Import awk-20110810.
[dragonfly.git] / contrib / awk / FIXES
CommitLineData
e90ffc14
SK
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
b12bae18
SW
28Aug 10, 2011:
29 another fix to avoid core dump with delete(ARGV); again, many thanks
30 to ruslan ermilov.
31
32Aug 7, 2011:
33 split(s, a, //) now behaves the same as split(s, a, "")
34
35Jun 12, 2011:
36 /pat/, \n /pat/ {...} is now legal, though bad style to use.
37
38 added checks to new -v code that permits -vnospace; thanks to
39 ruslan ermilov for spotting this and providing the patch.
40
41 removed fixed limit on number of open files; thanks to aleksey
42 cheusov and christos zoulos.
43
44 fixed day 1 bug that resurrected deleted elements of ARGV when
45 used as filenames (in lib.c).
46
47 minor type fiddles to make gcc -Wall -pedantic happier (but not
48 totally so); turned on -fno-strict-aliasing in makefile.
49
0020174d
PA
50May 6, 2011:
51 added #ifdef for isblank.
52 now allows -ffoo as well as -f foo arguments.
53 (thanks, ruslan)
54
55May 1, 2011:
56 after advice from todd miller, kevin lo, ruslan ermilov,
57 and arnold robbins, changed srand() to return the previous
58 seed (which is 1 on the first call of srand). the seed is
59 an Awkfloat internally though converted to unsigned int to
60 pass to the library srand(). thanks, everyone.
61
62 fixed a subtle (and i hope low-probability) overflow error
63 in fldbld, by adding space for one extra \0. thanks to
64 robert bassett for spotting this one and providing a fix.
65
66 removed the files related to compilation on windows. i no
67 longer have anything like a current windows environment, so
68 i can't test any of it.
69
e90ffc14
SK
70May 23, 2010:
71 fixed long-standing overflow bug in run.c; many thanks to
72 nelson beebe for spotting it and providing the fix.
73
74 fixed bug that didn't parse -vd=1 properly; thanks to santiago
75 vila for spotting it.
76
77Feb 8, 2010:
78 i give up. replaced isblank with isspace in b.c; there are
79 no consistent header files.
80
81Nov 26, 2009:
82 fixed a long-standing issue with when FS takes effect. a
83 change to FS is now noticed immediately for subsequent splits.
84
85 changed the name getline() to awkgetline() to avoid yet another
86 name conflict somewhere.
87
88Feb 11, 2009:
89 temporarily for now defined HAS_ISBLANK, since that seems to
90 be the best way through the thicket. isblank arrived in C99,
91 but seems to be arriving at different systems at different
92 times.
93
94Oct 8, 2008:
95 fixed typo in b.c that set tmpvec wrongly. no one had ever
96 run into the problem, apparently. thanks to alistair crooks.
97
98Oct 23, 2007:
99 minor fix in lib.c: increase inputFS to 100, change malloc
100 for fields to n+1.
101
102 fixed memory fault caused by out of order test in setsval.
103
104 thanks to david o'brien, freebsd, for both fixes.
105
106May 1, 2007:
107 fiddle in makefile to fix for BSD make; thanks to igor sobrado.
108
109Mar 31, 2007:
110 fixed some null pointer refs calling adjbuf.
111
112Feb 21, 2007:
113 fixed a bug in matching the null RE in sub and gsub. thanks to al aho
114 who actually did the fix (in b.c), and to wolfgang seeberg for finding
115 it and providing a very compact test case.
116
117 fixed quotation in b.c; thanks to Hal Pratt and the Princeton Dante
118 Project.
119
120 removed some no-effect asserts in run.c.
121
122 fiddled maketab.c to not complain about bison-generated values.
123
124 removed the obsolete -V argument; fixed --version to print the
125 version and exit.
126
127 fixed wording and an outright error in the usage message; thanks to igor
128 sobrado and jason mcintyre.
129
130 fixed a bug in -d that caused core dump if no program followed.
131
132Jan 1, 2007:
133 dropped mac.code from makefile; there are few non-MacOSX
134 mac's these days.
135
136Jan 17, 2006:
137 system() not flagged as unsafe in the unadvertised -safe option.
138 found it while enhancing tests before shipping the ;login: article.
139 practice what you preach.
140
141 removed the 9-years-obsolete -mr and -mf flags.
142
143 added -version and --version options.
144
145 core dump on linux with BEGIN {nextfile}, now fixed.
146
147 removed some #ifdef's in run.c and lex.c that appear to no
148 longer be necessary.
149
150Apr 24, 2005:
151 modified lib.c so that values of $0 et al are preserved in the END
152 block, apparently as required by posix. thanks to havard eidnes
153 for the report and code.
154
155Jan 14, 2005:
156 fixed infinite loop in parsing, originally found by brian tsang.
157 thanks to arnold robbins for a suggestion that started me
158 rethinking it.
159
160Dec 31, 2004:
161 prevent overflow of -f array in main, head off potential error in
162 call of SYNTAX(), test malloc return in lib.c, all with thanks to
163 todd miller.
164
165Dec 22, 2004:
166 cranked up size of NCHARS; coverity thinks it can be overrun with
167 smaller size, and i think that's right. added some assertions to b.c
168 to catch places where it might overrun. the RE code is still fragile.
169
170Dec 5, 2004:
171 fixed a couple of overflow problems with ridiculous field numbers:
172 e.g., print $(2^32-1). thanks to ruslan ermilov, giorgos keramidas
173 and david o'brien at freebsd.org for patches. this really should
174 be re-done from scratch.
175
176Nov 21, 2004:
177 fixed another 25-year-old RE bug, in split. it's another failure
178 to (re-)initialize. thanks to steve fisher for spotting this and
179 providing a good test case.
180
181Nov 22, 2003:
182 fixed a bug in regular expressions that dates (so help me) from 1977;
183 it's been there from the beginning. an anchored longest match that
184 was longer than the number of states triggered a failure to initialize
185 the machine properly. many thanks to moinak ghosh for not only finding
186 this one but for providing a fix, in some of the most mysterious
187 code known to man.
188
189 fixed a storage leak in call() that appears to have been there since
190 1983 or so -- a function without an explicit return that assigns a
191 string to a parameter leaked a Cell. thanks to moinak ghosh for
192 spotting this very subtle one.
193
194Jul 31, 2003:
195 fixed, thanks to andrey chernov and ruslan ermilov, a bug in lex.c
196 that mis-handled the character 255 in input. (it was being compared
197 to EOF with a signed comparison.)
198
199Jul 29, 2003:
200 fixed (i think) the long-standing botch that included the beginning of
201 line state ^ for RE's in the set of valid characters; this led to a
202 variety of odd problems, including failure to properly match certain
203 regular expressions in non-US locales. thanks to ruslan for keeping
204 at this one.
205
206Jul 28, 2003:
207 n-th try at getting internationalization right, with thanks to volker
208 kiefel, arnold robbins and ruslan ermilov for advice, though they
209 should not be blamed for the outcome. according to posix, "." is the
210 radix character in programs and command line arguments regardless of
211 the locale; otherwise, the locale should prevail for input and output
212 of numbers. so it's intended to work that way.
213
214 i have rescinded the attempt to use strcoll in expanding shorthands in
215 regular expressions (cclenter). its properties are much too
216 surprising; for example [a-c] matches aAbBc in locale en_US but abBcC
217 in locale fr_CA. i can see how this might arise by implementation
218 but i cannot explain it to a human user. (this behavior can be seen
219 in gawk as well; we're leaning on the same library.)
220
221 the issue appears to be that strcoll is meant for sorting, where
222 merging upper and lower case may make sense (though note that unix
223 sort does not do this by default either). it is not appropriate
224 for regular expressions, where the goal is to match specific
225 patterns of characters. in any case, the notations [:lower:], etc.,
226 are available in awk, and they are more likely to work correctly in
227 most locales.
228
229 a moratorium is hereby declared on internationalization changes.
230 i apologize to friends and colleagues in other parts of the world.
231 i would truly like to get this "right", but i don't know what
232 that is, and i do not want to keep making changes until it's clear.
233
234Jul 4, 2003:
235 fixed bug that permitted non-terminated RE, as in "awk /x".
236
237Jun 1, 2003:
238 subtle change to split: if source is empty, number of elems
239 is always 0 and the array is not set.
240
241Mar 21, 2003:
242 added some parens to isblank, in another attempt to make things
243 internationally portable.
244
245Mar 14, 2003:
246 the internationalization changes, somewhat modified, are now
247 reinstated. in theory awk will now do character comparisons
248 and case conversions in national language, but "." will always
249 be the decimal point separator on input and output regardless
250 of national language. isblank(){} has an #ifndef.
251
252 this no longer compiles on windows: LC_MESSAGES isn't defined
253 in vc6++.
254
255 fixed subtle behavior in field and record splitting: if FS is
256 a single character and RS is not empty, \n is NOT a separator.
257 this tortuous reading is found in the awk book; behavior now
258 matches gawk and mawk.
259
260Dec 13, 2002:
261 for the moment, the internationalization changes of nov 29 are
262 rolled back -- programs like x = 1.2 don't work in some locales,
263 because the parser is expecting x = 1,2. until i understand this
264 better, this will have to wait.
265
266Nov 29, 2002:
267 modified b.c (with tiny changes in main and run) to support
268 locales, using strcoll and iswhatever tests for posix character
269 classes. thanks to ruslan ermilov (ru@freebsd.org) for code.
270 the function isblank doesn't seem to have propagated to any
271 header file near me, so it's there explicitly. not properly
272 tested on non-ascii character sets by me.
273
274Jun 28, 2002:
275 modified run/format() and tran/getsval() to do a slightly better
276 job on using OFMT for output from print and CONVFMT for other
277 number->string conversions, as promised by posix and done by
278 gawk and mawk. there are still places where it doesn't work
279 right if CONVFMT is changed; by then the STR attribute of the
280 variable has been irrevocably set. thanks to arnold robbins for
281 code and examples.
282
283 fixed subtle bug in format that could get core dump. thanks to
284 Jaromir Dolecek <jdolecek@NetBSD.org> for finding and fixing.
285 minor cleanup in run.c / format() at the same time.
286
287 added some tests for null pointers to debugging printf's, which
288 were never intended for external consumption. thanks to dave
289 kerns (dkerns@lucent.com) for pointing this out.
290
291 GNU compatibility: an empty regexp matches anything (thanks to
292 dag-erling smorgrav, des@ofug.org). subject to reversion if
293 this does more harm than good.
294
295 pervasive small changes to make things more const-correct, as
296 reported by gcc's -Wwrite-strings. as it says in the gcc manual,
297 this may be more nuisance than useful. provoked by a suggestion
298 and code from arnaud desitter, arnaud@nimbus.geog.ox.ac.uk
299
300 minor documentation changes to note that this now compiles out
301 of the box on Mac OS X.
302
303Feb 10, 2002:
304 changed types in posix chars structure to quiet solaris cc.
305
306Jan 1, 2002:
307 fflush() or fflush("") flushes all files and pipes.
308
309 length(arrayname) returns number of elements; thanks to
310 arnold robbins for suggestion.
311
312 added a makefile.win to make it easier to build on windows.
313 based on dan allen's buildwin.bat.
314
315Nov 16, 2001:
316 added support for posix character class names like [:digit:],
317 which are not exactly shorter than [0-9] and perhaps no more
318 portable. thanks to dag-erling smorgrav for code.
319
320Feb 16, 2001:
321 removed -m option; no longer needed, and it was actually
322 broken (noted thanks to volker kiefel).
323
324Feb 10, 2001:
325 fixed an appalling bug in gettok: any sequence of digits, +,-, E, e,
326 and period was accepted as a valid number if it started with a period.
327 this would never have happened with the lex version.
328
329 other 1-character botches, now fixed, include a bare $ and a
330 bare " at the end of the input.
331
332Feb 7, 2001:
333 more (const char *) casts in b.c and tran.c to silence warnings.
334
335Nov 15, 2000:
336 fixed a bug introduced in august 1997 that caused expressions
337 like $f[1] to be syntax errors. thanks to arnold robbins for
338 noticing this and providing a fix.
339
340Oct 30, 2000:
341 fixed some nextfile bugs: not handling all cases. thanks to
342 arnold robbins for pointing this out. new regressions added.
343
344 close() is now a function. it returns whatever the library
345 fclose returns, and -1 for closing a file or pipe that wasn't
346 opened.
347
348Sep 24, 2000:
349 permit \n explicitly in character classes; won't work right
350 if comes in as "[\n]" but ok as /[\n]/, because of multiple
351 processing of \'s. thanks to arnold robbins.
352
353July 5, 2000:
354 minor fiddles in tran.c to keep compilers happy about uschar.
355 thanks to norman wilson.
356
357May 25, 2000:
358 yet another attempt at making 8-bit input work, with another
359 band-aid in b.c (member()), and some (uschar) casts to head
360 off potential errors in subscripts (like isdigit). also
361 changed HAT to NCHARS-2. thanks again to santiago vila.
362
363 changed maketab.c to ignore apparently out of range definitions
364 instead of halting; new freeBSD generates one. thanks to
365 jon snader <jsnader@ix.netcom.com> for pointing out the problem.
366
367May 2, 2000:
368 fixed an 8-bit problem in b.c by making several char*'s into
369 unsigned char*'s. not clear i have them all yet. thanks to
370 Santiago Vila <sanvila@unex.es> for the bug report.
371
372Apr 21, 2000:
373 finally found and fixed a memory leak in function call; it's
374 been there since functions were added ~1983. thanks to
375 jon bentley for the test case that found it.
376
377 added test in envinit to catch environment "variables" with
378 names beginning with '='; thanks to Berend Hasselman.
379
380Jul 28, 1999:
381 added test in defn() to catch function foo(foo), which
382 otherwise recurses until core dump. thanks to arnold
383 robbins for noticing this.
384
385Jun 20, 1999:
386 added *bp in gettok in lex.c; appears possible to exit function
387 without terminating the string. thanks to russ cox.
388
389Jun 2, 1999:
390 added function stdinit() to run to initialize files[] array,
391 in case stdin, etc., are not constants; some compilers care.
392
393May 10, 1999:
394 replaced the ERROR ... FATAL, etc., macros with functions
395 based on vprintf, to avoid problems caused by overrunning
396 fixed-size errbuf array. thanks to ralph corderoy for the
397 impetus, and for pointing out a string termination bug in
398 qstring as well.
399
400Apr 21, 1999:
401 fixed bug that caused occasional core dumps with commandline
402 variable with value ending in \. (thanks to nelson beebe for
403 the test case.)
404
405Apr 16, 1999:
406 with code kindly provided by Bruce Lilly, awk now parses
407 /=/ and similar constructs more sensibly in more places.
408 Bruce also provided some helpful test cases.
409
410Apr 5, 1999:
411 changed true/false to True/False in run.c to make it
412 easier to compile with C++. Added some casts on malloc
413 and realloc to be honest about casts; ditto. changed
414 ltype int to long in struct rrow to reduce some 64-bit
415 complaints; other changes scattered throughout for the
416 same purpose. thanks to Nelson Beebe for these portability
417 improvements.
418
419 removed some horrible pointer-int casting in b.c and elsewhere
420 by adding ptoi and itonp to localize the casts, which are
421 all benign. fixed one incipient bug that showed up on sgi
422 in 64-bit mode.
423
424 reset lineno for new source file; include filename in error
425 message. also fixed line number error in continuation lines.
426 (thanks to Nelson Beebe for both of these.)
427
428Mar 24, 1999:
429 Nelson Beebe notes that irix 5.3 yacc dies with a bogus
430 error; use a newer version or switch to bison, since sgi
431 is unlikely to fix it.
432
433Mar 5, 1999:
434 changed isnumber to is_number to avoid the problem caused by
435 versions of ctype.h that include the name isnumber.
436
437 distribution now includes a script for building on a Mac,
438 thanks to Dan Allen.
439
440Feb 20, 1999:
441 fixed memory leaks in run.c (call) and tran.c (setfval).
442 thanks to Stephen Nutt for finding these and providing the fixes.
443
444Jan 13, 1999:
445 replaced srand argument by (unsigned int) in run.c;
446 avoids problem on Mac and potentially on Unix & Windows.
447 thanks to Dan Allen.
448
449 added a few (int) casts to silence useless compiler warnings.
450 e.g., errorflag= in run.c jump().
451
452 added proctab.c to the bundle outout; one less thing
453 to have to compile out of the box.
454
455 added calls to _popen and _pclose to the win95 stub for
456 pipes (thanks to Steve Adams for this helpful suggestion).
457 seems to work, though properties are not well understood
458 by me, and it appears that under some circumstances the
459 pipe output is truncated. Be careful.
460
461Oct 19, 1998:
462 fixed a couple of bugs in getrec: could fail to update $0
463 after a getline var; because inputFS wasn't initialized,
464 could split $0 on every character, a misleading diversion.
465
466 fixed caching bug in makedfa: LRU was actually removing
467 least often used.
468
469 thanks to ross ridge for finding these, and for providing
470 great bug reports.
471
472May 12, 1998:
473 fixed potential bug in readrec: might fail to update record
474 pointer after growing. thanks to dan levy for spotting this
475 and suggesting the fix.
476
477Mar 12, 1998:
478 added -V to print version number and die.
479
480Feb 11, 1998:
481 subtle silent bug in lex.c: if the program ended with a number
482 longer than 1 digit, part of the input would be pushed back and
483 parsed again because token buffer wasn't terminated right.
484 example: awk 'length($0) > 10'. blush. at least i found it
485 myself.
486
487Aug 31, 1997:
488 s/adelete/awkdelete/: SGI uses this in malloc.h.
489 thanks to nelson beebe for pointing this one out.
490
491Aug 21, 1997:
492 fixed some bugs in sub and gsub when replacement includes \\.
493 this is a dark, horrible corner, but at least now i believe that
494 the behavior is the same as gawk and the intended posix standard.
495 thanks to arnold robbins for advice here.
496
497Aug 9, 1997:
498 somewhat regretfully, replaced the ancient lex-based lexical
499 analyzer with one written in C. it's longer, generates less code,
500 and more portable; the old one depended too much on mysterious
501 properties of lex that were not preserved in other environments.
502 in theory these recognize the same language.
503
504 now using strtod to test whether a string is a number, instead of
505 the convoluted original function. should be more portable and
506 reliable if strtod is implemented right.
507
508 removed now-pointless optimization in makefile that tries to avoid
509 recompilation when awkgram.y is changed but symbols are not.
510
511 removed most fixed-size arrays, though a handful remain, some
512 of which are unchecked. you have been warned.
513
514Aug 4, 1997:
515 with some trepidation, replaced the ancient code that managed
516 fields and $0 in fixed-size arrays with arrays that grow on
517 demand. there is still some tension between trying to make this
518 run fast and making it clean; not sure it's right yet.
519
520 the ill-conceived -mr and -mf arguments are now useful only
521 for debugging. previous dynamic string code removed.
522
523 numerous other minor cleanups along the way.
524
525Jul 30, 1997:
526 using code provided by dan levy (to whom profuse thanks), replaced
527 fixed-size arrays and awkward kludges by a fairly uniform mechanism
528 to grow arrays as needed for printf, sub, gsub, etc.
529
530Jul 23, 1997:
531 falling off the end of a function returns "" and 0, not 0.
532 thanks to arnold robbins.
533
534Jun 17, 1997:
535 replaced several fixed-size arrays by dynamically-created ones
536 in run.c; added overflow tests to some previously unchecked cases.
537 getline, toupper, tolower.
538
539 getline code is still broken in that recursive calls may wind
540 up using the same space. [fixed later]
541
542 increased RECSIZE to 8192 to push problems further over the horizon.
543
544 added \r to \n as input line separator for programs, not data.
545 damn CRLFs.
546
547 modified format() to permit explicit printf("%c", 0) to include
548 a null byte in output. thanks to ken stailey for the fix.
549
550 added a "-safe" argument that disables file output (print >,
551 print >>), process creation (cmd|getline, print |, system), and
552 access to the environment (ENVIRON). this is a first approximation
553 to a "safe" version of awk, but don't rely on it too much. thanks
554 to joan feigenbaum and matt blaze for the inspiration long ago.
555
556Jul 8, 1996:
557 fixed long-standing bug in sub, gsub(/a/, "\\\\&"); thanks to
558 ralph corderoy.
559
560Jun 29, 1996:
561 fixed awful bug in new field splitting; didn't get all the places
562 where input was done.
563
564Jun 28, 1996:
565 changed field-splitting to conform to posix definition: fields are
566 split using the value of FS at the time of input; it used to be
567 the value when the field or NF was first referred to, a much less
568 predictable definition. thanks to arnold robbins for encouragement
569 to do the right thing.
570
571May 28, 1996:
572 fixed appalling but apparently unimportant bug in parsing octal
573 numbers in reg exprs.
574
575 explicit hex in reg exprs now limited to 2 chars: \xa, \xaa.
576
577May 27, 1996:
578 cleaned up some declarations so gcc -Wall is now almost silent.
579
580 makefile now includes backup copies of ytab.c and lexyy.c in case
581 one makes before looking; it also avoids recreating lexyy.c unless
582 really needed.
583
584 s/aprintf/awkprint, s/asprintf/awksprintf/ to avoid some name clashes
585 with unwisely-written header files.
586
587 thanks to jeffrey friedl for several of these.
588
589May 26, 1996:
590 an attempt to rationalize the (unsigned) char issue. almost all
591 instances of unsigned char have been removed; the handful of places
592 in b.c where chars are used as table indices have been hand-crafted.
593 added some latin-1 tests to the regression, but i'm not confident;
594 none of my compilers seem to care much. thanks to nelson beebe for
595 pointing out some others that do care.
596
597May 2, 1996:
598 removed all register declarations.
599
600 enhanced split(), as in gawk, etc: split(s, a, "") splits s into
601 a[1]...a[length(s)] with each character a single element.
602
603 made the same changes for field-splitting if FS is "".
604
605 added nextfile, as in gawk: causes immediate advance to next
606 input file. (thanks to arnold robbins for inspiration and code).
607
608 small fixes to regexpr code: can now handle []], [[], and
609 variants; [] is now a syntax error, rather than matching
610 everything; [z-a] is now empty, not z. far from complete
611 or correct, however. (thanks to jeffrey friedl for pointing out
612 some awful behaviors.)
613
614Apr 29, 1996:
615 replaced uchar by uschar everywhere; apparently some compilers
616 usurp this name and this causes conflicts.
617
618 fixed call to time in run.c (bltin); arg is time_t *.
619
620 replaced horrible pointer/long punning in b.c by a legitimate
621 union. should be safer on 64-bit machines and cleaner everywhere.
622 (thanks to nelson beebe for pointing out some of these problems.)
623
624 replaced nested comments by #if 0...#endif in run.c, lib.c.
625
626 removed getsval, setsval, execute macros from run.c and lib.c.
627 machines are 100x faster than they were when these macros were
628 first used.
629
630 revised filenames: awk.g.y => awkgram.y, awk.lx.l => awklex.l,
631 y.tab.[ch] => ytab.[ch], lex.yy.c => lexyy.c, all in the aid of
632 portability to nameless systems.
633
634 "make bundle" now includes yacc and lex output files for recipients
635 who don't have yacc or lex.
636
637Aug 15, 1995:
638 initialized Cells in setsymtab more carefully; some fields
639 were not set. (thanks to purify, all of whose complaints i
640 think i now understand.)
641
642 fixed at least one error in gsub that looked at -1-th element
643 of an array when substituting for a null match (e.g., $).
644
645 delete arrayname is now legal; it clears the elements but leaves
646 the array, which may not be the right behavior.
647
648 modified makefile: my current make can't cope with the test used
649 to avoid unnecessary yacc invocations.
650
651Jul 17, 1995:
652 added dynamically growing strings to awk.lx.l and b.c
653 to permit regular expressions to be much bigger.
654 the state arrays can still overflow.
655
656Aug 24, 1994:
657 detect duplicate arguments in function definitions (mdm).
658
659May 11, 1994:
660 trivial fix to printf to limit string size in sub().
661
662Apr 22, 1994:
663 fixed yet another subtle self-assignment problem:
664 $1 = $2; $1 = $1 clobbered $1.
665
666 Regression tests now use private echo, to avoid quoting problems.
667
668Feb 2, 1994:
669 changed error() to print line number as %d, not %g.
670
671Jul 23, 1993:
672 cosmetic changes: increased sizes of some arrays,
673 reworded some error messages.
674
675 added CONVFMT as in posix (just replaced OFMT in getsval)
676
677 FILENAME is now "" until the first thing that causes a file
678 to be opened.
679
680Nov 28, 1992:
681 deleted yyunput and yyoutput from proto.h;
682 different versions of lex give these different declarations.
683
684May 31, 1992:
685 added -mr N and -mf N options: more record and fields.
686 these really ought to adjust automatically.
687
688 cleaned up some error messages; "out of space" now means
689 malloc returned NULL in all cases.
690
691 changed rehash so that if it runs out, it just returns;
692 things will continue to run slow, but maybe a bit longer.
693
694Apr 24, 1992:
695 remove redundant close of stdin when using -f -.
696
697 got rid of core dump with -d; awk -d just prints date.
698
699Apr 12, 1992:
700 added explicit check for /dev/std(in,out,err) in redirection.
701 unlike gawk, no /dev/fd/n yet.
702
703 added (file/pipe) builtin. hard to test satisfactorily.
704 not posix.
705
706Feb 20, 1992:
707 recompile after abortive changes; should be unchanged.
708
709Dec 2, 1991:
710 die-casting time: converted to ansi C, installed that.
711
712Nov 30, 1991:
713 fixed storage leak in freefa, failing to recover [N]CCL.
714 thanks to Bill Jones (jones@cs.usask.ca)
715
716Nov 19, 1991:
717 use RAND_MAX instead of literal in builtin().
718
719Nov 12, 1991:
720 cranked up some fixed-size arrays in b.c, and added a test for
721 overflow in penter. thanks to mark larsen.
722
723Sep 24, 1991:
724 increased buffer in gsub. a very crude fix to a general problem.
725 and again on Sep 26.
726
727Aug 18, 1991:
728 enforce variable name syntax for commandline variables: has to
729 start with letter or _.
730
731Jul 27, 1991:
732 allow newline after ; in for statements.
733
734Jul 21, 1991:
735 fixed so that in self-assignment like $1=$1, side effects
736 like recomputing $0 take place. (this is getting subtle.)
737
738Jun 30, 1991:
739 better test for detecting too-long output record.
740
741Jun 2, 1991:
742 better defense against very long printf strings.
743 made break and continue illegal outside of loops.
744
745May 13, 1991:
746 removed extra arg on gettemp, tempfree. minor error message rewording.
747
748May 6, 1991:
749 fixed silly bug in hex parsing in hexstr().
750 removed an apparently unnecessary test in isnumber().
751 warn about weird printf conversions.
752 fixed unchecked array overwrite in relex().
753
754 changed for (i in array) to access elements in sorted order.
755 then unchanged it -- it really does run slower in too many cases.
756 left the code in place, commented out.
757
758Feb 10, 1991:
759 check error status on all writes, to avoid banging on full disks.
760
761Jan 28, 1991:
762 awk -f - reads the program from stdin.
763
764Jan 11, 1991:
765 failed to set numeric state on $0 in cmd|getline context in run.c.
766
767Nov 2, 1990:
768 fixed sleazy test for integrality in getsval; use modf.
769
770Oct 29, 1990:
771 fixed sleazy buggy code in lib.c that looked (incorrectly) for
772 too long input lines.
773
774Oct 14, 1990:
775 fixed the bug on p. 198 in which it couldn't deduce that an
776 argument was an array in some contexts. replaced the error
777 message in intest() by code that damn well makes it an array.
778
779Oct 8, 1990:
780 fixed horrible bug: types and values were not preserved in
781 some kinds of self-assignment. (in assign().)
782
783Aug 24, 1990:
784 changed NCHARS to 256 to handle 8-bit characters in strings
785 presented to match(), etc.
786
787Jun 26, 1990:
788 changed struct rrow (awk.h) to use long instead of int for lval,
789 since cfoll() stores a pointer in it. now works better when int's
790 are smaller than pointers!
791
792May 6, 1990:
793 AVA fixed the grammar so that ! is uniformly of the same precedence as
794 unary + and -. This renders illegal some constructs like !x=y, which
795 now has to be parenthesized as !(x=y), and makes others work properly:
796 !x+y is (!x)+y, and x!y is x !y, not two pattern-action statements.
797 (These problems were pointed out by Bob Lenk of Posix.)
798
799 Added \x to regular expressions (already in strings).
800 Limited octal to octal digits; \8 and \9 are not octal.
801 Centralized the code for parsing escapes in regular expressions.
802 Added a bunch of tests to T.re and T.sub to verify some of this.
803
804Feb 9, 1990:
805 fixed null pointer dereference bug in main.c: -F[nothing]. sigh.
806
807 restored srand behavior: it returns the current seed.
808
809Jan 18, 1990:
810 srand now returns previous seed value (0 to start).
811
812Jan 5, 1990:
813 fix potential problem in tran.c -- something was freed,
814 then used in freesymtab.
815
816Oct 18, 1989:
817 another try to get the max number of open files set with
818 relatively machine-independent code.
819
820 small fix to input() in case of multiple reads after EOF.
821
822Oct 11, 1989:
823 FILENAME is now defined in the BEGIN block -- too many old
824 programs broke.
825
826 "-" means stdin in getline as well as on the commandline.
827
828 added a bunch of casts to the code to tell the truth about
829 char * vs. unsigned char *, a right royal pain. added a
830 setlocale call to the front of main, though probably no one
831 has it usefully implemented yet.
832
833Aug 24, 1989:
834 removed redundant relational tests against nullnode if parse
835 tree already had a relational at that point.
836
837Aug 11, 1989:
838 fixed bug: commandline variable assignment has to look like
839 var=something. (consider the man page for =, in file =.1)
840
841 changed number of arguments to functions to static arrays
842 to avoid repeated malloc calls.
843
844Aug 2, 1989:
845 restored -F (space) separator
846
847Jul 30, 1989:
848 added -v x=1 y=2 ... for immediate commandline variable assignment;
849 done before the BEGIN block for sure. they have to precede the
850 program if the program is on the commandline.
851 Modified Aug 2 to require a separate -v for each assignment.
852
853Jul 10, 1989:
854 fixed ref-thru-zero bug in environment code in tran.c
855
856Jun 23, 1989:
857 add newline to usage message.
858
859Jun 14, 1989:
860 added some missing ansi printf conversion letters: %i %X %E %G.
861 no sensible meaning for h or L, so they may not do what one expects.
862
863 made %* conversions work.
864
865 changed x^y so that if n is a positive integer, it's done
866 by explicit multiplication, thus achieving maximum accuracy.
867 (this should be done by pow() but it seems not to be locally.)
868 done to x ^= y as well.
869
870Jun 4, 1989:
871 ENVIRON array contains environment: if shell variable V=thing,
872 ENVIRON["V"] is "thing"
873
874 multiple -f arguments permitted. error reporting is naive.
875 (they were permitted before, but only the last was used.)
876
877 fixed a really stupid botch in the debugging macro dprintf
878
879 fixed order of evaluation of commandline assignments to match
880 what the book claims: an argument of the form x=e is evaluated
881 at the time it would have been opened if it were a filename (p 63).
882 this invalidates the suggested answer to ex 4-1 (p 195).
883
884 removed some code that permitted -F (space) fieldseparator,
885 since it didn't quite work right anyway. (restored aug 2)
886
887Apr 27, 1989:
888 Line number now accumulated correctly for comment lines.
889
890Apr 26, 1989:
891 Debugging output now includes a version date,
892 if one compiles it into the source each time.
893
894Apr 9, 1989:
895 Changed grammar to prohibit constants as 3rd arg of sub and gsub;
896 prevents class of overwriting-a-constant errors. (Last one?)
897 This invalidates the "banana" example on page 43 of the book.
898
899 Added \a ("alert"), \v (vertical tab), \xhhh (hexadecimal),
900 as in ANSI, for strings. Rescinded the sloppiness that permitted
901 non-octal digits in \ooo. Warning: not all compilers and libraries
902 will be able to deal with \x correctly.
903
904Jan 9, 1989:
905 Fixed bug that caused tempcell list to contain a duplicate.
906 The fix is kludgy.
907
908Dec 17, 1988:
909 Catches some more commandline errors in main.
910 Removed redundant decl of modf in run.c (confuses some compilers).
911 Warning: there's no single declaration of malloc, etc., in awk.h
912 that seems to satisfy all compilers.
913
914Dec 7, 1988:
915 Added a bit of code to error printing to avoid printing nulls.
916 (Not clear that it actually would.)
917
918Nov 27, 1988:
919 With fear and trembling, modified the grammar to permit
920 multiple pattern-action statements on one line without
921 an explicit separator. By definition, this capitulation
922 to the ghost of ancient implementations remains undefined
923 and thus subject to change without notice or apology.
924 DO NOT COUNT ON IT.
925
926Oct 30, 1988:
927 Fixed bug in call() that failed to recover storage.
928
929 A warning is now generated if there are more arguments
930 in the call than in the definition (in lieu of fixing
931 another storage leak).
932
933Oct 20, 1988:
934 Fixed %c: if expr is numeric, use numeric value;
935 otherwise print 1st char of string value. still
936 doesn't work if the value is 0 -- won't print \0.
937
938 Added a few more checks for running out of malloc.
939
940Oct 12, 1988:
941 Fixed bug in call() that freed local arrays twice.
942
943 Fixed to handle deletion of non-existent array right;
944 complains about attempt to delete non-array element.
945
946Sep 30, 1988:
947 Now guarantees to evaluate all arguments of built-in
948 functions, as in C; the appearance is that arguments
949 are evaluated before the function is called. Places
950 affected are sub (gsub was ok), substr, printf, and
951 all the built-in arithmetic functions in bltin().
952 A warning is generated if a bltin() is called with
953 the wrong number of arguments.
954
955 This requires changing makeprof on p167 of the book.
956
957Aug 23, 1988:
958 setting FILENAME in BEGIN caused core dump, apparently
959 because it was freeing space not allocated by malloc.
960
961July 24, 1988:
962 fixed egregious error in toupper/tolower functions.
963 still subject to rescinding, however.
964
965July 2, 1988:
966 flush stdout before opening file or pipe
967
968July 2, 1988:
969 performance bug in b.c/cgoto(): not freeing some sets of states.
970 partial fix only right now, and the number of states increased
971 to make it less obvious.
972
973June 1, 1988:
974 check error status on close
975
976May 28, 1988:
977 srand returns seed value it's using.
978 see 1/18/90
979
980May 22, 1988:
981 Removed limit on depth of function calls.
982
983May 10, 1988:
984 Fixed lib.c to permit _ in commandline variable names.
985
986Mar 25, 1988:
987 main.c fixed to recognize -- as terminator of command-
988 line options. Illegal options flagged.
989 Error reporting slightly cleaned up.
990
991Dec 2, 1987:
992 Newer C compilers apply a strict scope rule to extern
993 declarations within functions. Two extern declarations in
994 lib.c and tran.c have been moved to obviate this problem.
995
996Oct xx, 1987:
997 Reluctantly added toupper and tolower functions.
998 Subject to rescinding without notice.
999
1000Sep 17, 1987:
1001 Error-message printer had printf(s) instead of
1002 printf("%s",s); got core dumps when the message
1003 included a %.
1004
1005Sep 12, 1987:
1006 Very long printf strings caused core dump;
1007 fixed aprintf, asprintf, format to catch them.
1008 Can still get a core dump in printf itself.
1009
1010