[BACK]Return to FIXES CVS log [TXT][DIR] Up to [local] / src / usr.bin / awk

Annotation of src/usr.bin/awk/FIXES, Revision 1.36

1.36    ! millert     1: /*     $OpenBSD: FIXES,v 1.35 2020/06/26 15:57:39 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.
1.36    ! millert    28:
        !            29: July 2, 2020:
        !            30:        Merge PRs 85 and 86 which fix regressions. Thanks to
        !            31:        Tim van der Molen for the fixes.
1.35      millert    32:
                     33: June 25, 2020:
                     34:        Merge PRs 82 and 84. The latter fixes issue #83. Thanks to
                     35:        Todd Miller and awkfan77.
1.34      millert    36:
                     37: June 12, 2020:
                     38:        Clear errno before calling errcheck to avoid any spurious errors
                     39:        left over from previous calls that may have set it. Thanks to
                     40:        Todd Miller for the fix, from PR #80.
                     41:
                     42:        Fix Issue #78 by allowing \r to follow floating point numbers in
                     43:        lib.c:is_number. Thanks to GitHub user ajcarr for the report
                     44:        and to Arnold Robbins for the fix.
1.33      millert    45:
                     46: June 5, 2020:
                     47:        In fldbld(), make sure that inputFS is set before trying to
                     48:        use it. Thanks to  Steffen Nurpmeso <steffen@sdaoden.eu>
                     49:        for the report.
                     50:
                     51: May 5, 2020:
                     52:        Fix checks for compilers that can handle noreturn. Thanks to
                     53:        GitHub user enh-google for pointing it out. Closes Issue #79.
                     54:
                     55: April 16, 2020:
                     56:        Handle old compilers that don't support C11 (for noreturn).
                     57:        Thanks to Arnold Robbins.
                     58:
                     59: April 5, 2020:
                     60:        Use <stdnoreturn.h> and noreturn instead of GCC attributes.
                     61:        Thanks to GitHub user awkfan77. Closes PR #77.
1.30      millert    62:
1.32      millert    63: February 28, 2020:
                     64:        More cleanups from Christos Zoulas: notably backslash continuation
                     65:        inside strings removes the newline and a fix for RS = "^a".
                     66:        Fix for address sanitizer-found problem. Thanks to GitHub user
                     67:        enh-google.
                     68:
                     69: February 19, 2020:
                     70:        More small cleanups from Christos Zoulas.
                     71:
                     72: February 18, 2020:
                     73:        Additional cleanups from Christos Zoulas. It's no longer necessary
                     74:        to use the -y flag to bison.
                     75:
                     76: February 6, 2020:
                     77:        Additional small cleanups from Christos Zoulas. awk is now
                     78:        a little more robust about reporting I/O errors upon exit.
                     79:
1.31      millert    80: January 31, 2020:
                     81:        Merge PR #70, which avoids use of variable length arrays. Thanks
                     82:        to GitHub user michaelforney.  Fix issue #60 ({0} in interval
1.32      millert    83:        expressions doesn't work).  Also get all tests working again.
                     84:        Thanks to Arnold Robbins.
1.31      millert    85:
                     86: January 24, 2020:
                     87:        A number of small cleanups from Christos Zoulas.  Add the close
                     88:        on exec flag to files/pipes opened for redirection; courtesy of
                     89:        Arnold Robbins.
                     90:
                     91: January 19, 2020:
                     92:        If POSIXLY_CORRECT is set in the environment, then sub and gsub
                     93:        use POSIX rules for multiple backslashes.  This fixes Issue #66,
                     94:        while maintaining backwards compatibility.
                     95:
                     96: January 9, 2020:
                     97:        Input/output errors on closing files are now fatal instead of
                     98:        mere warnings. Thanks to Martijn Dekker <martijn@inlv.org>.
                     99:
1.30      millert   100: January 5, 2020:
                    101:        Fix a bug in the concatentation of two string constants into
                    102:        one done in the grammar.  Fixes GitHub issue #61.  Thanks
                    103:        to GitHub user awkfan77 for pointing out the direction for
                    104:        the fix.  New test T.concat added to the test suite.
                    105:        Fix a few memory leaks reported by valgrind, as well.
                    106:
                    107: December 27, 2019:
                    108:        Fix a bug whereby a{0,3} could match four a's.  Thanks to
                    109:        "Anonymous AWK fan" for the report.
                    110:
                    111: December 11, 2019:
                    112:        Further printf-related fixes for 32 bit systems.
                    113:        Thanks again to Christos Zoulas.
                    114:
                    115: December 8, 2019:
                    116:        Fix the return value of sprintf("%d") on 32 bit systems.
                    117:        Thanks to Jim Lowe for the report and to Christos Zoulas
                    118:        for the fix.
1.29      millert   119:
                    120: November 10, 2019:
                    121:        Convert a number of Boolean integer variables into
                    122:        actual bools. Convert compile_time variable into an
                    123:        enum and simplify some of the related code.  Thanks
                    124:        to Arnold Robbins.
                    125:
                    126: November 8, 2019:
                    127:        Fix from Ori Bernstein to get UTF-8 characters instead of
                    128:        bytes when FS = "".  This is currently the only bit of
                    129:        the One True Awk that understands multibyte characters.
                    130:        From Arnold Robbins, apply some cleanups in the test suite.
1.28      millert   131:
                    132: October 25, 2019:
                    133:        More fixes and cleanups from NetBSD, courtesy of Christos
                    134:        Zoulas. Merges PRs 54 and 55.
                    135:
                    136: October 24, 2019:
                    137:        Import second round of code cleanups from NetBSD. Much thanks
1.31      millert   138:        to Christos Zoulas (GitHub user zoulasc). Merges PR 53.
1.28      millert   139:        Add an optimization for string concatenation, also from
                    140:        Christos.
1.27      millert   141:
                    142: October 17, 2019:
                    143:        Import code cleanups from NetBSD. Much thanks to Christos
1.31      millert   144:        Zoulas (GitHub user zoulasc). Merges PR 51.
1.26      millert   145:
                    146: October 6, 2019:
                    147:        Import code from NetBSD awk that implements RS as a regular
                    148:        expression.
1.23      millert   149:
1.25      millert   150: September 10, 2019:
                    151:        Fixes for various array / memory overruns found via gcc's
1.31      millert   152:        -fsanitize=unknown. Thanks to Alexander Richardson (GitHub
1.25      millert   153:        user arichardson). Merges PRs 47 and 48.
                    154:
                    155: July 28, 2019:
                    156:        Import grammar optimization from NetBSD: Two string constants
                    157:        concatenated together get turned into a single string.
                    158:
                    159: July 26, 2019:
                    160:        Support POSIX-specified C-style escape sequences "\a" (alarm)
                    161:        and "\v" (vertical tab) in command line arguments and regular
                    162:        expressions, further to the support for them in strings added on
                    163:        Apr 9, 1989. These now no longer match as literal "a" and "v"
                    164:        characters (as they don't on other awk implementations).
                    165:        Thanks to Martijn Dekker.
                    166:
                    167: July 17, 2019:
                    168:        Pull in a number of code cleanups and minor fixes from
                    169:        Warner Losh's bsd-ota branch.  The only user visible change
                    170:        is the use of random(3) as the random number generator.
                    171:        Thanks to Warner Losh for collecting all these fixes in
                    172:        one easy place to get them from.
                    173:
                    174: July 16, 2019:
                    175:        Fix field splitting to use FS value as of the time a record
                    176:        was read or assigned to.  Thanks to GitHub user Cody Mello (melloc)
                    177:        for the fix. (Merged from his branch, via PR #42.) Updated
                    178:        testdir/T.split per said PR as well.
                    179:
                    180: June 24, 2019:
                    181:        Extract awktest.tar into testdir directory. Add some very
                    182:        simple mechanics to the makefile for running the tests and
                    183:        for cleaning up. No changes to awk itself.
                    184:
1.24      millert   185: June 17, 2019:
                    186:        Disallow deleting SYMTAB and its elements, which creates
                    187:        use-after-free bugs. Thanks to GitHub user Cody Mello (melloc)
                    188:        for the fix. (Merged from PR #43.)
                    189:
                    190: June 5, 2019:
                    191:        Allow unmatched right parenthesis in a regular expression to
                    192:        be treated literally. Fixes Issue #40. Thanks to GitHub user
                    193:        Warner Losh (bsdimp) for the report. Thanks to Arnold Robbins
                    194:        for the fix.
                    195:
                    196: May 29,2019:
                    197:        Fix check for command line arguments to no longer require that
                    198:        first character after '=' not be another '='. Reverts change of
                    199:        August 11, 1989. Thanks to GitHub user Jamie Landeg Jones for
                    200:        pointing out the issue; from Issue #38.
                    201:
                    202: Apr 7, 2019:
                    203:        Update awktest.tar(p.50) to use modern options to sort. Needed
                    204:        for Android development. Thanks to GitHub user mohd-akram (Mohamed
                    205:        Akram).  From Issue #33.
                    206:
                    207: Mar 12, 2019:
                    208:        Added very simplistic support for cross-compiling in the
                    209:        makefile.  We are NOT going to go in the direction of the
                    210:        autotools, though.  Thanks to GitHub user nee-san for
                    211:        the basic change. (Merged from PR #34.)
                    212:
1.23      millert   213: Mar 5, 2019:
                    214:        Added support for POSIX-standard interval expressions (a.k.a.
                    215:        bounds, a.k.a. repetition expressions) in regular expressions,
                    216:        backported (via NetBSD) from Apple awk-24 (20070501).
                    217:        Thanks to Martijn Dekker <martijn@inlv.org> for the port.
                    218:        (Merged from PR #30.)
                    219:
                    220: Mar 3, 2019:
                    221:        Merge PRs as follows:
                    222:        #12: Avoid undefined behaviour when using ctype(3) functions in
                    223:             relex(). Thanks to GitHub user iamleot.
                    224:        #31: Make getline handle numeric strings, and update FIXES. Thanks
1.24      millert   225:             to GitHub user arnoldrobbins.
1.23      millert   226:        #32: maketab: support build systems with read-only source. Thanks
                    227:             to GitHub user enh.
1.22      millert   228:
                    229: Jan 25, 2019:
                    230:        Make getline handle numeric strings properly in all cases.
                    231:        (Thanks, Arnold.)
                    232:
                    233: Jan 21, 2019:
                    234:        Merged a number of small fixes from GitHub pull requests.
                    235:        Thanks to GitHub users Arnold Robbins (arnoldrobbins),
                    236:        Cody Mello (melloc) and Christoph Junghans (junghans).
                    237:        PR numbers: 13-21, 23, 24, 27.
1.21      millert   238:
                    239: Oct 25, 2018:
                    240:        Added test in maketab.c to prevent generating a proctab entry
                    241:        for YYSTYPE_IS_DEFINED.  It was harmless but some gcc settings
                    242:        generated a warning message.  Thanks to Nan Xiao for report.
1.20      millert   243:
                    244: Aug 27, 2018:
                    245:        Disallow '$' in printf formats; arguments evaluated in order
                    246:        and printed in order.
                    247:
                    248:        Added some casts to silence warnings on debugging printfs.
                    249:        (Thanks, Arnold.)
1.18      millert   250:
1.19      millert   251: Aug 23, 2018:
                    252:         A long list of fixes courtesy of Arnold Robbins,
                    253:         to whom profound thanks.
                    254:
                    255:         1. ofs-rebuild: OFS value used to rebuild the record was incorrect.
                    256:         Fixed August 19, 2014. Revised fix August 2018.
                    257:
                    258:         2. system-status: Instead of a floating-point division by 256, use
                    259:         the wait(2) macros to create a reasonable exit status.
                    260:         Fixed March 12, 2016.
                    261:
                    262:         3. space: Use provided xisblank() function instead of ispace() for
                    263:         matching [[:blank:]].
                    264:
                    265:         4. a-format: Add POSIX standard %a and %A to supported formats. Check
                    266:         at runtime that this format is available.
                    267:
                    268:         5. decr-NF: Decrementing NF did not change $0. This is a decades-old
                    269:         bug. There are interactions with the old and new value of OFS as well.
                    270:         Most of the fix came from the NetBSD awk.
                    271:
                    272:         6. string-conv: String conversions of scalars were sticky.  Once a
                    273:         conversion to string happened, even with OFMT, that value was used until
                    274:         a new numeric value was assigned, even if OFMT differed from CONVFMT,
                    275:         and also if CONVFMT changed.
                    276:
                    277:         7. unary-plus: Unary plus on a string constant returned the string.
                    278:         Instead, it should convert the value to numeric and give that value.
                    279:
                    280:        Also added Arnold's tests for these to awktest.tar as T.arnold.
                    281:
                    282: Aug 15, 2018:
                    283:        fixed mangled awktest.tar (thanks, Arnold), posted all
                    284:        current (very minor) fixes to github / onetrueawk
                    285:
1.18      millert   286: Jun 7, 2018:
                    287:        (yes, a long layoff)
                    288:        Updated some broken tests (beebe.tar, T.lilly)
                    289:        [thanks to Arnold Robbins]
                    290:
                    291: Mar 26, 2015:
                    292:        buffer overflow in error reporting; thanks to tobias ulmer
                    293:        and john-mark gurney for spotting it and the fix.
                    294:
                    295: Feb 4, 2013:
                    296:        cleaned up a handful of tests that didn't seem to actually
                    297:        test for correct behavior: T.latin1, T.gawk.
                    298:
                    299: Jan 5, 2013:
                    300:        added ,NULL initializer to static Cells in run.c; not really
                    301:        needed but cleaner.  Thanks to Michael Bombardieri.
1.17      millert   302:
                    303: Dec 20, 2012:
                    304:        fiddled makefile to get correct yacc and bison flags.  pick yacc
                    305:        (linux) or bison (mac) as necessary.
                    306:
                    307:        added  __attribute__((__noreturn__)) to a couple of lines in
                    308:        proto.h, to silence someone's enthusiastic checker.
                    309:
                    310:        fixed obscure call by value bug in split(a[1],a) reported on
                    311:        9fans.  the management of temporary values is just a mess; i
                    312:        took a shortcut by making an extra string copy.  thanks
                    313:        to paul patience and arnold robbins for passing it on and for
                    314:        proposed patches.
                    315:
                    316:        tiny fiddle in setfval to eliminate -0 results in T.expr, which
                    317:        has irritated me for 20+ years.
1.16      millert   318:
                    319: Aug 10, 2011:
                    320:        another fix to avoid core dump with delete(ARGV); again, many thanks
                    321:        to ruslan ermilov.
                    322:
                    323: Aug 7, 2011:
                    324:        split(s, a, //) now behaves the same as split(s, a, "")
                    325:
                    326: Jun 12, 2011:
                    327:        /pat/, \n /pat/ {...} is now legal, though bad style to use.
                    328:
                    329:        added checks to new -v code that permits -vnospace; thanks to
1.25      millert   330:        ruslan ermilov for spotting this and providing the patch.
1.16      millert   331:
                    332:        removed fixed limit on number of open files; thanks to aleksey
1.25      millert   333:        cheusov and christos zoulos.
1.16      millert   334:
                    335:        fixed day 1 bug that resurrected deleted elements of ARGV when
                    336:        used as filenames (in lib.c).
                    337:
                    338:        minor type fiddles to make gcc -Wall -pedantic happier (but not
                    339:        totally so); turned on -fno-strict-aliasing in makefile.
                    340:
                    341: May 6, 2011:
                    342:        added #ifdef for isblank.
                    343:        now allows -ffoo as well as -f foo arguments.
                    344:        (thanks, ruslan)
                    345:
                    346: May 1, 2011:
                    347:        after advice from todd miller, kevin lo, ruslan ermilov,
                    348:        and arnold robbins, changed srand() to return the previous
                    349:        seed (which is 1 on the first call of srand).  the seed is
                    350:        an Awkfloat internally though converted to unsigned int to
1.25      millert   351:        pass to the library srand().  thanks, everyone.
1.16      millert   352:
                    353:        fixed a subtle (and i hope low-probability) overflow error
1.25      millert   354:        in fldbld, by adding space for one extra \0.  thanks to
1.16      millert   355:        robert bassett for spotting this one and providing a fix.
                    356:
                    357:        removed the files related to compilation on windows.  i no
                    358:        longer have anything like a current windows environment, so
                    359:        i can't test any of it.
1.15      millert   360:
                    361: May 23, 2010:
                    362:        fixed long-standing overflow bug in run.c; many thanks to
                    363:        nelson beebe for spotting it and providing the fix.
                    364:
                    365:        fixed bug that didn't parse -vd=1 properly; thanks to santiago
                    366:        vila for spotting it.
                    367:
                    368: Feb 8, 2010:
                    369:        i give up.  replaced isblank with isspace in b.c; there are
                    370:        no consistent header files.
                    371:
                    372: Nov 26, 2009:
                    373:        fixed a long-standing issue with when FS takes effect.  a
                    374:        change to FS is now noticed immediately for subsequent splits.
                    375:
                    376:        changed the name getline() to awkgetline() to avoid yet another
                    377:        name conflict somewhere.
                    378:
                    379: Feb 11, 2009:
                    380:        temporarily for now defined HAS_ISBLANK, since that seems to
                    381:        be the best way through the thicket.  isblank arrived in C99,
                    382:        but seems to be arriving at different systems at different
                    383:        times.
                    384:
                    385: Oct 8, 2008:
                    386:        fixed typo in b.c that set tmpvec wrongly.  no one had ever
                    387:        run into the problem, apparently.  thanks to alistair crooks.
1.14      millert   388:
                    389: Oct 23, 2007:
                    390:        minor fix in lib.c: increase inputFS to 100, change malloc
1.25      millert   391:        for fields to n+1.
1.14      millert   392:
                    393:        fixed memory fault caused by out of order test in setsval.
                    394:
                    395:        thanks to david o'brien, freebsd, for both fixes.
                    396:
                    397: May 1, 2007:
                    398:        fiddle in makefile to fix for BSD make; thanks to igor sobrado.
                    399:
                    400: Mar 31, 2007:
                    401:        fixed some null pointer refs calling adjbuf.
                    402:
                    403: Feb 21, 2007:
                    404:        fixed a bug in matching the null RE in sub and gsub.  thanks to al aho
                    405:        who actually did the fix (in b.c), and to wolfgang seeberg for finding
                    406:        it and providing a very compact test case.
                    407:
                    408:        fixed quotation in b.c; thanks to Hal Pratt and the Princeton Dante
                    409:        Project.
                    410:
                    411:        removed some no-effect asserts in run.c.
                    412:
                    413:        fiddled maketab.c to not complain about bison-generated values.
                    414:
                    415:        removed the obsolete -V argument; fixed --version to print the
                    416:        version and exit.
                    417:
                    418:        fixed wording and an outright error in the usage message; thanks to igor
                    419:        sobrado and jason mcintyre.
                    420:
                    421:        fixed a bug in -d that caused core dump if no program followed.
                    422:
                    423: Jan 1, 2007:
                    424:        dropped mac.code from makefile; there are few non-MacOSX
                    425:        mac's these days.
                    426:
                    427: Jan 17, 2006:
                    428:        system() not flagged as unsafe in the unadvertised -safe option.
                    429:        found it while enhancing tests before shipping the ;login: article.
                    430:        practice what you preach.
                    431:
                    432:        removed the 9-years-obsolete -mr and -mf flags.
                    433:
                    434:        added -version and --version options.
                    435:
                    436:        core dump on linux with BEGIN {nextfile}, now fixed.
                    437:
1.25      millert   438:        removed some #ifdef's in run.c and lex.c that appear to no
1.14      millert   439:        longer be necessary.
                    440:
                    441: Apr 24, 2005:
                    442:        modified lib.c so that values of $0 et al are preserved in the END
                    443:        block, apparently as required by posix.  thanks to havard eidnes
                    444:        for the report and code.
                    445:
                    446: Jan 14, 2005:
                    447:        fixed infinite loop in parsing, originally found by brian tsang.
                    448:        thanks to arnold robbins for a suggestion that started me
                    449:        rethinking it.
                    450:
                    451: Dec 31, 2004:
1.25      millert   452:        prevent overflow of -f array in main, head off potential error in
                    453:        call of SYNTAX(), test malloc return in lib.c, all with thanks to
1.14      millert   454:        todd miller.
1.13      millert   455:
                    456: Dec 22, 2004:
                    457:        cranked up size of NCHARS; coverity thinks it can be overrun with
                    458:        smaller size, and i think that's right.  added some assertions to b.c
                    459:        to catch places where it might overrun.  the RE code is still fragile.
                    460:
                    461: Dec 5, 2004:
                    462:        fixed a couple of overflow problems with ridiculous field numbers:
                    463:        e.g., print $(2^32-1).  thanks to ruslan ermilov, giorgos keramidas
                    464:        and david o'brien at freebsd.org for patches.  this really should
                    465:        be re-done from scratch.
                    466:
                    467: Nov 21, 2004:
                    468:        fixed another 25-year-old RE bug, in split.  it's another failure
                    469:        to (re-)initialize.  thanks to steve fisher for spotting this and
                    470:        providing a good test case.
                    471:
                    472: Nov 22, 2003:
                    473:        fixed a bug in regular expressions that dates (so help me) from 1977;
                    474:        it's been there from the beginning.  an anchored longest match that
                    475:        was longer than the number of states triggered a failure to initialize
                    476:        the machine properly.  many thanks to moinak ghosh for not only finding
                    477:        this one but for providing a fix, in some of the most mysterious
                    478:        code known to man.
                    479:
                    480:        fixed a storage leak in call() that appears to have been there since
1.25      millert   481:        1983 or so -- a function without an explicit return that assigns a
                    482:        string to a parameter leaked a Cell.  thanks to moinak ghosh for
1.13      millert   483:        spotting this very subtle one.
                    484:
                    485: Jul 31, 2003:
                    486:        fixed, thanks to andrey chernov and ruslan ermilov, a bug in lex.c
                    487:        that mis-handled the character 255 in input.  (it was being compared
                    488:        to EOF with a signed comparison.)
                    489:
                    490: Jul 29, 2003:
                    491:        fixed (i think) the long-standing botch that included the beginning of
                    492:        line state ^ for RE's in the set of valid characters; this led to a
                    493:        variety of odd problems, including failure to properly match certain
                    494:        regular expressions in non-US locales.  thanks to ruslan for keeping
                    495:        at this one.
                    496:
                    497: Jul 28, 2003:
                    498:        n-th try at getting internationalization right, with thanks to volker
                    499:        kiefel, arnold robbins and ruslan ermilov for advice, though they
                    500:        should not be blamed for the outcome.  according to posix, "."  is the
                    501:        radix character in programs and command line arguments regardless of
                    502:        the locale; otherwise, the locale should prevail for input and output
                    503:        of numbers.  so it's intended to work that way.
1.25      millert   504:
1.13      millert   505:        i have rescinded the attempt to use strcoll in expanding shorthands in
                    506:        regular expressions (cclenter).  its properties are much too
                    507:        surprising; for example [a-c] matches aAbBc in locale en_US but abBcC
                    508:        in locale fr_CA.  i can see how this might arise by implementation
                    509:        but i cannot explain it to a human user.  (this behavior can be seen
                    510:        in gawk as well; we're leaning on the same library.)
                    511:
                    512:        the issue appears to be that strcoll is meant for sorting, where
                    513:        merging upper and lower case may make sense (though note that unix
                    514:        sort does not do this by default either).  it is not appropriate
                    515:        for regular expressions, where the goal is to match specific
                    516:        patterns of characters.  in any case, the notations [:lower:], etc.,
                    517:        are available in awk, and they are more likely to work correctly in
                    518:        most locales.
                    519:
                    520:        a moratorium is hereby declared on internationalization changes.
                    521:        i apologize to friends and colleagues in other parts of the world.
                    522:        i would truly like to get this "right", but i don't know what
                    523:        that is, and i do not want to keep making changes until it's clear.
                    524:
                    525: Jul 4, 2003:
                    526:        fixed bug that permitted non-terminated RE, as in "awk /x".
                    527:
                    528: Jun 1, 2003:
                    529:        subtle change to split: if source is empty, number of elems
                    530:        is always 0 and the array is not set.
                    531:
                    532: Mar 21, 2003:
                    533:        added some parens to isblank, in another attempt to make things
                    534:        internationally portable.
                    535:
                    536: Mar 14, 2003:
                    537:        the internationalization changes, somewhat modified, are now
                    538:        reinstated.  in theory awk will now do character comparisons
                    539:        and case conversions in national language, but "." will always
                    540:        be the decimal point separator on input and output regardless
                    541:        of national language.  isblank(){} has an #ifndef.
                    542:
                    543:        this no longer compiles on windows: LC_MESSAGES isn't defined
                    544:        in vc6++.
                    545:
                    546:        fixed subtle behavior in field and record splitting: if FS is
                    547:        a single character and RS is not empty, \n is NOT a separator.
                    548:        this tortuous reading is found in the awk book; behavior now
                    549:        matches gawk and mawk.
1.11      millert   550:
                    551: Dec 13, 2002:
                    552:        for the moment, the internationalization changes of nov 29 are
                    553:        rolled back -- programs like x = 1.2 don't work in some locales,
                    554:        because the parser is expecting x = 1,2.  until i understand this
                    555:        better, this will have to wait.
                    556:
                    557: Nov 29, 2002:
                    558:        modified b.c (with tiny changes in main and run) to support
                    559:        locales, using strcoll and iswhatever tests for posix character
                    560:        classes.  thanks to ruslan ermilov (ru@freebsd.org) for code.
                    561:        the function isblank doesn't seem to have propagated to any
                    562:        header file near me, so it's there explicitly.  not properly
                    563:        tested on non-ascii character sets by me.
                    564:
                    565: Jun 28, 2002:
                    566:        modified run/format() and tran/getsval() to do a slightly better
                    567:        job on using OFMT for output from print and CONVFMT for other
1.25      millert   568:        number->string conversions, as promised by posix and done by
1.11      millert   569:        gawk and mawk.  there are still places where it doesn't work
                    570:        right if CONVFMT is changed; by then the STR attribute of the
                    571:        variable has been irrevocably set.  thanks to arnold robbins for
                    572:        code and examples.
                    573:
                    574:        fixed subtle bug in format that could get core dump.  thanks to
                    575:        Jaromir Dolecek <jdolecek@NetBSD.org> for finding and fixing.
                    576:        minor cleanup in run.c / format() at the same time.
                    577:
                    578:        added some tests for null pointers to debugging printf's, which
                    579:        were never intended for external consumption.  thanks to dave
                    580:        kerns (dkerns@lucent.com) for pointing this out.
                    581:
                    582:        GNU compatibility: an empty regexp matches anything (thanks to
                    583:        dag-erling smorgrav, des@ofug.org).  subject to reversion if
                    584:        this does more harm than good.
                    585:
                    586:        pervasive small changes to make things more const-correct, as
                    587:        reported by gcc's -Wwrite-strings.  as it says in the gcc manual,
                    588:        this may be more nuisance than useful.  provoked by a suggestion
                    589:        and code from arnaud desitter, arnaud@nimbus.geog.ox.ac.uk
                    590:
                    591:        minor documentation changes to note that this now compiles out
                    592:        of the box on Mac OS X.
                    593:
                    594: Feb 10, 2002:
                    595:        changed types in posix chars structure to quiet solaris cc.
                    596:
                    597: Jan 1, 2002:
                    598:        fflush() or fflush("") flushes all files and pipes.
                    599:
1.25      millert   600:        length(arrayname) returns number of elements; thanks to
1.11      millert   601:        arnold robbins for suggestion.
                    602:
                    603:        added a makefile.win to make it easier to build on windows.
                    604:        based on dan allen's buildwin.bat.
                    605:
                    606: Nov 16, 2001:
                    607:        added support for posix character class names like [:digit:],
                    608:        which are not exactly shorter than [0-9] and perhaps no more
                    609:        portable.  thanks to dag-erling smorgrav for code.
                    610:
                    611: Feb 16, 2001:
                    612:        removed -m option; no longer needed, and it was actually
                    613:        broken (noted thanks to volker kiefel).
                    614:
                    615: Feb 10, 2001:
                    616:        fixed an appalling bug in gettok: any sequence of digits, +,-, E, e,
                    617:        and period was accepted as a valid number if it started with a period.
                    618:        this would never have happened with the lex version.
                    619:
                    620:        other 1-character botches, now fixed, include a bare $ and a
                    621:        bare " at the end of the input.
                    622:
                    623: Feb 7, 2001:
                    624:        more (const char *) casts in b.c and tran.c to silence warnings.
1.8       millert   625:
1.10      millert   626: Nov 15, 2000:
                    627:        fixed a bug introduced in august 1997 that caused expressions
                    628:        like $f[1] to be syntax errors.  thanks to arnold robbins for
                    629:        noticing this and providing a fix.
                    630:
                    631: Oct 30, 2000:
                    632:        fixed some nextfile bugs: not handling all cases.  thanks to
                    633:        arnold robbins for pointing this out.  new regressions added.
                    634:
                    635:        close() is now a function.  it returns whatever the library
                    636:        fclose returns, and -1 for closing a file or pipe that wasn't
                    637:        opened.
                    638:
                    639: Sep 24, 2000:
                    640:        permit \n explicitly in character classes; won't work right
                    641:        if comes in as "[\n]" but ok as /[\n]/, because of multiple
                    642:        processing of \'s.  thanks to arnold robbins.
                    643:
                    644: July 5, 2000:
                    645:        minor fiddles in tran.c to keep compilers happy about uschar.
                    646:        thanks to norman wilson.
                    647:
                    648: May 25, 2000:
                    649:        yet another attempt at making 8-bit input work, with another
1.25      millert   650:        band-aid in b.c (member()), and some (uschar) casts to head
1.10      millert   651:        off potential errors in subscripts (like isdigit).  also
                    652:        changed HAT to NCHARS-2.  thanks again to santiago vila.
                    653:
                    654:        changed maketab.c to ignore apparently out of range definitions
                    655:        instead of halting; new freeBSD generates one.  thanks to
                    656:        jon snader <jsnader@ix.netcom.com> for pointing out the problem.
                    657:
                    658: May 2, 2000:
                    659:        fixed an 8-bit problem in b.c by making several char*'s into
                    660:        unsigned char*'s.  not clear i have them all yet.  thanks to
                    661:        Santiago Vila <sanvila@unex.es> for the bug report.
                    662:
                    663: Apr 21, 2000:
                    664:        finally found and fixed a memory leak in function call; it's
                    665:        been there since functions were added ~1983.  thanks to
                    666:        jon bentley for the test case that found it.
                    667:
                    668:        added test in envinit to catch environment "variables" with
1.12      otto      669:        names beginning with '='; thanks to Berend Hasselman.
1.10      millert   670:
                    671: Jul 28, 1999:
                    672:        added test in defn() to catch function foo(foo), which
                    673:        otherwise recurses until core dump.  thanks to arnold
                    674:        robbins for noticing this.
                    675:
1.9       millert   676: Jun 20, 1999:
                    677:        added *bp in gettok in lex.c; appears possible to exit function
                    678:        without terminating the string.  thanks to russ cox.
                    679:
                    680: Jun 2, 1999:
                    681:        added function stdinit() to run to initialize files[] array,
                    682:        in case stdin, etc., are not constants; some compilers care.
                    683:
                    684: May 10, 1999:
                    685:        replaced the ERROR ... FATAL, etc., macros with functions
                    686:        based on vprintf, to avoid problems caused by overrunning
                    687:        fixed-size errbuf array.  thanks to ralph corderoy for the
                    688:        impetus, and for pointing out a string termination bug in
                    689:        qstring as well.
                    690:
                    691: Apr 21, 1999:
                    692:        fixed bug that caused occasional core dumps with commandline
                    693:        variable with value ending in \.  (thanks to nelson beebe for
                    694:        the test case.)
                    695:
1.8       millert   696: Apr 16, 1999:
1.25      millert   697:        with code kindly provided by Bruce Lilly, awk now parses
1.8       millert   698:        /=/ and similar constructs more sensibly in more places.
1.10      millert   699:        Bruce also provided some helpful test cases.
1.8       millert   700:
                    701: Apr 5, 1999:
                    702:        changed true/false to True/False in run.c to make it
                    703:        easier to compile with C++.  Added some casts on malloc
                    704:        and realloc to be honest about casts; ditto.  changed
                    705:        ltype int to long in struct rrow to reduce some 64-bit
                    706:        complaints; other changes scattered throughout for the
                    707:        same purpose.  thanks to Nelson Beebe for these portability
                    708:        improvements.
                    709:
                    710:        removed some horrible pointer-int casting in b.c and elsewhere
                    711:        by adding ptoi and itonp to localize the casts, which are
                    712:        all benign.  fixed one incipient bug that showed up on sgi
                    713:        in 64-bit mode.
                    714:
                    715:        reset lineno for new source file; include filename in error
                    716:        message.  also fixed line number error in continuation lines.
                    717:        (thanks to Nelson Beebe for both of these.)
                    718:
                    719: Mar 24, 1999:
                    720:        Nelson Beebe notes that irix 5.3 yacc dies with a bogus
                    721:        error; use a newer version or switch to bison, since sgi
                    722:        is unlikely to fix it.
1.1       tholo     723:
1.7       millert   724: Mar 5, 1999:
1.9       millert   725:        changed isnumber to is_number to avoid the problem caused by
                    726:        versions of ctype.h that include the name isnumber.
1.7       millert   727:
                    728:        distribution now includes a script for building on a Mac,
                    729:        thanks to Dan Allen.
                    730:
                    731: Feb 20, 1999:
                    732:        fixed memory leaks in run.c (call) and tran.c (setfval).
                    733:        thanks to Stephen Nutt for finding these and providing the fixes.
                    734:
                    735: Jan 13, 1999:
                    736:        replaced srand argument by (unsigned int) in run.c;
                    737:        avoids problem on Mac and potentially on Unix & Windows.
                    738:        thanks to Dan Allen.
                    739:
                    740:        added a few (int) casts to silence useless compiler warnings.
                    741:        e.g., errorflag= in run.c jump().
                    742:
                    743:        added proctab.c to the bundle outout; one less thing
                    744:        to have to compile out of the box.
                    745:
                    746:        added calls to _popen and _pclose to the win95 stub for
                    747:        pipes (thanks to Steve Adams for this helpful suggestion).
                    748:        seems to work, though properties are not well understood
                    749:        by me, and it appears that under some circumstances the
                    750:        pipe output is truncated.  Be careful.
                    751:
                    752: Oct 19, 1998:
                    753:        fixed a couple of bugs in getrec: could fail to update $0
1.25      millert   754:        after a getline var; because inputFS wasn't initialized,
1.7       millert   755:        could split $0 on every character, a misleading diversion.
                    756:
                    757:        fixed caching bug in makedfa: LRU was actually removing
                    758:        least often used.
                    759:
                    760:        thanks to ross ridge for finding these, and for providing
                    761:        great bug reports.
                    762:
                    763: May 12, 1998:
                    764:        fixed potential bug in readrec: might fail to update record
                    765:        pointer after growing.  thanks to dan levy for spotting this
                    766:        and suggesting the fix.
                    767:
                    768: Mar 12, 1998:
                    769:        added -V to print version number and die.
1.19      millert   770:
                    771: [notify dave kerns, dkerns@dacsoup.ih.lucent.com]
1.7       millert   772:
                    773: Feb 11, 1998:
                    774:        subtle silent bug in lex.c: if the program ended with a number
                    775:        longer than 1 digit, part of the input would be pushed back and
                    776:        parsed again because token buffer wasn't terminated right.
                    777:        example:  awk 'length($0) > 10'.  blush.  at least i found it
                    778:        myself.
                    779:
                    780: Aug 31, 1997:
                    781:        s/adelete/awkdelete/: SGI uses this in malloc.h.
                    782:        thanks to nelson beebe for pointing this one out.
                    783:
                    784: Aug 21, 1997:
                    785:        fixed some bugs in sub and gsub when replacement includes \\.
                    786:        this is a dark, horrible corner, but at least now i believe that
                    787:        the behavior is the same as gawk and the intended posix standard.
                    788:        thanks to arnold robbins for advice here.
                    789:
                    790: Aug 9, 1997:
                    791:        somewhat regretfully, replaced the ancient lex-based lexical
                    792:        analyzer with one written in C.  it's longer, generates less code,
                    793:        and more portable; the old one depended too much on mysterious
                    794:        properties of lex that were not preserved in other environments.
                    795:        in theory these recognize the same language.
                    796:
                    797:        now using strtod to test whether a string is a number, instead of
                    798:        the convoluted original function.  should be more portable and
                    799:        reliable if strtod is implemented right.
                    800:
                    801:        removed now-pointless optimization in makefile that tries to avoid
                    802:        recompilation when awkgram.y is changed but symbols are not.
                    803:
                    804:        removed most fixed-size arrays, though a handful remain, some
                    805:        of which are unchecked.  you have been warned.
                    806:
                    807: Aug 4, 1997:
                    808:        with some trepidation, replaced the ancient code that managed
                    809:        fields and $0 in fixed-size arrays with arrays that grow on
                    810:        demand.  there is still some tension between trying to make this
                    811:        run fast and making it clean; not sure it's right yet.
                    812:
                    813:        the ill-conceived -mr and -mf arguments are now useful only
                    814:        for debugging.  previous dynamic string code removed.
                    815:
                    816:        numerous other minor cleanups along the way.
1.1       tholo     817:
1.7       millert   818: Jul 30, 1997:
                    819:        using code provided by dan levy (to whom profuse thanks), replaced
                    820:        fixed-size arrays and awkward kludges by a fairly uniform mechanism
                    821:        to grow arrays as needed for printf, sub, gsub, etc.
1.1       tholo     822:
1.7       millert   823: Jul 23, 1997:
                    824:        falling off the end of a function returns "" and 0, not 0.
                    825:        thanks to arnold robbins.
1.1       tholo     826:
1.7       millert   827: Jun 17, 1997:
                    828:        replaced several fixed-size arrays by dynamically-created ones
                    829:        in run.c; added overflow tests to some previously unchecked cases.
                    830:        getline, toupper, tolower.
1.1       tholo     831:
1.7       millert   832:        getline code is still broken in that recursive calls may wind
                    833:        up using the same space.  [fixed later]
1.1       tholo     834:
1.7       millert   835:        increased RECSIZE to 8192 to push problems further over the horizon.
1.1       tholo     836:
1.7       millert   837:        added \r to \n as input line separator for programs, not data.
                    838:        damn CRLFs.
1.1       tholo     839:
1.7       millert   840:        modified format() to permit explicit printf("%c", 0) to include
                    841:        a null byte in output.  thanks to ken stailey for the fix.
1.1       tholo     842:
1.7       millert   843:        added a "-safe" argument that disables file output (print >,
                    844:        print >>), process creation (cmd|getline, print |, system), and
                    845:        access to the environment (ENVIRON).  this is a first approximation
                    846:        to a "safe" version of awk, but don't rely on it too much.  thanks
                    847:        to joan feigenbaum and matt blaze for the inspiration long ago.
1.1       tholo     848:
1.7       millert   849: Jul 8, 1996:
                    850:        fixed long-standing bug in sub, gsub(/a/, "\\\\&"); thanks to
                    851:        ralph corderoy.
1.1       tholo     852:
1.7       millert   853: Jun 29, 1996:
                    854:        fixed awful bug in new field splitting; didn't get all the places
                    855:        where input was done.
1.1       tholo     856:
1.7       millert   857: Jun 28, 1996:
                    858:        changed field-splitting to conform to posix definition: fields are
                    859:        split using the value of FS at the time of input; it used to be
                    860:        the value when the field or NF was first referred to, a much less
                    861:        predictable definition.  thanks to arnold robbins for encouragement
                    862:        to do the right thing.
1.1       tholo     863:
1.7       millert   864: May 28, 1996:
                    865:        fixed appalling but apparently unimportant bug in parsing octal
                    866:        numbers in reg exprs.
1.1       tholo     867:
1.7       millert   868:        explicit hex in reg exprs now limited to 2 chars: \xa, \xaa.
1.1       tholo     869:
1.7       millert   870: May 27, 1996:
                    871:        cleaned up some declarations so gcc -Wall is now almost silent.
1.1       tholo     872:
1.7       millert   873:        makefile now includes backup copies of ytab.c and lexyy.c in case
                    874:        one makes before looking; it also avoids recreating lexyy.c unless
                    875:        really needed.
1.1       tholo     876:
1.7       millert   877:        s/aprintf/awkprint, s/asprintf/awksprintf/ to avoid some name clashes
                    878:        with unwisely-written header files.
1.1       tholo     879:
1.7       millert   880:        thanks to jeffrey friedl for several of these.
1.1       tholo     881:
1.7       millert   882: May 26, 1996:
                    883:        an attempt to rationalize the (unsigned) char issue.  almost all
                    884:        instances of unsigned char have been removed; the handful of places
                    885:        in b.c where chars are used as table indices have been hand-crafted.
                    886:        added some latin-1 tests to the regression, but i'm not confident;
                    887:        none of my compilers seem to care much.  thanks to nelson beebe for
                    888:        pointing out some others that do care.
1.1       tholo     889:
1.7       millert   890: May 2, 1996:
                    891:        removed all register declarations.
1.1       tholo     892:
1.7       millert   893:        enhanced split(), as in gawk, etc:  split(s, a, "") splits s into
                    894:        a[1]...a[length(s)] with each character a single element.
1.1       tholo     895:
1.7       millert   896:        made the same changes for field-splitting if FS is "".
1.1       tholo     897:
1.7       millert   898:        added nextfile, as in gawk: causes immediate advance to next
                    899:        input file. (thanks to arnold robbins for inspiration and code).
1.1       tholo     900:
1.7       millert   901:        small fixes to regexpr code:  can now handle []], [[], and
1.25      millert   902:        variants;  [] is now a syntax error, rather than matching
1.7       millert   903:        everything;  [z-a] is now empty, not z.  far from complete
                    904:        or correct, however.  (thanks to jeffrey friedl for pointing out
                    905:        some awful behaviors.)
1.1       tholo     906:
1.7       millert   907: Apr 29, 1996:
1.12      otto      908:        replaced uchar by uschar everywhere; apparently some compilers
1.7       millert   909:        usurp this name and this causes conflicts.
1.1       tholo     910:
1.7       millert   911:        fixed call to time in run.c (bltin); arg is time_t *.
1.1       tholo     912:
1.7       millert   913:        replaced horrible pointer/long punning in b.c by a legitimate
                    914:        union.  should be safer on 64-bit machines and cleaner everywhere.
                    915:        (thanks to nelson beebe for pointing out some of these problems.)
1.1       tholo     916:
1.7       millert   917:        replaced nested comments by #if 0...#endif in run.c, lib.c.
1.1       tholo     918:
1.7       millert   919:        removed getsval, setsval, execute macros from run.c and lib.c.
                    920:        machines are 100x faster than they were when these macros were
                    921:        first used.
1.1       tholo     922:
1.7       millert   923:        revised filenames: awk.g.y => awkgram.y, awk.lx.l => awklex.l,
                    924:        y.tab.[ch] => ytab.[ch], lex.yy.c => lexyy.c, all in the aid of
                    925:        portability to nameless systems.
1.1       tholo     926:
1.7       millert   927:        "make bundle" now includes yacc and lex output files for recipients
                    928:        who don't have yacc or lex.
1.1       tholo     929:
1.7       millert   930: Aug 15, 1995:
                    931:        initialized Cells in setsymtab more carefully; some fields
                    932:        were not set.  (thanks to purify, all of whose complaints i
                    933:        think i now understand.)
1.1       tholo     934:
1.7       millert   935:        fixed at least one error in gsub that looked at -1-th element
                    936:        of an array when substituting for a null match (e.g., $).
1.1       tholo     937:
1.7       millert   938:        delete arrayname is now legal; it clears the elements but leaves
                    939:        the array, which may not be the right behavior.
1.1       tholo     940:
1.7       millert   941:        modified makefile: my current make can't cope with the test used
                    942:        to avoid unnecessary yacc invocations.
1.1       tholo     943:
1.7       millert   944: Jul 17, 1995:
                    945:        added dynamically growing strings to awk.lx.l and b.c
                    946:        to permit regular expressions to be much bigger.
                    947:        the state arrays can still overflow.
1.1       tholo     948:
1.7       millert   949: Aug 24, 1994:
                    950:        detect duplicate arguments in function definitions (mdm).
1.1       tholo     951:
1.7       millert   952: May 11, 1994:
                    953:        trivial fix to printf to limit string size in sub().
1.1       tholo     954:
1.7       millert   955: Apr 22, 1994:
                    956:        fixed yet another subtle self-assignment problem:
                    957:        $1 = $2; $1 = $1 clobbered $1.
1.1       tholo     958:
1.7       millert   959:        Regression tests now use private echo, to avoid quoting problems.
1.1       tholo     960:
1.7       millert   961: Feb 2, 1994:
                    962:        changed error() to print line number as %d, not %g.
1.1       tholo     963:
1.7       millert   964: Jul 23, 1993:
                    965:        cosmetic changes: increased sizes of some arrays,
                    966:        reworded some error messages.
1.1       tholo     967:
1.7       millert   968:        added CONVFMT as in posix (just replaced OFMT in getsval)
1.1       tholo     969:
1.7       millert   970:        FILENAME is now "" until the first thing that causes a file
                    971:        to be opened.
1.1       tholo     972:
1.7       millert   973: Nov 28, 1992:
                    974:        deleted yyunput and yyoutput from proto.h;
                    975:        different versions of lex give these different declarations.
1.1       tholo     976:
1.7       millert   977: May 31, 1992:
                    978:        added -mr N and -mf N options: more record and fields.
                    979:        these really ought to adjust automatically.
1.1       tholo     980:
1.7       millert   981:        cleaned up some error messages; "out of space" now means
                    982:        malloc returned NULL in all cases.
1.1       tholo     983:
1.7       millert   984:        changed rehash so that if it runs out, it just returns;
                    985:        things will continue to run slow, but maybe a bit longer.
1.1       tholo     986:
1.7       millert   987: Apr 24, 1992:
                    988:        remove redundant close of stdin when using -f -.
1.1       tholo     989:
1.7       millert   990:        got rid of core dump with -d; awk -d just prints date.
1.1       tholo     991:
1.7       millert   992: Apr 12, 1992:
                    993:        added explicit check for /dev/std(in,out,err) in redirection.
                    994:        unlike gawk, no /dev/fd/n yet.
1.1       tholo     995:
1.7       millert   996:        added (file/pipe) builtin.  hard to test satisfactorily.
                    997:        not posix.
1.1       tholo     998:
1.7       millert   999: Feb 20, 1992:
                   1000:        recompile after abortive changes;  should be unchanged.
1.1       tholo    1001:
1.7       millert  1002: Dec 2, 1991:
                   1003:        die-casting time:  converted to ansi C, installed that.
1.1       tholo    1004:
1.7       millert  1005: Nov 30, 1991:
                   1006:        fixed storage leak in freefa, failing to recover [N]CCL.
                   1007:        thanks to Bill Jones (jones@cs.usask.ca)
1.1       tholo    1008:
1.7       millert  1009: Nov 19, 1991:
                   1010:        use RAND_MAX instead of literal in builtin().
1.1       tholo    1011:
1.7       millert  1012: Nov 12, 1991:
                   1013:        cranked up some fixed-size arrays in b.c, and added a test for
                   1014:        overflow in penter.  thanks to mark larsen.
1.1       tholo    1015:
1.7       millert  1016: Sep 24, 1991:
                   1017:        increased buffer in gsub.  a very crude fix to a general problem.
                   1018:        and again on Sep 26.
1.1       tholo    1019:
1.7       millert  1020: Aug 18, 1991:
                   1021:        enforce variable name syntax for commandline variables: has to
                   1022:        start with letter or _.
1.1       tholo    1023:
1.7       millert  1024: Jul 27, 1991:
                   1025:        allow newline after ; in for statements.
1.1       tholo    1026:
1.7       millert  1027: Jul 21, 1991:
                   1028:        fixed so that in self-assignment like $1=$1, side effects
                   1029:        like recomputing $0 take place.  (this is getting subtle.)
1.1       tholo    1030:
1.7       millert  1031: Jun 30, 1991:
                   1032:        better test for detecting too-long output record.
1.1       tholo    1033:
1.7       millert  1034: Jun 2, 1991:
                   1035:        better defense against very long printf strings.
                   1036:        made break and continue illegal outside of loops.
1.1       tholo    1037:
1.7       millert  1038: May 13, 1991:
                   1039:        removed extra arg on gettemp, tempfree.  minor error message rewording.
1.1       tholo    1040:
                   1041: May 6, 1991:
                   1042:        fixed silly bug in hex parsing in hexstr().
                   1043:        removed an apparently unnecessary test in isnumber().
                   1044:        warn about weird printf conversions.
                   1045:        fixed unchecked array overwrite in relex().
                   1046:
                   1047:        changed for (i in array) to access elements in sorted order.
                   1048:        then unchanged it -- it really does run slower in too many cases.
                   1049:        left the code in place, commented out.
                   1050:
1.7       millert  1051: Feb 10, 1991:
                   1052:        check error status on all writes, to avoid banging on full disks.
1.1       tholo    1053:
1.7       millert  1054: Jan 28, 1991:
                   1055:        awk -f - reads the program from stdin.
1.1       tholo    1056:
1.7       millert  1057: Jan 11, 1991:
                   1058:        failed to set numeric state on $0 in cmd|getline context in run.c.
1.1       tholo    1059:
1.7       millert  1060: Nov 2, 1990:
                   1061:        fixed sleazy test for integrality in getsval;  use modf.
1.1       tholo    1062:
1.7       millert  1063: Oct 29, 1990:
                   1064:        fixed sleazy buggy code in lib.c that looked (incorrectly) for
                   1065:        too long input lines.
1.1       tholo    1066:
1.7       millert  1067: Oct 14, 1990:
                   1068:        fixed the bug on p. 198 in which it couldn't deduce that an
                   1069:        argument was an array in some contexts.  replaced the error
                   1070:        message in intest() by code that damn well makes it an array.
1.1       tholo    1071:
1.7       millert  1072: Oct 8, 1990:
                   1073:        fixed horrible bug:  types and values were not preserved in
                   1074:        some kinds of self-assignment. (in assign().)
1.1       tholo    1075:
1.7       millert  1076: Aug 24, 1990:
                   1077:        changed NCHARS to 256 to handle 8-bit characters in strings
                   1078:        presented to match(), etc.
1.1       tholo    1079:
1.7       millert  1080: Jun 26, 1990:
                   1081:        changed struct rrow (awk.h) to use long instead of int for lval,
                   1082:        since cfoll() stores a pointer in it.  now works better when int's
                   1083:        are smaller than pointers!
1.1       tholo    1084:
1.7       millert  1085: May 6, 1990:
                   1086:        AVA fixed the grammar so that ! is uniformly of the same precedence as
                   1087:        unary + and -.  This renders illegal some constructs like !x=y, which
                   1088:        now has to be parenthesized as !(x=y), and makes others work properly:
                   1089:        !x+y is (!x)+y, and x!y is x !y, not two pattern-action statements.
                   1090:        (These problems were pointed out by Bob Lenk of Posix.)
1.1       tholo    1091:
1.7       millert  1092:        Added \x to regular expressions (already in strings).
                   1093:        Limited octal to octal digits; \8 and \9 are not octal.
                   1094:        Centralized the code for parsing escapes in regular expressions.
                   1095:        Added a bunch of tests to T.re and T.sub to verify some of this.
1.1       tholo    1096:
1.7       millert  1097: Feb 9, 1990:
                   1098:        fixed null pointer dereference bug in main.c:  -F[nothing].  sigh.
1.1       tholo    1099:
1.7       millert  1100:        restored srand behavior:  it returns the current seed.
1.1       tholo    1101:
1.7       millert  1102: Jan 18, 1990:
                   1103:        srand now returns previous seed value (0 to start).
1.1       tholo    1104:
1.7       millert  1105: Jan 5, 1990:
                   1106:        fix potential problem in tran.c -- something was freed,
                   1107:        then used in freesymtab.
1.1       tholo    1108:
1.7       millert  1109: Oct 18, 1989:
                   1110:        another try to get the max number of open files set with
                   1111:        relatively machine-independent code.
1.1       tholo    1112:
1.7       millert  1113:        small fix to input() in case of multiple reads after EOF.
1.1       tholo    1114:
1.7       millert  1115: Oct 11, 1989:
                   1116:        FILENAME is now defined in the BEGIN block -- too many old
                   1117:        programs broke.
1.1       tholo    1118:
1.7       millert  1119:        "-" means stdin in getline as well as on the commandline.
1.1       tholo    1120:
1.7       millert  1121:        added a bunch of casts to the code to tell the truth about
                   1122:        char * vs. unsigned char *, a right royal pain.  added a
                   1123:        setlocale call to the front of main, though probably no one
                   1124:        has it usefully implemented yet.
1.1       tholo    1125:
1.7       millert  1126: Aug 24, 1989:
                   1127:        removed redundant relational tests against nullnode if parse
                   1128:        tree already had a relational at that point.
1.1       tholo    1129:
1.7       millert  1130: Aug 11, 1989:
                   1131:        fixed bug:  commandline variable assignment has to look like
                   1132:        var=something.  (consider the man page for =, in file =.1)
1.1       tholo    1133:
1.7       millert  1134:        changed number of arguments to functions to static arrays
                   1135:        to avoid repeated malloc calls.
1.1       tholo    1136:
1.7       millert  1137: Aug 2, 1989:
                   1138:        restored -F (space) separator
1.1       tholo    1139:
1.7       millert  1140: Jul 30, 1989:
                   1141:        added -v x=1 y=2 ... for immediate commandline variable assignment;
                   1142:        done before the BEGIN block for sure.  they have to precede the
                   1143:        program if the program is on the commandline.
                   1144:        Modified Aug 2 to require a separate -v for each assignment.
1.1       tholo    1145:
1.7       millert  1146: Jul 10, 1989:
                   1147:        fixed ref-thru-zero bug in environment code in tran.c
1.1       tholo    1148:
1.7       millert  1149: Jun 23, 1989:
                   1150:        add newline to usage message.
1.1       tholo    1151:
1.7       millert  1152: Jun 14, 1989:
                   1153:        added some missing ansi printf conversion letters: %i %X %E %G.
                   1154:        no sensible meaning for h or L, so they may not do what one expects.
1.1       tholo    1155:
1.7       millert  1156:        made %* conversions work.
1.1       tholo    1157:
1.7       millert  1158:        changed x^y so that if n is a positive integer, it's done
                   1159:        by explicit multiplication, thus achieving maximum accuracy.
                   1160:        (this should be done by pow() but it seems not to be locally.)
                   1161:        done to x ^= y as well.
1.1       tholo    1162:
1.7       millert  1163: Jun 4, 1989:
                   1164:        ENVIRON array contains environment: if shell variable V=thing,
                   1165:                ENVIRON["V"] is "thing"
1.1       tholo    1166:
1.7       millert  1167:        multiple -f arguments permitted.  error reporting is naive.
                   1168:        (they were permitted before, but only the last was used.)
1.1       tholo    1169:
1.7       millert  1170:        fixed a really stupid botch in the debugging macro dprintf
1.1       tholo    1171:
1.7       millert  1172:        fixed order of evaluation of commandline assignments to match
                   1173:        what the book claims:  an argument of the form x=e is evaluated
                   1174:        at the time it would have been opened if it were a filename (p 63).
                   1175:        this invalidates the suggested answer to ex 4-1 (p 195).
1.1       tholo    1176:
1.7       millert  1177:        removed some code that permitted -F (space) fieldseparator,
                   1178:        since it didn't quite work right anyway.  (restored aug 2)
1.1       tholo    1179:
1.7       millert  1180: Apr 27, 1989:
                   1181:        Line number now accumulated correctly for comment lines.
1.1       tholo    1182:
1.7       millert  1183: Apr 26, 1989:
                   1184:        Debugging output now includes a version date,
                   1185:        if one compiles it into the source each time.
1.1       tholo    1186:
1.7       millert  1187: Apr 9, 1989:
                   1188:        Changed grammar to prohibit constants as 3rd arg of sub and gsub;
                   1189:        prevents class of overwriting-a-constant errors.  (Last one?)
                   1190:        This invalidates the "banana" example on page 43 of the book.
1.1       tholo    1191:
1.7       millert  1192:        Added \a ("alert"), \v (vertical tab), \xhhh (hexadecimal),
                   1193:        as in ANSI, for strings.  Rescinded the sloppiness that permitted
                   1194:        non-octal digits in \ooo.  Warning:  not all compilers and libraries
                   1195:        will be able to deal with \x correctly.
1.1       tholo    1196:
1.7       millert  1197: Jan 9, 1989:
                   1198:        Fixed bug that caused tempcell list to contain a duplicate.
                   1199:        The fix is kludgy.
1.1       tholo    1200:
1.7       millert  1201: Dec 17, 1988:
                   1202:        Catches some more commandline errors in main.
                   1203:        Removed redundant decl of modf in run.c (confuses some compilers).
                   1204:        Warning:  there's no single declaration of malloc, etc., in awk.h
                   1205:        that seems to satisfy all compilers.
1.1       tholo    1206:
1.7       millert  1207: Dec 7, 1988:
                   1208:        Added a bit of code to error printing to avoid printing nulls.
                   1209:        (Not clear that it actually would.)
1.1       tholo    1210:
1.7       millert  1211: Nov 27, 1988:
                   1212:        With fear and trembling, modified the grammar to permit
                   1213:        multiple pattern-action statements on one line without
                   1214:        an explicit separator.  By definition, this capitulation
                   1215:        to the ghost of ancient implementations remains undefined
                   1216:        and thus subject to change without notice or apology.
                   1217:        DO NOT COUNT ON IT.
1.1       tholo    1218:
1.7       millert  1219: Oct 30, 1988:
                   1220:        Fixed bug in call() that failed to recover storage.
1.1       tholo    1221:
1.7       millert  1222:        A warning is now generated if there are more arguments
                   1223:        in the call than in the definition (in lieu of fixing
                   1224:        another storage leak).
1.2       millert  1225:
1.7       millert  1226: Oct 20, 1988:
                   1227:        Fixed %c:  if expr is numeric, use numeric value;
                   1228:        otherwise print 1st char of string value.  still
                   1229:        doesn't work if the value is 0 -- won't print \0.
1.2       millert  1230:
1.7       millert  1231:        Added a few more checks for running out of malloc.
1.2       millert  1232:
1.7       millert  1233: Oct 12, 1988:
                   1234:        Fixed bug in call() that freed local arrays twice.
1.5       kstailey 1235:
1.7       millert  1236:        Fixed to handle deletion of non-existent array right;
                   1237:        complains about attempt to delete non-array element.
1.6       kstailey 1238:
1.7       millert  1239: Sep 30, 1988:
                   1240:        Now guarantees to evaluate all arguments of built-in
                   1241:        functions, as in C;  the appearance is that arguments
                   1242:        are evaluated before the function is called.  Places
                   1243:        affected are sub (gsub was ok), substr, printf, and
                   1244:        all the built-in arithmetic functions in bltin().
                   1245:        A warning is generated if a bltin() is called with
                   1246:        the wrong number of arguments.
1.6       kstailey 1247:
1.7       millert  1248:        This requires changing makeprof on p167 of the book.
1.6       kstailey 1249:
1.7       millert  1250: Aug 23, 1988:
                   1251:        setting FILENAME in BEGIN caused core dump, apparently
                   1252:        because it was freeing space not allocated by malloc.
1.6       kstailey 1253:
1.7       millert  1254: July 24, 1988:
                   1255:        fixed egregious error in toupper/tolower functions.
                   1256:        still subject to rescinding, however.
1.6       kstailey 1257:
1.7       millert  1258: July 2, 1988:
                   1259:        flush stdout before opening file or pipe
1.6       kstailey 1260:
1.7       millert  1261: July 2, 1988:
                   1262:        performance bug in b.c/cgoto(): not freeing some sets of states.
                   1263:        partial fix only right now, and the number of states increased
                   1264:        to make it less obvious.
1.6       kstailey 1265:
1.7       millert  1266: June 1, 1988:
                   1267:        check error status on close
1.6       kstailey 1268:
1.7       millert  1269: May 28, 1988:
                   1270:        srand returns seed value it's using.
                   1271:        see 1/18/90
1.6       kstailey 1272:
1.7       millert  1273: May 22, 1988:
                   1274:        Removed limit on depth of function calls.
1.6       kstailey 1275:
1.7       millert  1276: May 10, 1988:
                   1277:        Fixed lib.c to permit _ in commandline variable names.
1.6       kstailey 1278:
1.7       millert  1279: Mar 25, 1988:
                   1280:        main.c fixed to recognize -- as terminator of command-
                   1281:        line options.  Illegal options flagged.
                   1282:        Error reporting slightly cleaned up.
1.6       kstailey 1283:
1.7       millert  1284: Dec 2, 1987:
                   1285:        Newer C compilers apply a strict scope rule to extern
                   1286:        declarations within functions.  Two extern declarations in
                   1287:        lib.c and tran.c have been moved to obviate this problem.
1.6       kstailey 1288:
1.7       millert  1289: Oct xx, 1987:
                   1290:        Reluctantly added toupper and tolower functions.
                   1291:        Subject to rescinding without notice.
1.6       kstailey 1292:
1.7       millert  1293: Sep 17, 1987:
                   1294:        Error-message printer had printf(s) instead of
                   1295:        printf("%s",s);  got core dumps when the message
                   1296:        included a %.
1.6       kstailey 1297:
1.7       millert  1298: Sep 12, 1987:
                   1299:        Very long printf strings caused core dump;
                   1300:        fixed aprintf, asprintf, format to catch them.
                   1301:        Can still get a core dump in printf itself.
1.6       kstailey 1302:
                   1303: