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