging info is better (source file tracking).
3805 Add new source file type.c.
3807 * Makefile.in: For nextstep3, link f771 with -segaddr __DATA
3808 6000000. Fix typo. Change deps-kinda target to handle building
3809 from gcc/. Update dependencies.
3811 * bld-op.def, bld.h, bld.c: Remove opBACKEND and all related
3813 Remove consistency tests that cause compiler warnings.
3815 * cktyps.c: Remove all typing checking.
3817 * com-rt.def: Change all rttypeFLOAT_ intrinsics to rttypeDOUBLE_,
3818 to precisely match how they're declared in libf2c.
3820 * com.h, com.c: Revise to more elegantly track related stuff
3821 in the version of f2c.h used to build libf2c.
3823 * com.c: Increase FFECOM_sizeMAXSTACKITEM, and if 0 or undefined
3824 when checked to determine where to put entity, treat as infinite.
3825 Rewrite temporary mechanism to be based on trees instead of
3826 ffeinfo stuff, and make it much simpler. Change interface
3828 Fixes to better track types of things, make appropriate
3829 conversions, etc. E.g. when making an arg for a libf2c
3830 function, make sure it's of the right type (such as ftnlen).
3831 Delete opBACKEND transformation code.
3832 (ffecom_init_0): Smoother initialization of types, especially
3833 paying attention to using consistent rules for making INTEGER,
3834 REAL, DOUBLE PRECISION, etc., and for deciding their "*N"
3835 and kind values that will work across all g77 platforms.
3836 No longer require per-target configuration info in target.h
3837 or config/*/*; use new type module to store size, alignment.
3838 (ffecom_member_phase2): Declare COMMON/EQUIVALENCE group members
3839 so debugger sees them.
3840 (ffecom_finish_progunit): Transform all symbols in program unit,
3841 so -g will show they all exist.
3843 * expr.c (ffeexpr_collapse_substr): Handle strange substring
3846 * info.h, info.c: Provide connection to new type module.
3847 Remove tests that yield compiler warnings.
3849 * intrin.c (ffeintrin_is_intrinsic): Properly handle deleted
3852 * lex.c (ffelex_file_fixed): Remove redundant/buggy code.
3854 * stc.c (ffestc_kindtype_kind_, ffestc_kindtype_star_): Replace
3855 boring switch stmt with simple call to new type module. This
3856 sort of thing is a reason to get up in the morning.
3858 * ste.c: Update to handle new interface for
3859 ffecom_push/pop_tempvar.
3860 Fixes to better track types of things.
3861 Fixes to not crash for certain diagnosed constructs.
3862 (ffeste_begin_iterdo_): Check only constants for overflow to avoid
3863 spurious diagnostics.
3864 Don't convert larger integer (say, INTEGER*8) to canonical integer
3865 for iteration count.
3867 * stw.h: Track DO iteration count temporary variable.
3869 * symbol.c: Remove consistency tests that cause compiler warnings.
3871 * target.c (ffetarget_aggregate_info): Replace big switch with
3872 little call to new type module.
3873 (ffetarget_layout): Remove consistency tests that cause
3875 (ffetarget_convert_character1_typeless): Pick up length of
3876 typeless type from new type module.
3878 * target.h: Crash build if target float bit pattern cannot be
3879 precisely determined.
3880 Remove all the type cruft now determined by ffecom_init_0
3881 at invocation time and maintained in new type module.
3882 Put casts on second arg of all REAL_VALUE_TO_TARGET_DOUBLE
3883 uses so compiler warnings avoided (requires target float bit
3884 pattern to be precisely determined, hence code to crash build).
3886 * top.c: Add inits/terminates for new type module.
3888 * type.h, type.c: New module.
3890 * gbe/2.6.2.diff: Remove all patches to files in gcc/config/
3891 directory and its subdirectories.
3893 Mon Jan 9 19:23:25 1995 Dave Love <d.love@dl.ac.uk>
3895 * com.h (FFECOM_F2C_INTEGER_TYPE_NODE): Define and use instead of
3896 long_integer_type_node where appropriate.
3898 Tue Jan 3 14:56:18 1995 Dave Love <d.love@dl.ac.uk>
3900 * com.h: Make ffecom_f2c_logical_type_node long, consistent with
3903 Fri Dec 2 20:07:37 1994 Dave Love <d.love@dl.ac.uk>
3905 * config-lang.in (stagestuff): Add f2c conditionally.
3906 * Make-lang.in: Add f2c and related targets.
3907 * f2c: Add the directory.
3909 Fri Nov 25 22:17:26 1994 Dave Love <d.love@dl.ac.uk>
3911 * Makefile.in (FLAGS_TO_PASS): pass $(CROSS)
3912 * Make-lang.in: more changes to runtime targets
3914 Thu Nov 24 18:03:21 1994 Dave Love <d.love@dl.ac.uk>
3916 * Makefile.in (FLAGS_TO_PASS): define for sub-makes
3918 * g77.c (main): change f77-cpp-output to f77-cpp-input (.F files)
3920 Wed Nov 23 15:22:53 1994 Dave Love <d.love@dl.ac.uk>
3922 * bad.c (ffebad_finish): kluge to fool emacs19 into finding errors:
3923 add trailing space to <file>:<line>:
3925 Tue Nov 22 11:30:50 1994 Dave Love <d.love@dl.ac.uk>
3927 * runtime/libF77/signal_.c (RETSIGTYPE): added
3929 Mon Nov 21 13:04:13 1994 Dave Love <d.love@dl.ac.uk>
3931 * Makefile.in (compiler): add runtime
3933 * config-lang.in (stagestuff): add libf2c.a to stagestuff
3936 G77STAGESTUFF <- MORESTAGESTUFF
3937 f77-runtime: new target, plus supporting ones
3939 * runtime: add the directory, containing libI77, libF77 and autoconf
3944 * g77.1: minor fixes
3946 Thu Nov 17 15:18:05 1994 Craig Burley (burley@gnu.ai.mit.edu)
3948 * Version 0.5.2 released.
3950 * bad.def: Modify wording of FFEBAD_UNIMPL_STMT to indicate
3951 that it covers a wide array of possible problems (that, someday,
3952 should be handled via separate diagnostics).
3954 * lex.c: Allow $ in identifiers if -fdollar-ok.
3955 * top.c: Support -fdollar-ok.
3956 * top.h: Support -fdollar-ok.
3957 * target.h: Support -fdollar-ok.
3958 * DOC: Describe -fdollar-ok.
3960 * std.c (ffestd_R1229_finish): Fix bug so stand-alone build works.
3961 * ste.c (ffeste_R819A): Fix bug so stand-alone build works.
3963 * Make: Improvements for stand-alone build.
3965 * Makefile.in: Fix copyright text at top of file.
3967 * LINK, SRCS, UNLINK: Removed. Not particularly useful now that
3968 g77 sources live in their own subdirectory.
3970 * g77.c (main): Cast arg to bzero to avoid warning. (This is
3971 identical to Kenner's fix to cp/g++.c.)
3973 * gbe/: New subdirectory, to contain .diff files for various
3974 versions of the GNU CC back end.
3976 * gbe/README: New file.
3977 * gbe/2.6.2.diff: New file.
3979 Tue Nov 8 10:23:10 1994 Dave Love <d.love@dl.ac.uk>
3981 * Make-lang.in: don't install as f77 as well as g77 to avoid
3982 confusion with system's compiler (especially while testing)
3984 * g77.c (main): use -lf2c and -lm; fix sense of test for .f/.F files
3986 Fri Oct 28 09:45:00 1994 Craig Burley (burley@gnu.ai.mit.edu)
3988 * Version 0.5.1 released.
3990 * gcc.c: Invoke f771 instead of f-771.
3992 Fri Oct 28 02:00:44 1994 Craig Burley (burley@gnu.ai.mit.edu)
3994 * Version 0.5.0 released.
3996 Fri Oct 14 15:03:35 1994 Craig Burley (burley@gnu.ai.mit.edu)
3998 * Makefile.in: Handle the Fortran-77 front-end in a subdirectory.
3999 * f-*: Move Fortran-77 front-end to f/*.