• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1/****************************************************************
2Copyright (C) Lucent Technologies 1997
3All Rights Reserved
4
5Permission to use, copy, modify, and distribute this software and
6its documentation for any purpose and without fee is hereby
7granted, provided that the above copyright notice appear in all
8copies and that both that the copyright notice and this
9permission notice and warranty disclaimer appear in supporting
10documentation, and that the name Lucent Technologies or any of
11its entities not be used in advertising or publicity pertaining
12to distribution of the software without specific, written prior
13permission.
14
15LUCENT DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
16INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS.
17IN NO EVENT SHALL LUCENT OR ANY OF ITS ENTITIES BE LIABLE FOR ANY
18SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
19WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER
20IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION,
21ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF
22THIS SOFTWARE.
23****************************************************************/
24
25This file lists all bug fixes, changes, etc., made since the AWK book
26was sent to the printers in August, 1987.
27
28December 8, 2021:
29	The error handling in closefile and closeall was mangled. Long
30	standing warnings had been made fatal and some fatal errors went
31	undetected. Thanks to Miguel Pineiro Jr. <mpj@pineiro.cc>.
32
33Nov 03, 2021:
34        getline accesses uninitialized data after getrec()
35	returns 0 on EOF and leaves the contents of buf unchanged.
36	Thanks to Volodymyr Gubarkov, and Todd C Miller.
37
38Oct 12, 2021:
39	The fix for #83 changed the code to insert 2 chars, but the
40	call to adjbuf just above it only allows for 1 char. This can
41	cause a heap buffer overflow.
42
43July 27, 2021:
44	As per IEEE Std 1003.1-2008, -F "str" is now consistent with
45	-v FS="str" when str is null. Thanks to Warner Losh.
46
47July 24, 2021:
48	Fix readrec's definition of a record. This fixes an issue
49	with NetBSD's RS regular expression support that can cause
50	an infinite read loop. Thanks to Miguel Pineiro Jr.
51
52	Fix regular expression RS ^-anchoring. RS ^-anchoring needs to
53	know if it is reading the first record of a file. This change
54	restores a missing line that was overlooked when porting NetBSD's
55	RS regex functionality. Thanks to Miguel Pineiro Jr.
56
57	Fix size computation in replace_repeat() for special case
58	REPEAT_WITH_Q. Thanks to Todd C. Miller.
59
60February 15, 2021:
61	Small fix so that awk will compile again with g++. Thanks to
62	Arnold Robbins.
63
64January 06, 2021:
65	Fix a decision bug with trailing stuff in lib.c:is_valid_number
66	after recent changes. Thanks to Ozan Yigit.
67
68December 18, 2020:
69	Fix problems converting inf and NaN values in lib.c:is_valid_number.
70	Enhance number to string conversion to do the right thing for
71	NaN and inf values.  Things are now pretty much the same as in
72	gawk.  (Found a gawk bug while we're at it.) Added a torture
73	test for these values.  Thanks to Arnold Robbins.  Allows closing
74	of PR #101.
75
76December 15, 2020:
77	Merge PR #99, which gets the right header for strcasecmp.
78	Thanks to GitHub user michaelforney.
79
80December 8, 2020:
81	Merge PR #98: Disallow hex data. Allow only +nan, -nan,
82	+inf, -inf (case independent) to give NaN and infinity values.
83	Improve things so that string to double conversion is only
84	done once, yielding something of a speedup.  This obviate
85	PR #95. Thanks to Arnold Robbins.
86
87December 3, 2020:
88	Fix to argument parsing to avoid printing spurious newlines.
89	Thanks to Todd Miller. Merges PR #97.
90
91October 13, 2020:
92	Add casts before all the calls to malloc/calloc/realloc in order
93	to get it to compile with g++. Thanks to Arnold Robbins.
94
95August 16, 2020:
96	Additional fixes for DJGPP. Thanks to Eli Zaretskii for
97	the testing.
98
99August 7, 2020:
100	Merge PR #93, which adds casts to (void*) for debug prints
101	using the %p format specifier. Thanks to GitHub user YongHaoWu
102	("Chris") for the fixes.
103
104August 4, 2020:
105	In run.c, use non-restartable multibyte routines to attain
106	portability to DJGPP. Should fix Issue 92. Thanks to Albert Wik
107	for the report and to Todd Miller for the suggested fix.
108
109July 30, 2020:
110	Merge PRs 88-91 which fix small bugs. Thanks to Todd Miller and
111	Tim van der Molen for the fixes.
112
113	In order to make life easier, we move exclusively to bison
114	as the parser generator.
115
116July 2, 2020:
117	Merge PRs 85 and 86 which fix regressions. Thanks to
118	Tim van der Molen for the fixes.
119
120June 25, 2020:
121	Merge PRs 82 and 84. The latter fixes issue #83. Thanks to
122	Todd Miller and awkfan77.
123
124June 12, 2020:
125	Clear errno before calling errcheck to avoid any spurious errors
126	left over from previous calls that may have set it. Thanks to
127	Todd Miller for the fix, from PR #80.
128
129	Fix Issue #78 by allowing \r to follow floating point numbers in
130	lib.c:is_number. Thanks to GitHub user ajcarr for the report
131	and to Arnold Robbins for the fix.
132
133June 5, 2020:
134	In fldbld(), make sure that inputFS is set before trying to
135	use it. Thanks to  Steffen Nurpmeso <steffen@sdaoden.eu>
136	for the report.
137
138May 5, 2020:
139	Fix checks for compilers that can handle noreturn. Thanks to
140	GitHub user enh-google for pointing it out. Closes Issue #79.
141
142April 16, 2020:
143	Handle old compilers that don't support C11 (for noreturn).
144	Thanks to Arnold Robbins.
145
146April 5, 2020:
147	Use <stdnoreturn.h> and noreturn instead of GCC attributes.
148	Thanks to GitHub user awkfan77. Closes PR #77.
149
150February 28, 2020:
151	More cleanups from Christos Zoulas: notably backslash continuation
152	inside strings removes the newline and a fix for RS = "^a".
153	Fix for address sanitizer-found problem. Thanks to GitHub user
154	enh-google.
155
156February 19, 2020:
157	More small cleanups from Christos Zoulas.
158
159February 18, 2020:
160	Additional cleanups from Christos Zoulas. It's no longer necessary
161	to use the -y flag to bison.
162
163February 6, 2020:
164	Additional small cleanups from Christos Zoulas. awk is now
165	a little more robust about reporting I/O errors upon exit.
166
167January 31, 2020:
168	Merge PR #70, which avoids use of variable length arrays. Thanks
169	to GitHub user michaelforney.  Fix issue #60 ({0} in interval
170	expressions doesn't work).  Also get all tests working again.
171	Thanks to Arnold Robbins.
172
173January 24, 2020:
174	A number of small cleanups from Christos Zoulas.  Add the close
175	on exec flag to files/pipes opened for redirection; courtesy of
176	Arnold Robbins.
177
178January 19, 2020:
179	If POSIXLY_CORRECT is set in the environment, then sub and gsub
180	use POSIX rules for multiple backslashes.  This fixes Issue #66,
181	while maintaining backwards compatibility.
182
183January 9, 2020:
184	Input/output errors on closing files are now fatal instead of
185	mere warnings. Thanks to Martijn Dekker <martijn@inlv.org>.
186
187January 5, 2020:
188	Fix a bug in the concatentation of two string constants into
189	one done in the grammar.  Fixes GitHub issue #61.  Thanks
190	to GitHub user awkfan77 for pointing out the direction for
191	the fix.  New test T.concat added to the test suite.
192	Fix a few memory leaks reported by valgrind, as well.
193
194December 27, 2019:
195	Fix a bug whereby a{0,3} could match four a's.  Thanks to
196	"Anonymous AWK fan" for the report.
197
198December 11, 2019:
199	Further printf-related fixes for 32 bit systems.
200	Thanks again to Christos Zoulas.
201
202December 8, 2019:
203	Fix the return value of sprintf("%d") on 32 bit systems.
204	Thanks to Jim Lowe for the report and to Christos Zoulas
205	for the fix.
206
207November 10, 2019:
208	Convert a number of Boolean integer variables into
209	actual bools. Convert compile_time variable into an
210	enum and simplify some of the related code.  Thanks
211	to Arnold Robbins.
212
213November 8, 2019:
214	Fix from Ori Bernstein to get UTF-8 characters instead of
215	bytes when FS = "".  This is currently the only bit of
216	the One True Awk that understands multibyte characters.
217	From Arnold Robbins, apply some cleanups in the test suite.
218
219October 25, 2019:
220	More fixes and cleanups from NetBSD, courtesy of Christos
221	Zoulas. Merges PRs 54 and 55.
222
223October 24, 2019:
224	Import second round of code cleanups from NetBSD. Much thanks
225	to Christos Zoulas (GitHub user zoulasc). Merges PR 53.
226	Add an optimization for string concatenation, also from
227	Christos.
228
229October 17, 2019:
230	Import code cleanups from NetBSD. Much thanks to Christos
231	Zoulas (GitHub user zoulasc). Merges PR 51.
232
233October 6, 2019:
234	Import code from NetBSD awk that implements RS as a regular
235	expression.
236
237September 10, 2019:
238	Fixes for various array / memory overruns found via gcc's
239	-fsanitize=unknown. Thanks to Alexander Richardson (GitHub
240	user arichardson). Merges PRs 47 and 48.
241
242July 28, 2019:
243	Import grammar optimization from NetBSD: Two string constants
244	concatenated together get turned into a single string.
245
246July 26, 2019:
247	Support POSIX-specified C-style escape sequences "\a" (alarm)
248	and "\v" (vertical tab) in command line arguments and regular
249	expressions, further to the support for them in strings added on
250	Apr 9, 1989. These now no longer match as literal "a" and "v"
251	characters (as they don't on other awk implementations).
252	Thanks to Martijn Dekker.
253
254July 17, 2019:
255	Pull in a number of code cleanups and minor fixes from
256	Warner Losh's bsd-ota branch.  The only user visible change
257	is the use of random(3) as the random number generator.
258	Thanks to Warner Losh for collecting all these fixes in
259	one easy place to get them from.
260
261July 16, 2019:
262	Fix field splitting to use FS value as of the time a record
263	was read or assigned to.  Thanks to GitHub user Cody Mello (melloc)
264	for the fix. (Merged from his branch, via PR #42.) Updated
265	testdir/T.split per said PR as well.
266
267June 24, 2019:
268	Extract awktest.tar into testdir directory. Add some very
269	simple mechanics to the makefile for running the tests and
270	for cleaning up. No changes to awk itself.
271
272June 17, 2019:
273	Disallow deleting SYMTAB and its elements, which creates
274	use-after-free bugs. Thanks to GitHub user Cody Mello (melloc)
275	for the fix. (Merged from PR #43.)
276
277June 5, 2019:
278	Allow unmatched right parenthesis in a regular expression to
279	be treated literally. Fixes Issue #40. Thanks to GitHub user
280	Warner Losh (bsdimp) for the report. Thanks to Arnold Robbins
281	for the fix.
282
283May 29,2019:
284	Fix check for command line arguments to no longer require that
285	first character after '=' not be another '='. Reverts change of
286	August 11, 1989. Thanks to GitHub user Jamie Landeg Jones for
287	pointing out the issue; from Issue #38.
288
289Apr 7, 2019:
290	Update awktest.tar(p.50) to use modern options to sort. Needed
291	for Android development. Thanks to GitHub user mohd-akram (Mohamed
292	Akram).  From Issue #33.
293
294Mar 12, 2019:
295	Added very simplistic support for cross-compiling in the
296	makefile.  We are NOT going to go in the direction of the
297	autotools, though.  Thanks to GitHub user nee-san for
298	the basic change. (Merged from PR #34.)
299
300Mar 5, 2019:
301	Added support for POSIX-standard interval expressions (a.k.a.
302	bounds, a.k.a. repetition expressions) in regular expressions,
303	backported (via NetBSD) from Apple awk-24 (20070501).
304	Thanks to Martijn Dekker <martijn@inlv.org> for the port.
305	(Merged from PR #30.)
306
307Mar 3, 2019:
308	Merge PRs as follows:
309	#12: Avoid undefined behaviour when using ctype(3) functions in
310	     relex(). Thanks to GitHub user iamleot.
311	#31: Make getline handle numeric strings, and update FIXES. Thanks
312	     to GitHub user arnoldrobbins.
313	#32: maketab: support build systems with read-only source. Thanks
314	     to GitHub user enh.
315
316Jan 25, 2019:
317	Make getline handle numeric strings properly in all cases.
318	(Thanks, Arnold.)
319
320Jan 21, 2019:
321	Merged a number of small fixes from GitHub pull requests.
322	Thanks to GitHub users Arnold Robbins (arnoldrobbins),
323	Cody Mello (melloc) and Christoph Junghans (junghans).
324	PR numbers: 13-21, 23, 24, 27.
325
326Oct 25, 2018:
327	Added test in maketab.c to prevent generating a proctab entry
328	for YYSTYPE_IS_DEFINED.  It was harmless but some gcc settings
329	generated a warning message.  Thanks to Nan Xiao for report.
330
331Aug 27, 2018:
332	Disallow '$' in printf formats; arguments evaluated in order
333	and printed in order.
334
335	Added some casts to silence warnings on debugging printfs.
336	(Thanks, Arnold.)
337
338Aug 23, 2018:
339        A long list of fixes courtesy of Arnold Robbins,
340        to whom profound thanks.
341
342        1. ofs-rebuild: OFS value used to rebuild the record was incorrect.
343        Fixed August 19, 2014. Revised fix August 2018.
344
345        2. system-status: Instead of a floating-point division by 256, use
346        the wait(2) macros to create a reasonable exit status.
347        Fixed March 12, 2016.
348
349        3. space: Use provided xisblank() function instead of ispace() for
350        matching [[:blank:]].
351
352        4. a-format: Add POSIX standard %a and %A to supported formats. Check
353        at runtime that this format is available.
354
355        5. decr-NF: Decrementing NF did not change $0. This is a decades-old
356        bug. There are interactions with the old and new value of OFS as well.
357        Most of the fix came from the NetBSD awk.
358
359        6. string-conv: String conversions of scalars were sticky.  Once a
360        conversion to string happened, even with OFMT, that value was used until
361        a new numeric value was assigned, even if OFMT differed from CONVFMT,
362        and also if CONVFMT changed.
363
364        7. unary-plus: Unary plus on a string constant returned the string.
365        Instead, it should convert the value to numeric and give that value.
366
367	Also added Arnold's tests for these to awktest.tar as T.arnold.
368
369Aug 15, 2018:
370	fixed mangled awktest.tar (thanks, Arnold), posted all
371	current (very minor) fixes to github / onetrueawk
372
373Jun 7, 2018:
374	(yes, a long layoff)
375	Updated some broken tests (beebe.tar, T.lilly)
376	[thanks to Arnold Robbins]
377
378Mar 26, 2015:
379	buffer overflow in error reporting; thanks to tobias ulmer
380	and john-mark gurney for spotting it and the fix.
381
382Feb 4, 2013:
383	cleaned up a handful of tests that didn't seem to actually
384	test for correct behavior: T.latin1, T.gawk.
385
386Jan 5, 2013:
387	added ,NULL initializer to static Cells in run.c; not really
388	needed but cleaner.  Thanks to Michael Bombardieri.
389
390Dec 20, 2012:
391	fiddled makefile to get correct yacc and bison flags.  pick yacc
392	(linux) or bison (mac) as necessary.
393
394	added  __attribute__((__noreturn__)) to a couple of lines in
395	proto.h, to silence someone's enthusiastic checker.
396
397	fixed obscure call by value bug in split(a[1],a) reported on
398	9fans.  the management of temporary values is just a mess; i
399	took a shortcut by making an extra string copy.  thanks
400	to paul patience and arnold robbins for passing it on and for
401	proposed patches.
402
403	tiny fiddle in setfval to eliminate -0 results in T.expr, which
404	has irritated me for 20+ years.
405
406Aug 10, 2011:
407	another fix to avoid core dump with delete(ARGV); again, many thanks
408	to ruslan ermilov.
409
410Aug 7, 2011:
411	split(s, a, //) now behaves the same as split(s, a, "")
412
413Jun 12, 2011:
414	/pat/, \n /pat/ {...} is now legal, though bad style to use.
415
416	added checks to new -v code that permits -vnospace; thanks to
417	ruslan ermilov for spotting this and providing the patch.
418
419	removed fixed limit on number of open files; thanks to aleksey
420	cheusov and christos zoulos.
421
422	fixed day 1 bug that resurrected deleted elements of ARGV when
423	used as filenames (in lib.c).
424
425	minor type fiddles to make gcc -Wall -pedantic happier (but not
426	totally so); turned on -fno-strict-aliasing in makefile.
427
428May 6, 2011:
429	added #ifdef for isblank.
430	now allows -ffoo as well as -f foo arguments.
431	(thanks, ruslan)
432
433May 1, 2011:
434	after advice from todd miller, kevin lo, ruslan ermilov,
435	and arnold robbins, changed srand() to return the previous
436	seed (which is 1 on the first call of srand).  the seed is
437	an Awkfloat internally though converted to unsigned int to
438	pass to the library srand().  thanks, everyone.
439
440	fixed a subtle (and i hope low-probability) overflow error
441	in fldbld, by adding space for one extra \0.  thanks to
442	robert bassett for spotting this one and providing a fix.
443
444	removed the files related to compilation on windows.  i no
445	longer have anything like a current windows environment, so
446	i can't test any of it.
447
448May 23, 2010:
449	fixed long-standing overflow bug in run.c; many thanks to
450	nelson beebe for spotting it and providing the fix.
451
452	fixed bug that didn't parse -vd=1 properly; thanks to santiago
453	vila for spotting it.
454
455Feb 8, 2010:
456	i give up.  replaced isblank with isspace in b.c; there are
457	no consistent header files.
458
459Nov 26, 2009:
460	fixed a long-standing issue with when FS takes effect.  a
461	change to FS is now noticed immediately for subsequent splits.
462
463	changed the name getline() to awkgetline() to avoid yet another
464	name conflict somewhere.
465
466Feb 11, 2009:
467	temporarily for now defined HAS_ISBLANK, since that seems to
468	be the best way through the thicket.  isblank arrived in C99,
469	but seems to be arriving at different systems at different
470	times.
471
472Oct 8, 2008:
473	fixed typo in b.c that set tmpvec wrongly.  no one had ever
474	run into the problem, apparently.  thanks to alistair crooks.
475
476Oct 23, 2007:
477	minor fix in lib.c: increase inputFS to 100, change malloc
478	for fields to n+1.
479
480	fixed memory fault caused by out of order test in setsval.
481
482	thanks to david o'brien, freebsd, for both fixes.
483
484May 1, 2007:
485	fiddle in makefile to fix for BSD make; thanks to igor sobrado.
486
487Mar 31, 2007:
488	fixed some null pointer refs calling adjbuf.
489
490Feb 21, 2007:
491	fixed a bug in matching the null RE in sub and gsub.  thanks to al aho
492	who actually did the fix (in b.c), and to wolfgang seeberg for finding
493	it and providing a very compact test case.
494
495	fixed quotation in b.c; thanks to Hal Pratt and the Princeton Dante
496	Project.
497
498	removed some no-effect asserts in run.c.
499
500	fiddled maketab.c to not complain about bison-generated values.
501
502	removed the obsolete -V argument; fixed --version to print the
503	version and exit.
504
505	fixed wording and an outright error in the usage message; thanks to igor
506	sobrado and jason mcintyre.
507
508	fixed a bug in -d that caused core dump if no program followed.
509
510Jan 1, 2007:
511	dropped mac.code from makefile; there are few non-MacOSX
512	mac's these days.
513
514Jan 17, 2006:
515	system() not flagged as unsafe in the unadvertised -safe option.
516	found it while enhancing tests before shipping the ;login: article.
517	practice what you preach.
518
519	removed the 9-years-obsolete -mr and -mf flags.
520
521	added -version and --version options.
522
523	core dump on linux with BEGIN {nextfile}, now fixed.
524
525	removed some #ifdef's in run.c and lex.c that appear to no
526	longer be necessary.
527
528Apr 24, 2005:
529	modified lib.c so that values of $0 et al are preserved in the END
530	block, apparently as required by posix.  thanks to havard eidnes
531	for the report and code.
532
533Jan 14, 2005:
534	fixed infinite loop in parsing, originally found by brian tsang.
535	thanks to arnold robbins for a suggestion that started me
536	rethinking it.
537
538Dec 31, 2004:
539	prevent overflow of -f array in main, head off potential error in
540	call of SYNTAX(), test malloc return in lib.c, all with thanks to
541	todd miller.
542
543Dec 22, 2004:
544	cranked up size of NCHARS; coverity thinks it can be overrun with
545	smaller size, and i think that's right.  added some assertions to b.c
546	to catch places where it might overrun.  the RE code is still fragile.
547
548Dec 5, 2004:
549	fixed a couple of overflow problems with ridiculous field numbers:
550	e.g., print $(2^32-1).  thanks to ruslan ermilov, giorgos keramidas
551	and david o'brien at freebsd.org for patches.  this really should
552	be re-done from scratch.
553
554Nov 21, 2004:
555	fixed another 25-year-old RE bug, in split.  it's another failure
556	to (re-)initialize.  thanks to steve fisher for spotting this and
557	providing a good test case.
558
559Nov 22, 2003:
560	fixed a bug in regular expressions that dates (so help me) from 1977;
561	it's been there from the beginning.  an anchored longest match that
562	was longer than the number of states triggered a failure to initialize
563	the machine properly.  many thanks to moinak ghosh for not only finding
564	this one but for providing a fix, in some of the most mysterious
565	code known to man.
566
567	fixed a storage leak in call() that appears to have been there since
568	1983 or so -- a function without an explicit return that assigns a
569	string to a parameter leaked a Cell.  thanks to moinak ghosh for
570	spotting this very subtle one.
571
572Jul 31, 2003:
573	fixed, thanks to andrey chernov and ruslan ermilov, a bug in lex.c
574	that mis-handled the character 255 in input.  (it was being compared
575	to EOF with a signed comparison.)
576
577Jul 29, 2003:
578	fixed (i think) the long-standing botch that included the beginning of
579	line state ^ for RE's in the set of valid characters; this led to a
580	variety of odd problems, including failure to properly match certain
581	regular expressions in non-US locales.  thanks to ruslan for keeping
582	at this one.
583
584Jul 28, 2003:
585	n-th try at getting internationalization right, with thanks to volker
586	kiefel, arnold robbins and ruslan ermilov for advice, though they
587	should not be blamed for the outcome.  according to posix, "."  is the
588	radix character in programs and command line arguments regardless of
589	the locale; otherwise, the locale should prevail for input and output
590	of numbers.  so it's intended to work that way.
591
592	i have rescinded the attempt to use strcoll in expanding shorthands in
593	regular expressions (cclenter).  its properties are much too
594	surprising; for example [a-c] matches aAbBc in locale en_US but abBcC
595	in locale fr_CA.  i can see how this might arise by implementation
596	but i cannot explain it to a human user.  (this behavior can be seen
597	in gawk as well; we're leaning on the same library.)
598
599	the issue appears to be that strcoll is meant for sorting, where
600	merging upper and lower case may make sense (though note that unix
601	sort does not do this by default either).  it is not appropriate
602	for regular expressions, where the goal is to match specific
603	patterns of characters.  in any case, the notations [:lower:], etc.,
604	are available in awk, and they are more likely to work correctly in
605	most locales.
606
607	a moratorium is hereby declared on internationalization changes.
608	i apologize to friends and colleagues in other parts of the world.
609	i would truly like to get this "right", but i don't know what
610	that is, and i do not want to keep making changes until it's clear.
611
612Jul 4, 2003:
613	fixed bug that permitted non-terminated RE, as in "awk /x".
614
615Jun 1, 2003:
616	subtle change to split: if source is empty, number of elems
617	is always 0 and the array is not set.
618
619Mar 21, 2003:
620	added some parens to isblank, in another attempt to make things
621	internationally portable.
622
623Mar 14, 2003:
624	the internationalization changes, somewhat modified, are now
625	reinstated.  in theory awk will now do character comparisons
626	and case conversions in national language, but "." will always
627	be the decimal point separator on input and output regardless
628	of national language.  isblank(){} has an #ifndef.
629
630	this no longer compiles on windows: LC_MESSAGES isn't defined
631	in vc6++.
632
633	fixed subtle behavior in field and record splitting: if FS is
634	a single character and RS is not empty, \n is NOT a separator.
635	this tortuous reading is found in the awk book; behavior now
636	matches gawk and mawk.
637
638Dec 13, 2002:
639	for the moment, the internationalization changes of nov 29 are
640	rolled back -- programs like x = 1.2 don't work in some locales,
641	because the parser is expecting x = 1,2.  until i understand this
642	better, this will have to wait.
643
644Nov 29, 2002:
645	modified b.c (with tiny changes in main and run) to support
646	locales, using strcoll and iswhatever tests for posix character
647	classes.  thanks to ruslan ermilov (ru@freebsd.org) for code.
648	the function isblank doesn't seem to have propagated to any
649	header file near me, so it's there explicitly.  not properly
650	tested on non-ascii character sets by me.
651
652Jun 28, 2002:
653	modified run/format() and tran/getsval() to do a slightly better
654	job on using OFMT for output from print and CONVFMT for other
655	number->string conversions, as promised by posix and done by
656	gawk and mawk.  there are still places where it doesn't work
657	right if CONVFMT is changed; by then the STR attribute of the
658	variable has been irrevocably set.  thanks to arnold robbins for
659	code and examples.
660
661	fixed subtle bug in format that could get core dump.  thanks to
662	Jaromir Dolecek <jdolecek@NetBSD.org> for finding and fixing.
663	minor cleanup in run.c / format() at the same time.
664
665	added some tests for null pointers to debugging printf's, which
666	were never intended for external consumption.  thanks to dave
667	kerns (dkerns@lucent.com) for pointing this out.
668
669	GNU compatibility: an empty regexp matches anything (thanks to
670	dag-erling smorgrav, des@ofug.org).  subject to reversion if
671	this does more harm than good.
672
673	pervasive small changes to make things more const-correct, as
674	reported by gcc's -Wwrite-strings.  as it says in the gcc manual,
675	this may be more nuisance than useful.  provoked by a suggestion
676	and code from arnaud desitter, arnaud@nimbus.geog.ox.ac.uk
677
678	minor documentation changes to note that this now compiles out
679	of the box on Mac OS X.
680
681Feb 10, 2002:
682	changed types in posix chars structure to quiet solaris cc.
683
684Jan 1, 2002:
685	fflush() or fflush("") flushes all files and pipes.
686
687	length(arrayname) returns number of elements; thanks to
688	arnold robbins for suggestion.
689
690	added a makefile.win to make it easier to build on windows.
691	based on dan allen's buildwin.bat.
692
693Nov 16, 2001:
694	added support for posix character class names like [:digit:],
695	which are not exactly shorter than [0-9] and perhaps no more
696	portable.  thanks to dag-erling smorgrav for code.
697
698Feb 16, 2001:
699	removed -m option; no longer needed, and it was actually
700	broken (noted thanks to volker kiefel).
701
702Feb 10, 2001:
703	fixed an appalling bug in gettok: any sequence of digits, +,-, E, e,
704	and period was accepted as a valid number if it started with a period.
705	this would never have happened with the lex version.
706
707	other 1-character botches, now fixed, include a bare $ and a
708	bare " at the end of the input.
709
710Feb 7, 2001:
711	more (const char *) casts in b.c and tran.c to silence warnings.
712
713Nov 15, 2000:
714	fixed a bug introduced in august 1997 that caused expressions
715	like $f[1] to be syntax errors.  thanks to arnold robbins for
716	noticing this and providing a fix.
717
718Oct 30, 2000:
719	fixed some nextfile bugs: not handling all cases.  thanks to
720	arnold robbins for pointing this out.  new regressions added.
721
722	close() is now a function.  it returns whatever the library
723	fclose returns, and -1 for closing a file or pipe that wasn't
724	opened.
725
726Sep 24, 2000:
727	permit \n explicitly in character classes; won't work right
728	if comes in as "[\n]" but ok as /[\n]/, because of multiple
729	processing of \'s.  thanks to arnold robbins.
730
731July 5, 2000:
732	minor fiddles in tran.c to keep compilers happy about uschar.
733	thanks to norman wilson.
734
735May 25, 2000:
736	yet another attempt at making 8-bit input work, with another
737	band-aid in b.c (member()), and some (uschar) casts to head
738	off potential errors in subscripts (like isdigit).  also
739	changed HAT to NCHARS-2.  thanks again to santiago vila.
740
741	changed maketab.c to ignore apparently out of range definitions
742	instead of halting; new freeBSD generates one.  thanks to
743	jon snader <jsnader@ix.netcom.com> for pointing out the problem.
744
745May 2, 2000:
746	fixed an 8-bit problem in b.c by making several char*'s into
747	unsigned char*'s.  not clear i have them all yet.  thanks to
748	Santiago Vila <sanvila@unex.es> for the bug report.
749
750Apr 21, 2000:
751	finally found and fixed a memory leak in function call; it's
752	been there since functions were added ~1983.  thanks to
753	jon bentley for the test case that found it.
754
755	added test in envinit to catch environment "variables" with
756	names beginning with '='; thanks to Berend Hasselman.
757
758Jul 28, 1999:
759	added test in defn() to catch function foo(foo), which
760	otherwise recurses until core dump.  thanks to arnold
761	robbins for noticing this.
762
763Jun 20, 1999:
764	added *bp in gettok in lex.c; appears possible to exit function
765	without terminating the string.  thanks to russ cox.
766
767Jun 2, 1999:
768	added function stdinit() to run to initialize files[] array,
769	in case stdin, etc., are not constants; some compilers care.
770
771May 10, 1999:
772	replaced the ERROR ... FATAL, etc., macros with functions
773	based on vprintf, to avoid problems caused by overrunning
774	fixed-size errbuf array.  thanks to ralph corderoy for the
775	impetus, and for pointing out a string termination bug in
776	qstring as well.
777
778Apr 21, 1999:
779	fixed bug that caused occasional core dumps with commandline
780	variable with value ending in \.  (thanks to nelson beebe for
781	the test case.)
782
783Apr 16, 1999:
784	with code kindly provided by Bruce Lilly, awk now parses
785	/=/ and similar constructs more sensibly in more places.
786	Bruce also provided some helpful test cases.
787
788Apr 5, 1999:
789	changed true/false to True/False in run.c to make it
790	easier to compile with C++.  Added some casts on malloc
791	and realloc to be honest about casts; ditto.  changed
792	ltype int to long in struct rrow to reduce some 64-bit
793	complaints; other changes scattered throughout for the
794	same purpose.  thanks to Nelson Beebe for these portability
795	improvements.
796
797	removed some horrible pointer-int casting in b.c and elsewhere
798	by adding ptoi and itonp to localize the casts, which are
799	all benign.  fixed one incipient bug that showed up on sgi
800	in 64-bit mode.
801
802	reset lineno for new source file; include filename in error
803	message.  also fixed line number error in continuation lines.
804	(thanks to Nelson Beebe for both of these.)
805
806Mar 24, 1999:
807	Nelson Beebe notes that irix 5.3 yacc dies with a bogus
808	error; use a newer version or switch to bison, since sgi
809	is unlikely to fix it.
810
811Mar 5, 1999:
812	changed isnumber to is_number to avoid the problem caused by
813	versions of ctype.h that include the name isnumber.
814
815	distribution now includes a script for building on a Mac,
816	thanks to Dan Allen.
817
818Feb 20, 1999:
819	fixed memory leaks in run.c (call) and tran.c (setfval).
820	thanks to Stephen Nutt for finding these and providing the fixes.
821
822Jan 13, 1999:
823	replaced srand argument by (unsigned int) in run.c;
824	avoids problem on Mac and potentially on Unix & Windows.
825	thanks to Dan Allen.
826
827	added a few (int) casts to silence useless compiler warnings.
828	e.g., errorflag= in run.c jump().
829
830	added proctab.c to the bundle outout; one less thing
831	to have to compile out of the box.
832
833	added calls to _popen and _pclose to the win95 stub for
834	pipes (thanks to Steve Adams for this helpful suggestion).
835	seems to work, though properties are not well understood
836	by me, and it appears that under some circumstances the
837	pipe output is truncated.  Be careful.
838
839Oct 19, 1998:
840	fixed a couple of bugs in getrec: could fail to update $0
841	after a getline var; because inputFS wasn't initialized,
842	could split $0 on every character, a misleading diversion.
843
844	fixed caching bug in makedfa: LRU was actually removing
845	least often used.
846
847	thanks to ross ridge for finding these, and for providing
848	great bug reports.
849
850May 12, 1998:
851	fixed potential bug in readrec: might fail to update record
852	pointer after growing.  thanks to dan levy for spotting this
853	and suggesting the fix.
854
855Mar 12, 1998:
856	added -V to print version number and die.
857
858[notify dave kerns, dkerns@dacsoup.ih.lucent.com]
859
860Feb 11, 1998:
861	subtle silent bug in lex.c: if the program ended with a number
862	longer than 1 digit, part of the input would be pushed back and
863	parsed again because token buffer wasn't terminated right.
864	example:  awk 'length($0) > 10'.  blush.  at least i found it
865	myself.
866
867Aug 31, 1997:
868	s/adelete/awkdelete/: SGI uses this in malloc.h.
869	thanks to nelson beebe for pointing this one out.
870
871Aug 21, 1997:
872	fixed some bugs in sub and gsub when replacement includes \\.
873	this is a dark, horrible corner, but at least now i believe that
874	the behavior is the same as gawk and the intended posix standard.
875	thanks to arnold robbins for advice here.
876
877Aug 9, 1997:
878	somewhat regretfully, replaced the ancient lex-based lexical
879	analyzer with one written in C.  it's longer, generates less code,
880	and more portable; the old one depended too much on mysterious
881	properties of lex that were not preserved in other environments.
882	in theory these recognize the same language.
883
884	now using strtod to test whether a string is a number, instead of
885	the convoluted original function.  should be more portable and
886	reliable if strtod is implemented right.
887
888	removed now-pointless optimization in makefile that tries to avoid
889	recompilation when awkgram.y is changed but symbols are not.
890
891	removed most fixed-size arrays, though a handful remain, some
892	of which are unchecked.  you have been warned.
893
894Aug 4, 1997:
895	with some trepidation, replaced the ancient code that managed
896	fields and $0 in fixed-size arrays with arrays that grow on
897	demand.  there is still some tension between trying to make this
898	run fast and making it clean; not sure it's right yet.
899
900	the ill-conceived -mr and -mf arguments are now useful only
901	for debugging.  previous dynamic string code removed.
902
903	numerous other minor cleanups along the way.
904
905Jul 30, 1997:
906	using code provided by dan levy (to whom profuse thanks), replaced
907	fixed-size arrays and awkward kludges by a fairly uniform mechanism
908	to grow arrays as needed for printf, sub, gsub, etc.
909
910Jul 23, 1997:
911	falling off the end of a function returns "" and 0, not 0.
912	thanks to arnold robbins.
913
914Jun 17, 1997:
915	replaced several fixed-size arrays by dynamically-created ones
916	in run.c; added overflow tests to some previously unchecked cases.
917	getline, toupper, tolower.
918
919	getline code is still broken in that recursive calls may wind
920	up using the same space.  [fixed later]
921
922	increased RECSIZE to 8192 to push problems further over the horizon.
923
924	added \r to \n as input line separator for programs, not data.
925	damn CRLFs.
926
927	modified format() to permit explicit printf("%c", 0) to include
928	a null byte in output.  thanks to ken stailey for the fix.
929
930	added a "-safe" argument that disables file output (print >,
931	print >>), process creation (cmd|getline, print |, system), and
932	access to the environment (ENVIRON).  this is a first approximation
933	to a "safe" version of awk, but don't rely on it too much.  thanks
934	to joan feigenbaum and matt blaze for the inspiration long ago.
935
936Jul 8, 1996:
937	fixed long-standing bug in sub, gsub(/a/, "\\\\&"); thanks to
938	ralph corderoy.
939
940Jun 29, 1996:
941	fixed awful bug in new field splitting; didn't get all the places
942	where input was done.
943
944Jun 28, 1996:
945	changed field-splitting to conform to posix definition: fields are
946	split using the value of FS at the time of input; it used to be
947	the value when the field or NF was first referred to, a much less
948	predictable definition.  thanks to arnold robbins for encouragement
949	to do the right thing.
950
951May 28, 1996:
952	fixed appalling but apparently unimportant bug in parsing octal
953	numbers in reg exprs.
954
955	explicit hex in reg exprs now limited to 2 chars: \xa, \xaa.
956
957May 27, 1996:
958	cleaned up some declarations so gcc -Wall is now almost silent.
959
960	makefile now includes backup copies of ytab.c and lexyy.c in case
961	one makes before looking; it also avoids recreating lexyy.c unless
962	really needed.
963
964	s/aprintf/awkprint, s/asprintf/awksprintf/ to avoid some name clashes
965	with unwisely-written header files.
966
967	thanks to jeffrey friedl for several of these.
968
969May 26, 1996:
970	an attempt to rationalize the (unsigned) char issue.  almost all
971	instances of unsigned char have been removed; the handful of places
972	in b.c where chars are used as table indices have been hand-crafted.
973	added some latin-1 tests to the regression, but i'm not confident;
974	none of my compilers seem to care much.  thanks to nelson beebe for
975	pointing out some others that do care.
976
977May 2, 1996:
978	removed all register declarations.
979
980	enhanced split(), as in gawk, etc:  split(s, a, "") splits s into
981	a[1]...a[length(s)] with each character a single element.
982
983	made the same changes for field-splitting if FS is "".
984
985	added nextfile, as in gawk: causes immediate advance to next
986	input file. (thanks to arnold robbins for inspiration and code).
987
988	small fixes to regexpr code:  can now handle []], [[], and
989	variants;  [] is now a syntax error, rather than matching
990	everything;  [z-a] is now empty, not z.  far from complete
991	or correct, however.  (thanks to jeffrey friedl for pointing out
992	some awful behaviors.)
993
994Apr 29, 1996:
995	replaced uchar by uschar everywhere; apparently some compilers
996	usurp this name and this causes conflicts.
997
998	fixed call to time in run.c (bltin); arg is time_t *.
999
1000	replaced horrible pointer/long punning in b.c by a legitimate
1001	union.  should be safer on 64-bit machines and cleaner everywhere.
1002	(thanks to nelson beebe for pointing out some of these problems.)
1003
1004	replaced nested comments by #if 0...#endif in run.c, lib.c.
1005
1006	removed getsval, setsval, execute macros from run.c and lib.c.
1007	machines are 100x faster than they were when these macros were
1008	first used.
1009
1010	revised filenames: awk.g.y => awkgram.y, awk.lx.l => awklex.l,
1011	y.tab.[ch] => ytab.[ch], lex.yy.c => lexyy.c, all in the aid of
1012	portability to nameless systems.
1013
1014	"make bundle" now includes yacc and lex output files for recipients
1015	who don't have yacc or lex.
1016
1017Aug 15, 1995:
1018	initialized Cells in setsymtab more carefully; some fields
1019	were not set.  (thanks to purify, all of whose complaints i
1020	think i now understand.)
1021
1022	fixed at least one error in gsub that looked at -1-th element
1023	of an array when substituting for a null match (e.g., $).
1024
1025	delete arrayname is now legal; it clears the elements but leaves
1026	the array, which may not be the right behavior.
1027
1028	modified makefile: my current make can't cope with the test used
1029	to avoid unnecessary yacc invocations.
1030
1031Jul 17, 1995:
1032	added dynamically growing strings to awk.lx.l and b.c
1033	to permit regular expressions to be much bigger.
1034	the state arrays can still overflow.
1035
1036Aug 24, 1994:
1037	detect duplicate arguments in function definitions (mdm).
1038
1039May 11, 1994:
1040	trivial fix to printf to limit string size in sub().
1041
1042Apr 22, 1994:
1043	fixed yet another subtle self-assignment problem:
1044	$1 = $2; $1 = $1 clobbered $1.
1045
1046	Regression tests now use private echo, to avoid quoting problems.
1047
1048Feb 2, 1994:
1049	changed error() to print line number as %d, not %g.
1050
1051Jul 23, 1993:
1052	cosmetic changes: increased sizes of some arrays,
1053	reworded some error messages.
1054
1055	added CONVFMT as in posix (just replaced OFMT in getsval)
1056
1057	FILENAME is now "" until the first thing that causes a file
1058	to be opened.
1059
1060Nov 28, 1992:
1061	deleted yyunput and yyoutput from proto.h;
1062	different versions of lex give these different declarations.
1063
1064May 31, 1992:
1065	added -mr N and -mf N options: more record and fields.
1066	these really ought to adjust automatically.
1067
1068	cleaned up some error messages; "out of space" now means
1069	malloc returned NULL in all cases.
1070
1071	changed rehash so that if it runs out, it just returns;
1072	things will continue to run slow, but maybe a bit longer.
1073
1074Apr 24, 1992:
1075	remove redundant close of stdin when using -f -.
1076
1077	got rid of core dump with -d; awk -d just prints date.
1078
1079Apr 12, 1992:
1080	added explicit check for /dev/std(in,out,err) in redirection.
1081	unlike gawk, no /dev/fd/n yet.
1082
1083	added (file/pipe) builtin.  hard to test satisfactorily.
1084	not posix.
1085
1086Feb 20, 1992:
1087	recompile after abortive changes;  should be unchanged.
1088
1089Dec 2, 1991:
1090	die-casting time:  converted to ansi C, installed that.
1091
1092Nov 30, 1991:
1093	fixed storage leak in freefa, failing to recover [N]CCL.
1094	thanks to Bill Jones (jones@cs.usask.ca)
1095
1096Nov 19, 1991:
1097	use RAND_MAX instead of literal in builtin().
1098
1099Nov 12, 1991:
1100	cranked up some fixed-size arrays in b.c, and added a test for
1101	overflow in penter.  thanks to mark larsen.
1102
1103Sep 24, 1991:
1104	increased buffer in gsub.  a very crude fix to a general problem.
1105	and again on Sep 26.
1106
1107Aug 18, 1991:
1108	enforce variable name syntax for commandline variables: has to
1109	start with letter or _.
1110
1111Jul 27, 1991:
1112	allow newline after ; in for statements.
1113
1114Jul 21, 1991:
1115	fixed so that in self-assignment like $1=$1, side effects
1116	like recomputing $0 take place.  (this is getting subtle.)
1117
1118Jun 30, 1991:
1119	better test for detecting too-long output record.
1120
1121Jun 2, 1991:
1122	better defense against very long printf strings.
1123	made break and continue illegal outside of loops.
1124
1125May 13, 1991:
1126	removed extra arg on gettemp, tempfree.  minor error message rewording.
1127
1128May 6, 1991:
1129	fixed silly bug in hex parsing in hexstr().
1130	removed an apparently unnecessary test in isnumber().
1131	warn about weird printf conversions.
1132	fixed unchecked array overwrite in relex().
1133
1134	changed for (i in array) to access elements in sorted order.
1135	then unchanged it -- it really does run slower in too many cases.
1136	left the code in place, commented out.
1137
1138Feb 10, 1991:
1139	check error status on all writes, to avoid banging on full disks.
1140
1141Jan 28, 1991:
1142	awk -f - reads the program from stdin.
1143
1144Jan 11, 1991:
1145	failed to set numeric state on $0 in cmd|getline context in run.c.
1146
1147Nov 2, 1990:
1148	fixed sleazy test for integrality in getsval;  use modf.
1149
1150Oct 29, 1990:
1151	fixed sleazy buggy code in lib.c that looked (incorrectly) for
1152	too long input lines.
1153
1154Oct 14, 1990:
1155	fixed the bug on p. 198 in which it couldn't deduce that an
1156	argument was an array in some contexts.  replaced the error
1157	message in intest() by code that damn well makes it an array.
1158
1159Oct 8, 1990:
1160	fixed horrible bug:  types and values were not preserved in
1161	some kinds of self-assignment. (in assign().)
1162
1163Aug 24, 1990:
1164	changed NCHARS to 256 to handle 8-bit characters in strings
1165	presented to match(), etc.
1166
1167Jun 26, 1990:
1168	changed struct rrow (awk.h) to use long instead of int for lval,
1169	since cfoll() stores a pointer in it.  now works better when int's
1170	are smaller than pointers!
1171
1172May 6, 1990:
1173	AVA fixed the grammar so that ! is uniformly of the same precedence as
1174	unary + and -.  This renders illegal some constructs like !x=y, which
1175	now has to be parenthesized as !(x=y), and makes others work properly:
1176	!x+y is (!x)+y, and x!y is x !y, not two pattern-action statements.
1177	(These problems were pointed out by Bob Lenk of Posix.)
1178
1179	Added \x to regular expressions (already in strings).
1180	Limited octal to octal digits; \8 and \9 are not octal.
1181	Centralized the code for parsing escapes in regular expressions.
1182	Added a bunch of tests to T.re and T.sub to verify some of this.
1183
1184Feb 9, 1990:
1185	fixed null pointer dereference bug in main.c:  -F[nothing].  sigh.
1186
1187	restored srand behavior:  it returns the current seed.
1188
1189Jan 18, 1990:
1190	srand now returns previous seed value (0 to start).
1191
1192Jan 5, 1990:
1193	fix potential problem in tran.c -- something was freed,
1194	then used in freesymtab.
1195
1196Oct 18, 1989:
1197	another try to get the max number of open files set with
1198	relatively machine-independent code.
1199
1200	small fix to input() in case of multiple reads after EOF.
1201
1202Oct 11, 1989:
1203	FILENAME is now defined in the BEGIN block -- too many old
1204	programs broke.
1205
1206	"-" means stdin in getline as well as on the commandline.
1207
1208	added a bunch of casts to the code to tell the truth about
1209	char * vs. unsigned char *, a right royal pain.  added a
1210	setlocale call to the front of main, though probably no one
1211	has it usefully implemented yet.
1212
1213Aug 24, 1989:
1214	removed redundant relational tests against nullnode if parse
1215	tree already had a relational at that point.
1216
1217Aug 11, 1989:
1218	fixed bug:  commandline variable assignment has to look like
1219	var=something.  (consider the man page for =, in file =.1)
1220
1221	changed number of arguments to functions to static arrays
1222	to avoid repeated malloc calls.
1223
1224Aug 2, 1989:
1225	restored -F (space) separator
1226
1227Jul 30, 1989:
1228	added -v x=1 y=2 ... for immediate commandline variable assignment;
1229	done before the BEGIN block for sure.  they have to precede the
1230	program if the program is on the commandline.
1231	Modified Aug 2 to require a separate -v for each assignment.
1232
1233Jul 10, 1989:
1234	fixed ref-thru-zero bug in environment code in tran.c
1235
1236Jun 23, 1989:
1237	add newline to usage message.
1238
1239Jun 14, 1989:
1240	added some missing ansi printf conversion letters: %i %X %E %G.
1241	no sensible meaning for h or L, so they may not do what one expects.
1242
1243	made %* conversions work.
1244
1245	changed x^y so that if n is a positive integer, it's done
1246	by explicit multiplication, thus achieving maximum accuracy.
1247	(this should be done by pow() but it seems not to be locally.)
1248	done to x ^= y as well.
1249
1250Jun 4, 1989:
1251	ENVIRON array contains environment: if shell variable V=thing,
1252		ENVIRON["V"] is "thing"
1253
1254	multiple -f arguments permitted.  error reporting is naive.
1255	(they were permitted before, but only the last was used.)
1256
1257	fixed a really stupid botch in the debugging macro dprintf
1258
1259	fixed order of evaluation of commandline assignments to match
1260	what the book claims:  an argument of the form x=e is evaluated
1261	at the time it would have been opened if it were a filename (p 63).
1262	this invalidates the suggested answer to ex 4-1 (p 195).
1263
1264	removed some code that permitted -F (space) fieldseparator,
1265	since it didn't quite work right anyway.  (restored aug 2)
1266
1267Apr 27, 1989:
1268	Line number now accumulated correctly for comment lines.
1269
1270Apr 26, 1989:
1271	Debugging output now includes a version date,
1272	if one compiles it into the source each time.
1273
1274Apr 9, 1989:
1275	Changed grammar to prohibit constants as 3rd arg of sub and gsub;
1276	prevents class of overwriting-a-constant errors.  (Last one?)
1277	This invalidates the "banana" example on page 43 of the book.
1278
1279	Added \a ("alert"), \v (vertical tab), \xhhh (hexadecimal),
1280	as in ANSI, for strings.  Rescinded the sloppiness that permitted
1281	non-octal digits in \ooo.  Warning:  not all compilers and libraries
1282	will be able to deal with \x correctly.
1283
1284Jan 9, 1989:
1285	Fixed bug that caused tempcell list to contain a duplicate.
1286	The fix is kludgy.
1287
1288Dec 17, 1988:
1289	Catches some more commandline errors in main.
1290	Removed redundant decl of modf in run.c (confuses some compilers).
1291	Warning:  there's no single declaration of malloc, etc., in awk.h
1292	that seems to satisfy all compilers.
1293
1294Dec 7, 1988:
1295	Added a bit of code to error printing to avoid printing nulls.
1296	(Not clear that it actually would.)
1297
1298Nov 27, 1988:
1299	With fear and trembling, modified the grammar to permit
1300	multiple pattern-action statements on one line without
1301	an explicit separator.  By definition, this capitulation
1302	to the ghost of ancient implementations remains undefined
1303	and thus subject to change without notice or apology.
1304	DO NOT COUNT ON IT.
1305
1306Oct 30, 1988:
1307	Fixed bug in call() that failed to recover storage.
1308
1309	A warning is now generated if there are more arguments
1310	in the call than in the definition (in lieu of fixing
1311	another storage leak).
1312
1313Oct 20, 1988:
1314	Fixed %c:  if expr is numeric, use numeric value;
1315	otherwise print 1st char of string value.  still
1316	doesn't work if the value is 0 -- won't print \0.
1317
1318	Added a few more checks for running out of malloc.
1319
1320Oct 12, 1988:
1321	Fixed bug in call() that freed local arrays twice.
1322
1323	Fixed to handle deletion of non-existent array right;
1324	complains about attempt to delete non-array element.
1325
1326Sep 30, 1988:
1327	Now guarantees to evaluate all arguments of built-in
1328	functions, as in C;  the appearance is that arguments
1329	are evaluated before the function is called.  Places
1330	affected are sub (gsub was ok), substr, printf, and
1331	all the built-in arithmetic functions in bltin().
1332	A warning is generated if a bltin() is called with
1333	the wrong number of arguments.
1334
1335	This requires changing makeprof on p167 of the book.
1336
1337Aug 23, 1988:
1338	setting FILENAME in BEGIN caused core dump, apparently
1339	because it was freeing space not allocated by malloc.
1340
1341July 24, 1988:
1342	fixed egregious error in toupper/tolower functions.
1343	still subject to rescinding, however.
1344
1345July 2, 1988:
1346	flush stdout before opening file or pipe
1347
1348July 2, 1988:
1349	performance bug in b.c/cgoto(): not freeing some sets of states.
1350	partial fix only right now, and the number of states increased
1351	to make it less obvious.
1352
1353June 1, 1988:
1354	check error status on close
1355
1356May 28, 1988:
1357	srand returns seed value it's using.
1358	see 1/18/90
1359
1360May 22, 1988:
1361	Removed limit on depth of function calls.
1362
1363May 10, 1988:
1364	Fixed lib.c to permit _ in commandline variable names.
1365
1366Mar 25, 1988:
1367	main.c fixed to recognize -- as terminator of command-
1368	line options.  Illegal options flagged.
1369	Error reporting slightly cleaned up.
1370
1371Dec 2, 1987:
1372	Newer C compilers apply a strict scope rule to extern
1373	declarations within functions.  Two extern declarations in
1374	lib.c and tran.c have been moved to obviate this problem.
1375
1376Oct xx, 1987:
1377	Reluctantly added toupper and tolower functions.
1378	Subject to rescinding without notice.
1379
1380Sep 17, 1987:
1381	Error-message printer had printf(s) instead of
1382	printf("%s",s);  got core dumps when the message
1383	included a %.
1384
1385Sep 12, 1987:
1386	Very long printf strings caused core dump;
1387	fixed aprintf, asprintf, format to catch them.
1388	Can still get a core dump in printf itself.
1389
1390
1391