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