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 Fix check for command line arguments to no longer require that
30 first character after '=' not be another '='. Reverts change of
31 August 11, 1989. Thanks to GitHub user Jamie Landeg Jones for
32 pointing out the issue; from Issue #38.
35 Update awktest.tar(p.50) to use modern options to sort. Needed
36 for Android development. Thanks to GitHub user mohd-akram (Mohamed
37 Akram). From Comment #33.
40 Added very simplistic support for cross-compiling in the
41 makefile. We are NOT going to go in the direction of the
42 autotools, though. Thanks to GitHub user nee-san for
43 the basic change. (Merged from PR #34.)
46 Added support for POSIX-standard interval expressions (a.k.a.
47 bounds, a.k.a. repetition expressions) in regular expressions,
48 backported (via NetBSD) from Apple awk-24 (20070501).
49 Thanks to Martijn Dekker <martijn@inlv.org> for the port.
54 #12: Avoid undefined behaviour when using ctype(3) functions in
55 relex(). Thanks to GitHub user iamleot.
56 #31: Make getline handle numeric strings, and update FIXES. Thanks
57 to GitHub user arnoldrobbins
58 #32: maketab: support build systems with read-only source. Thanks
62 Make getline handle numeric strings properly in all cases.
66 Merged a number of small fixes from GitHub pull requests.
67 Thanks to GitHub users Arnold Robbins (arnoldrobbins),
68 Cody Mello (melloc) and Christoph Junghans (junghans).
69 PR numbers: 13-21, 23, 24, 27.
72 Added test in maketab.c to prevent generating a proctab entry
73 for YYSTYPE_IS_DEFINED. It was harmless but some gcc settings
74 generated a warning message. Thanks to Nan Xiao for report.
77 Disallow '$' in printf formats; arguments evaluated in order
80 Added some casts to silence warnings on debugging printfs.
84 A long list of fixes courtesy of Arnold Robbins,
85 to whom profound thanks.
87 1. ofs-rebuild: OFS value used to rebuild the record was incorrect.
88 Fixed August 19, 2014. Revised fix August 2018.
90 2. system-status: Instead of a floating-point division by 256, use
91 the wait(2) macros to create a reasonable exit status.
94 3. space: Use provided xisblank() function instead of ispace() for
97 4. a-format: Add POSIX standard %a and %A to supported formats. Check
98 at runtime that this format is available.
100 5. decr-NF: Decrementing NF did not change $0. This is a decades-old
101 bug. There are interactions with the old and new value of OFS as well.
102 Most of the fix came from the NetBSD awk.
104 6. string-conv: String conversions of scalars were sticky. Once a
105 conversion to string happened, even with OFMT, that value was used until
106 a new numeric value was assigned, even if OFMT differed from CONVFMT,
107 and also if CONVFMT changed.
109 7. unary-plus: Unary plus on a string constant returned the string.
110 Instead, it should convert the value to numeric and give that value.
112 Also added Arnold's tests for these to awktest.tar as T.arnold.
115 fixed mangled awktest.tar (thanks, Arnold), posted all
116 current (very minor) fixes to github / onetrueawk
120 Updated some broken tests (beebe.tar, T.lilly)
121 [thanks to Arnold Robbins]
124 buffer overflow in error reporting; thanks to tobias ulmer
125 and john-mark gurney for spotting it and the fix.
128 cleaned up a handful of tests that didn't seem to actually
129 test for correct behavior: T.latin1, T.gawk.
132 added ,NULL initializer to static Cells in run.c; not really
133 needed but cleaner. Thanks to Michael Bombardieri.
136 fiddled makefile to get correct yacc and bison flags. pick yacc
137 (linux) or bison (mac) as necessary.
139 added __attribute__((__noreturn__)) to a couple of lines in
140 proto.h, to silence someone's enthusiastic checker.
142 fixed obscure call by value bug in split(a[1],a) reported on
143 9fans. the management of temporary values is just a mess; i
144 took a shortcut by making an extra string copy. thanks
145 to paul patience and arnold robbins for passing it on and for
148 tiny fiddle in setfval to eliminate -0 results in T.expr, which
149 has irritated me for 20+ years.
152 another fix to avoid core dump with delete(ARGV); again, many thanks
156 split(s, a, //) now behaves the same as split(s, a, "")
159 /pat/, \n /pat/ {...} is now legal, though bad style to use.
161 added checks to new -v code that permits -vnospace; thanks to
162 ruslan ermilov for spotting this and providing the patch.
164 removed fixed limit on number of open files; thanks to aleksey
165 cheusov and christos zoulos.
167 fixed day 1 bug that resurrected deleted elements of ARGV when
168 used as filenames (in lib.c).
170 minor type fiddles to make gcc -Wall -pedantic happier (but not
171 totally so); turned on -fno-strict-aliasing in makefile.
174 added #ifdef for isblank.
175 now allows -ffoo as well as -f foo arguments.
179 after advice from todd miller, kevin lo, ruslan ermilov,
180 and arnold robbins, changed srand() to return the previous
181 seed (which is 1 on the first call of srand). the seed is
182 an Awkfloat internally though converted to unsigned int to
183 pass to the library srand(). thanks, everyone.
185 fixed a subtle (and i hope low-probability) overflow error
186 in fldbld, by adding space for one extra \0. thanks to
187 robert bassett for spotting this one and providing a fix.
189 removed the files related to compilation on windows. i no
190 longer have anything like a current windows environment, so
191 i can't test any of it.
194 fixed long-standing overflow bug in run.c; many thanks to
195 nelson beebe for spotting it and providing the fix.
197 fixed bug that didn't parse -vd=1 properly; thanks to santiago
198 vila for spotting it.
201 i give up. replaced isblank with isspace in b.c; there are
202 no consistent header files.
205 fixed a long-standing issue with when FS takes effect. a
206 change to FS is now noticed immediately for subsequent splits.
208 changed the name getline() to awkgetline() to avoid yet another
209 name conflict somewhere.
212 temporarily for now defined HAS_ISBLANK, since that seems to
213 be the best way through the thicket. isblank arrived in C99,
214 but seems to be arriving at different systems at different
218 fixed typo in b.c that set tmpvec wrongly. no one had ever
219 run into the problem, apparently. thanks to alistair crooks.
222 minor fix in lib.c: increase inputFS to 100, change malloc
225 fixed memory fault caused by out of order test in setsval.
227 thanks to david o'brien, freebsd, for both fixes.
230 fiddle in makefile to fix for BSD make; thanks to igor sobrado.
233 fixed some null pointer refs calling adjbuf.
236 fixed a bug in matching the null RE in sub and gsub. thanks to al aho
237 who actually did the fix (in b.c), and to wolfgang seeberg for finding
238 it and providing a very compact test case.
240 fixed quotation in b.c; thanks to Hal Pratt and the Princeton Dante
243 removed some no-effect asserts in run.c.
245 fiddled maketab.c to not complain about bison-generated values.
247 removed the obsolete -V argument; fixed --version to print the
250 fixed wording and an outright error in the usage message; thanks to igor
251 sobrado and jason mcintyre.
253 fixed a bug in -d that caused core dump if no program followed.
256 dropped mac.code from makefile; there are few non-MacOSX
260 system() not flagged as unsafe in the unadvertised -safe option.
261 found it while enhancing tests before shipping the ;login: article.
262 practice what you preach.
264 removed the 9-years-obsolete -mr and -mf flags.
266 added -version and --version options.
268 core dump on linux with BEGIN {nextfile}, now fixed.
270 removed some #ifdef's in run.c and lex.c that appear to no
274 modified lib.c so that values of $0 et al are preserved in the END
275 block, apparently as required by posix. thanks to havard eidnes
276 for the report and code.
279 fixed infinite loop in parsing, originally found by brian tsang.
280 thanks to arnold robbins for a suggestion that started me
284 prevent overflow of -f array in main, head off potential error in
285 call of SYNTAX(), test malloc return in lib.c, all with thanks to
289 cranked up size of NCHARS; coverity thinks it can be overrun with
290 smaller size, and i think that's right. added some assertions to b.c
291 to catch places where it might overrun. the RE code is still fragile.
294 fixed a couple of overflow problems with ridiculous field numbers:
295 e.g., print $(2^32-1). thanks to ruslan ermilov, giorgos keramidas
296 and david o'brien at freebsd.org for patches. this really should
297 be re-done from scratch.
300 fixed another 25-year-old RE bug, in split. it's another failure
301 to (re-)initialize. thanks to steve fisher for spotting this and
302 providing a good test case.
305 fixed a bug in regular expressions that dates (so help me) from 1977;
306 it's been there from the beginning. an anchored longest match that
307 was longer than the number of states triggered a failure to initialize
308 the machine properly. many thanks to moinak ghosh for not only finding
309 this one but for providing a fix, in some of the most mysterious
312 fixed a storage leak in call() that appears to have been there since
313 1983 or so -- a function without an explicit return that assigns a
314 string to a parameter leaked a Cell. thanks to moinak ghosh for
315 spotting this very subtle one.
318 fixed, thanks to andrey chernov and ruslan ermilov, a bug in lex.c
319 that mis-handled the character 255 in input. (it was being compared
320 to EOF with a signed comparison.)
323 fixed (i think) the long-standing botch that included the beginning of
324 line state ^ for RE's in the set of valid characters; this led to a
325 variety of odd problems, including failure to properly match certain
326 regular expressions in non-US locales. thanks to ruslan for keeping
330 n-th try at getting internationalization right, with thanks to volker
331 kiefel, arnold robbins and ruslan ermilov for advice, though they
332 should not be blamed for the outcome. according to posix, "." is the
333 radix character in programs and command line arguments regardless of
334 the locale; otherwise, the locale should prevail for input and output
335 of numbers. so it's intended to work that way.
337 i have rescinded the attempt to use strcoll in expanding shorthands in
338 regular expressions (cclenter). its properties are much too
339 surprising; for example [a-c] matches aAbBc in locale en_US but abBcC
340 in locale fr_CA. i can see how this might arise by implementation
341 but i cannot explain it to a human user. (this behavior can be seen
342 in gawk as well; we're leaning on the same library.)
344 the issue appears to be that strcoll is meant for sorting, where
345 merging upper and lower case may make sense (though note that unix
346 sort does not do this by default either). it is not appropriate
347 for regular expressions, where the goal is to match specific
348 patterns of characters. in any case, the notations [:lower:], etc.,
349 are available in awk, and they are more likely to work correctly in
352 a moratorium is hereby declared on internationalization changes.
353 i apologize to friends and colleagues in other parts of the world.
354 i would truly like to get this "right", but i don't know what
355 that is, and i do not want to keep making changes until it's clear.
358 fixed bug that permitted non-terminated RE, as in "awk /x".
361 subtle change to split: if source is empty, number of elems
362 is always 0 and the array is not set.
365 added some parens to isblank, in another attempt to make things
366 internationally portable.
369 the internationalization changes, somewhat modified, are now
370 reinstated. in theory awk will now do character comparisons
371 and case conversions in national language, but "." will always
372 be the decimal point separator on input and output regardless
373 of national language. isblank(){} has an #ifndef.
375 this no longer compiles on windows: LC_MESSAGES isn't defined
378 fixed subtle behavior in field and record splitting: if FS is
379 a single character and RS is not empty, \n is NOT a separator.
380 this tortuous reading is found in the awk book; behavior now
381 matches gawk and mawk.
384 for the moment, the internationalization changes of nov 29 are
385 rolled back -- programs like x = 1.2 don't work in some locales,
386 because the parser is expecting x = 1,2. until i understand this
387 better, this will have to wait.
390 modified b.c (with tiny changes in main and run) to support
391 locales, using strcoll and iswhatever tests for posix character
392 classes. thanks to ruslan ermilov (ru@freebsd.org) for code.
393 the function isblank doesn't seem to have propagated to any
394 header file near me, so it's there explicitly. not properly
395 tested on non-ascii character sets by me.
398 modified run/format() and tran/getsval() to do a slightly better
399 job on using OFMT for output from print and CONVFMT for other
400 number->string conversions, as promised by posix and done by
401 gawk and mawk. there are still places where it doesn't work
402 right if CONVFMT is changed; by then the STR attribute of the
403 variable has been irrevocably set. thanks to arnold robbins for
406 fixed subtle bug in format that could get core dump. thanks to
407 Jaromir Dolecek <jdolecek@NetBSD.org> for finding and fixing.
408 minor cleanup in run.c / format() at the same time.
410 added some tests for null pointers to debugging printf's, which
411 were never intended for external consumption. thanks to dave
412 kerns (dkerns@lucent.com) for pointing this out.
414 GNU compatibility: an empty regexp matches anything (thanks to
415 dag-erling smorgrav, des@ofug.org). subject to reversion if
416 this does more harm than good.
418 pervasive small changes to make things more const-correct, as
419 reported by gcc's -Wwrite-strings. as it says in the gcc manual,
420 this may be more nuisance than useful. provoked by a suggestion
421 and code from arnaud desitter, arnaud@nimbus.geog.ox.ac.uk
423 minor documentation changes to note that this now compiles out
424 of the box on Mac OS X.
427 changed types in posix chars structure to quiet solaris cc.
430 fflush() or fflush("") flushes all files and pipes.
432 length(arrayname) returns number of elements; thanks to
433 arnold robbins for suggestion.
435 added a makefile.win to make it easier to build on windows.
436 based on dan allen's buildwin.bat.
439 added support for posix character class names like [:digit:],
440 which are not exactly shorter than [0-9] and perhaps no more
441 portable. thanks to dag-erling smorgrav for code.
444 removed -m option; no longer needed, and it was actually
445 broken (noted thanks to volker kiefel).
448 fixed an appalling bug in gettok: any sequence of digits, +,-, E, e,
449 and period was accepted as a valid number if it started with a period.
450 this would never have happened with the lex version.
452 other 1-character botches, now fixed, include a bare $ and a
453 bare " at the end of the input.
456 more (const char *) casts in b.c and tran.c to silence warnings.
459 fixed a bug introduced in august 1997 that caused expressions
460 like $f[1] to be syntax errors. thanks to arnold robbins for
461 noticing this and providing a fix.
464 fixed some nextfile bugs: not handling all cases. thanks to
465 arnold robbins for pointing this out. new regressions added.
467 close() is now a function. it returns whatever the library
468 fclose returns, and -1 for closing a file or pipe that wasn't
472 permit \n explicitly in character classes; won't work right
473 if comes in as "[\n]" but ok as /[\n]/, because of multiple
474 processing of \'s. thanks to arnold robbins.
477 minor fiddles in tran.c to keep compilers happy about uschar.
478 thanks to norman wilson.
481 yet another attempt at making 8-bit input work, with another
482 band-aid in b.c (member()), and some (uschar) casts to head
483 off potential errors in subscripts (like isdigit). also
484 changed HAT to NCHARS-2. thanks again to santiago vila.
486 changed maketab.c to ignore apparently out of range definitions
487 instead of halting; new freeBSD generates one. thanks to
488 jon snader <jsnader@ix.netcom.com> for pointing out the problem.
491 fixed an 8-bit problem in b.c by making several char*'s into
492 unsigned char*'s. not clear i have them all yet. thanks to
493 Santiago Vila <sanvila@unex.es> for the bug report.
496 finally found and fixed a memory leak in function call; it's
497 been there since functions were added ~1983. thanks to
498 jon bentley for the test case that found it.
500 added test in envinit to catch environment "variables" with
501 names beginning with '='; thanks to Berend Hasselman.
504 added test in defn() to catch function foo(foo), which
505 otherwise recurses until core dump. thanks to arnold
506 robbins for noticing this.
509 added *bp in gettok in lex.c; appears possible to exit function
510 without terminating the string. thanks to russ cox.
513 added function stdinit() to run to initialize files[] array,
514 in case stdin, etc., are not constants; some compilers care.
517 replaced the ERROR ... FATAL, etc., macros with functions
518 based on vprintf, to avoid problems caused by overrunning
519 fixed-size errbuf array. thanks to ralph corderoy for the
520 impetus, and for pointing out a string termination bug in
524 fixed bug that caused occasional core dumps with commandline
525 variable with value ending in \. (thanks to nelson beebe for
529 with code kindly provided by Bruce Lilly, awk now parses
530 /=/ and similar constructs more sensibly in more places.
531 Bruce also provided some helpful test cases.
534 changed true/false to True/False in run.c to make it
535 easier to compile with C++. Added some casts on malloc
536 and realloc to be honest about casts; ditto. changed
537 ltype int to long in struct rrow to reduce some 64-bit
538 complaints; other changes scattered throughout for the
539 same purpose. thanks to Nelson Beebe for these portability
542 removed some horrible pointer-int casting in b.c and elsewhere
543 by adding ptoi and itonp to localize the casts, which are
544 all benign. fixed one incipient bug that showed up on sgi
547 reset lineno for new source file; include filename in error
548 message. also fixed line number error in continuation lines.
549 (thanks to Nelson Beebe for both of these.)
552 Nelson Beebe notes that irix 5.3 yacc dies with a bogus
553 error; use a newer version or switch to bison, since sgi
554 is unlikely to fix it.
557 changed isnumber to is_number to avoid the problem caused by
558 versions of ctype.h that include the name isnumber.
560 distribution now includes a script for building on a Mac,
564 fixed memory leaks in run.c (call) and tran.c (setfval).
565 thanks to Stephen Nutt for finding these and providing the fixes.
568 replaced srand argument by (unsigned int) in run.c;
569 avoids problem on Mac and potentially on Unix & Windows.
572 added a few (int) casts to silence useless compiler warnings.
573 e.g., errorflag= in run.c jump().
575 added proctab.c to the bundle outout; one less thing
576 to have to compile out of the box.
578 added calls to _popen and _pclose to the win95 stub for
579 pipes (thanks to Steve Adams for this helpful suggestion).
580 seems to work, though properties are not well understood
581 by me, and it appears that under some circumstances the
582 pipe output is truncated. Be careful.
585 fixed a couple of bugs in getrec: could fail to update $0
586 after a getline var; because inputFS wasn't initialized,
587 could split $0 on every character, a misleading diversion.
589 fixed caching bug in makedfa: LRU was actually removing
592 thanks to ross ridge for finding these, and for providing
596 fixed potential bug in readrec: might fail to update record
597 pointer after growing. thanks to dan levy for spotting this
598 and suggesting the fix.
601 added -V to print version number and die.
603 [notify dave kerns, dkerns@dacsoup.ih.lucent.com]
606 subtle silent bug in lex.c: if the program ended with a number
607 longer than 1 digit, part of the input would be pushed back and
608 parsed again because token buffer wasn't terminated right.
609 example: awk 'length($0) > 10'. blush. at least i found it
613 s/adelete/awkdelete/: SGI uses this in malloc.h.
614 thanks to nelson beebe for pointing this one out.
617 fixed some bugs in sub and gsub when replacement includes \\.
618 this is a dark, horrible corner, but at least now i believe that
619 the behavior is the same as gawk and the intended posix standard.
620 thanks to arnold robbins for advice here.
623 somewhat regretfully, replaced the ancient lex-based lexical
624 analyzer with one written in C. it's longer, generates less code,
625 and more portable; the old one depended too much on mysterious
626 properties of lex that were not preserved in other environments.
627 in theory these recognize the same language.
629 now using strtod to test whether a string is a number, instead of
630 the convoluted original function. should be more portable and
631 reliable if strtod is implemented right.
633 removed now-pointless optimization in makefile that tries to avoid
634 recompilation when awkgram.y is changed but symbols are not.
636 removed most fixed-size arrays, though a handful remain, some
637 of which are unchecked. you have been warned.
640 with some trepidation, replaced the ancient code that managed
641 fields and $0 in fixed-size arrays with arrays that grow on
642 demand. there is still some tension between trying to make this
643 run fast and making it clean; not sure it's right yet.
645 the ill-conceived -mr and -mf arguments are now useful only
646 for debugging. previous dynamic string code removed.
648 numerous other minor cleanups along the way.
651 using code provided by dan levy (to whom profuse thanks), replaced
652 fixed-size arrays and awkward kludges by a fairly uniform mechanism
653 to grow arrays as needed for printf, sub, gsub, etc.
656 falling off the end of a function returns "" and 0, not 0.
657 thanks to arnold robbins.
660 replaced several fixed-size arrays by dynamically-created ones
661 in run.c; added overflow tests to some previously unchecked cases.
662 getline, toupper, tolower.
664 getline code is still broken in that recursive calls may wind
665 up using the same space. [fixed later]
667 increased RECSIZE to 8192 to push problems further over the horizon.
669 added \r to \n as input line separator for programs, not data.
672 modified format() to permit explicit printf("%c", 0) to include
673 a null byte in output. thanks to ken stailey for the fix.
675 added a "-safe" argument that disables file output (print >,
676 print >>), process creation (cmd|getline, print |, system), and
677 access to the environment (ENVIRON). this is a first approximation
678 to a "safe" version of awk, but don't rely on it too much. thanks
679 to joan feigenbaum and matt blaze for the inspiration long ago.
682 fixed long-standing bug in sub, gsub(/a/, "\\\\&"); thanks to
686 fixed awful bug in new field splitting; didn't get all the places
687 where input was done.
690 changed field-splitting to conform to posix definition: fields are
691 split using the value of FS at the time of input; it used to be
692 the value when the field or NF was first referred to, a much less
693 predictable definition. thanks to arnold robbins for encouragement
694 to do the right thing.
697 fixed appalling but apparently unimportant bug in parsing octal
698 numbers in reg exprs.
700 explicit hex in reg exprs now limited to 2 chars: \xa, \xaa.
703 cleaned up some declarations so gcc -Wall is now almost silent.
705 makefile now includes backup copies of ytab.c and lexyy.c in case
706 one makes before looking; it also avoids recreating lexyy.c unless
709 s/aprintf/awkprint, s/asprintf/awksprintf/ to avoid some name clashes
710 with unwisely-written header files.
712 thanks to jeffrey friedl for several of these.
715 an attempt to rationalize the (unsigned) char issue. almost all
716 instances of unsigned char have been removed; the handful of places
717 in b.c where chars are used as table indices have been hand-crafted.
718 added some latin-1 tests to the regression, but i'm not confident;
719 none of my compilers seem to care much. thanks to nelson beebe for
720 pointing out some others that do care.
723 removed all register declarations.
725 enhanced split(), as in gawk, etc: split(s, a, "") splits s into
726 a[1]...a[length(s)] with each character a single element.
728 made the same changes for field-splitting if FS is "".
730 added nextfile, as in gawk: causes immediate advance to next
731 input file. (thanks to arnold robbins for inspiration and code).
733 small fixes to regexpr code: can now handle []], [[], and
734 variants; [] is now a syntax error, rather than matching
735 everything; [z-a] is now empty, not z. far from complete
736 or correct, however. (thanks to jeffrey friedl for pointing out
737 some awful behaviors.)
740 replaced uchar by uschar everywhere; apparently some compilers
741 usurp this name and this causes conflicts.
743 fixed call to time in run.c (bltin); arg is time_t *.
745 replaced horrible pointer/long punning in b.c by a legitimate
746 union. should be safer on 64-bit machines and cleaner everywhere.
747 (thanks to nelson beebe for pointing out some of these problems.)
749 replaced nested comments by #if 0...#endif in run.c, lib.c.
751 removed getsval, setsval, execute macros from run.c and lib.c.
752 machines are 100x faster than they were when these macros were
755 revised filenames: awk.g.y => awkgram.y, awk.lx.l => awklex.l,
756 y.tab.[ch] => ytab.[ch], lex.yy.c => lexyy.c, all in the aid of
757 portability to nameless systems.
759 "make bundle" now includes yacc and lex output files for recipients
760 who don't have yacc or lex.
763 initialized Cells in setsymtab more carefully; some fields
764 were not set. (thanks to purify, all of whose complaints i
765 think i now understand.)
767 fixed at least one error in gsub that looked at -1-th element
768 of an array when substituting for a null match (e.g., $).
770 delete arrayname is now legal; it clears the elements but leaves
771 the array, which may not be the right behavior.
773 modified makefile: my current make can't cope with the test used
774 to avoid unnecessary yacc invocations.
777 added dynamically growing strings to awk.lx.l and b.c
778 to permit regular expressions to be much bigger.
779 the state arrays can still overflow.
782 detect duplicate arguments in function definitions (mdm).
785 trivial fix to printf to limit string size in sub().
788 fixed yet another subtle self-assignment problem:
789 $1 = $2; $1 = $1 clobbered $1.
791 Regression tests now use private echo, to avoid quoting problems.
794 changed error() to print line number as %d, not %g.
797 cosmetic changes: increased sizes of some arrays,
798 reworded some error messages.
800 added CONVFMT as in posix (just replaced OFMT in getsval)
802 FILENAME is now "" until the first thing that causes a file
806 deleted yyunput and yyoutput from proto.h;
807 different versions of lex give these different declarations.
810 added -mr N and -mf N options: more record and fields.
811 these really ought to adjust automatically.
813 cleaned up some error messages; "out of space" now means
814 malloc returned NULL in all cases.
816 changed rehash so that if it runs out, it just returns;
817 things will continue to run slow, but maybe a bit longer.
820 remove redundant close of stdin when using -f -.
822 got rid of core dump with -d; awk -d just prints date.
825 added explicit check for /dev/std(in,out,err) in redirection.
826 unlike gawk, no /dev/fd/n yet.
828 added (file/pipe) builtin. hard to test satisfactorily.
832 recompile after abortive changes; should be unchanged.
835 die-casting time: converted to ansi C, installed that.
838 fixed storage leak in freefa, failing to recover [N]CCL.
839 thanks to Bill Jones (jones@cs.usask.ca)
842 use RAND_MAX instead of literal in builtin().
845 cranked up some fixed-size arrays in b.c, and added a test for
846 overflow in penter. thanks to mark larsen.
849 increased buffer in gsub. a very crude fix to a general problem.
853 enforce variable name syntax for commandline variables: has to
854 start with letter or _.
857 allow newline after ; in for statements.
860 fixed so that in self-assignment like $1=$1, side effects
861 like recomputing $0 take place. (this is getting subtle.)
864 better test for detecting too-long output record.
867 better defense against very long printf strings.
868 made break and continue illegal outside of loops.
871 removed extra arg on gettemp, tempfree. minor error message rewording.
874 fixed silly bug in hex parsing in hexstr().
875 removed an apparently unnecessary test in isnumber().
876 warn about weird printf conversions.
877 fixed unchecked array overwrite in relex().
879 changed for (i in array) to access elements in sorted order.
880 then unchanged it -- it really does run slower in too many cases.
881 left the code in place, commented out.
884 check error status on all writes, to avoid banging on full disks.
887 awk -f - reads the program from stdin.
890 failed to set numeric state on $0 in cmd|getline context in run.c.
893 fixed sleazy test for integrality in getsval; use modf.
896 fixed sleazy buggy code in lib.c that looked (incorrectly) for
897 too long input lines.
900 fixed the bug on p. 198 in which it couldn't deduce that an
901 argument was an array in some contexts. replaced the error
902 message in intest() by code that damn well makes it an array.
905 fixed horrible bug: types and values were not preserved in
906 some kinds of self-assignment. (in assign().)
909 changed NCHARS to 256 to handle 8-bit characters in strings
910 presented to match(), etc.
913 changed struct rrow (awk.h) to use long instead of int for lval,
914 since cfoll() stores a pointer in it. now works better when int's
915 are smaller than pointers!
918 AVA fixed the grammar so that ! is uniformly of the same precedence as
919 unary + and -. This renders illegal some constructs like !x=y, which
920 now has to be parenthesized as !(x=y), and makes others work properly:
921 !x+y is (!x)+y, and x!y is x !y, not two pattern-action statements.
922 (These problems were pointed out by Bob Lenk of Posix.)
924 Added \x to regular expressions (already in strings).
925 Limited octal to octal digits; \8 and \9 are not octal.
926 Centralized the code for parsing escapes in regular expressions.
927 Added a bunch of tests to T.re and T.sub to verify some of this.
930 fixed null pointer dereference bug in main.c: -F[nothing]. sigh.
932 restored srand behavior: it returns the current seed.
935 srand now returns previous seed value (0 to start).
938 fix potential problem in tran.c -- something was freed,
939 then used in freesymtab.
942 another try to get the max number of open files set with
943 relatively machine-independent code.
945 small fix to input() in case of multiple reads after EOF.
948 FILENAME is now defined in the BEGIN block -- too many old
951 "-" means stdin in getline as well as on the commandline.
953 added a bunch of casts to the code to tell the truth about
954 char * vs. unsigned char *, a right royal pain. added a
955 setlocale call to the front of main, though probably no one
956 has it usefully implemented yet.
959 removed redundant relational tests against nullnode if parse
960 tree already had a relational at that point.
963 fixed bug: commandline variable assignment has to look like
964 var=something. (consider the man page for =, in file =.1)
966 changed number of arguments to functions to static arrays
967 to avoid repeated malloc calls.
970 restored -F (space) separator
973 added -v x=1 y=2 ... for immediate commandline variable assignment;
974 done before the BEGIN block for sure. they have to precede the
975 program if the program is on the commandline.
976 Modified Aug 2 to require a separate -v for each assignment.
979 fixed ref-thru-zero bug in environment code in tran.c
982 add newline to usage message.
985 added some missing ansi printf conversion letters: %i %X %E %G.
986 no sensible meaning for h or L, so they may not do what one expects.
988 made %* conversions work.
990 changed x^y so that if n is a positive integer, it's done
991 by explicit multiplication, thus achieving maximum accuracy.
992 (this should be done by pow() but it seems not to be locally.)
993 done to x ^= y as well.
996 ENVIRON array contains environment: if shell variable V=thing,
997 ENVIRON["V"] is "thing"
999 multiple -f arguments permitted. error reporting is naive.
1000 (they were permitted before, but only the last was used.)
1002 fixed a really stupid botch in the debugging macro dprintf
1004 fixed order of evaluation of commandline assignments to match
1005 what the book claims: an argument of the form x=e is evaluated
1006 at the time it would have been opened if it were a filename (p 63).
1007 this invalidates the suggested answer to ex 4-1 (p 195).
1009 removed some code that permitted -F (space) fieldseparator,
1010 since it didn't quite work right anyway. (restored aug 2)
1013 Line number now accumulated correctly for comment lines.
1016 Debugging output now includes a version date,
1017 if one compiles it into the source each time.
1020 Changed grammar to prohibit constants as 3rd arg of sub and gsub;
1021 prevents class of overwriting-a-constant errors. (Last one?)
1022 This invalidates the "banana" example on page 43 of the book.
1024 Added \a ("alert"), \v (vertical tab), \xhhh (hexadecimal),
1025 as in ANSI, for strings. Rescinded the sloppiness that permitted
1026 non-octal digits in \ooo. Warning: not all compilers and libraries
1027 will be able to deal with \x correctly.
1030 Fixed bug that caused tempcell list to contain a duplicate.
1034 Catches some more commandline errors in main.
1035 Removed redundant decl of modf in run.c (confuses some compilers).
1036 Warning: there's no single declaration of malloc, etc., in awk.h
1037 that seems to satisfy all compilers.
1040 Added a bit of code to error printing to avoid printing nulls.
1041 (Not clear that it actually would.)
1044 With fear and trembling, modified the grammar to permit
1045 multiple pattern-action statements on one line without
1046 an explicit separator. By definition, this capitulation
1047 to the ghost of ancient implementations remains undefined
1048 and thus subject to change without notice or apology.
1052 Fixed bug in call() that failed to recover storage.
1054 A warning is now generated if there are more arguments
1055 in the call than in the definition (in lieu of fixing
1056 another storage leak).
1059 Fixed %c: if expr is numeric, use numeric value;
1060 otherwise print 1st char of string value. still
1061 doesn't work if the value is 0 -- won't print \0.
1063 Added a few more checks for running out of malloc.
1066 Fixed bug in call() that freed local arrays twice.
1068 Fixed to handle deletion of non-existent array right;
1069 complains about attempt to delete non-array element.
1072 Now guarantees to evaluate all arguments of built-in
1073 functions, as in C; the appearance is that arguments
1074 are evaluated before the function is called. Places
1075 affected are sub (gsub was ok), substr, printf, and
1076 all the built-in arithmetic functions in bltin().
1077 A warning is generated if a bltin() is called with
1078 the wrong number of arguments.
1080 This requires changing makeprof on p167 of the book.
1083 setting FILENAME in BEGIN caused core dump, apparently
1084 because it was freeing space not allocated by malloc.
1087 fixed egregious error in toupper/tolower functions.
1088 still subject to rescinding, however.
1091 flush stdout before opening file or pipe
1094 performance bug in b.c/cgoto(): not freeing some sets of states.
1095 partial fix only right now, and the number of states increased
1096 to make it less obvious.
1099 check error status on close
1102 srand returns seed value it's using.
1106 Removed limit on depth of function calls.
1109 Fixed lib.c to permit _ in commandline variable names.
1112 main.c fixed to recognize -- as terminator of command-
1113 line options. Illegal options flagged.
1114 Error reporting slightly cleaned up.
1117 Newer C compilers apply a strict scope rule to extern
1118 declarations within functions. Two extern declarations in
1119 lib.c and tran.c have been moved to obviate this problem.
1122 Reluctantly added toupper and tolower functions.
1123 Subject to rescinding without notice.
1126 Error-message printer had printf(s) instead of
1127 printf("%s",s); got core dumps when the message
1131 Very long printf strings caused core dump;
1132 fixed aprintf, asprintf, format to catch them.
1133 Can still get a core dump in printf itself.