Annotation of src/usr.bin/awk/FIXES, Revision 1.22
1.22 ! millert 1: /* $OpenBSD: FIXES,v 1.21 2020/06/10 21:01:13 millert Exp $ */
1.1 tholo 2: /****************************************************************
1.6 kstailey 3: Copyright (C) Lucent Technologies 1997
1.1 tholo 4: All Rights Reserved
5:
6: Permission to use, copy, modify, and distribute this software and
7: its documentation for any purpose and without fee is hereby
8: granted, provided that the above copyright notice appear in all
9: copies and that both that the copyright notice and this
10: permission notice and warranty disclaimer appear in supporting
1.6 kstailey 11: documentation, and that the name Lucent Technologies or any of
12: its entities not be used in advertising or publicity pertaining
13: to distribution of the software without specific, written prior
14: permission.
15:
16: LUCENT DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
17: INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS.
18: IN NO EVENT SHALL LUCENT OR ANY OF ITS ENTITIES BE LIABLE FOR ANY
19: SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
20: WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER
21: IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION,
22: ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF
23: THIS SOFTWARE.
1.1 tholo 24: ****************************************************************/
25:
26: This file lists all bug fixes, changes, etc., made since the AWK book
27: was sent to the printers in August, 1987.
1.22 ! millert 28:
! 29: Jan 25, 2019:
! 30: Make getline handle numeric strings properly in all cases.
! 31: (Thanks, Arnold.)
! 32:
! 33: Jan 21, 2019:
! 34: Merged a number of small fixes from GitHub pull requests.
! 35: Thanks to GitHub users Arnold Robbins (arnoldrobbins),
! 36: Cody Mello (melloc) and Christoph Junghans (junghans).
! 37: PR numbers: 13-21, 23, 24, 27.
1.21 millert 38:
39: Oct 25, 2018:
40: Added test in maketab.c to prevent generating a proctab entry
41: for YYSTYPE_IS_DEFINED. It was harmless but some gcc settings
42: generated a warning message. Thanks to Nan Xiao for report.
1.20 millert 43:
44: Aug 27, 2018:
45: Disallow '$' in printf formats; arguments evaluated in order
46: and printed in order.
47:
48: Added some casts to silence warnings on debugging printfs.
49: (Thanks, Arnold.)
1.18 millert 50:
1.19 millert 51: Aug 23, 2018:
52: A long list of fixes courtesy of Arnold Robbins,
53: to whom profound thanks.
54:
55: 1. ofs-rebuild: OFS value used to rebuild the record was incorrect.
56: Fixed August 19, 2014. Revised fix August 2018.
57:
58: 2. system-status: Instead of a floating-point division by 256, use
59: the wait(2) macros to create a reasonable exit status.
60: Fixed March 12, 2016.
61:
62: 3. space: Use provided xisblank() function instead of ispace() for
63: matching [[:blank:]].
64:
65: 4. a-format: Add POSIX standard %a and %A to supported formats. Check
66: at runtime that this format is available.
67:
68: 5. decr-NF: Decrementing NF did not change $0. This is a decades-old
69: bug. There are interactions with the old and new value of OFS as well.
70: Most of the fix came from the NetBSD awk.
71:
72: 6. string-conv: String conversions of scalars were sticky. Once a
73: conversion to string happened, even with OFMT, that value was used until
74: a new numeric value was assigned, even if OFMT differed from CONVFMT,
75: and also if CONVFMT changed.
76:
77: 7. unary-plus: Unary plus on a string constant returned the string.
78: Instead, it should convert the value to numeric and give that value.
79:
80: Also added Arnold's tests for these to awktest.tar as T.arnold.
81:
82: Aug 15, 2018:
83: fixed mangled awktest.tar (thanks, Arnold), posted all
84: current (very minor) fixes to github / onetrueawk
85:
1.18 millert 86: Jun 7, 2018:
87: (yes, a long layoff)
88: Updated some broken tests (beebe.tar, T.lilly)
89: [thanks to Arnold Robbins]
90:
91: Mar 26, 2015:
92: buffer overflow in error reporting; thanks to tobias ulmer
93: and john-mark gurney for spotting it and the fix.
94:
95: Feb 4, 2013:
96: cleaned up a handful of tests that didn't seem to actually
97: test for correct behavior: T.latin1, T.gawk.
98:
99: Jan 5, 2013:
100: added ,NULL initializer to static Cells in run.c; not really
101: needed but cleaner. Thanks to Michael Bombardieri.
1.17 millert 102:
103: Dec 20, 2012:
104: fiddled makefile to get correct yacc and bison flags. pick yacc
105: (linux) or bison (mac) as necessary.
106:
107: added __attribute__((__noreturn__)) to a couple of lines in
108: proto.h, to silence someone's enthusiastic checker.
109:
110: fixed obscure call by value bug in split(a[1],a) reported on
111: 9fans. the management of temporary values is just a mess; i
112: took a shortcut by making an extra string copy. thanks
113: to paul patience and arnold robbins for passing it on and for
114: proposed patches.
115:
116: tiny fiddle in setfval to eliminate -0 results in T.expr, which
117: has irritated me for 20+ years.
1.16 millert 118:
119: Aug 10, 2011:
120: another fix to avoid core dump with delete(ARGV); again, many thanks
121: to ruslan ermilov.
122:
123: Aug 7, 2011:
124: split(s, a, //) now behaves the same as split(s, a, "")
125:
126: Jun 12, 2011:
127: /pat/, \n /pat/ {...} is now legal, though bad style to use.
128:
129: added checks to new -v code that permits -vnospace; thanks to
130: ruslan ermilov for spotting this and providing the patch.
131:
132: removed fixed limit on number of open files; thanks to aleksey
133: cheusov and christos zoulos.
134:
135: fixed day 1 bug that resurrected deleted elements of ARGV when
136: used as filenames (in lib.c).
137:
138: minor type fiddles to make gcc -Wall -pedantic happier (but not
139: totally so); turned on -fno-strict-aliasing in makefile.
140:
141: May 6, 2011:
142: added #ifdef for isblank.
143: now allows -ffoo as well as -f foo arguments.
144: (thanks, ruslan)
145:
146: May 1, 2011:
147: after advice from todd miller, kevin lo, ruslan ermilov,
148: and arnold robbins, changed srand() to return the previous
149: seed (which is 1 on the first call of srand). the seed is
150: an Awkfloat internally though converted to unsigned int to
151: pass to the library srand(). thanks, everyone.
152:
153: fixed a subtle (and i hope low-probability) overflow error
154: in fldbld, by adding space for one extra \0. thanks to
155: robert bassett for spotting this one and providing a fix.
156:
157: removed the files related to compilation on windows. i no
158: longer have anything like a current windows environment, so
159: i can't test any of it.
1.15 millert 160:
161: May 23, 2010:
162: fixed long-standing overflow bug in run.c; many thanks to
163: nelson beebe for spotting it and providing the fix.
164:
165: fixed bug that didn't parse -vd=1 properly; thanks to santiago
166: vila for spotting it.
167:
168: Feb 8, 2010:
169: i give up. replaced isblank with isspace in b.c; there are
170: no consistent header files.
171:
172: Nov 26, 2009:
173: fixed a long-standing issue with when FS takes effect. a
174: change to FS is now noticed immediately for subsequent splits.
175:
176: changed the name getline() to awkgetline() to avoid yet another
177: name conflict somewhere.
178:
179: Feb 11, 2009:
180: temporarily for now defined HAS_ISBLANK, since that seems to
181: be the best way through the thicket. isblank arrived in C99,
182: but seems to be arriving at different systems at different
183: times.
184:
185: Oct 8, 2008:
186: fixed typo in b.c that set tmpvec wrongly. no one had ever
187: run into the problem, apparently. thanks to alistair crooks.
1.14 millert 188:
189: Oct 23, 2007:
190: minor fix in lib.c: increase inputFS to 100, change malloc
191: for fields to n+1.
192:
193: fixed memory fault caused by out of order test in setsval.
194:
195: thanks to david o'brien, freebsd, for both fixes.
196:
197: May 1, 2007:
198: fiddle in makefile to fix for BSD make; thanks to igor sobrado.
199:
200: Mar 31, 2007:
201: fixed some null pointer refs calling adjbuf.
202:
203: Feb 21, 2007:
204: fixed a bug in matching the null RE in sub and gsub. thanks to al aho
205: who actually did the fix (in b.c), and to wolfgang seeberg for finding
206: it and providing a very compact test case.
207:
208: fixed quotation in b.c; thanks to Hal Pratt and the Princeton Dante
209: Project.
210:
211: removed some no-effect asserts in run.c.
212:
213: fiddled maketab.c to not complain about bison-generated values.
214:
215: removed the obsolete -V argument; fixed --version to print the
216: version and exit.
217:
218: fixed wording and an outright error in the usage message; thanks to igor
219: sobrado and jason mcintyre.
220:
221: fixed a bug in -d that caused core dump if no program followed.
222:
223: Jan 1, 2007:
224: dropped mac.code from makefile; there are few non-MacOSX
225: mac's these days.
226:
227: Jan 17, 2006:
228: system() not flagged as unsafe in the unadvertised -safe option.
229: found it while enhancing tests before shipping the ;login: article.
230: practice what you preach.
231:
232: removed the 9-years-obsolete -mr and -mf flags.
233:
234: added -version and --version options.
235:
236: core dump on linux with BEGIN {nextfile}, now fixed.
237:
238: removed some #ifdef's in run.c and lex.c that appear to no
239: longer be necessary.
240:
241: Apr 24, 2005:
242: modified lib.c so that values of $0 et al are preserved in the END
243: block, apparently as required by posix. thanks to havard eidnes
244: for the report and code.
245:
246: Jan 14, 2005:
247: fixed infinite loop in parsing, originally found by brian tsang.
248: thanks to arnold robbins for a suggestion that started me
249: rethinking it.
250:
251: Dec 31, 2004:
252: prevent overflow of -f array in main, head off potential error in
253: call of SYNTAX(), test malloc return in lib.c, all with thanks to
254: todd miller.
1.13 millert 255:
256: Dec 22, 2004:
257: cranked up size of NCHARS; coverity thinks it can be overrun with
258: smaller size, and i think that's right. added some assertions to b.c
259: to catch places where it might overrun. the RE code is still fragile.
260:
261: Dec 5, 2004:
262: fixed a couple of overflow problems with ridiculous field numbers:
263: e.g., print $(2^32-1). thanks to ruslan ermilov, giorgos keramidas
264: and david o'brien at freebsd.org for patches. this really should
265: be re-done from scratch.
266:
267: Nov 21, 2004:
268: fixed another 25-year-old RE bug, in split. it's another failure
269: to (re-)initialize. thanks to steve fisher for spotting this and
270: providing a good test case.
271:
272: Nov 22, 2003:
273: fixed a bug in regular expressions that dates (so help me) from 1977;
274: it's been there from the beginning. an anchored longest match that
275: was longer than the number of states triggered a failure to initialize
276: the machine properly. many thanks to moinak ghosh for not only finding
277: this one but for providing a fix, in some of the most mysterious
278: code known to man.
279:
280: fixed a storage leak in call() that appears to have been there since
281: 1983 or so -- a function without an explicit return that assigns a
282: string to a parameter leaked a Cell. thanks to moinak ghosh for
283: spotting this very subtle one.
284:
285: Jul 31, 2003:
286: fixed, thanks to andrey chernov and ruslan ermilov, a bug in lex.c
287: that mis-handled the character 255 in input. (it was being compared
288: to EOF with a signed comparison.)
289:
290: Jul 29, 2003:
291: fixed (i think) the long-standing botch that included the beginning of
292: line state ^ for RE's in the set of valid characters; this led to a
293: variety of odd problems, including failure to properly match certain
294: regular expressions in non-US locales. thanks to ruslan for keeping
295: at this one.
296:
297: Jul 28, 2003:
298: n-th try at getting internationalization right, with thanks to volker
299: kiefel, arnold robbins and ruslan ermilov for advice, though they
300: should not be blamed for the outcome. according to posix, "." is the
301: radix character in programs and command line arguments regardless of
302: the locale; otherwise, the locale should prevail for input and output
303: of numbers. so it's intended to work that way.
304:
305: i have rescinded the attempt to use strcoll in expanding shorthands in
306: regular expressions (cclenter). its properties are much too
307: surprising; for example [a-c] matches aAbBc in locale en_US but abBcC
308: in locale fr_CA. i can see how this might arise by implementation
309: but i cannot explain it to a human user. (this behavior can be seen
310: in gawk as well; we're leaning on the same library.)
311:
312: the issue appears to be that strcoll is meant for sorting, where
313: merging upper and lower case may make sense (though note that unix
314: sort does not do this by default either). it is not appropriate
315: for regular expressions, where the goal is to match specific
316: patterns of characters. in any case, the notations [:lower:], etc.,
317: are available in awk, and they are more likely to work correctly in
318: most locales.
319:
320: a moratorium is hereby declared on internationalization changes.
321: i apologize to friends and colleagues in other parts of the world.
322: i would truly like to get this "right", but i don't know what
323: that is, and i do not want to keep making changes until it's clear.
324:
325: Jul 4, 2003:
326: fixed bug that permitted non-terminated RE, as in "awk /x".
327:
328: Jun 1, 2003:
329: subtle change to split: if source is empty, number of elems
330: is always 0 and the array is not set.
331:
332: Mar 21, 2003:
333: added some parens to isblank, in another attempt to make things
334: internationally portable.
335:
336: Mar 14, 2003:
337: the internationalization changes, somewhat modified, are now
338: reinstated. in theory awk will now do character comparisons
339: and case conversions in national language, but "." will always
340: be the decimal point separator on input and output regardless
341: of national language. isblank(){} has an #ifndef.
342:
343: this no longer compiles on windows: LC_MESSAGES isn't defined
344: in vc6++.
345:
346: fixed subtle behavior in field and record splitting: if FS is
347: a single character and RS is not empty, \n is NOT a separator.
348: this tortuous reading is found in the awk book; behavior now
349: matches gawk and mawk.
1.11 millert 350:
351: Dec 13, 2002:
352: for the moment, the internationalization changes of nov 29 are
353: rolled back -- programs like x = 1.2 don't work in some locales,
354: because the parser is expecting x = 1,2. until i understand this
355: better, this will have to wait.
356:
357: Nov 29, 2002:
358: modified b.c (with tiny changes in main and run) to support
359: locales, using strcoll and iswhatever tests for posix character
360: classes. thanks to ruslan ermilov (ru@freebsd.org) for code.
361: the function isblank doesn't seem to have propagated to any
362: header file near me, so it's there explicitly. not properly
363: tested on non-ascii character sets by me.
364:
365: Jun 28, 2002:
366: modified run/format() and tran/getsval() to do a slightly better
367: job on using OFMT for output from print and CONVFMT for other
368: number->string conversions, as promised by posix and done by
369: gawk and mawk. there are still places where it doesn't work
370: right if CONVFMT is changed; by then the STR attribute of the
371: variable has been irrevocably set. thanks to arnold robbins for
372: code and examples.
373:
374: fixed subtle bug in format that could get core dump. thanks to
375: Jaromir Dolecek <jdolecek@NetBSD.org> for finding and fixing.
376: minor cleanup in run.c / format() at the same time.
377:
378: added some tests for null pointers to debugging printf's, which
379: were never intended for external consumption. thanks to dave
380: kerns (dkerns@lucent.com) for pointing this out.
381:
382: GNU compatibility: an empty regexp matches anything (thanks to
383: dag-erling smorgrav, des@ofug.org). subject to reversion if
384: this does more harm than good.
385:
386: pervasive small changes to make things more const-correct, as
387: reported by gcc's -Wwrite-strings. as it says in the gcc manual,
388: this may be more nuisance than useful. provoked by a suggestion
389: and code from arnaud desitter, arnaud@nimbus.geog.ox.ac.uk
390:
391: minor documentation changes to note that this now compiles out
392: of the box on Mac OS X.
393:
394: Feb 10, 2002:
395: changed types in posix chars structure to quiet solaris cc.
396:
397: Jan 1, 2002:
398: fflush() or fflush("") flushes all files and pipes.
399:
400: length(arrayname) returns number of elements; thanks to
401: arnold robbins for suggestion.
402:
403: added a makefile.win to make it easier to build on windows.
404: based on dan allen's buildwin.bat.
405:
406: Nov 16, 2001:
407: added support for posix character class names like [:digit:],
408: which are not exactly shorter than [0-9] and perhaps no more
409: portable. thanks to dag-erling smorgrav for code.
410:
411: Feb 16, 2001:
412: removed -m option; no longer needed, and it was actually
413: broken (noted thanks to volker kiefel).
414:
415: Feb 10, 2001:
416: fixed an appalling bug in gettok: any sequence of digits, +,-, E, e,
417: and period was accepted as a valid number if it started with a period.
418: this would never have happened with the lex version.
419:
420: other 1-character botches, now fixed, include a bare $ and a
421: bare " at the end of the input.
422:
423: Feb 7, 2001:
424: more (const char *) casts in b.c and tran.c to silence warnings.
1.8 millert 425:
1.10 millert 426: Nov 15, 2000:
427: fixed a bug introduced in august 1997 that caused expressions
428: like $f[1] to be syntax errors. thanks to arnold robbins for
429: noticing this and providing a fix.
430:
431: Oct 30, 2000:
432: fixed some nextfile bugs: not handling all cases. thanks to
433: arnold robbins for pointing this out. new regressions added.
434:
435: close() is now a function. it returns whatever the library
436: fclose returns, and -1 for closing a file or pipe that wasn't
437: opened.
438:
439: Sep 24, 2000:
440: permit \n explicitly in character classes; won't work right
441: if comes in as "[\n]" but ok as /[\n]/, because of multiple
442: processing of \'s. thanks to arnold robbins.
443:
444: July 5, 2000:
445: minor fiddles in tran.c to keep compilers happy about uschar.
446: thanks to norman wilson.
447:
448: May 25, 2000:
449: yet another attempt at making 8-bit input work, with another
450: band-aid in b.c (member()), and some (uschar) casts to head
451: off potential errors in subscripts (like isdigit). also
452: changed HAT to NCHARS-2. thanks again to santiago vila.
453:
454: changed maketab.c to ignore apparently out of range definitions
455: instead of halting; new freeBSD generates one. thanks to
456: jon snader <jsnader@ix.netcom.com> for pointing out the problem.
457:
458: May 2, 2000:
459: fixed an 8-bit problem in b.c by making several char*'s into
460: unsigned char*'s. not clear i have them all yet. thanks to
461: Santiago Vila <sanvila@unex.es> for the bug report.
462:
463: Apr 21, 2000:
464: finally found and fixed a memory leak in function call; it's
465: been there since functions were added ~1983. thanks to
466: jon bentley for the test case that found it.
467:
468: added test in envinit to catch environment "variables" with
1.12 otto 469: names beginning with '='; thanks to Berend Hasselman.
1.10 millert 470:
471: Jul 28, 1999:
472: added test in defn() to catch function foo(foo), which
473: otherwise recurses until core dump. thanks to arnold
474: robbins for noticing this.
475:
1.9 millert 476: Jun 20, 1999:
477: added *bp in gettok in lex.c; appears possible to exit function
478: without terminating the string. thanks to russ cox.
479:
480: Jun 2, 1999:
481: added function stdinit() to run to initialize files[] array,
482: in case stdin, etc., are not constants; some compilers care.
483:
484: May 10, 1999:
485: replaced the ERROR ... FATAL, etc., macros with functions
486: based on vprintf, to avoid problems caused by overrunning
487: fixed-size errbuf array. thanks to ralph corderoy for the
488: impetus, and for pointing out a string termination bug in
489: qstring as well.
490:
491: Apr 21, 1999:
492: fixed bug that caused occasional core dumps with commandline
493: variable with value ending in \. (thanks to nelson beebe for
494: the test case.)
495:
1.8 millert 496: Apr 16, 1999:
497: with code kindly provided by Bruce Lilly, awk now parses
498: /=/ and similar constructs more sensibly in more places.
1.10 millert 499: Bruce also provided some helpful test cases.
1.8 millert 500:
501: Apr 5, 1999:
502: changed true/false to True/False in run.c to make it
503: easier to compile with C++. Added some casts on malloc
504: and realloc to be honest about casts; ditto. changed
505: ltype int to long in struct rrow to reduce some 64-bit
506: complaints; other changes scattered throughout for the
507: same purpose. thanks to Nelson Beebe for these portability
508: improvements.
509:
510: removed some horrible pointer-int casting in b.c and elsewhere
511: by adding ptoi and itonp to localize the casts, which are
512: all benign. fixed one incipient bug that showed up on sgi
513: in 64-bit mode.
514:
515: reset lineno for new source file; include filename in error
516: message. also fixed line number error in continuation lines.
517: (thanks to Nelson Beebe for both of these.)
518:
519: Mar 24, 1999:
520: Nelson Beebe notes that irix 5.3 yacc dies with a bogus
521: error; use a newer version or switch to bison, since sgi
522: is unlikely to fix it.
1.1 tholo 523:
1.7 millert 524: Mar 5, 1999:
1.9 millert 525: changed isnumber to is_number to avoid the problem caused by
526: versions of ctype.h that include the name isnumber.
1.7 millert 527:
528: distribution now includes a script for building on a Mac,
529: thanks to Dan Allen.
530:
531: Feb 20, 1999:
532: fixed memory leaks in run.c (call) and tran.c (setfval).
533: thanks to Stephen Nutt for finding these and providing the fixes.
534:
535: Jan 13, 1999:
536: replaced srand argument by (unsigned int) in run.c;
537: avoids problem on Mac and potentially on Unix & Windows.
538: thanks to Dan Allen.
539:
540: added a few (int) casts to silence useless compiler warnings.
541: e.g., errorflag= in run.c jump().
542:
543: added proctab.c to the bundle outout; one less thing
544: to have to compile out of the box.
545:
546: added calls to _popen and _pclose to the win95 stub for
547: pipes (thanks to Steve Adams for this helpful suggestion).
548: seems to work, though properties are not well understood
549: by me, and it appears that under some circumstances the
550: pipe output is truncated. Be careful.
551:
552: Oct 19, 1998:
553: fixed a couple of bugs in getrec: could fail to update $0
554: after a getline var; because inputFS wasn't initialized,
555: could split $0 on every character, a misleading diversion.
556:
557: fixed caching bug in makedfa: LRU was actually removing
558: least often used.
559:
560: thanks to ross ridge for finding these, and for providing
561: great bug reports.
562:
563: May 12, 1998:
564: fixed potential bug in readrec: might fail to update record
565: pointer after growing. thanks to dan levy for spotting this
566: and suggesting the fix.
567:
568: Mar 12, 1998:
569: added -V to print version number and die.
1.19 millert 570:
571: [notify dave kerns, dkerns@dacsoup.ih.lucent.com]
1.7 millert 572:
573: Feb 11, 1998:
574: subtle silent bug in lex.c: if the program ended with a number
575: longer than 1 digit, part of the input would be pushed back and
576: parsed again because token buffer wasn't terminated right.
577: example: awk 'length($0) > 10'. blush. at least i found it
578: myself.
579:
580: Aug 31, 1997:
581: s/adelete/awkdelete/: SGI uses this in malloc.h.
582: thanks to nelson beebe for pointing this one out.
583:
584: Aug 21, 1997:
585: fixed some bugs in sub and gsub when replacement includes \\.
586: this is a dark, horrible corner, but at least now i believe that
587: the behavior is the same as gawk and the intended posix standard.
588: thanks to arnold robbins for advice here.
589:
590: Aug 9, 1997:
591: somewhat regretfully, replaced the ancient lex-based lexical
592: analyzer with one written in C. it's longer, generates less code,
593: and more portable; the old one depended too much on mysterious
594: properties of lex that were not preserved in other environments.
595: in theory these recognize the same language.
596:
597: now using strtod to test whether a string is a number, instead of
598: the convoluted original function. should be more portable and
599: reliable if strtod is implemented right.
600:
601: removed now-pointless optimization in makefile that tries to avoid
602: recompilation when awkgram.y is changed but symbols are not.
603:
604: removed most fixed-size arrays, though a handful remain, some
605: of which are unchecked. you have been warned.
606:
607: Aug 4, 1997:
608: with some trepidation, replaced the ancient code that managed
609: fields and $0 in fixed-size arrays with arrays that grow on
610: demand. there is still some tension between trying to make this
611: run fast and making it clean; not sure it's right yet.
612:
613: the ill-conceived -mr and -mf arguments are now useful only
614: for debugging. previous dynamic string code removed.
615:
616: numerous other minor cleanups along the way.
1.1 tholo 617:
1.7 millert 618: Jul 30, 1997:
619: using code provided by dan levy (to whom profuse thanks), replaced
620: fixed-size arrays and awkward kludges by a fairly uniform mechanism
621: to grow arrays as needed for printf, sub, gsub, etc.
1.1 tholo 622:
1.7 millert 623: Jul 23, 1997:
624: falling off the end of a function returns "" and 0, not 0.
625: thanks to arnold robbins.
1.1 tholo 626:
1.7 millert 627: Jun 17, 1997:
628: replaced several fixed-size arrays by dynamically-created ones
629: in run.c; added overflow tests to some previously unchecked cases.
630: getline, toupper, tolower.
1.1 tholo 631:
1.7 millert 632: getline code is still broken in that recursive calls may wind
633: up using the same space. [fixed later]
1.1 tholo 634:
1.7 millert 635: increased RECSIZE to 8192 to push problems further over the horizon.
1.1 tholo 636:
1.7 millert 637: added \r to \n as input line separator for programs, not data.
638: damn CRLFs.
1.1 tholo 639:
1.7 millert 640: modified format() to permit explicit printf("%c", 0) to include
641: a null byte in output. thanks to ken stailey for the fix.
1.1 tholo 642:
1.7 millert 643: added a "-safe" argument that disables file output (print >,
644: print >>), process creation (cmd|getline, print |, system), and
645: access to the environment (ENVIRON). this is a first approximation
646: to a "safe" version of awk, but don't rely on it too much. thanks
647: to joan feigenbaum and matt blaze for the inspiration long ago.
1.1 tholo 648:
1.7 millert 649: Jul 8, 1996:
650: fixed long-standing bug in sub, gsub(/a/, "\\\\&"); thanks to
651: ralph corderoy.
1.1 tholo 652:
1.7 millert 653: Jun 29, 1996:
654: fixed awful bug in new field splitting; didn't get all the places
655: where input was done.
1.1 tholo 656:
1.7 millert 657: Jun 28, 1996:
658: changed field-splitting to conform to posix definition: fields are
659: split using the value of FS at the time of input; it used to be
660: the value when the field or NF was first referred to, a much less
661: predictable definition. thanks to arnold robbins for encouragement
662: to do the right thing.
1.1 tholo 663:
1.7 millert 664: May 28, 1996:
665: fixed appalling but apparently unimportant bug in parsing octal
666: numbers in reg exprs.
1.1 tholo 667:
1.7 millert 668: explicit hex in reg exprs now limited to 2 chars: \xa, \xaa.
1.1 tholo 669:
1.7 millert 670: May 27, 1996:
671: cleaned up some declarations so gcc -Wall is now almost silent.
1.1 tholo 672:
1.7 millert 673: makefile now includes backup copies of ytab.c and lexyy.c in case
674: one makes before looking; it also avoids recreating lexyy.c unless
675: really needed.
1.1 tholo 676:
1.7 millert 677: s/aprintf/awkprint, s/asprintf/awksprintf/ to avoid some name clashes
678: with unwisely-written header files.
1.1 tholo 679:
1.7 millert 680: thanks to jeffrey friedl for several of these.
1.1 tholo 681:
1.7 millert 682: May 26, 1996:
683: an attempt to rationalize the (unsigned) char issue. almost all
684: instances of unsigned char have been removed; the handful of places
685: in b.c where chars are used as table indices have been hand-crafted.
686: added some latin-1 tests to the regression, but i'm not confident;
687: none of my compilers seem to care much. thanks to nelson beebe for
688: pointing out some others that do care.
1.1 tholo 689:
1.7 millert 690: May 2, 1996:
691: removed all register declarations.
1.1 tholo 692:
1.7 millert 693: enhanced split(), as in gawk, etc: split(s, a, "") splits s into
694: a[1]...a[length(s)] with each character a single element.
1.1 tholo 695:
1.7 millert 696: made the same changes for field-splitting if FS is "".
1.1 tholo 697:
1.7 millert 698: added nextfile, as in gawk: causes immediate advance to next
699: input file. (thanks to arnold robbins for inspiration and code).
1.1 tholo 700:
1.7 millert 701: small fixes to regexpr code: can now handle []], [[], and
702: variants; [] is now a syntax error, rather than matching
703: everything; [z-a] is now empty, not z. far from complete
704: or correct, however. (thanks to jeffrey friedl for pointing out
705: some awful behaviors.)
1.1 tholo 706:
1.7 millert 707: Apr 29, 1996:
1.12 otto 708: replaced uchar by uschar everywhere; apparently some compilers
1.7 millert 709: usurp this name and this causes conflicts.
1.1 tholo 710:
1.7 millert 711: fixed call to time in run.c (bltin); arg is time_t *.
1.1 tholo 712:
1.7 millert 713: replaced horrible pointer/long punning in b.c by a legitimate
714: union. should be safer on 64-bit machines and cleaner everywhere.
715: (thanks to nelson beebe for pointing out some of these problems.)
1.1 tholo 716:
1.7 millert 717: replaced nested comments by #if 0...#endif in run.c, lib.c.
1.1 tholo 718:
1.7 millert 719: removed getsval, setsval, execute macros from run.c and lib.c.
720: machines are 100x faster than they were when these macros were
721: first used.
1.1 tholo 722:
1.7 millert 723: revised filenames: awk.g.y => awkgram.y, awk.lx.l => awklex.l,
724: y.tab.[ch] => ytab.[ch], lex.yy.c => lexyy.c, all in the aid of
725: portability to nameless systems.
1.1 tholo 726:
1.7 millert 727: "make bundle" now includes yacc and lex output files for recipients
728: who don't have yacc or lex.
1.1 tholo 729:
1.7 millert 730: Aug 15, 1995:
731: initialized Cells in setsymtab more carefully; some fields
732: were not set. (thanks to purify, all of whose complaints i
733: think i now understand.)
1.1 tholo 734:
1.7 millert 735: fixed at least one error in gsub that looked at -1-th element
736: of an array when substituting for a null match (e.g., $).
1.1 tholo 737:
1.7 millert 738: delete arrayname is now legal; it clears the elements but leaves
739: the array, which may not be the right behavior.
1.1 tholo 740:
1.7 millert 741: modified makefile: my current make can't cope with the test used
742: to avoid unnecessary yacc invocations.
1.1 tholo 743:
1.7 millert 744: Jul 17, 1995:
745: added dynamically growing strings to awk.lx.l and b.c
746: to permit regular expressions to be much bigger.
747: the state arrays can still overflow.
1.1 tholo 748:
1.7 millert 749: Aug 24, 1994:
750: detect duplicate arguments in function definitions (mdm).
1.1 tholo 751:
1.7 millert 752: May 11, 1994:
753: trivial fix to printf to limit string size in sub().
1.1 tholo 754:
1.7 millert 755: Apr 22, 1994:
756: fixed yet another subtle self-assignment problem:
757: $1 = $2; $1 = $1 clobbered $1.
1.1 tholo 758:
1.7 millert 759: Regression tests now use private echo, to avoid quoting problems.
1.1 tholo 760:
1.7 millert 761: Feb 2, 1994:
762: changed error() to print line number as %d, not %g.
1.1 tholo 763:
1.7 millert 764: Jul 23, 1993:
765: cosmetic changes: increased sizes of some arrays,
766: reworded some error messages.
1.1 tholo 767:
1.7 millert 768: added CONVFMT as in posix (just replaced OFMT in getsval)
1.1 tholo 769:
1.7 millert 770: FILENAME is now "" until the first thing that causes a file
771: to be opened.
1.1 tholo 772:
1.7 millert 773: Nov 28, 1992:
774: deleted yyunput and yyoutput from proto.h;
775: different versions of lex give these different declarations.
1.1 tholo 776:
1.7 millert 777: May 31, 1992:
778: added -mr N and -mf N options: more record and fields.
779: these really ought to adjust automatically.
1.1 tholo 780:
1.7 millert 781: cleaned up some error messages; "out of space" now means
782: malloc returned NULL in all cases.
1.1 tholo 783:
1.7 millert 784: changed rehash so that if it runs out, it just returns;
785: things will continue to run slow, but maybe a bit longer.
1.1 tholo 786:
1.7 millert 787: Apr 24, 1992:
788: remove redundant close of stdin when using -f -.
1.1 tholo 789:
1.7 millert 790: got rid of core dump with -d; awk -d just prints date.
1.1 tholo 791:
1.7 millert 792: Apr 12, 1992:
793: added explicit check for /dev/std(in,out,err) in redirection.
794: unlike gawk, no /dev/fd/n yet.
1.1 tholo 795:
1.7 millert 796: added (file/pipe) builtin. hard to test satisfactorily.
797: not posix.
1.1 tholo 798:
1.7 millert 799: Feb 20, 1992:
800: recompile after abortive changes; should be unchanged.
1.1 tholo 801:
1.7 millert 802: Dec 2, 1991:
803: die-casting time: converted to ansi C, installed that.
1.1 tholo 804:
1.7 millert 805: Nov 30, 1991:
806: fixed storage leak in freefa, failing to recover [N]CCL.
807: thanks to Bill Jones (jones@cs.usask.ca)
1.1 tholo 808:
1.7 millert 809: Nov 19, 1991:
810: use RAND_MAX instead of literal in builtin().
1.1 tholo 811:
1.7 millert 812: Nov 12, 1991:
813: cranked up some fixed-size arrays in b.c, and added a test for
814: overflow in penter. thanks to mark larsen.
1.1 tholo 815:
1.7 millert 816: Sep 24, 1991:
817: increased buffer in gsub. a very crude fix to a general problem.
818: and again on Sep 26.
1.1 tholo 819:
1.7 millert 820: Aug 18, 1991:
821: enforce variable name syntax for commandline variables: has to
822: start with letter or _.
1.1 tholo 823:
1.7 millert 824: Jul 27, 1991:
825: allow newline after ; in for statements.
1.1 tholo 826:
1.7 millert 827: Jul 21, 1991:
828: fixed so that in self-assignment like $1=$1, side effects
829: like recomputing $0 take place. (this is getting subtle.)
1.1 tholo 830:
1.7 millert 831: Jun 30, 1991:
832: better test for detecting too-long output record.
1.1 tholo 833:
1.7 millert 834: Jun 2, 1991:
835: better defense against very long printf strings.
836: made break and continue illegal outside of loops.
1.1 tholo 837:
1.7 millert 838: May 13, 1991:
839: removed extra arg on gettemp, tempfree. minor error message rewording.
1.1 tholo 840:
841: May 6, 1991:
842: fixed silly bug in hex parsing in hexstr().
843: removed an apparently unnecessary test in isnumber().
844: warn about weird printf conversions.
845: fixed unchecked array overwrite in relex().
846:
847: changed for (i in array) to access elements in sorted order.
848: then unchanged it -- it really does run slower in too many cases.
849: left the code in place, commented out.
850:
1.7 millert 851: Feb 10, 1991:
852: check error status on all writes, to avoid banging on full disks.
1.1 tholo 853:
1.7 millert 854: Jan 28, 1991:
855: awk -f - reads the program from stdin.
1.1 tholo 856:
1.7 millert 857: Jan 11, 1991:
858: failed to set numeric state on $0 in cmd|getline context in run.c.
1.1 tholo 859:
1.7 millert 860: Nov 2, 1990:
861: fixed sleazy test for integrality in getsval; use modf.
1.1 tholo 862:
1.7 millert 863: Oct 29, 1990:
864: fixed sleazy buggy code in lib.c that looked (incorrectly) for
865: too long input lines.
1.1 tholo 866:
1.7 millert 867: Oct 14, 1990:
868: fixed the bug on p. 198 in which it couldn't deduce that an
869: argument was an array in some contexts. replaced the error
870: message in intest() by code that damn well makes it an array.
1.1 tholo 871:
1.7 millert 872: Oct 8, 1990:
873: fixed horrible bug: types and values were not preserved in
874: some kinds of self-assignment. (in assign().)
1.1 tholo 875:
1.7 millert 876: Aug 24, 1990:
877: changed NCHARS to 256 to handle 8-bit characters in strings
878: presented to match(), etc.
1.1 tholo 879:
1.7 millert 880: Jun 26, 1990:
881: changed struct rrow (awk.h) to use long instead of int for lval,
882: since cfoll() stores a pointer in it. now works better when int's
883: are smaller than pointers!
1.1 tholo 884:
1.7 millert 885: May 6, 1990:
886: AVA fixed the grammar so that ! is uniformly of the same precedence as
887: unary + and -. This renders illegal some constructs like !x=y, which
888: now has to be parenthesized as !(x=y), and makes others work properly:
889: !x+y is (!x)+y, and x!y is x !y, not two pattern-action statements.
890: (These problems were pointed out by Bob Lenk of Posix.)
1.1 tholo 891:
1.7 millert 892: Added \x to regular expressions (already in strings).
893: Limited octal to octal digits; \8 and \9 are not octal.
894: Centralized the code for parsing escapes in regular expressions.
895: Added a bunch of tests to T.re and T.sub to verify some of this.
1.1 tholo 896:
1.7 millert 897: Feb 9, 1990:
898: fixed null pointer dereference bug in main.c: -F[nothing]. sigh.
1.1 tholo 899:
1.7 millert 900: restored srand behavior: it returns the current seed.
1.1 tholo 901:
1.7 millert 902: Jan 18, 1990:
903: srand now returns previous seed value (0 to start).
1.1 tholo 904:
1.7 millert 905: Jan 5, 1990:
906: fix potential problem in tran.c -- something was freed,
907: then used in freesymtab.
1.1 tholo 908:
1.7 millert 909: Oct 18, 1989:
910: another try to get the max number of open files set with
911: relatively machine-independent code.
1.1 tholo 912:
1.7 millert 913: small fix to input() in case of multiple reads after EOF.
1.1 tholo 914:
1.7 millert 915: Oct 11, 1989:
916: FILENAME is now defined in the BEGIN block -- too many old
917: programs broke.
1.1 tholo 918:
1.7 millert 919: "-" means stdin in getline as well as on the commandline.
1.1 tholo 920:
1.7 millert 921: added a bunch of casts to the code to tell the truth about
922: char * vs. unsigned char *, a right royal pain. added a
923: setlocale call to the front of main, though probably no one
924: has it usefully implemented yet.
1.1 tholo 925:
1.7 millert 926: Aug 24, 1989:
927: removed redundant relational tests against nullnode if parse
928: tree already had a relational at that point.
1.1 tholo 929:
1.7 millert 930: Aug 11, 1989:
931: fixed bug: commandline variable assignment has to look like
932: var=something. (consider the man page for =, in file =.1)
1.1 tholo 933:
1.7 millert 934: changed number of arguments to functions to static arrays
935: to avoid repeated malloc calls.
1.1 tholo 936:
1.7 millert 937: Aug 2, 1989:
938: restored -F (space) separator
1.1 tholo 939:
1.7 millert 940: Jul 30, 1989:
941: added -v x=1 y=2 ... for immediate commandline variable assignment;
942: done before the BEGIN block for sure. they have to precede the
943: program if the program is on the commandline.
944: Modified Aug 2 to require a separate -v for each assignment.
1.1 tholo 945:
1.7 millert 946: Jul 10, 1989:
947: fixed ref-thru-zero bug in environment code in tran.c
1.1 tholo 948:
1.7 millert 949: Jun 23, 1989:
950: add newline to usage message.
1.1 tholo 951:
1.7 millert 952: Jun 14, 1989:
953: added some missing ansi printf conversion letters: %i %X %E %G.
954: no sensible meaning for h or L, so they may not do what one expects.
1.1 tholo 955:
1.7 millert 956: made %* conversions work.
1.1 tholo 957:
1.7 millert 958: changed x^y so that if n is a positive integer, it's done
959: by explicit multiplication, thus achieving maximum accuracy.
960: (this should be done by pow() but it seems not to be locally.)
961: done to x ^= y as well.
1.1 tholo 962:
1.7 millert 963: Jun 4, 1989:
964: ENVIRON array contains environment: if shell variable V=thing,
965: ENVIRON["V"] is "thing"
1.1 tholo 966:
1.7 millert 967: multiple -f arguments permitted. error reporting is naive.
968: (they were permitted before, but only the last was used.)
1.1 tholo 969:
1.7 millert 970: fixed a really stupid botch in the debugging macro dprintf
1.1 tholo 971:
1.7 millert 972: fixed order of evaluation of commandline assignments to match
973: what the book claims: an argument of the form x=e is evaluated
974: at the time it would have been opened if it were a filename (p 63).
975: this invalidates the suggested answer to ex 4-1 (p 195).
1.1 tholo 976:
1.7 millert 977: removed some code that permitted -F (space) fieldseparator,
978: since it didn't quite work right anyway. (restored aug 2)
1.1 tholo 979:
1.7 millert 980: Apr 27, 1989:
981: Line number now accumulated correctly for comment lines.
1.1 tholo 982:
1.7 millert 983: Apr 26, 1989:
984: Debugging output now includes a version date,
985: if one compiles it into the source each time.
1.1 tholo 986:
1.7 millert 987: Apr 9, 1989:
988: Changed grammar to prohibit constants as 3rd arg of sub and gsub;
989: prevents class of overwriting-a-constant errors. (Last one?)
990: This invalidates the "banana" example on page 43 of the book.
1.1 tholo 991:
1.7 millert 992: Added \a ("alert"), \v (vertical tab), \xhhh (hexadecimal),
993: as in ANSI, for strings. Rescinded the sloppiness that permitted
994: non-octal digits in \ooo. Warning: not all compilers and libraries
995: will be able to deal with \x correctly.
1.1 tholo 996:
1.7 millert 997: Jan 9, 1989:
998: Fixed bug that caused tempcell list to contain a duplicate.
999: The fix is kludgy.
1.1 tholo 1000:
1.7 millert 1001: Dec 17, 1988:
1002: Catches some more commandline errors in main.
1003: Removed redundant decl of modf in run.c (confuses some compilers).
1004: Warning: there's no single declaration of malloc, etc., in awk.h
1005: that seems to satisfy all compilers.
1.1 tholo 1006:
1.7 millert 1007: Dec 7, 1988:
1008: Added a bit of code to error printing to avoid printing nulls.
1009: (Not clear that it actually would.)
1.1 tholo 1010:
1.7 millert 1011: Nov 27, 1988:
1012: With fear and trembling, modified the grammar to permit
1013: multiple pattern-action statements on one line without
1014: an explicit separator. By definition, this capitulation
1015: to the ghost of ancient implementations remains undefined
1016: and thus subject to change without notice or apology.
1017: DO NOT COUNT ON IT.
1.1 tholo 1018:
1.7 millert 1019: Oct 30, 1988:
1020: Fixed bug in call() that failed to recover storage.
1.1 tholo 1021:
1.7 millert 1022: A warning is now generated if there are more arguments
1023: in the call than in the definition (in lieu of fixing
1024: another storage leak).
1.2 millert 1025:
1.7 millert 1026: Oct 20, 1988:
1027: Fixed %c: if expr is numeric, use numeric value;
1028: otherwise print 1st char of string value. still
1029: doesn't work if the value is 0 -- won't print \0.
1.2 millert 1030:
1.7 millert 1031: Added a few more checks for running out of malloc.
1.2 millert 1032:
1.7 millert 1033: Oct 12, 1988:
1034: Fixed bug in call() that freed local arrays twice.
1.5 kstailey 1035:
1.7 millert 1036: Fixed to handle deletion of non-existent array right;
1037: complains about attempt to delete non-array element.
1.6 kstailey 1038:
1.7 millert 1039: Sep 30, 1988:
1040: Now guarantees to evaluate all arguments of built-in
1041: functions, as in C; the appearance is that arguments
1042: are evaluated before the function is called. Places
1043: affected are sub (gsub was ok), substr, printf, and
1044: all the built-in arithmetic functions in bltin().
1045: A warning is generated if a bltin() is called with
1046: the wrong number of arguments.
1.6 kstailey 1047:
1.7 millert 1048: This requires changing makeprof on p167 of the book.
1.6 kstailey 1049:
1.7 millert 1050: Aug 23, 1988:
1051: setting FILENAME in BEGIN caused core dump, apparently
1052: because it was freeing space not allocated by malloc.
1.6 kstailey 1053:
1.7 millert 1054: July 24, 1988:
1055: fixed egregious error in toupper/tolower functions.
1056: still subject to rescinding, however.
1.6 kstailey 1057:
1.7 millert 1058: July 2, 1988:
1059: flush stdout before opening file or pipe
1.6 kstailey 1060:
1.7 millert 1061: July 2, 1988:
1062: performance bug in b.c/cgoto(): not freeing some sets of states.
1063: partial fix only right now, and the number of states increased
1064: to make it less obvious.
1.6 kstailey 1065:
1.7 millert 1066: June 1, 1988:
1067: check error status on close
1.6 kstailey 1068:
1.7 millert 1069: May 28, 1988:
1070: srand returns seed value it's using.
1071: see 1/18/90
1.6 kstailey 1072:
1.7 millert 1073: May 22, 1988:
1074: Removed limit on depth of function calls.
1.6 kstailey 1075:
1.7 millert 1076: May 10, 1988:
1077: Fixed lib.c to permit _ in commandline variable names.
1.6 kstailey 1078:
1.7 millert 1079: Mar 25, 1988:
1080: main.c fixed to recognize -- as terminator of command-
1081: line options. Illegal options flagged.
1082: Error reporting slightly cleaned up.
1.6 kstailey 1083:
1.7 millert 1084: Dec 2, 1987:
1085: Newer C compilers apply a strict scope rule to extern
1086: declarations within functions. Two extern declarations in
1087: lib.c and tran.c have been moved to obviate this problem.
1.6 kstailey 1088:
1.7 millert 1089: Oct xx, 1987:
1090: Reluctantly added toupper and tolower functions.
1091: Subject to rescinding without notice.
1.6 kstailey 1092:
1.7 millert 1093: Sep 17, 1987:
1094: Error-message printer had printf(s) instead of
1095: printf("%s",s); got core dumps when the message
1096: included a %.
1.6 kstailey 1097:
1.7 millert 1098: Sep 12, 1987:
1099: Very long printf strings caused core dump;
1100: fixed aprintf, asprintf, format to catch them.
1101: Can still get a core dump in printf itself.
1.6 kstailey 1102:
1103: