-
Notifications
You must be signed in to change notification settings - Fork 84
/
ChangeLog.2
6653 lines (4671 loc) · 245 KB
/
ChangeLog.2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
2000-06-22 Paul D. Smith <[email protected]>
* job.c (start_job_command): Increment commands_started before the
special check for ":" (empty command) to avoid spurious "is up to
date" messages. Also move the test for question_flag after we
expand arguments, and only stop if the expansion provided an
actual command to run, not just whitespace. This fixes PR/1780.
2000-06-21 Paul D. Smith <[email protected]>
* read.c (read_makefile): If we find a semicolon in the target
definition, remember where it was. If the line turns out to be a
target-specific variable, add back the semicolon and everything
after it. Fixes PR/1709.
2000-06-19 Paul D. Smith <[email protected]>
* config.h-vms.template: #define uintmax_t for this system.
* config.ami.template: Ditto.
* config.h.W32.template: Ditto.
* configure.in: We don't use select(2) anymore, so don't bother
checking for it.
* acconfig.h: Ditto.
* acinclude.m4: Ditto.
* file.c (all_secondary): New static global; if 1 it means
.SECONDARY with no prerequisites was seen in the makefile.
(snap_deps): Set it appropriately.
(remove_intermediates): Check it.
(num_intermediates): Remove this global, it's not used anywhere.
(considered): Move this to remake.c and make it static.
* NEWS: Document the change to .SECONDARY.
* make.texinfo (Special Targets): Document the change to .SECONDARY.
* implicit.c (pattern_search): Remove the increment of
num_intermediates; it's not used.
* filedef.h: Remove num_intermediates and considered.
* function.c (handle_function): If the last argument was empty, we
were pretending it didn't exist rather than providing an empty
value. Keep looking until we're past the end, not just at the end.
* implicit.c (pattern_search): Multi-target implicit rules weren't
expanding the "also made" targets correctly if the pattern didn't
contain a slash but the target did; in that case the directory
part wasn't being added back to the stem on the "also made"
targets. Reported by Seth M LaForge <[email protected]>, with
a patch.
2000-06-17 Eli Zaretskii <[email protected]>
* Makefile.DOS.template (DESTDIR, bindir, datadir, libdir)
(infodir, mandir, includedir): Support installation under a
non-default DESTDIR.
* remake.c (f_mtime): Fix the spelling of __MSDOS__.
* configh.DOS.template (HAVE_FDOPEN, HAVE_MKSTEMP): Define.
2000-06-14 Paul D. Smith <[email protected]>
* acinclude.m4 (pds_WITH_GETTEXT): rewrite fp_WITH_GETTEXT and
rename it to avoid confusion. This version is very specific: it
won't accept any gettext that isn't GNU. If the user doesn't
explicitly ask for the included gettext, we look to see if the
system gettext is GNU (testing both the actual libintl library,
and the libintl.h header file). Only if the system gettext is
really GNU gettext will we allow it to be used.
(pds_CHECK_SYSTEM_GETTEXT): A helper function.
2000-06-13 Paul D. Smith <[email protected]>
* gettext.h: If we have libintl.h, use that instead of any of the
contents of gettext.h. We won't check for libintl.h unless we're
using the system gettext.
* function.c (func_word): Clarify error message.
2000-06-10 Paul Eggert <[email protected]>
Support nanosecond resolution on hosts with 64-bit time_t and
uintmax_t (e.g. 64-bit Sparc Solaris), by splitting
FILE_TIMESTAMP into a 30-bit part for nanoseconds, with the
rest for seconds, if FILE_TIMESTAMP is at least 64 bits wide.
* make.h: Always define FILE_TIMESTAMP to be uintmax_t, for
simplicity.
* filedef.h (FILE_TIMESTAMP_HI_RES, FILE_TIMESTAMP_LO_BITS)
(UNKNOWN_MTIME, NONEXISTENT_MTIME, OLD_MTIME)
(ORDINARY_MTIME_MIN, ORDINARY_MTIME_MAX): New macros.
(FILE_TIMESTAMP_STAT_MODTIME): Now takes fname arg. All uses changed.
(FILE_TIMESTAMP_DIV, FILE_TIMESTAMP_MOD)
(FILE_TIMESTAMP_FROM_S_AND_NS): Remove.
(FILE_TIMESTAMP_S, FILE_TIMESTAMP_NS): Use shifts instead of
multiplication and division. Offset the timestamps by
ORDINARY_MTIME_MIN.
(file_timestamp_cons): New decl.
(NEW_MTIME): Now just the maximal timestamp value, as we no longer use
-1 to refer to nonexistent files.
* file.c (snap_deps, print_file): Use NONEXISTENT_MTIME,
UNKNOWN_MTIME, and OLD_MTIME instead of magic constants.
* filedef.h (file_mtime_1): Likewise.
* main.c (main): Likewise.
* remake.c (update_file_1, notice_finished_file, check_dep)
(f_mtime, name_mtime, library_search): Likewise.
* vpath.c (selective_vpath_search): Likewise.
* remake.c (f_mtime): Do not assume that (time_t) -1 equals
NONEXISTENT_MTIME. When futzing with time stamps, adjust by
multiples of 2**30, not 10**9. Do not calculate timestamp
adjustments on DOS unless they are needed.
* commands.c (delete_target): Do not assume that
FILE_TIMESTAMP_S yields -1 for a nonexistent file, as that is
no longer true with the new representation.
* file.c (file_timestamp_cons): New function, replacing
FILE_TIMESTAMP_FROM_S_AND_NS. All uses changed.
(file_timestamp_now): Use FILE_TIMESTAMP_HI_RES instead of 1 <
FILE_TIMESTAMPS_PER_S to determine whether we're using hi-res
timestamps.
(print_file): Print OLD_MTIME values as "very old" instead of
as a timestamp.
2000-05-31 Paul Eggert <[email protected]>
* remake.c (name_mtime): Check for stat failures. Retry if EINTR.
2000-05-24 Paul D. Smith <[email protected]>
* main.c (decode_switches): The "positive_int" switch uses atoi()
which succeeds for any input, and doesn't notice if extra,
non-digit text is after the number. This causes make to mis-parse
command lines like "make -j 5foo" as "make -j5" (ignoring "foo"
completely) instead of "make -j0 5foo" (where "5foo" is a
target). Fix this by checking the value by hand. We could use
strtol() if we were sure of having it; this is the only
questionable use of atoi() I found so we'll just stick with that.
Fixes PR/1716.
* i18n/ja.po, i18n/nl.po, i18n/pt_BR.po: New translation files.
* configure.in (ALL_LINGUAS): Added pt_BR.
2000-05-22 Paul Eggert <[email protected]>
* remake.c (f_mtime): Fix bug when handling future odd
timestamps in the WINDOWS32 case. Do not bother initializing
static var to zero. Simplify code that works around WINDOWS32
and __MSDOS__ time skew brain damage.
2000-05-22 Paul Eggert <[email protected]>
* job.c: Don't include time.h, as make.h already does this.
2000-05-22 Paul Eggert <[email protected]>
* configure.in (AC_CHECK_HEADERS): Add sys/time.h.
(AC_HEADER_TIME): Add.
(clock_gettime): Prefer -lrt to -lposix4, for Solaris 7.
(gettimeofday): Add check for standard version of gettimeofday.
This merges changes written by Paul D. Smith.
* file.c (file_timestamp_now): Use gettimeofday if available
and if clock_gettime does not work. Don't bother with
high-resolution clocks if file timestamps have only one-second
resolution.
* make.h <sys/time.h>: Include, conditionally on the usual
TIME_WITH_SYS_TIME and HAVE_SYS_TIME_H macros. This is needed
for gettimeofday.
2000-05-20 Paul D. Smith <[email protected]>
* read.c (read_makefile): We weren't keeping makefile names around
unless there was a rule defined in them; but now we need to keep
them for variables as well. Forget trying to be fancy: just keep
every makefile name we successfully open.
* remote-cstms.c (start_remote_job_p): Change DB_EXTRA (?) to DB_JOBS.
2000-05-17 Paul Eggert <[email protected]>
* commands.c (chop_commands): Ensure ctype macro args are nonnegative.
* expand.c (variable_expand_string): Likewise.
* function.c (subst_expand, lookup_function, msdos_openpipe):
Likewise.
* job.c (vms_redirect, start_job_command, new_job, child_execute_job,
construct_command_argv_internal, construct_command_argv): Likewise.
* main.c (decode_env_switches, quote_for_env): Likewise.
* misc.c (collapse_continuations, end_of_token, end_of_token_w32,
next_token): Likewise.
* read.c (read_makefile, do_define, conditional_line,
find_char_unquote,get_next_mword): Likewise.
* variable.c (try_variable_definition): Likewise.
* vpath.c (construct_vpath_list): Likewise.
* w32/pathstuff.c (convert_vpath_to_windows32): Likewise.
2000-05-10 Eli Zaretskii <[email protected]>
* main.c (main) [__MSDOS__]: Add SIGFPE to signals we block when
running child programs, to prevent Make from dying on Windows 9X
when the child triggers an FP exception.
2000-05-08 Paul D. Smith <[email protected]>
* dir.c (find_directory) [WINDOWS32]: If we strip a trailing "\"
from the directory name, remember to add it back. The argument
might really be inside a longer string (e.g. %Path%) and if you
don't restore the "\" it'll be truncated permanently. Fixes PR/1722.
Reported by <[email protected]>
2000-05-02 Paul D. Smith <[email protected]>
* job.c (construct_command_argv_internal) [WINDOWS32]: Added "rd"
and "rmdir" to the list of command.com commands.
Reported by Elod Horvath <[email protected]>
2000-04-24 Paul D. Smith <[email protected]>
* i18n/ja.po: New translation file from the Japanese language team.
2000-04-18 Paul D. Smith <[email protected]>
* remake.c (f_mtime): If ar_member_date() returns -1 (the member
doesn't exist), then return (FILE_TIMESTAMP)-1 rather than
returning the timestamp calculated from the value -1. Fixes PR/1696.
Reported by Gilles Bourhis <[email protected]>.
2000-04-17 Paul D. Smith <[email protected]>
* config.h.W32.template: Add LOCALEDIR macro resolving to "".
* w32/subproc/sub_proc.c (process_begin): Remove reference to
debug_flag; change it to a DB() call. Fixes PR/1700.
Reported by Jim Smith <[email protected]>
2000-04-17 Bruno Haible <[email protected]>
* arscan.c [BeOS]: Add replacement for nonexistent <ar.h> from GNU
binutils.
2000-04-11 Paul D. Smith <[email protected]>
* function.c (expand_builtin_function): If no arguments were
provided, just quit early rather than changing each function to
test for this.
(function_table[]): Change the min # of arguments to 0 for all
those functions for which it makes sense (currently everything
that used to take a minimum of 1 argument, except $(call ...)).
Fixes PR/1689.
2000-04-09 Eli Zaretskii <[email protected]>
* README.DOS: Add instructions to install a binary distro.
Mention latest versions of Windows.
2000-04-07 Eli Zaretskii <[email protected]>
* main.c (main): Rename TMP_TEMPLATE into DEFAULT_TMPDIR, and use
it for the directory of the temporary file. If P_tmpdir is
defined, use it in preference to "/tmp/". Try $TMPDIR, $TEMP, and
$TMP in the environment before defaulting to DEFAULT_TMPDIR.
(print_version): Add year 2000 to the Copyright line.
2000-04-04 Paul D. Smith <[email protected]>
* Version 3.79 released.
* make.texinfo: Update documentation with new features for 3.79.
* function.c (func_wordlist): Don't re-order arguments to
wordlist.
2000-04-03 Paul D. Smith <[email protected]>
* remake.c (f_mtime): Archive member timestamps are stored as
time_t, without nanoseconds. But, f_mtime() wants to return
nanosecond info on those systems that support it. So, convert the
return value of ar_member_date() into a FILE_TIMESTAMP, using 0 as
the nanoseconds.
2000-03-28 Paul D. Smith <[email protected]>
* Version 3.78.92 released.
* build.template: Updates for gettext support; some bugs fixed.
2000-03-27 Paul D. Smith <[email protected]>
* config.guess, config.sub: Updated from config CVS archive at
:pserver:[email protected]:/home/cvs as of today.
* read.c (record_files): Check if expanding a static pattern
rule's prerequisite pattern leaves an empty string as the
prerequisite, and issue an error if so. Fixes PR/1670.
(read_makefile): Store the starting linenumber for a rule in
TGTS_STARTED.
(record_waiting_files): Use the TGTS_STARTED value for the file
location passed to record_file() instead of the current
linenumber, so error messages list the line where the target was
defined instead of the line after the end of the rule definition.
* remake.c (start_updating, finish_updating, is_updating): Fix
PR/1671; circular dependencies in double-colon rules are not
diagnosed. These macros set the updating flag in the root
double-colon file instead of the current one, if it's part of a
double-colon list. This solution provided by Tim Magill
<[email protected]>; I just changed the macro names :).
(update_file_1): Call them.
(check_dep): Call them.
The change to not automatically evaluate the $(call ...)
function's arguments breaks recursive use of call. Although using
$(if ...) and $(foreach ...) in $(call ...) macros is important,
the error conditions generated are simply to obscure for me to
feel comfortable with. If a method is devised to get both
working, we'll revisit. For now, remove this change.
* function.c (function_table): Turn on the expand bit for func_call.
(func_call): Don't expand arguments for builtin functions; that
will have already been done.
2000-03-26 Paul D. Smith <[email protected]>
* file.c (remove_intermediates): Never remove targets explicitly
requested on the command-line by checking the cmd_target flag.
Fixed PR/1669.
2000-03-23 Paul Eggert <[email protected]>
* filedef.h (FILE_TIMESTAMP_STAT_MODTIME): Use st_mtime instead of
st_mtim.tv_sec; the latter doesn't work on Unixware.
2000-03-18 Paul D. Smith <[email protected]>
* file.c (file_hash_enter): If we're trying to change a file into
itself, just return. We used to assert this wasn't true, but
someone came up with a weird case involving archives. After
playing with it for a while I decided it was OK to ignore it.
* default.c: Define COFLAGS to empty to avoid spurious warnings.
* filedef.h: Change #if ST_MTIM_NSEC to #ifdef; this is a macro
containing the name of the nsec field, not true/false.
* make.h: Ditto.
Reported by Marco Franzen <[email protected]>.
2000-03-08 Tim Magill <[email protected]>
* remake.c (update_file): Return the exit status of the pruned
file when pruning, not just 0. Fixes PR/1634.
2000-02-24 Paul D. Smith <[email protected]>
* configure.in: Close a minor potential security hole; if you're
reading makefiles from stdin (who does that?) you could run into a
race condition with the temp file using mktemp() or tmpnam(). Add
a check for mkstemp() and fdopen().
* main.c (open_tmpfile): New function to open a temporary file.
If we have mkstemp() (and fdopen()), use that. If not use
mktemp() or tmpnam(). If we have fdopen(), use open() to open the
file O_CREAT|O_EXCL. If not, fall back to normal fopen() (insecure).
(main): Call it.
* job.c (child_execute_job) [VMS]: Call it.
* variable.c (lookup_variable): If we find a variable which is
being expanded, then note it but keep looking through the rest of
the set list to see if we can find one that isn't. If we do,
return that. If we don't, return the original. Fix for PR/1610.
While implementing this I realized that it also solves PR/1380 in
a much more elegant way. I don't know what I was smoking before.
So, remove the hackage surrounding the original fix for that (see
below). Change this function back to lookup_variable and remove
the extra setlist argument.
* variable.h (recursively_expand_setlist): Remove the macro,
rename the prototype, and remove the extra setlist argument.
(lookup_variable): Ditto.
* expand.c (recursively_expand): Rename and remove the extra
setlist argument.
(reference_variable): Use lookup_variable() again.
(allocated_variable_append): Remove the extra setlist argument.
2000-02-21 Paul D. Smith <[email protected]>
* README.template: A few updates.
* i18n/de.po: New version from the German translation team.
2000-02-09 Paul D. Smith <[email protected]>
* Version 3.78.91 released.
2000-02-07 Paul D. Smith <[email protected]>
* read.c (read_makefile): Reset *p2 to ':', not *colonp. If any
filenames contained backslashes the resulting output (without
backslashes) will be shorter, so setting *colonp doesn't change
the right character. Fix for PR/1586.
For += target-specific variables we need to remember which
variable set we found the variable in, so we can start looking
from there in the next iteration (otherwise we might see it again
in recursively_expand and fail!). This is turning into a hack; if
it gets any worse we'll have to rethink this entire algorithm...
implementing expansion of these references separately from the
"normal" expansion, say, instead of using the same codepath.
Actually, it's already "worse enough" :-/.
Fix for PR/1380.
* variable.h (recursively_expand_setlist): Rename
recursively_expand to add a struct variable_set_list argument, and
make a macro for recursively_expand.
(lookup_variable_setlist): Rename lookup_variable to add a struct
variable_set_list argument, and make a macro for lookup_variable.
* expand.c (recursively_expand_setlist): Take an extra struct
variable_set_list argument and pass it to allocated_variable_append().
(reference_variable): Use lookup_variable_setlist() and pass the
returned variable_set_list to recursively_expand_setlist.
(allocated_variable_append): Take an extra setlist argument and
use this as the starting place when searching for the appended
expansion. If it's null, use current_variable_set_list as before.
* variable.c (lookup_variable_setlist): If the LISTP argument is
not nil, set it to the list containing the variable we found.
2000-02-04 Paul D. Smith <[email protected]>
* variable.c (print_variable): Write out filename/linenumber
information for the variable definition if present.
(define_variable_in_set): Store filename information if provided.
(define_variable, define_variable_for_file): Removed.
(try_variable_definition): Use define_variable_loc() to keep
variable definition location information.
* read.c (read_makefile): Keep variable definition location info.
(do_define): Ditto.
(record_target_var): Ditto.
* variable.h (define_variable_in_set): New fileinfo argument.
(define_variable, define_variable_loc, define_variable_for_file):
Declare new macros.
Fix PR/1407:
* filedef.h (struct file): Rename patvar to pat_variables and make
it just a variable_set_list; we need our own copy of the pattern
variable's variable set list here to avoid overwriting the global
one.
* variable.c (initialize_file_variables): Move the instantiation
of the pat_variables pointer here. Only do the search after we're
done reading the makefiles so we don't search too early. If
there's a pat_variables value, set up the variables next ptr.
* expand.c (variable_expand_for_file): Remove the setup of the
pat_variables info; it's done earlier now to ensure the parent's
pattern variables are set up correctly as well.
2000-02-03 Paul D. Smith <[email protected]>
* job.c (sh_chars_dos) [WINDOWS32]: Add "&" as a shell
metacharacter for the W32 DOS shell.
Reported by Warren Jones <[email protected]>.
2000-02-02 Paul D. Smith <[email protected]>
Fixes for the OpenVMS port from Hartmut Becker <[email protected]>
* config.h-vms [VMS]: Define LOCALEDIR to something; needed for
the expansion of bindtextdomain() even though it's a no-op.
* vmsfunctions.c (strcmpi): Remove duplicate definition of strcmpi().
(readdir): Use DB() instead of testing debug_flag.
* dir.c (file_impossible) [VMS]: Search "p" not "name".
* job.c [VMS]: Switch from debug_flag to the new DB macro. Add
some i18n _() macros (even though VMS doesn't yet support it).
* function.c (patsubst_expand): Change "len" to not be unsigned to
avoid type mismatches.
* main.c (main): Declare signame_init() if we're going to call it.
2000-01-29 Eli Zaretskii <[email protected]>
* Makefile.DOS.template: Track changes in Makefile.in
(install-recursive, uninstall-recursive): Add missing targets.
(DESTDIR): Define.
(install-binPROGRAMS, uninstall-binPROGRAMS): Use $(DESTDIR).
* default.c (default_variables) [__MSDOS__]: Define CXX to gpp.
2000-01-27 Paul D. Smith <[email protected]>
* gettext.c: Some warning cleanups, and a fix for systems which
don't define HAVE_ALLOCA (the workaround code was included
twice).
2000-01-26 Paul D. Smith <[email protected]>
* Version 3.78.90 released.
2000-01-25 Paul D. Smith <[email protected]>
Change gettext support to use the simplified version in libit 0.7.
* getopt.c, make.h: Use gettext.h instead of libintl.h.
* ABOUT-NLS, gettext.h, gettext.c: New files from libit 0.7.
Modified to remove some static declarations which aren't defined.
* acconfig.h: Use new gettext #defines.
* acinclude.m4: Add fp_WITH_GETTEXT; remove AM_GNU_GETTEXT.
* configure.in: Call fp_WITH_GETTEXT instead.
* Makefile.am: New gettext stuff. Also force inclusion of glob
files for systems which have LIBC glob.
* i18n/Makefile.am, i18n/.cvsignore: New dir for translation files.
* i18n/de.po, i18n/es.po, i18n/fr.po, i18n/ko.po, i18n/nl.po:
* i18n/pl.po, i18n/ru.po: Import translations already done for
earlier versions of GNU make. Thanks for that work!!
* po/Makefile.in.in, po/POTFILES.in: Removed.
2000-01-23 Paul D. Smith <[email protected]>
* main.c (decode_debug_flags): If debug_flag is set, enable all
debugging levels.
(debug_flag): Resurrect this flag variable.
(switches): Make -d give the old behavior of turning on all
debugging. Change --debug alone to emit basic debugging and take
optional arguments to expand debugging.
* NEWS: Document the new debugging options.
* remake.c (no_rule_error): Remove this function. This tries to
fix a real problem--see the description with the introduction of
this function below. However, the cure is worse than the disease
and this approach won't work.
(remake_file): Put the code from no_rule_error back here.
(update_file_1): Remove call to no_rule_error.
* filedef.h (struct file): Remove mfile_status field.
2000-01-22 Paul D. Smith <[email protected]>
Integrate GNU gettext support.
* configure.in: Add AM_GNU_GETTEXT.
* Makefile.am: Add options for setting LOCALEDIR, -Iintl, etc.
* acinclude.m4: Add gettext autoconf macros.
* acconfig.h: Add new gettext #defines.
* make.h: Include libintl.h. Make sure _() and N_() macros are
declared. Make gettext() an empty macro is NLS is disabled.
* main.c (struct command_switch switches[]): Can't initialize
static data with _() (gettext calls), so use N_() there then use
gettext() directly when printing the strings.
* remake.c (no_rule_error): The string constants can't be static
when initializing _() macros.
* file.c (print_file): Reformat a few strings to work better for
translation.
* po/POTFILES.in, po/Makefile.in.in: New files. Take
Makefile.in.in from the latest GNU tar distribution, as that
version works better than the one that comes with gettext.
* NEWS: Mention i18n ability.
2000-01-21 Paul D. Smith <[email protected]>
Installed patches for the VMS port.
Patches provided by: Hartmut Becker <[email protected]>
* readme.vms, arscan.c, config.h-vms, default.c, dir.c, file.c:
* implicit.c, job.c, make.h, makefile.com, makefile.vms, rule.c:
* variable.c, vmsdir.h, vmsfunctions.c, vmsify.c, glob/glob.c:
* glob/glob.h: Installed patches. See readme.vms for details.
2000-01-14 Andreas Schwab <[email protected]>
* dir.c (read_dirstream): Initialize d_type if it exists.
2000-01-11 Paul D. Smith <[email protected]>
Resolve PR/xxxx: don't automatically evaluate the $(call ...)
function's arguments. While we're here, clean up argument passing
protocol to always use simple nul-terminated strings, instead of
sometimes using offset pointers to mark the end of arguments.
This change also fixes PR/1517.
Reported by Damien GIBOU <[email protected]>.
* function.c (struct function_table_entry): Remove the negative
required_args hack; put in explicit min and max # of arguments.
(function_table): Add in the max value. Turn off the expand bit
for func_call.
(expand_builtin_function): Test against minimum_args instead of
the obsolete required_args.
(handle_function): Rewrite this. We don't try to be fancy and
pass one style of arguments to expanded functions and another
style to non-expanded functions: pass pointers to nul-terminated
strings to all functions.
(func_call): Rewrite this. If we are invoking a builtin function
and it's supposed to have its arguments expanded, do that (since
it's not done by handle_function for $(call ...) anymore). For
non-builtins, just add the variables as before but mark them as
recursive so they'll be expanded later, as needed.
(func_if): All arguments are vanilla nul-terminated strings:
remove trickery with "argv[1]-1".
(func_foreach): Ditto.
* expand.c (expand_argument): If the second arg is NULL, expand
the entire first argument.
* job.c (new_job): Zero the child struct. This change was just
made to keep some heap checking software happy, not because there
was an actual bug (the important memory was being cleared properly).
1999-12-15 Paul D. Smith <[email protected]>
* variable.c (print_variable): Print the variable with += if the
append flag is set.
* implicit.c (pattern_search): Remove the extra check of the
implicit flag added on 8/24/1998. This causes problems and the
reason for the change was better resolved by the change made to
check_deps() on 1998-08-26. This fixes PR/1423.
1999-12-08 Paul D. Smith <[email protected]>
* dir.c (dir_setup_glob): On 64 bit ReliantUNIX (5.44 and above)
in LFS mode, stat() is actually a macro for stat64(). Assignment
doesn't work in that case. So, stat is a macro, make a local
wrapper function to invoke it.
(local_stat): Wrapper function, if needed.
Reported by Andrej Borsenkow <[email protected]>.
1999-12-02 Paul D. Smith <[email protected]>
* remake.c (update_file): Move the considered test outside the
double-colon loop, _but_ make sure we test the double_colon target
not the "current" target. If we stop early because one
double-colon target is running, mark all the rest considered and
try to start their prerequisites (so they're marked considered).
Fix for PR/1476 suggested by Tim Magill <[email protected]>.
1999-11-22 Rob Tulloh <[email protected]>
* function.c (windows32_openpipe, func_shell): Correct Windows32
problem where $(shell nosuchfile) would incorrectly exit make. The
fix is to print the error and let make continue.
Reported by David Masterson <[email protected]>.
* w32/subproc/misc.c (arr2envblk): Memory leak fix.
1999-11-21 Paul D. Smith <[email protected]>
Rework GNU make debugging to provide different levels of output.
* NEWS: mention it.
* debug.h: New file. Define various debugging levels and macros.
* function.c, implicit.c, job.c, main.c, misc.c, read.c, remake.c
* remote-cstms.c, vmsfunctions.c: Replace all code depending on
debug_flag with invocations of debugging macros.
* make.h: Remove debug_flag and DEBUGPR, add db_level.
1999-11-18 Paul Eggert <[email protected]>
* acinclude.m4 (AC_SYS_LARGEFILE_FLAGS): Work around a problem
with the QNX 4.25 shell, which doesn't propagate exit status of
failed commands inside shell assignments.
1999-11-17 Paul D. Smith <[email protected]>
* function.c (func_if): Find the end of the arg list by testing
the next item for NULL; any other test is not correct.
Reported by Graham Reed <[email protected]> (PR/1429).
Fix += when used in a target-specific variable context.
* variable.h: New bitfield APPEND set if we have a +=
target-specific variable.
* variable.c (try_variable_definition): Add an argument to specify
if we're trying a target-specific variable. If we are and it's an
append style, don't append it, record it as normal recursive, but
set the APPEND flag so it'll be expanded later.
* main.c (handle_non_switch_argument): Use new
try_variable_definition() signature.
* read.c (read_makefile,record_target_var): Ditto.
* expand.c (allocated_variable_append): New function: like
allocated_variable_expand(), but we expand the same variable name
in the context of the ``next'' variable set, then we append this
expanded value.
(recursively_expand): Invoke it, if the APPEND bit is set.
1999-11-10 Paul D. Smith <[email protected]>
* file.c (snap_deps): If the .NOTPARALLEL target is defined, turn
off parallel builds for this make only (still allow submakes to be
run in parallel).
* main.c: New variable, ``not_parallel''.
* make.h: Add an extern for it.
* job.c (new_job): Test NOT_PARALLEL as well as JOB_SLOTS.
* NEWS: Add info on .NOTPARALLEL.
* make.texinfo (Special Targets): Document it.
* configure.in (GLOBDIR): Set to "glob" if we need to build the
glob library.
* Makefile.am (SUBDIRS): Use the GLOBDIR variable instead of
"glob" so we don't try to build glob if we don't need to (if we
have GLIBC glob). Reported by Lars Hecking <[email protected]>.
* main.c (main): Don't put "***" in the clock skew warning
message. Reported by [email protected].
* make.h: Remove unneeded signal setup.
* signame.c: Remove extraneous #includes; some versions of Ultrix
don't protect against multiple inclusions and it causes compile
errors. Reported by Simon Burge <[email protected]>.
1999-10-15 Paul D. Smith <[email protected]>
* main.c (quote_for_env): Rename from quote_as_word().
* make.h, *.c: Prefer strchr() and strrchr() in the code
rather than index() and rindex(). Define strchr/strrchr in terms
of index/rindex if the former aren't supported.
* default.c (CHECKOUT,v): Replace the fancy, complicated
patsubst/filter expression with a simple $(if ...) expression.
* main.c (print_usage): Add the bug reporting mailing address to
the --help output, as per the GNU coding standards.
Reported by Paul Eggert <[email protected]>.
* README.customs: Installed information on running Customs-ized
GNU make and setuid root, collected by Ted Stern <[email protected]>.
* read.c (read_all_makefiles): PR/1394: Mark the end of the next
token in the MAKEFILES value string _before_ we dup it.
1999-10-13 Paul D. Smith <[email protected]>
* configure.in (make_cv_sys_gnu_glob): We used to add the -Iglob
flag to CPPFLAGS, but that loses if the user specifies his own
CPPFLAGS; this one gets added _after_ his and if he happens to
have an old or broken glob.h--boom. Instead, put it in GLOBINC
and SUBST it.
* Makefile.am (INCLUDES): Add @GLOBINC@ to the INCLUDES macro;
these things get on the compile line well before the user's
CPPFLAGS.
1999-10-12 Paul D. Smith <[email protected]>
* remake.c (notice_finished_file): If we get here and -n is set,
see if all the command lines are marked recursive. If so, then we
ran every command there is, so check the mtime on this file just
like we would normally. If not, we assume the command we didn't
run would updates the target and set mtime of the target to "very new".
* job.c (start_job_command): Update lines_flags in the file's cmds
structure with any per-line tokens we found (`@', `-', `+').
1999-10-08 Paul D. Smith <[email protected]>
* variable.c (initialize_file_variables): Always recurse to
initialize the parent's file variables: the parent might not have
any rules to run so it might not have been initialized before
this--we need this to set up the chain properly for
target-specific variables.
1999-09-29 Paul Eggert <[email protected]>
* main.c (quote_as_word): Always quote for decode_env_switches
instead of for the shell, so that arguments with strange
characters are are passed to submakes correctly. Remove
double_dollars arg; we always double dollars now. All callers
changed.
(decode_env_switches): Don't run off the end of an environment
variable whose contents ends in a unescaped backslash.
1999-09-23 Paul D. Smith <[email protected]>
* commands.c, function.c, job.c, read.c: Cast arguments to
ctype.h functions/macros to _unsigned_ char for portability.
* remake.c, function.c: Compiler warning fixes: the second
argument to find_next_token() should be an _unsigned_ int*.
Reported by Han-Wen Nienhuys <[email protected]>.
1999-09-23 Paul D. Smith <[email protected]>
* Version 3.78.1 released.
* make.texinfo: Update version/date stamp.
* main.c (main): Argh. For some reason we were closing _all_ the
jobserver pipes before we re-exec'd due to changed makefiles.
This means that any re-exec got a "jobserver unavailable" error :-/.
I can't believe we didn't notice this before.
1999-09-22 Paul D. Smith <[email protected]>
* Version 3.78 released.
* main.c (main): Only fail on multiple --jobserver-fds options if
they aren't all the same. Some makefiles use things like
$(MAKE) $(MFLAGS) which will cause multiple, identical copies of
--jobserver-fds to show up.
1999-09-16 Paul D. Smith <[email protected]>
* main.c (define_makeflags): Zero out FLAGSTRING to avoid
uninitialized memory reads when checking *p != '-' in the loop.
1999-09-15 Paul D. Smith <[email protected]>
* Version 3.77.97 released.
* configure.in (MAKE_HOST): AC_SUBST this so it will go into the
makefile.
* Makefile.am (check-local): Print a success banner if the check
succeeds.
(check-regression): A bit of fine-tuning.
1999-09-15 Eli Zaretskii <[email protected]>
* README.DOS.template: Document requirements for the test suite.
* Makefile.DOS.template: Updates to allow the test suite to run
from "make check".
* main.c (main): Handle it if argv[0] isn't an absolute path.
1999-09-13 Paul D. Smith <[email protected]>
* Version 3.77.96 released.
* Makefile.am (loadavg): Use CPPFLAGS, etc. to make sure we get
all the right #defines to compile.
(check-regression): Look for the regression test suite in the make
package itself. If we're building remotely, use symlinks to make
a local copy.
(dist-hook): Put the test suite into the tar file.
* configure.in: Look for perl for the test suite.
1999-09-10 Paul Eggert <[email protected]>
* acinclude.m4 (AC_SYS_LARGEFILE_FLAGS): If on HP-UX 10.20 or
later, and using GCC, define __STDC_EXT__; this works around a
bug in GCC 2.95.1.
1999-09-08 Paul D. Smith <[email protected]>
* main.c (print_version): Ugh. GLIBC's configure tries to check
make version strings and is too aggressive with their matching
expressions. I've struck a deal with them to leave the version
output as-is for 3.78, and they'll change their configure checks
so that I can change this back in the future.
1999-09-07 Eli Zaretskii <[email protected]>
* job.c (construct_command_argv_internal) [__MSDOS__]: Add "echo"
and "unset" to the list of builtin shell commands.
* configh.DOS.template (MAKE_HOST): Define to "i386-pc-msdosdjgpp"
which is the canonical name of the DJGPP host.
1999-09-05 Paul D. Smith <[email protected]>
* Version 3.77.95 released.
* make.texinfo (Make Errors): Document some new jobserver error
messages.
1999-09-04 Eli Zaretskii <[email protected]>
* make.texinfo (Make Errors): Document the hint about 8 spaces
instead of a TAB.
(Call Function, Quick Reference): Use @code{$(1)}, not @var.
* main.c (main) [__MSDOS__]: Say "on this platform" instead of "on
MS-DOS", since the MSDOS version could run on Windows.
1999-09-03 Paul D. Smith <[email protected]>
* remake.c (notice_finished_file): Always set mtime_before_update
if it's not been set, not just if we ran some rules. Otherwise we
may have a situation where a target's prerequisite was rebuilt but
not changed, so this target's rules weren't run, then
update_goal_chain() sees mtime_before_update != last_mtime and
thinks that the top-level target changed when it really didn't.
This can cause an infinite loop when remaking makefiles.
(update_goal_chain): If we get back to the top and we don't know
what the goal's last_mtime was, find it now. We need to know so
we can compare it to mtime_before_update later (this is only
crucial when remaking makefiles--should we only do it then?)
1999-09-02 Paul D. Smith <[email protected]>
* read.c (read_makefile): If "override" appears as the first
prerequisite, look further to ensure this is really a
target-specific variable definition, and not just some
prerequisite named "override".
1999-09-01 Paul D. Smith <[email protected]>
* function.c (IS_PATHSEP) [WINDOWS32]: Allow backslash separators
for W32 platforms.
* read.c (record_files) [WINDOWS32]: Allow backslash separators
for W32 platforms.
* implicit.c (pattern_search) [WINDOWS32]: Allow backslash
separators for W32 platforms.
* configure.in (MAKE_HOST): Define it to be the canonical build
host info, now that we need AC_CANONICAL_HOST anyway (for large
file support).
* version.c (make_host): Define a variable to MAKE_HOST so we're
sure to get it from the local config.h.
* main.c (print_version): Use it in the version information.
* config.ami.template: Add MAKE_HOST.
* configh.dos.template: Ditto.
* config.h.W32.template: Ditto.
* config.h-vms.template: Ditto.
* main.c (main): Close the jobserver file descriptors if we need
to re-exec ourselves.
Also print more reasonable error if users force -jN for submakes.
This may be common for a while until people use the jobserver
feature. If it happens, we ignore the existing jobserver stuff
and use whatever they specified on the commandline.
(define_makeflags): Fixed a long-standing bug: if a long name
only option comes immediately after a single letter option with no
argument, then the option string is constructed incorrectly. For
example, with -w and --jobserver-fds you get "-w-jobserver-fds..."
instead of "-w --jobserver-fds..."; add in an extra " -".
* make.texinfo (Phony Targets): Add another example of using
.PHONY with subdirectories/recursive make.
1999-08-30 Paul D. Smith <[email protected]>
* README.W32.template: Renamed from README.W32 so it's
autogenerated during the dist. A few minor modifications.
* configure.in: Check for kstat_open before AC_FUNC_GETLOADAVG
since the latter needs to know whether the former exists to give
an accurate result.
1999-08-26 Rob Tulloh <[email protected]>
* NMakefile [WINDOWS32]: Now more robust. If you change a file
under w32/subproc, the make.exe will be relinked. Also added some
tests to make sure erase commands won't fail when executed in a
pristine build environment.
* w32/subproc/sub_proc.c [WINDOWS32]: Added support for
HAVE_CYGWIN_SHELL. If you are using the Cygwin B20.1 release, it
is now possible to have have native support for this shell without
having to rely on klutzy BATCH_MODE_ONLY_SHELL.
* config.h.W32 [WINDOWS32]: Added HAVE_CYGWIN_SHELL macro which
users can define if they want to build make to use this shell.
* README.W32 [WINDOWS32]: Added informaton about
HAVE_CYGWIN_SHELL. Cleaned up text a bit to make it more current.
1999-08-26 Paul Eggert <[email protected]>
Support large files in AIX, HP-UX, and IRIX.
* acinclude.m4 (AC_LFS): Remove. Superseded by AC_SYS_LARGEFILE.
(AC_SYS_LARGEFILE_FLAGS, AC_SYS_LARGEFILE_SPACE_APPEND,
AC_SYS_LARGEFILE_MACRO_VALUE, AC_SYS_LARGEFILE): New macros.
(jm_AC_TYPE_UINTMAX_T): Check for busted compilers that can't
shift or divide unsigned long long.
(AM_PROG_CC_STDC): New macro; a temporary workaround of a bug in
automake 1.4.
* configure.in (AC_CANONICAL_HOST): Add; required by new
AC_SYS_LARGEFILE.
(AC_SYS_LARGEFILE): Renamed from AC_LFS.
(AM_PROG_CC_STDC): Add.
* config.guess, config.sub: New files, needed for AC_CANONICAL_HOST.
1999-08-25 Paul Eggert <[email protected]>
* make.h (CHAR_MAX): New macro.
* main.c (struct command_switch): c is now int,
so that it can store values greater than CHAR_MAX.
(switches): Replace small numbers N with CHAR_MAX+N-1,
to avoid problems with non-ASCII character sets.
(short_option): New macro.
(init_switches, print_usage, define_makeflags): Use it instead of
isalnum.
1999-08-25 Paul D. Smith <[email protected]>
* Version 3.77.94 released.
* main.c (main) [__MSDOS__]: If the user uses -j, warn that it's
not supported and reset it.