FIXES revision 170331
185587Sobrien/**************************************************************** 285587SobrienCopyright (C) Lucent Technologies 1997 385587SobrienAll Rights Reserved 485587Sobrien 585587SobrienPermission to use, copy, modify, and distribute this software and 685587Sobrienits documentation for any purpose and without fee is hereby 785587Sobriengranted, provided that the above copyright notice appear in all 885587Sobriencopies and that both that the copyright notice and this 985587Sobrienpermission notice and warranty disclaimer appear in supporting 1085587Sobriendocumentation, and that the name Lucent Technologies or any of 1185587Sobrienits entities not be used in advertising or publicity pertaining 1285587Sobriento distribution of the software without specific, written prior 1385587Sobrienpermission. 1485587Sobrien 1585587SobrienLUCENT DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, 1685587SobrienINCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS. 1785587SobrienIN NO EVENT SHALL LUCENT OR ANY OF ITS ENTITIES BE LIABLE FOR ANY 1885587SobrienSPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES 1985587SobrienWHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER 2085587SobrienIN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, 2185587SobrienARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF 2285587SobrienTHIS SOFTWARE. 2385587Sobrien****************************************************************/ 2485587Sobrien 2585587SobrienThis file lists all bug fixes, changes, etc., made since the AWK book 2685587Sobrienwas sent to the printers in August, 1987. 2785587Sobrien 28170331SrafanMay 1, 2007: 29170331Srafan fiddle in makefile to fix for BSD make; thanks to igor sobrado. 30170331Srafan 31170331SrafanMar 31, 2007: 32170331Srafan fixed some null pointer refs calling adjbuf. 33170331Srafan 34170331SrafanFeb 21, 2007: 35170331Srafan fixed a bug in matching the null RE in sub and gsub. thanks to al aho 36170331Srafan who actually did the fix (in b.c), and to wolfgang seeberg for finding 37170331Srafan it and providing a very compact test case. 38170331Srafan 39170331Srafan fixed quotation in b.c; thanks to Hal Pratt and the Princeton Dante 40170331Srafan Project. 41170331Srafan 42170331Srafan removed some no-effect asserts in run.c. 43170331Srafan 44170331Srafan fiddled maketab.c to not complain about bison-generated values. 45170331Srafan 46170331Srafan removed the obsolete -V argument; fixed --version to print the 47170331Srafan version and exit. 48170331Srafan 49170331Srafan fixed wording and an outright error in the usage message; thanks to igor 50170331Srafan sobrado and jason mcintyre. 51170331Srafan 52170331Srafan fixed a bug in -d that caused core dump if no program followed. 53170331Srafan 54170331SrafanJan 1, 2007: 55170331Srafan dropped mac.code from makefile; there are few non-MacOSX 56170331Srafan mac's these days. 57170331Srafan 58170331SrafanJan 17, 2006: 59170331Srafan system() not flagged as unsafe in the unadvertised -safe option. 60170331Srafan found it while enhancing tests before shipping the ;login: article. 61170331Srafan practice what you preach. 62170331Srafan 63170331Srafan removed the 9-years-obsolete -mr and -mf flags. 64170331Srafan 65170331Srafan added -version and --version options. 66170331Srafan 67170331Srafan core dump on linux with BEGIN {nextfile}, now fixed. 68170331Srafan 69170331Srafan removed some #ifdef's in run.c and lex.c that appear to no 70170331Srafan longer be necessary. 71170331Srafan 72146299SruApr 24, 2005: 73146299Sru modified lib.c so that values of $0 et al are preserved in the END 74146299Sru block, apparently as required by posix. thanks to havard eidnes 75146299Sru for the report and code. 76146299Sru 77146299SruJan 14, 2005: 78146299Sru fixed infinite loop in parsing, originally found by brian tsang. 79146299Sru thanks to arnold robbins for a suggestion that started me 80146299Sru rethinking it. 81146299Sru 82146299SruDec 31, 2004: 83146299Sru prevent overflow of -f array in main, head off potential error in 84146299Sru call of SYNTAX(), test malloc return in lib.c, all with thanks to 85146299Sru todd miller. 86146299Sru 87146299SruDec 22, 2004: 88146299Sru cranked up size of NCHARS; coverity thinks it can be overrun with 89146299Sru smaller size, and i think that's right. added some assertions to b.c 90146299Sru to catch places where it might overrun. the RE code is still fragile. 91146299Sru 92146299SruDec 5, 2004: 93146299Sru fixed a couple of overflow problems with ridiculous field numbers: 94146299Sru e.g., print $(2^32-1). thanks to ruslan ermilov, giorgos keramidas 95146299Sru and david o'brien at freebsd.org for patches. this really should 96146299Sru be re-done from scratch. 97146299Sru 98146299SruNov 21, 2004: 99146299Sru fixed another 25-year-old RE bug, in split. it's another failure 100146299Sru to (re-)initialize. thanks to steve fisher for spotting this and 101146299Sru providing a good test case. 102146299Sru 103125601SruNov 22, 2003: 104125601Sru fixed a bug in regular expressions that dates (so help me) from 1977; 105125601Sru it's been there from the beginning. an anchored longest match that 106125601Sru was longer than the number of states triggered a failure to initialize 107125601Sru the machine properly. many thanks to moinak ghosh for not only finding 108125601Sru this one but for providing a fix, in some of the most mysterious 109125601Sru code known to man. 110125601Sru 111125601Sru fixed a storage leak in call() that appears to have been there since 112125601Sru 1983 or so -- a function without an explicit return that assigns a 113125601Sru string to a parameter leaked a Cell. thanks to moinak ghosh for 114125601Sru spotting this very subtle one. 115125601Sru 116125601SruJul 31, 2003: 117125601Sru fixed, thanks to andrey chernov and ruslan ermilov, a bug in lex.c 118125601Sru that mis-handled the character 255 in input. (it was being compared 119125601Sru to EOF with a signed comparison.) 120125601Sru 121118194SruJul 29, 2003: 122118194Sru fixed (i think) the long-standing botch that included the beginning of 123118194Sru line state ^ for RE's in the set of valid characters; this led to a 124118194Sru variety of odd problems, including failure to properly match certain 125118194Sru regular expressions in non-US locales. thanks to ruslan for keeping 126118194Sru at this one. 127118194Sru 128118194SruJul 28, 2003: 129118194Sru n-th try at getting internationalization right, with thanks to volker 130118194Sru kiefel, arnold robbins and ruslan ermilov for advice, though they 131118194Sru should not be blamed for the outcome. according to posix, "." is the 132118194Sru radix character in programs and command line arguments regardless of 133118194Sru the locale; otherwise, the locale should prevail for input and output 134118194Sru of numbers. so it's intended to work that way. 135118194Sru 136118194Sru i have rescinded the attempt to use strcoll in expanding shorthands in 137118194Sru regular expressions (cclenter). its properties are much too 138118194Sru surprising; for example [a-c] matches aAbBc in locale en_US but abBcC 139118194Sru in locale fr_CA. i can see how this might arise by implementation 140118194Sru but i cannot explain it to a human user. (this behavior can be seen 141118194Sru in gawk as well; we're leaning on the same library.) 142118194Sru 143118194Sru the issue appears to be that strcoll is meant for sorting, where 144118194Sru merging upper and lower case may make sense (though note that unix 145118194Sru sort does not do this by default either). it is not appropriate 146118194Sru for regular expressions, where the goal is to match specific 147118194Sru patterns of characters. in any case, the notations [:lower:], etc., 148118194Sru are available in awk, and they are more likely to work correctly in 149118194Sru most locales. 150118194Sru 151118194Sru a moratorium is hereby declared on internationalization changes. 152118194Sru i apologize to friends and colleagues in other parts of the world. 153118194Sru i would truly like to get this "right", but i don't know what 154118194Sru that is, and i do not want to keep making changes until it's clear. 155118194Sru 156118194SruJul 4, 2003: 157118194Sru fixed bug that permitted non-terminated RE, as in "awk /x". 158118194Sru 159118194SruJun 1, 2003: 160118194Sru subtle change to split: if source is empty, number of elems 161118194Sru is always 0 and the array is not set. 162118194Sru 163118194SruMar 21, 2003: 164118194Sru added some parens to isblank, in another attempt to make things 165118194Sru internationally portable. 166118194Sru 167112336SobrienMar 14, 2003: 168112336Sobrien the internationalization changes, somewhat modified, are now 169112336Sobrien reinstated. in theory awk will now do character comparisons 170112336Sobrien and case conversions in national language, but "." will always 171112336Sobrien be the decimal point separator on input and output regardless 172112336Sobrien of national language. isblank(){} has an #ifndef. 173112336Sobrien 174112336Sobrien this no longer compiles on windows: LC_MESSAGES isn't defined 175112336Sobrien in vc6++. 176112336Sobrien 177112336Sobrien fixed subtle behavior in field and record splitting: if FS is 178112336Sobrien a single character and RS is not empty, \n is NOT a separator. 179112336Sobrien this tortuous reading is found in the awk book; behavior now 180112336Sobrien matches gawk and mawk. 181112336Sobrien 182108072SobrienDec 13, 2002: 183108072Sobrien for the moment, the internationalization changes of nov 29 are 184108072Sobrien rolled back -- programs like x = 1.2 don't work in some locales, 185108072Sobrien because the parser is expecting x = 1,2. until i understand this 186108072Sobrien better, this will have to wait. 187108072Sobrien 188107806SobrienNov 29, 2002: 189107806Sobrien modified b.c (with tiny changes in main and run) to support 190107806Sobrien locales, using strcoll and iswhatever tests for posix character 191107806Sobrien classes. thanks to ruslan ermilov (ru@freebsd.org) for code. 192107806Sobrien the function isblank doesn't seem to have propagated to any 193107806Sobrien header file near me, so it's there explicitly. not properly 194107806Sobrien tested on non-ascii character sets by me. 195107806Sobrien 196107806SobrienJun 28, 2002: 197107806Sobrien modified run/format() and tran/getsval() to do a slightly better 198107806Sobrien job on using OFMT for output from print and CONVFMT for other 199107806Sobrien number->string conversions, as promised by posix and done by 200107806Sobrien gawk and mawk. there are still places where it doesn't work 201107806Sobrien right if CONVFMT is changed; by then the STR attribute of the 202107806Sobrien variable has been irrevocably set. thanks to arnold robbins for 203107806Sobrien code and examples. 204107806Sobrien 205107806Sobrien fixed subtle bug in format that could get core dump. thanks to 206107806Sobrien Jaromir Dolecek <jdolecek@NetBSD.org> for finding and fixing. 207107806Sobrien minor cleanup in run.c / format() at the same time. 208107806Sobrien 209107806Sobrien added some tests for null pointers to debugging printf's, which 210107806Sobrien were never intended for external consumption. thanks to dave 211107806Sobrien kerns (dkerns@lucent.com) for pointing this out. 212107806Sobrien 213107806Sobrien GNU compatibility: an empty regexp matches anything (thanks to 214107806Sobrien dag-erling smorgrav, des@ofug.org). subject to reversion if 215107806Sobrien this does more harm than good. 216107806Sobrien 217107806Sobrien pervasive small changes to make things more const-correct, as 218107806Sobrien reported by gcc's -Wwrite-strings. as it says in the gcc manual, 219107806Sobrien this may be more nuisance than useful. provoked by a suggestion 220107806Sobrien and code from arnaud desitter, arnaud@nimbus.geog.ox.ac.uk 221107806Sobrien 222107806Sobrien minor documentation changes to note that this now compiles out 223107806Sobrien of the box on Mac OS X. 224107806Sobrien 22590902SdesFeb 10, 2002: 22690902Sdes changed types in posix chars structure to quiet solaris cc. 22790902Sdes 22890902SdesJan 1, 2002: 22990902Sdes fflush() or fflush("") flushes all files and pipes. 23090902Sdes 23190902Sdes length(arrayname) returns number of elements; thanks to 23290902Sdes arnold robbins for suggestion. 23390902Sdes 23490902Sdes added a makefile.win to make it easier to build on windows. 23590902Sdes based on dan allen's buildwin.bat. 23690902Sdes 23790902SdesNov 16, 2001: 23890902Sdes added support for posix character class names like [:digit:], 23990902Sdes which are not exactly shorter than [0-9] and perhaps no more 24090902Sdes portable. thanks to dag-erling smorgrav for code. 24190902Sdes 24290902SdesFeb 16, 2001: 24390902Sdes removed -m option; no longer needed, and it was actually 24490902Sdes broken (noted thanks to volker kiefel). 24590902Sdes 24690902SdesFeb 10, 2001: 24790902Sdes fixed an appalling bug in gettok: any sequence of digits, +,-, E, e, 24890902Sdes and period was accepted as a valid number if it started with a period. 24990902Sdes this would never have happened with the lex version. 25090902Sdes 25190902Sdes other 1-character botches, now fixed, include a bare $ and a 25290902Sdes bare " at the end of the input. 25390902Sdes 25490902SdesFeb 7, 2001: 25590902Sdes more (const char *) casts in b.c and tran.c to silence warnings. 25690902Sdes 25785587SobrienNov 15, 2000: 25885587Sobrien fixed a bug introduced in august 1997 that caused expressions 25985587Sobrien like $f[1] to be syntax errors. thanks to arnold robbins for 26085587Sobrien noticing this and providing a fix. 26185587Sobrien 26285587SobrienOct 30, 2000: 26385587Sobrien fixed some nextfile bugs: not handling all cases. thanks to 26485587Sobrien arnold robbins for pointing this out. new regressions added. 26585587Sobrien 26685587Sobrien close() is now a function. it returns whatever the library 26785587Sobrien fclose returns, and -1 for closing a file or pipe that wasn't 26885587Sobrien opened. 26985587Sobrien 27085587SobrienSep 24, 2000: 27185587Sobrien permit \n explicitly in character classes; won't work right 27285587Sobrien if comes in as "[\n]" but ok as /[\n]/, because of multiple 27385587Sobrien processing of \'s. thanks to arnold robbins. 27485587Sobrien 27585587SobrienJuly 5, 2000: 27685587Sobrien minor fiddles in tran.c to keep compilers happy about uschar. 27785587Sobrien thanks to norman wilson. 27885587Sobrien 27985587SobrienMay 25, 2000: 28085587Sobrien yet another attempt at making 8-bit input work, with another 28185587Sobrien band-aid in b.c (member()), and some (uschar) casts to head 28285587Sobrien off potential errors in subscripts (like isdigit). also 28385587Sobrien changed HAT to NCHARS-2. thanks again to santiago vila. 28485587Sobrien 28585587Sobrien changed maketab.c to ignore apparently out of range definitions 28685587Sobrien instead of halting; new freeBSD generates one. thanks to 28785587Sobrien jon snader <jsnader@ix.netcom.com> for pointing out the problem. 28885587Sobrien 28985587SobrienMay 2, 2000: 29085587Sobrien fixed an 8-bit problem in b.c by making several char*'s into 29185587Sobrien unsigned char*'s. not clear i have them all yet. thanks to 29285587Sobrien Santiago Vila <sanvila@unex.es> for the bug report. 29385587Sobrien 29485587SobrienApr 21, 2000: 29585587Sobrien finally found and fixed a memory leak in function call; it's 29685587Sobrien been there since functions were added ~1983. thanks to 29785587Sobrien jon bentley for the test case that found it. 29885587Sobrien 29985587Sobrien added test in envinit to catch environment "variables" with 300146299Sru names beginning with '='; thanks to Berend Hasselman. 30185587Sobrien 30285587SobrienJul 28, 1999: 30385587Sobrien added test in defn() to catch function foo(foo), which 30485587Sobrien otherwise recurses until core dump. thanks to arnold 30585587Sobrien robbins for noticing this. 30685587Sobrien 30785587SobrienJun 20, 1999: 30885587Sobrien added *bp in gettok in lex.c; appears possible to exit function 30985587Sobrien without terminating the string. thanks to russ cox. 31085587Sobrien 31185587SobrienJun 2, 1999: 31285587Sobrien added function stdinit() to run to initialize files[] array, 31385587Sobrien in case stdin, etc., are not constants; some compilers care. 31485587Sobrien 31585587SobrienMay 10, 1999: 31685587Sobrien replaced the ERROR ... FATAL, etc., macros with functions 31785587Sobrien based on vprintf, to avoid problems caused by overrunning 31885587Sobrien fixed-size errbuf array. thanks to ralph corderoy for the 31985587Sobrien impetus, and for pointing out a string termination bug in 32085587Sobrien qstring as well. 32185587Sobrien 32285587SobrienApr 21, 1999: 32385587Sobrien fixed bug that caused occasional core dumps with commandline 32485587Sobrien variable with value ending in \. (thanks to nelson beebe for 32585587Sobrien the test case.) 32685587Sobrien 32785587SobrienApr 16, 1999: 32885587Sobrien with code kindly provided by Bruce Lilly, awk now parses 32985587Sobrien /=/ and similar constructs more sensibly in more places. 33085587Sobrien Bruce also provided some helpful test cases. 33185587Sobrien 33285587SobrienApr 5, 1999: 33385587Sobrien changed true/false to True/False in run.c to make it 33485587Sobrien easier to compile with C++. Added some casts on malloc 33585587Sobrien and realloc to be honest about casts; ditto. changed 33685587Sobrien ltype int to long in struct rrow to reduce some 64-bit 33785587Sobrien complaints; other changes scattered throughout for the 33885587Sobrien same purpose. thanks to Nelson Beebe for these portability 33985587Sobrien improvements. 34085587Sobrien 34185587Sobrien removed some horrible pointer-int casting in b.c and elsewhere 34285587Sobrien by adding ptoi and itonp to localize the casts, which are 34385587Sobrien all benign. fixed one incipient bug that showed up on sgi 34485587Sobrien in 64-bit mode. 34585587Sobrien 34685587Sobrien reset lineno for new source file; include filename in error 34785587Sobrien message. also fixed line number error in continuation lines. 34885587Sobrien (thanks to Nelson Beebe for both of these.) 34985587Sobrien 35085587SobrienMar 24, 1999: 35185587Sobrien Nelson Beebe notes that irix 5.3 yacc dies with a bogus 35285587Sobrien error; use a newer version or switch to bison, since sgi 35385587Sobrien is unlikely to fix it. 35485587Sobrien 35585587SobrienMar 5, 1999: 35685587Sobrien changed isnumber to is_number to avoid the problem caused by 35785587Sobrien versions of ctype.h that include the name isnumber. 35885587Sobrien 35985587Sobrien distribution now includes a script for building on a Mac, 36085587Sobrien thanks to Dan Allen. 36185587Sobrien 36285587SobrienFeb 20, 1999: 36385587Sobrien fixed memory leaks in run.c (call) and tran.c (setfval). 36485587Sobrien thanks to Stephen Nutt for finding these and providing the fixes. 36585587Sobrien 36685587SobrienJan 13, 1999: 36785587Sobrien replaced srand argument by (unsigned int) in run.c; 36885587Sobrien avoids problem on Mac and potentially on Unix & Windows. 36985587Sobrien thanks to Dan Allen. 37085587Sobrien 37185587Sobrien added a few (int) casts to silence useless compiler warnings. 37285587Sobrien e.g., errorflag= in run.c jump(). 37385587Sobrien 37485587Sobrien added proctab.c to the bundle outout; one less thing 37585587Sobrien to have to compile out of the box. 37685587Sobrien 37785587Sobrien added calls to _popen and _pclose to the win95 stub for 37885587Sobrien pipes (thanks to Steve Adams for this helpful suggestion). 37985587Sobrien seems to work, though properties are not well understood 38085587Sobrien by me, and it appears that under some circumstances the 38185587Sobrien pipe output is truncated. Be careful. 38285587Sobrien 38385587SobrienOct 19, 1998: 38485587Sobrien fixed a couple of bugs in getrec: could fail to update $0 38585587Sobrien after a getline var; because inputFS wasn't initialized, 38685587Sobrien could split $0 on every character, a misleading diversion. 38785587Sobrien 38885587Sobrien fixed caching bug in makedfa: LRU was actually removing 38985587Sobrien least often used. 39085587Sobrien 39185587Sobrien thanks to ross ridge for finding these, and for providing 39285587Sobrien great bug reports. 39385587Sobrien 39485587SobrienMay 12, 1998: 39585587Sobrien fixed potential bug in readrec: might fail to update record 39685587Sobrien pointer after growing. thanks to dan levy for spotting this 39785587Sobrien and suggesting the fix. 39885587Sobrien 39985587SobrienMar 12, 1998: 40085587Sobrien added -V to print version number and die. 40185587Sobrien 40285587SobrienFeb 11, 1998: 40385587Sobrien subtle silent bug in lex.c: if the program ended with a number 40485587Sobrien longer than 1 digit, part of the input would be pushed back and 40585587Sobrien parsed again because token buffer wasn't terminated right. 40685587Sobrien example: awk 'length($0) > 10'. blush. at least i found it 40785587Sobrien myself. 40885587Sobrien 40985587SobrienAug 31, 1997: 41085587Sobrien s/adelete/awkdelete/: SGI uses this in malloc.h. 41185587Sobrien thanks to nelson beebe for pointing this one out. 41285587Sobrien 41385587SobrienAug 21, 1997: 41485587Sobrien fixed some bugs in sub and gsub when replacement includes \\. 41585587Sobrien this is a dark, horrible corner, but at least now i believe that 41685587Sobrien the behavior is the same as gawk and the intended posix standard. 41785587Sobrien thanks to arnold robbins for advice here. 41885587Sobrien 41985587SobrienAug 9, 1997: 42085587Sobrien somewhat regretfully, replaced the ancient lex-based lexical 42185587Sobrien analyzer with one written in C. it's longer, generates less code, 42285587Sobrien and more portable; the old one depended too much on mysterious 42385587Sobrien properties of lex that were not preserved in other environments. 42485587Sobrien in theory these recognize the same language. 42585587Sobrien 42685587Sobrien now using strtod to test whether a string is a number, instead of 42785587Sobrien the convoluted original function. should be more portable and 42885587Sobrien reliable if strtod is implemented right. 42985587Sobrien 43085587Sobrien removed now-pointless optimization in makefile that tries to avoid 43185587Sobrien recompilation when awkgram.y is changed but symbols are not. 43285587Sobrien 43385587Sobrien removed most fixed-size arrays, though a handful remain, some 43485587Sobrien of which are unchecked. you have been warned. 43585587Sobrien 43685587SobrienAug 4, 1997: 43785587Sobrien with some trepidation, replaced the ancient code that managed 43885587Sobrien fields and $0 in fixed-size arrays with arrays that grow on 43985587Sobrien demand. there is still some tension between trying to make this 44085587Sobrien run fast and making it clean; not sure it's right yet. 44185587Sobrien 44285587Sobrien the ill-conceived -mr and -mf arguments are now useful only 44385587Sobrien for debugging. previous dynamic string code removed. 44485587Sobrien 44585587Sobrien numerous other minor cleanups along the way. 44685587Sobrien 44785587SobrienJul 30, 1997: 44885587Sobrien using code provided by dan levy (to whom profuse thanks), replaced 44985587Sobrien fixed-size arrays and awkward kludges by a fairly uniform mechanism 45085587Sobrien to grow arrays as needed for printf, sub, gsub, etc. 45185587Sobrien 45285587SobrienJul 23, 1997: 45385587Sobrien falling off the end of a function returns "" and 0, not 0. 45485587Sobrien thanks to arnold robbins. 45585587Sobrien 45685587SobrienJun 17, 1997: 45785587Sobrien replaced several fixed-size arrays by dynamically-created ones 45885587Sobrien in run.c; added overflow tests to some previously unchecked cases. 45985587Sobrien getline, toupper, tolower. 46085587Sobrien 46185587Sobrien getline code is still broken in that recursive calls may wind 46285587Sobrien up using the same space. [fixed later] 46385587Sobrien 46485587Sobrien increased RECSIZE to 8192 to push problems further over the horizon. 46585587Sobrien 46685587Sobrien added \r to \n as input line separator for programs, not data. 46785587Sobrien damn CRLFs. 46885587Sobrien 46985587Sobrien modified format() to permit explicit printf("%c", 0) to include 47085587Sobrien a null byte in output. thanks to ken stailey for the fix. 47185587Sobrien 47285587Sobrien added a "-safe" argument that disables file output (print >, 47385587Sobrien print >>), process creation (cmd|getline, print |, system), and 47485587Sobrien access to the environment (ENVIRON). this is a first approximation 47585587Sobrien to a "safe" version of awk, but don't rely on it too much. thanks 47685587Sobrien to joan feigenbaum and matt blaze for the inspiration long ago. 47785587Sobrien 47885587SobrienJul 8, 1996: 47985587Sobrien fixed long-standing bug in sub, gsub(/a/, "\\\\&"); thanks to 48085587Sobrien ralph corderoy. 48185587Sobrien 48285587SobrienJun 29, 1996: 48385587Sobrien fixed awful bug in new field splitting; didn't get all the places 48485587Sobrien where input was done. 48585587Sobrien 48685587SobrienJun 28, 1996: 48785587Sobrien changed field-splitting to conform to posix definition: fields are 48885587Sobrien split using the value of FS at the time of input; it used to be 48985587Sobrien the value when the field or NF was first referred to, a much less 49085587Sobrien predictable definition. thanks to arnold robbins for encouragement 49185587Sobrien to do the right thing. 49285587Sobrien 49385587SobrienMay 28, 1996: 49485587Sobrien fixed appalling but apparently unimportant bug in parsing octal 49585587Sobrien numbers in reg exprs. 49685587Sobrien 49785587Sobrien explicit hex in reg exprs now limited to 2 chars: \xa, \xaa. 49885587Sobrien 49985587SobrienMay 27, 1996: 50085587Sobrien cleaned up some declarations so gcc -Wall is now almost silent. 50185587Sobrien 50285587Sobrien makefile now includes backup copies of ytab.c and lexyy.c in case 50385587Sobrien one makes before looking; it also avoids recreating lexyy.c unless 50485587Sobrien really needed. 50585587Sobrien 50685587Sobrien s/aprintf/awkprint, s/asprintf/awksprintf/ to avoid some name clashes 50785587Sobrien with unwisely-written header files. 50885587Sobrien 50985587Sobrien thanks to jeffrey friedl for several of these. 51085587Sobrien 51185587SobrienMay 26, 1996: 51285587Sobrien an attempt to rationalize the (unsigned) char issue. almost all 51385587Sobrien instances of unsigned char have been removed; the handful of places 51485587Sobrien in b.c where chars are used as table indices have been hand-crafted. 51585587Sobrien added some latin-1 tests to the regression, but i'm not confident; 51685587Sobrien none of my compilers seem to care much. thanks to nelson beebe for 51785587Sobrien pointing out some others that do care. 51885587Sobrien 51985587SobrienMay 2, 1996: 52085587Sobrien removed all register declarations. 52185587Sobrien 52285587Sobrien enhanced split(), as in gawk, etc: split(s, a, "") splits s into 52385587Sobrien a[1]...a[length(s)] with each character a single element. 52485587Sobrien 52585587Sobrien made the same changes for field-splitting if FS is "". 52685587Sobrien 52785587Sobrien added nextfile, as in gawk: causes immediate advance to next 52885587Sobrien input file. (thanks to arnold robbins for inspiration and code). 52985587Sobrien 53085587Sobrien small fixes to regexpr code: can now handle []], [[], and 53185587Sobrien variants; [] is now a syntax error, rather than matching 53285587Sobrien everything; [z-a] is now empty, not z. far from complete 53385587Sobrien or correct, however. (thanks to jeffrey friedl for pointing out 53485587Sobrien some awful behaviors.) 53585587Sobrien 53685587SobrienApr 29, 1996: 537146299Sru replaced uchar by uschar everywhere; apparently some compilers 53885587Sobrien usurp this name and this causes conflicts. 53985587Sobrien 54085587Sobrien fixed call to time in run.c (bltin); arg is time_t *. 54185587Sobrien 54285587Sobrien replaced horrible pointer/long punning in b.c by a legitimate 54385587Sobrien union. should be safer on 64-bit machines and cleaner everywhere. 54485587Sobrien (thanks to nelson beebe for pointing out some of these problems.) 54585587Sobrien 54685587Sobrien replaced nested comments by #if 0...#endif in run.c, lib.c. 54785587Sobrien 54885587Sobrien removed getsval, setsval, execute macros from run.c and lib.c. 54985587Sobrien machines are 100x faster than they were when these macros were 55085587Sobrien first used. 55185587Sobrien 55285587Sobrien revised filenames: awk.g.y => awkgram.y, awk.lx.l => awklex.l, 55385587Sobrien y.tab.[ch] => ytab.[ch], lex.yy.c => lexyy.c, all in the aid of 55485587Sobrien portability to nameless systems. 55585587Sobrien 55685587Sobrien "make bundle" now includes yacc and lex output files for recipients 55785587Sobrien who don't have yacc or lex. 55885587Sobrien 55985587SobrienAug 15, 1995: 56085587Sobrien initialized Cells in setsymtab more carefully; some fields 56185587Sobrien were not set. (thanks to purify, all of whose complaints i 56285587Sobrien think i now understand.) 56385587Sobrien 56485587Sobrien fixed at least one error in gsub that looked at -1-th element 56585587Sobrien of an array when substituting for a null match (e.g., $). 56685587Sobrien 56785587Sobrien delete arrayname is now legal; it clears the elements but leaves 56885587Sobrien the array, which may not be the right behavior. 56985587Sobrien 57085587Sobrien modified makefile: my current make can't cope with the test used 57185587Sobrien to avoid unnecessary yacc invocations. 57285587Sobrien 57385587SobrienJul 17, 1995: 57485587Sobrien added dynamically growing strings to awk.lx.l and b.c 57585587Sobrien to permit regular expressions to be much bigger. 57685587Sobrien the state arrays can still overflow. 57785587Sobrien 57885587SobrienAug 24, 1994: 57985587Sobrien detect duplicate arguments in function definitions (mdm). 58085587Sobrien 58185587SobrienMay 11, 1994: 58285587Sobrien trivial fix to printf to limit string size in sub(). 58385587Sobrien 58485587SobrienApr 22, 1994: 58585587Sobrien fixed yet another subtle self-assignment problem: 58685587Sobrien $1 = $2; $1 = $1 clobbered $1. 58785587Sobrien 58885587Sobrien Regression tests now use private echo, to avoid quoting problems. 58985587Sobrien 59085587SobrienFeb 2, 1994: 59185587Sobrien changed error() to print line number as %d, not %g. 59285587Sobrien 59385587SobrienJul 23, 1993: 59485587Sobrien cosmetic changes: increased sizes of some arrays, 59585587Sobrien reworded some error messages. 59685587Sobrien 59785587Sobrien added CONVFMT as in posix (just replaced OFMT in getsval) 59885587Sobrien 59985587Sobrien FILENAME is now "" until the first thing that causes a file 60085587Sobrien to be opened. 60185587Sobrien 60285587SobrienNov 28, 1992: 60385587Sobrien deleted yyunput and yyoutput from proto.h; 60485587Sobrien different versions of lex give these different declarations. 60585587Sobrien 60685587SobrienMay 31, 1992: 60785587Sobrien added -mr N and -mf N options: more record and fields. 60885587Sobrien these really ought to adjust automatically. 60985587Sobrien 61085587Sobrien cleaned up some error messages; "out of space" now means 61185587Sobrien malloc returned NULL in all cases. 61285587Sobrien 61385587Sobrien changed rehash so that if it runs out, it just returns; 61485587Sobrien things will continue to run slow, but maybe a bit longer. 61585587Sobrien 61685587SobrienApr 24, 1992: 61785587Sobrien remove redundant close of stdin when using -f -. 61885587Sobrien 61985587Sobrien got rid of core dump with -d; awk -d just prints date. 62085587Sobrien 62185587SobrienApr 12, 1992: 62285587Sobrien added explicit check for /dev/std(in,out,err) in redirection. 62385587Sobrien unlike gawk, no /dev/fd/n yet. 62485587Sobrien 62585587Sobrien added (file/pipe) builtin. hard to test satisfactorily. 62685587Sobrien not posix. 62785587Sobrien 62885587SobrienFeb 20, 1992: 62985587Sobrien recompile after abortive changes; should be unchanged. 63085587Sobrien 63185587SobrienDec 2, 1991: 63285587Sobrien die-casting time: converted to ansi C, installed that. 63385587Sobrien 63485587SobrienNov 30, 1991: 63585587Sobrien fixed storage leak in freefa, failing to recover [N]CCL. 63685587Sobrien thanks to Bill Jones (jones@cs.usask.ca) 63785587Sobrien 63885587SobrienNov 19, 1991: 63985587Sobrien use RAND_MAX instead of literal in builtin(). 64085587Sobrien 64185587SobrienNov 12, 1991: 64285587Sobrien cranked up some fixed-size arrays in b.c, and added a test for 64385587Sobrien overflow in penter. thanks to mark larsen. 64485587Sobrien 64585587SobrienSep 24, 1991: 64685587Sobrien increased buffer in gsub. a very crude fix to a general problem. 64785587Sobrien and again on Sep 26. 64885587Sobrien 64985587SobrienAug 18, 1991: 65085587Sobrien enforce variable name syntax for commandline variables: has to 65185587Sobrien start with letter or _. 65285587Sobrien 65385587SobrienJul 27, 1991: 65485587Sobrien allow newline after ; in for statements. 65585587Sobrien 65685587SobrienJul 21, 1991: 65785587Sobrien fixed so that in self-assignment like $1=$1, side effects 65885587Sobrien like recomputing $0 take place. (this is getting subtle.) 65985587Sobrien 66085587SobrienJun 30, 1991: 66185587Sobrien better test for detecting too-long output record. 66285587Sobrien 66385587SobrienJun 2, 1991: 66485587Sobrien better defense against very long printf strings. 66585587Sobrien made break and continue illegal outside of loops. 66685587Sobrien 66785587SobrienMay 13, 1991: 66885587Sobrien removed extra arg on gettemp, tempfree. minor error message rewording. 66985587Sobrien 67085587SobrienMay 6, 1991: 67185587Sobrien fixed silly bug in hex parsing in hexstr(). 67285587Sobrien removed an apparently unnecessary test in isnumber(). 67385587Sobrien warn about weird printf conversions. 67485587Sobrien fixed unchecked array overwrite in relex(). 67585587Sobrien 67685587Sobrien changed for (i in array) to access elements in sorted order. 67785587Sobrien then unchanged it -- it really does run slower in too many cases. 67885587Sobrien left the code in place, commented out. 67985587Sobrien 68085587SobrienFeb 10, 1991: 68185587Sobrien check error status on all writes, to avoid banging on full disks. 68285587Sobrien 68385587SobrienJan 28, 1991: 68485587Sobrien awk -f - reads the program from stdin. 68585587Sobrien 68685587SobrienJan 11, 1991: 68785587Sobrien failed to set numeric state on $0 in cmd|getline context in run.c. 68885587Sobrien 68985587SobrienNov 2, 1990: 69085587Sobrien fixed sleazy test for integrality in getsval; use modf. 69185587Sobrien 69285587SobrienOct 29, 1990: 69385587Sobrien fixed sleazy buggy code in lib.c that looked (incorrectly) for 69485587Sobrien too long input lines. 69585587Sobrien 69685587SobrienOct 14, 1990: 69785587Sobrien fixed the bug on p. 198 in which it couldn't deduce that an 69885587Sobrien argument was an array in some contexts. replaced the error 69985587Sobrien message in intest() by code that damn well makes it an array. 70085587Sobrien 70185587SobrienOct 8, 1990: 70285587Sobrien fixed horrible bug: types and values were not preserved in 70385587Sobrien some kinds of self-assignment. (in assign().) 70485587Sobrien 70585587SobrienAug 24, 1990: 70685587Sobrien changed NCHARS to 256 to handle 8-bit characters in strings 70785587Sobrien presented to match(), etc. 70885587Sobrien 70985587SobrienJun 26, 1990: 71085587Sobrien changed struct rrow (awk.h) to use long instead of int for lval, 71185587Sobrien since cfoll() stores a pointer in it. now works better when int's 71285587Sobrien are smaller than pointers! 71385587Sobrien 71485587SobrienMay 6, 1990: 71585587Sobrien AVA fixed the grammar so that ! is uniformly of the same precedence as 71685587Sobrien unary + and -. This renders illegal some constructs like !x=y, which 71785587Sobrien now has to be parenthesized as !(x=y), and makes others work properly: 71885587Sobrien !x+y is (!x)+y, and x!y is x !y, not two pattern-action statements. 71985587Sobrien (These problems were pointed out by Bob Lenk of Posix.) 72085587Sobrien 72185587Sobrien Added \x to regular expressions (already in strings). 72285587Sobrien Limited octal to octal digits; \8 and \9 are not octal. 72385587Sobrien Centralized the code for parsing escapes in regular expressions. 72485587Sobrien Added a bunch of tests to T.re and T.sub to verify some of this. 72585587Sobrien 72685587SobrienFeb 9, 1990: 72785587Sobrien fixed null pointer dereference bug in main.c: -F[nothing]. sigh. 72885587Sobrien 72985587Sobrien restored srand behavior: it returns the current seed. 73085587Sobrien 73185587SobrienJan 18, 1990: 73285587Sobrien srand now returns previous seed value (0 to start). 73385587Sobrien 73485587SobrienJan 5, 1990: 73585587Sobrien fix potential problem in tran.c -- something was freed, 73685587Sobrien then used in freesymtab. 73785587Sobrien 73885587SobrienOct 18, 1989: 73985587Sobrien another try to get the max number of open files set with 74085587Sobrien relatively machine-independent code. 74185587Sobrien 74285587Sobrien small fix to input() in case of multiple reads after EOF. 74385587Sobrien 74485587SobrienOct 11, 1989: 74585587Sobrien FILENAME is now defined in the BEGIN block -- too many old 74685587Sobrien programs broke. 74785587Sobrien 74885587Sobrien "-" means stdin in getline as well as on the commandline. 74985587Sobrien 75085587Sobrien added a bunch of casts to the code to tell the truth about 75185587Sobrien char * vs. unsigned char *, a right royal pain. added a 75285587Sobrien setlocale call to the front of main, though probably no one 75385587Sobrien has it usefully implemented yet. 75485587Sobrien 75585587SobrienAug 24, 1989: 75685587Sobrien removed redundant relational tests against nullnode if parse 75785587Sobrien tree already had a relational at that point. 75885587Sobrien 75985587SobrienAug 11, 1989: 76085587Sobrien fixed bug: commandline variable assignment has to look like 76185587Sobrien var=something. (consider the man page for =, in file =.1) 76285587Sobrien 76385587Sobrien changed number of arguments to functions to static arrays 76485587Sobrien to avoid repeated malloc calls. 76585587Sobrien 76685587SobrienAug 2, 1989: 76785587Sobrien restored -F (space) separator 76885587Sobrien 76985587SobrienJul 30, 1989: 77085587Sobrien added -v x=1 y=2 ... for immediate commandline variable assignment; 77185587Sobrien done before the BEGIN block for sure. they have to precede the 77285587Sobrien program if the program is on the commandline. 77385587Sobrien Modified Aug 2 to require a separate -v for each assignment. 77485587Sobrien 77585587SobrienJul 10, 1989: 77685587Sobrien fixed ref-thru-zero bug in environment code in tran.c 77785587Sobrien 77885587SobrienJun 23, 1989: 77985587Sobrien add newline to usage message. 78085587Sobrien 78185587SobrienJun 14, 1989: 78285587Sobrien added some missing ansi printf conversion letters: %i %X %E %G. 78385587Sobrien no sensible meaning for h or L, so they may not do what one expects. 78485587Sobrien 78585587Sobrien made %* conversions work. 78685587Sobrien 78785587Sobrien changed x^y so that if n is a positive integer, it's done 78885587Sobrien by explicit multiplication, thus achieving maximum accuracy. 78985587Sobrien (this should be done by pow() but it seems not to be locally.) 79085587Sobrien done to x ^= y as well. 79185587Sobrien 79285587SobrienJun 4, 1989: 79385587Sobrien ENVIRON array contains environment: if shell variable V=thing, 79485587Sobrien ENVIRON["V"] is "thing" 79585587Sobrien 79685587Sobrien multiple -f arguments permitted. error reporting is naive. 79785587Sobrien (they were permitted before, but only the last was used.) 79885587Sobrien 79985587Sobrien fixed a really stupid botch in the debugging macro dprintf 80085587Sobrien 80185587Sobrien fixed order of evaluation of commandline assignments to match 80285587Sobrien what the book claims: an argument of the form x=e is evaluated 80385587Sobrien at the time it would have been opened if it were a filename (p 63). 80485587Sobrien this invalidates the suggested answer to ex 4-1 (p 195). 80585587Sobrien 80685587Sobrien removed some code that permitted -F (space) fieldseparator, 80785587Sobrien since it didn't quite work right anyway. (restored aug 2) 80885587Sobrien 80985587SobrienApr 27, 1989: 81085587Sobrien Line number now accumulated correctly for comment lines. 81185587Sobrien 81285587SobrienApr 26, 1989: 81385587Sobrien Debugging output now includes a version date, 81485587Sobrien if one compiles it into the source each time. 81585587Sobrien 81685587SobrienApr 9, 1989: 81785587Sobrien Changed grammar to prohibit constants as 3rd arg of sub and gsub; 81885587Sobrien prevents class of overwriting-a-constant errors. (Last one?) 81985587Sobrien This invalidates the "banana" example on page 43 of the book. 82085587Sobrien 82185587Sobrien Added \a ("alert"), \v (vertical tab), \xhhh (hexadecimal), 82285587Sobrien as in ANSI, for strings. Rescinded the sloppiness that permitted 82385587Sobrien non-octal digits in \ooo. Warning: not all compilers and libraries 82485587Sobrien will be able to deal with \x correctly. 82585587Sobrien 82685587SobrienJan 9, 1989: 82785587Sobrien Fixed bug that caused tempcell list to contain a duplicate. 82885587Sobrien The fix is kludgy. 82985587Sobrien 83085587SobrienDec 17, 1988: 83185587Sobrien Catches some more commandline errors in main. 83285587Sobrien Removed redundant decl of modf in run.c (confuses some compilers). 83385587Sobrien Warning: there's no single declaration of malloc, etc., in awk.h 83485587Sobrien that seems to satisfy all compilers. 83585587Sobrien 83685587SobrienDec 7, 1988: 83785587Sobrien Added a bit of code to error printing to avoid printing nulls. 83885587Sobrien (Not clear that it actually would.) 83985587Sobrien 84085587SobrienNov 27, 1988: 84185587Sobrien With fear and trembling, modified the grammar to permit 84285587Sobrien multiple pattern-action statements on one line without 84385587Sobrien an explicit separator. By definition, this capitulation 84485587Sobrien to the ghost of ancient implementations remains undefined 84585587Sobrien and thus subject to change without notice or apology. 84685587Sobrien DO NOT COUNT ON IT. 84785587Sobrien 84885587SobrienOct 30, 1988: 84985587Sobrien Fixed bug in call() that failed to recover storage. 85085587Sobrien 85185587Sobrien A warning is now generated if there are more arguments 85285587Sobrien in the call than in the definition (in lieu of fixing 85385587Sobrien another storage leak). 85485587Sobrien 85585587SobrienOct 20, 1988: 85685587Sobrien Fixed %c: if expr is numeric, use numeric value; 85785587Sobrien otherwise print 1st char of string value. still 85885587Sobrien doesn't work if the value is 0 -- won't print \0. 85985587Sobrien 86085587Sobrien Added a few more checks for running out of malloc. 86185587Sobrien 86285587SobrienOct 12, 1988: 86385587Sobrien Fixed bug in call() that freed local arrays twice. 86485587Sobrien 86585587Sobrien Fixed to handle deletion of non-existent array right; 86685587Sobrien complains about attempt to delete non-array element. 86785587Sobrien 86885587SobrienSep 30, 1988: 86985587Sobrien Now guarantees to evaluate all arguments of built-in 87085587Sobrien functions, as in C; the appearance is that arguments 87185587Sobrien are evaluated before the function is called. Places 87285587Sobrien affected are sub (gsub was ok), substr, printf, and 87385587Sobrien all the built-in arithmetic functions in bltin(). 87485587Sobrien A warning is generated if a bltin() is called with 87585587Sobrien the wrong number of arguments. 87685587Sobrien 87785587Sobrien This requires changing makeprof on p167 of the book. 87885587Sobrien 87985587SobrienAug 23, 1988: 88085587Sobrien setting FILENAME in BEGIN caused core dump, apparently 88185587Sobrien because it was freeing space not allocated by malloc. 88285587Sobrien 88385587SobrienJuly 24, 1988: 88485587Sobrien fixed egregious error in toupper/tolower functions. 88585587Sobrien still subject to rescinding, however. 88685587Sobrien 88785587SobrienJuly 2, 1988: 88885587Sobrien flush stdout before opening file or pipe 88985587Sobrien 89085587SobrienJuly 2, 1988: 89185587Sobrien performance bug in b.c/cgoto(): not freeing some sets of states. 89285587Sobrien partial fix only right now, and the number of states increased 89385587Sobrien to make it less obvious. 89485587Sobrien 89585587SobrienJune 1, 1988: 89685587Sobrien check error status on close 89785587Sobrien 89885587SobrienMay 28, 1988: 89985587Sobrien srand returns seed value it's using. 90085587Sobrien see 1/18/90 90185587Sobrien 90285587SobrienMay 22, 1988: 90385587Sobrien Removed limit on depth of function calls. 90485587Sobrien 90585587SobrienMay 10, 1988: 90685587Sobrien Fixed lib.c to permit _ in commandline variable names. 90785587Sobrien 90885587SobrienMar 25, 1988: 90985587Sobrien main.c fixed to recognize -- as terminator of command- 91085587Sobrien line options. Illegal options flagged. 91185587Sobrien Error reporting slightly cleaned up. 91285587Sobrien 91385587SobrienDec 2, 1987: 91485587Sobrien Newer C compilers apply a strict scope rule to extern 91585587Sobrien declarations within functions. Two extern declarations in 91685587Sobrien lib.c and tran.c have been moved to obviate this problem. 91785587Sobrien 91885587SobrienOct xx, 1987: 91985587Sobrien Reluctantly added toupper and tolower functions. 92085587Sobrien Subject to rescinding without notice. 92185587Sobrien 92285587SobrienSep 17, 1987: 92385587Sobrien Error-message printer had printf(s) instead of 92485587Sobrien printf("%s",s); got core dumps when the message 92585587Sobrien included a %. 92685587Sobrien 92785587SobrienSep 12, 1987: 92885587Sobrien Very long printf strings caused core dump; 92985587Sobrien fixed aprintf, asprintf, format to catch them. 93085587Sobrien Can still get a core dump in printf itself. 93185587Sobrien 93285587Sobrien 933