"Fossies" - the Fresh Open Source Software Archive

Member "AutoHotkey_L-1.1.33.09/source/lib_pcre/pcre/ChangeLog" (8 May 2021, 225923 Bytes) of package /windows/misc/AutoHotkey_L-1.1.33.09.zip:


As a special service "Fossies" has tried to format the requested text file into HTML format (style: standard) with prefixed line numbers. Alternatively you can here view or download the uninterpreted source code file.

    1 ChangeLog for PCRE
    2 ------------------
    3 
    4 Version 8.30 04-February-2012
    5 -----------------------------
    6 
    7 1.  Renamed "isnumber" as "is_a_number" because in some Mac environments this
    8     name is defined in ctype.h.
    9 
   10 2.  Fixed a bug in fixed-length calculation for lookbehinds that would show up
   11     only in quite long subpatterns.
   12 
   13 3.  Removed the function pcre_info(), which has been obsolete and deprecated
   14     since it was replaced by pcre_fullinfo() in February 2000.
   15 
   16 4.  For a non-anchored pattern, if (*SKIP) was given with a name that did not
   17     match a (*MARK), and the match failed at the start of the subject, a
   18     reference to memory before the start of the subject could occur. This bug
   19     was introduced by fix 17 of release 8.21.
   20 
   21 5.  A reference to an unset group with zero minimum repetition was giving
   22     totally wrong answers (in non-JavaScript-compatibility mode). For example,
   23     /(another)?(\1?)test/ matched against "hello world test". This bug was
   24     introduced in release 8.13.
   25 
   26 6.  Add support for 16-bit character strings (a large amount of work involving
   27     many changes and refactorings).
   28 
   29 7.  RunGrepTest failed on msys because \r\n was replaced by whitespace when the
   30     command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
   31     from a file.
   32 
   33 8.  Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
   34     rounding is not applied in this particular case).
   35 
   36 9.  The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
   37     if they appear, or are escaped, in patterns.
   38 
   39 10. Get rid of a number of -Wunused-but-set-variable warnings.
   40 
   41 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
   42     "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
   43     Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
   44     also returns the mark "x". This bug applied to capturing parentheses,
   45     non-capturing parentheses, and atomic parentheses. It also applied to some
   46     assertions.
   47 
   48 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
   49     information out of configure.ac instead of relying on pcre.h.generic, which
   50     is not stored in the repository.
   51 
   52 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
   53     -lreadline.
   54 
   55 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
   56 
   57 15. Applied Graycode's patch to put the top-level frame on the stack rather
   58     than the heap when not using the stack for recursion. This gives a
   59     performance improvement in many cases when recursion is not deep.
   60 
   61 16. Experimental code added to "pcretest -C" to output the stack frame size.
   62 
   63 
   64 Version 8.21 12-Dec-2011
   65 ------------------------
   66 
   67 1.  Updating the JIT compiler.
   68 
   69 2.  JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
   70     are added as well.
   71 
   72 3.  Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
   73     PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
   74     calling _pcre_jit_exec. Some extra comments are added.
   75 
   76 4.  (*MARK) settings inside atomic groups that do not contain any capturing
   77     parentheses, for example, (?>a(*:m)), were not being passed out. This bug
   78     was introduced by change 18 for 8.20.
   79 
   80 5.  Supporting of \x, \U and \u in JavaScript compatibility mode based on the
   81     ECMA-262 standard.
   82 
   83 6.  Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
   84     erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
   85     This bug was probably introduced by change 9 of 8.13.
   86 
   87 7.  While fixing 6 above, I noticed that a number of other items were being
   88     incorrectly rejected as "not fixed length". This arose partly because newer
   89     opcodes had not been added to the fixed-length checking code. I have (a)
   90     corrected the bug and added tests for these items, and (b) arranged for an
   91     error to occur if an unknown opcode is encountered while checking for fixed
   92     length instead of just assuming "not fixed length". The items that were
   93     rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
   94     (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
   95     repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
   96 
   97 8.  A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
   98     being incorrectly compiled and would have given unpredicatble results.
   99 
  100 9.  A possessively repeated subpattern with minimum repeat count greater than
  101     one behaved incorrectly. For example, (A){2,}+ behaved as if it was
  102     (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
  103     the first (A) could occur when it should not.
  104 
  105 10. Add a cast and remove a redundant test from the code.
  106 
  107 11. JIT should use pcre_malloc/pcre_free for allocation.
  108 
  109 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
  110     best practice nowadays, and helps with cross-compiling. (If the exec_prefix
  111     is anything other than /usr, -L is still shown).
  112 
  113 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
  114 
  115 14. Perl does not support \N without a following name in a [] class; PCRE now
  116     also gives an error.
  117 
  118 15. If a forward reference was repeated with an upper limit of around 2000,
  119     it caused the error "internal error: overran compiling workspace". The
  120     maximum number of forward references (including repeats) was limited by the
  121     internal workspace, and dependent on the LINK_SIZE. The code has been
  122     rewritten so that the workspace expands (via pcre_malloc) if necessary, and
  123     the default depends on LINK_SIZE. There is a new upper limit (for safety)
  124     of around 200,000 forward references. While doing this, I also speeded up
  125     the filling in of repeated forward references.
  126 
  127 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
  128     incorrectly expecting the subject to contain another "a" after the start.
  129 
  130 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
  131     in the match, the SKIP should be ignored. This was not happening; instead
  132     the SKIP was being treated as NOMATCH. For patterns such as
  133     /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
  134     tested.
  135 
  136 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
  137     now much more compatible with Perl, in particular in cases where the result
  138     is a non-match for a non-anchored pattern. For example, if
  139     /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
  140     "m", where previously it did not return a name. A side effect of this
  141     change is that for partial matches, the last encountered mark name is
  142     returned, as for non matches. A number of tests that were previously not
  143     Perl-compatible have been moved into the Perl-compatible test files. The
  144     refactoring has had the pleasing side effect of removing one argument from
  145     the match() function, thus reducing its stack requirements.
  146 
  147 19. If the /S+ option was used in pcretest to study a pattern using JIT,
  148     subsequent uses of /S (without +) incorrectly behaved like /S+.
  149 
  150 21. Retrieve executable code size support for the JIT compiler and fixing
  151     some warnings.
  152 
  153 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
  154     not work when the shorter character appeared right at the end of the
  155     subject string.
  156 
  157 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
  158     systems.
  159 
  160 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
  161     output it when the /M option is used in pcretest.
  162 
  163 25. The CheckMan script was not being included in the distribution. Also, added
  164     an explicit "perl" to run Perl scripts from the PrepareRelease script
  165     because this is reportedly needed in Windows.
  166 
  167 26. If study data was being save in a file and studying had not found a set of
  168     "starts with" bytes for the pattern, the data written to the file (though
  169     never used) was taken from uninitialized memory and so caused valgrind to
  170     complain.
  171 
  172 27. Updated RunTest.bat as provided by Sheri Pierce.
  173 
  174 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
  175 
  176 29. Computation of memory usage for the table of capturing group names was
  177     giving an unnecessarily large value.
  178 
  179 
  180 Version 8.20 21-Oct-2011
  181 ------------------------
  182 
  183 1.  Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
  184     a POSIX class. After further experiments with Perl, which convinced me that
  185     Perl has bugs and confusions, a closing square bracket is no longer allowed
  186     in a POSIX name. This bug also affected patterns with classes that started
  187     with full stops.
  188 
  189 2.  If a pattern such as /(a)b|ac/ is matched against "ac", there is no
  190     captured substring, but while checking the failing first alternative,
  191     substring 1 is temporarily captured. If the output vector supplied to
  192     pcre_exec() was not big enough for this capture, the yield of the function
  193     was still zero ("insufficient space for captured substrings"). This cannot
  194     be totally fixed without adding another stack variable, which seems a lot
  195     of expense for a edge case. However, I have improved the situation in cases
  196     such as /(a)(b)x|abc/ matched against "abc", where the return code
  197     indicates that fewer than the maximum number of slots in the ovector have
  198     been set.
  199 
  200 3.  Related to (2) above: when there are more back references in a pattern than
  201     slots in the output vector, pcre_exec() uses temporary memory during
  202     matching, and copies in the captures as far as possible afterwards. It was
  203     using the entire output vector, but this conflicts with the specification
  204     that only 2/3 is used for passing back captured substrings. Now it uses
  205     only the first 2/3, for compatibility. This is, of course, another edge
  206     case.
  207 
  208 4.  Zoltan Herczeg's just-in-time compiler support has been integrated into the
  209     main code base, and can be used by building with --enable-jit. When this is
  210     done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
  211     runtime --no-jit option is given.
  212 
  213 5.  When the number of matches in a pcre_dfa_exec() run exactly filled the
  214     ovector, the return from the function was zero, implying that there were
  215     other matches that did not fit. The correct "exactly full" value is now
  216     returned.
  217 
  218 6.  If a subpattern that was called recursively or as a subroutine contained
  219     (*PRUNE) or any other control that caused it to give a non-standard return,
  220     invalid errors such as "Error -26 (nested recursion at the same subject
  221     position)" or even infinite loops could occur.
  222 
  223 7.  If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
  224     computing the minimum length on reaching *ACCEPT, and so ended up with the
  225     wrong value of 1 rather than 0. Further investigation indicates that
  226     computing a minimum subject length in the presence of *ACCEPT is difficult
  227     (think back references, subroutine calls), and so I have changed the code
  228     so that no minimum is registered for a pattern that contains *ACCEPT.
  229 
  230 8.  If (*THEN) was present in the first (true) branch of a conditional group,
  231     it was not handled as intended. [But see 16 below.]
  232 
  233 9.  Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
  234     Sheri Pierce.
  235 
  236 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
  237     the first byte in a match must be "a".
  238 
  239 11. Change 17 for 8.13 increased the recursion depth for patterns like
  240     /a(?:.)*?a/ drastically. I've improved things by remembering whether a
  241     pattern contains any instances of (*THEN). If it does not, the old
  242     optimizations are restored. It would be nice to do this on a per-group
  243     basis, but at the moment that is not feasible.
  244 
  245 12. In some environments, the output of pcretest -C is CRLF terminated. This
  246     broke RunTest's code that checks for the link size. A single white space
  247     character after the value is now allowed for.
  248 
  249 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
  250     For "fr", it uses the Windows-specific input and output files.
  251 
  252 14. If (*THEN) appeared in a group that was called recursively or as a
  253     subroutine, it did not work as intended. [But see next item.]
  254 
  255 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
  256     pattern fragments (but not containing any | characters). If A and B are
  257     matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
  258     was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
  259     D. In other words, Perl considers parentheses that do not contain any |
  260     characters to be part of a surrounding alternative, whereas PCRE was
  261     treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
  262     differently. PCRE now behaves in the same way as Perl, except in the case
  263     of subroutine/recursion calls such as (?1) which have in any case always
  264     been different (but PCRE had them first :-).
  265 
  266 16. Related to 15 above: Perl does not treat the | in a conditional group as
  267     creating alternatives. Such a group is treated in the same way as an
  268     ordinary group without any | characters when processing (*THEN). PCRE has
  269     been changed to match Perl's behaviour.
  270 
  271 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
  272     RunGrepTest script failed.
  273 
  274 18. Change 22 for version 13 caused atomic groups to use more stack. This is
  275     inevitable for groups that contain captures, but it can lead to a lot of
  276     stack use in large patterns. The old behaviour has been restored for atomic
  277     groups that do not contain any capturing parentheses.
  278 
  279 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
  280     suppress the check for a minimum subject length at run time. (If it was
  281     given to pcre_exec() or pcre_dfa_exec() it did work.)
  282 
  283 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
  284     fail to work when decoding hex characters in data strings in EBCDIC
  285     environments.
  286 
  287 21. It appears that in at least one Mac OS environment, the isxdigit() function
  288     is implemented as a macro that evaluates to its argument more than once,
  289     contravening the C 90 Standard (I haven't checked a later standard). There
  290     was an instance in pcretest which caused it to go wrong when processing
  291     \x{...} escapes in subject strings. The has been rewritten to avoid using
  292     things like p++ in the argument of isxdigit().
  293 
  294 
  295 Version 8.13 16-Aug-2011
  296 ------------------------
  297 
  298 1.  The Unicode data tables have been updated to Unicode 6.0.0.
  299 
  300 2.  Two minor typos in pcre_internal.h have been fixed.
  301 
  302 3.  Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
  303     pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
  304     in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
  305 
  306 4.  There were a number of related bugs in the code for matching backrefences
  307     caselessly in UTF-8 mode when codes for the characters concerned were
  308     different numbers of bytes. For example, U+023A and U+2C65 are an upper
  309     and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
  310     (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
  311     code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
  312     2-byte code at the end of the subject (it thought there wasn't enough data
  313     left).
  314 
  315 5.  Comprehensive information about what went wrong is now returned by
  316     pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
  317     as the output vector has at least 2 elements. The offset of the start of
  318     the failing character and a reason code are placed in the vector.
  319 
  320 6.  When the UTF-8 string check fails for pcre_compile(), the offset that is
  321     now returned is for the first byte of the failing character, instead of the
  322     last byte inspected. This is an incompatible change, but I hope it is small
  323     enough not to be a problem. It makes the returned offset consistent with
  324     pcre_exec() and pcre_dfa_exec().
  325 
  326 7.  pcretest now gives a text phrase as well as the error number when
  327     pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
  328     failure, the offset and reason code are output.
  329 
  330 8.  When \R was used with a maximizing quantifier it failed to skip backwards
  331     over a \r\n pair if the subsequent match failed. Instead, it just skipped
  332     back over a single character (\n). This seems wrong (because it treated the
  333     two characters as a single entity when going forwards), conflicts with the
  334     documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
  335     behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
  336     has been changed.
  337 
  338 9.  Some internal refactoring has changed the processing so that the handling
  339     of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
  340     time (the PCRE_DOTALL option was changed this way some time ago: version
  341     7.7 change 16). This has made it possible to abolish the OP_OPT op code,
  342     which was always a bit of a fudge. It also means that there is one less
  343     argument for the match() function, which reduces its stack requirements
  344     slightly. This change also fixes an incompatibility with Perl: the pattern
  345     (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
  346 
  347 10. More internal refactoring has drastically reduced the number of recursive
  348     calls to match() for possessively repeated groups such as (abc)++ when
  349     using pcre_exec().
  350 
  351 11. While implementing 10, a number of bugs in the handling of groups were
  352     discovered and fixed:
  353 
  354     (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
  355     (a|)*(?1) gave a compile-time internal error.
  356     ((a|)+)+  did not notice that the outer group could match an empty string.
  357     (^a|^)+   was not marked as anchored.
  358     (.*a|.*)+ was not marked as matching at start or after a newline.
  359 
  360 12. Yet more internal refactoring has removed another argument from the match()
  361     function. Special calls to this function are now indicated by setting a
  362     value in a variable in the "match data" data block.
  363 
  364 13. Be more explicit in pcre_study() instead of relying on "default" for
  365     opcodes that mean there is no starting character; this means that when new
  366     ones are added and accidentally left out of pcre_study(), testing should
  367     pick them up.
  368 
  369 14. The -s option of pcretest has been documented for ages as being an old
  370     synonym of -m (show memory usage). I have changed it to mean "force study
  371     for every regex", that is, assume /S for every regex. This is similar to -i
  372     and -d etc. It's slightly incompatible, but I'm hoping nobody is still
  373     using it. It makes it easier to run collections of tests with and without
  374     study enabled, and thereby test pcre_study() more easily. All the standard
  375     tests are now run with and without -s (but some patterns can be marked as
  376     "never study" - see 20 below).
  377 
  378 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
  379     restoration of the capturing data to the outer values was not happening
  380     correctly.
  381 
  382 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
  383     empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
  384     pattern had matched an empty string, and so incorrectly returned a no
  385     match.
  386 
  387 17. There was optimizing code for the last branch of non-capturing parentheses,
  388     and also for the obeyed branch of a conditional subexpression, which used
  389     tail recursion to cut down on stack usage. Unfortunately, now that there is
  390     the possibility of (*THEN) occurring in these branches, tail recursion is
  391     no longer possible because the return has to be checked for (*THEN). These
  392     two optimizations have therefore been removed. [But see 8.20/11 above.]
  393 
  394 18. If a pattern containing \R was studied, it was assumed that \R always
  395     matched two bytes, thus causing the minimum subject length to be
  396     incorrectly computed because \R can also match just one byte.
  397 
  398 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
  399     was incorrectly computed.
  400 
  401 20. If /S is present twice on a test pattern in pcretest input, it now
  402     *disables* studying, thereby overriding the use of -s on the command line
  403     (see 14 above). This is necessary for one or two tests to keep the output
  404     identical in both cases.
  405 
  406 21. When (*ACCEPT) was used in an assertion that matched an empty string and
  407     PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
  408 
  409 22. When an atomic group that contained a capturing parenthesis was
  410     successfully matched, but the branch in which it appeared failed, the
  411     capturing was not being forgotten if a higher numbered group was later
  412     captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
  413     group 1 to "a", when in fact it should be unset. This applied to multi-
  414     branched capturing and non-capturing groups, repeated or not, and also to
  415     positive assertions (capturing in negative assertions does not happen
  416     in PCRE) and also to nested atomic groups.
  417 
  418 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
  419     subject after a captured substring, to make it easier to tell which of a
  420     number of identical substrings has been captured.
  421 
  422 24. The way atomic groups are processed by pcre_exec() has been changed so that
  423     if they are repeated, backtracking one repetition now resets captured
  424     values correctly. For example, if ((?>(a+)b)+aabab) is matched against
  425     "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
  426     "aaa". Previously, it would have been "a". As part of this code
  427     refactoring, the way recursive calls are handled has also been changed.
  428 
  429 25. If an assertion condition captured any substrings, they were not passed
  430     back unless some other capturing happened later. For example, if
  431     (?(?=(a))a) was matched against "a", no capturing was returned.
  432 
  433 26. When studying a pattern that contained subroutine calls or assertions,
  434     the code for finding the minimum length of a possible match was handling
  435     direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
  436     group 1 called group 2 while simultaneously a separate group 2 called group
  437     1). A stack overflow occurred in this case. I have fixed this by limiting
  438     the recursion depth to 10.
  439 
  440 27. Updated RunTest.bat in the distribution to the version supplied by Tom
  441     Fortmann. This supports explicit test numbers on the command line, and has
  442     argument validation and error reporting.
  443 
  444 28. An instance of \X with an unlimited repeat could fail if at any point the
  445     first character it looked at was a mark character.
  446 
  447 29. Some minor code refactoring concerning Unicode properties and scripts
  448     should reduce the stack requirement of match() slightly.
  449 
  450 30. Added the '=' option to pcretest to check the setting of unused capturing
  451     slots at the end of the pattern, which are documented as being -1, but are
  452     not included in the return count.
  453 
  454 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
  455     compiled something random. Now it gives a compile-time error (as does
  456     Perl).
  457 
  458 32. A *MARK encountered during the processing of a positive assertion is now
  459     recorded and passed back (compatible with Perl).
  460 
  461 33. If --only-matching or --colour was set on a pcregrep call whose pattern
  462     had alternative anchored branches, the search for a second match in a line
  463     was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
  464     matched the line "0102" twice. The same bug affected patterns that started
  465     with a backwards assertion. For example /\b01|\b02/ also matched "0102"
  466     twice.
  467 
  468 34. Previously, PCRE did not allow quantification of assertions. However, Perl
  469     does, and because of capturing effects, quantifying parenthesized
  470     assertions may at times be useful. Quantifiers are now allowed for
  471     parenthesized assertions.
  472 
  473 35. A minor code tidy in pcre_compile() when checking options for \R usage.
  474 
  475 36. \g was being checked for fancy things in a character class, when it should
  476     just be a literal "g".
  477 
  478 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
  479     appearance of a nested POSIX class supersedes an apparent external class.
  480     For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
  481     unescaped square brackets may also appear as part of class names. For
  482     example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
  483     more like Perl. (But see 8.20/1 above.)
  484 
  485 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
  486     was because it thought it was \N{name}, which is not supported).
  487 
  488 39. Add minix to OS list not supporting the -S option in pcretest.
  489 
  490 40. PCRE tries to detect cases of infinite recursion at compile time, but it
  491     cannot analyze patterns in sufficient detail to catch mutual recursions
  492     such as ((?1))((?2)). There is now a runtime test that gives an error if a
  493     subgroup is called recursively as a subpattern for a second time at the
  494     same position in the subject string. In previous releases this might have
  495     been caught by the recursion limit, or it might have run out of stack.
  496 
  497 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
  498     happen only once. PCRE was, however incorrectly giving a compile time error
  499     "recursive call could loop indefinitely" because it cannot analyze the
  500     pattern in sufficient detail. The compile time test no longer happens when
  501     PCRE is compiling a conditional subpattern, but actual runaway loops are
  502     now caught at runtime (see 40 above).
  503 
  504 42. It seems that Perl allows any characters other than a closing parenthesis
  505     to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
  506     has been changed to be the same.
  507 
  508 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
  509     as not to get warnings when autogen.sh is called. Also changed
  510     AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
  511 
  512 44. To help people who use pcregrep to scan files containing exceedingly long
  513     lines, the following changes have been made:
  514 
  515     (a) The default value of the buffer size parameter has been increased from
  516         8K to 20K. (The actual buffer used is three times this size.)
  517 
  518     (b) The default can be changed by ./configure --with-pcregrep-bufsize when
  519         PCRE is built.
  520 
  521     (c) A --buffer-size=n option has been added to pcregrep, to allow the size
  522         to be set at run time.
  523 
  524     (d) Numerical values in pcregrep options can be followed by K or M, for
  525         example --buffer-size=50K.
  526 
  527     (e) If a line being scanned overflows pcregrep's buffer, an error is now
  528         given and the return code is set to 2.
  529 
  530 45. Add a pointer to the latest mark to the callout data block.
  531 
  532 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
  533     partial match of an empty string instead of no match. This was specific to
  534     the use of ".".
  535 
  536 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
  537     complete match instead of a partial match. This bug was dependent on both
  538     the PCRE_UTF8 and PCRE_DOTALL options being set.
  539 
  540 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
  541     starting byte set, because \b was not being ignored.
  542 
  543 
  544 Version 8.12 15-Jan-2011
  545 ------------------------
  546 
  547 1.  Fixed some typos in the markup of the man pages, and wrote a script that
  548     checks for such things as part of the documentation building process.
  549 
  550 2.  On a big-endian 64-bit system, pcregrep did not correctly process the
  551     --match-limit and --recursion-limit options (added for 8.11). In
  552     particular, this made one of the standard tests fail. (The integer value
  553     went into the wrong half of a long int.)
  554 
  555 3.  If the --colour option was given to pcregrep with -v (invert match), it
  556     did strange things, either producing crazy output, or crashing. It should,
  557     of course, ignore a request for colour when reporting lines that do not
  558     match.
  559 
  560 4.  Another pcregrep bug caused similar problems if --colour was specified with
  561     -M (multiline) and the pattern match finished with a line ending.
  562 
  563 5.  In pcregrep, when a pattern that ended with a literal newline sequence was
  564     matched in multiline mode, the following line was shown as part of the
  565     match. This seems wrong, so I have changed it.
  566 
  567 6.  Another pcregrep bug in multiline mode, when --colour was specified, caused
  568     the check for further matches in the same line (so they could be coloured)
  569     to overrun the end of the current line. If another match was found, it was
  570     incorrectly shown (and then shown again when found in the next line).
  571 
  572 7.  If pcregrep was compiled under Windows, there was a reference to the
  573     function pcregrep_exit() before it was defined. I am assuming this was
  574     the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
  575     reported by a user. I've moved the definition above the reference.
  576 
  577 
  578 Version 8.11 10-Dec-2010
  579 ------------------------
  580 
  581 1.  (*THEN) was not working properly if there were untried alternatives prior
  582     to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
  583     backtracked to try for "b" instead of moving to the next alternative branch
  584     at the same level (in this case, to look for "c"). The Perl documentation
  585     is clear that when (*THEN) is backtracked onto, it goes to the "next
  586     alternative in the innermost enclosing group".
  587 
  588 2.  (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
  589     such as   (A(*COMMIT)B(*THEN)C|D)  any failure after matching A should
  590     result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
  591     (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
  592     (*THEN).
  593 
  594 3.  If \s appeared in a character class, it removed the VT character from
  595     the class, even if it had been included by some previous item, for example
  596     in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
  597     of \s, but is part of the POSIX "space" class.)
  598 
  599 4.  A partial match never returns an empty string (because you can always
  600     match an empty string at the end of the subject); however the checking for
  601     an empty string was starting at the "start of match" point. This has been
  602     changed to the "earliest inspected character" point, because the returned
  603     data for a partial match starts at this character. This means that, for
  604     example, /(?<=abc)def/ gives a partial match for the subject "abc"
  605     (previously it gave "no match").
  606 
  607 5.  Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
  608     of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
  609     previously a full match would be given. However, setting PCRE_PARTIAL_HARD
  610     has an implication that the given string is incomplete (because a partial
  611     match is preferred over a full match). For this reason, these items now
  612     give a partial match in this situation. [Aside: previously, the one case
  613     /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
  614     match rather than a full match, which was wrong by the old rules, but is
  615     now correct.]
  616 
  617 6.  There was a bug in the handling of #-introduced comments, recognized when
  618     PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
  619     If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
  620     UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
  621     scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
  622     but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
  623     places in pcre_compile().
  624 
  625 7.  Related to (6) above, when pcre_compile() was skipping #-introduced
  626     comments when looking ahead for named forward references to subpatterns,
  627     the only newline sequence it recognized was NL. It now handles newlines
  628     according to the set newline convention.
  629 
  630 8.  SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
  631     former, but used strtoul(), whereas pcretest avoided strtoul() but did not
  632     cater for a lack of strerror(). These oversights have been fixed.
  633 
  634 9.  Added --match-limit and --recursion-limit to pcregrep.
  635 
  636 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
  637 
  638 11. When the -o option was used, pcregrep was setting a return code of 1, even
  639     when matches were found, and --line-buffered was not being honoured.
  640 
  641 12. Added an optional parentheses number to the -o and --only-matching options
  642     of pcregrep.
  643 
  644 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
  645     can match an empty string. The code to do it in pcretest and pcredemo
  646     needed fixing:
  647 
  648     (a) When the newline convention was "crlf", pcretest got it wrong, skipping
  649         only one byte after an empty string match just before CRLF (this case
  650         just got forgotten; "any" and "anycrlf" were OK).
  651 
  652     (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
  653         mode when an empty string match preceded an ASCII character followed by
  654         a non-ASCII character. (The code for advancing by one character rather
  655         than one byte was nonsense.)
  656 
  657     (c) The pcredemo.c sample program did not have any code at all to handle
  658         the cases when CRLF is a valid newline sequence.
  659 
  660 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
  661     as a starting offset was within the subject string. There is now a new
  662     error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
  663     negative or greater than the length of the string. In order to test this,
  664     pcretest is extended to allow the setting of negative starting offsets.
  665 
  666 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
  667     starting offset points to the beginning of a UTF-8 character was
  668     unnecessarily clumsy. I tidied it up.
  669 
  670 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
  671     bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
  672 
  673 17. Nobody had reported that the --include_dir option, which was added in
  674     release 7.7 should have been called --include-dir (hyphen, not underscore)
  675     for compatibility with GNU grep. I have changed it to --include-dir, but
  676     left --include_dir as an undocumented synonym, and the same for
  677     --exclude-dir, though that is not available in GNU grep, at least as of
  678     release 2.5.4.
  679 
  680 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
  681     characters from a string of bytes) have been redefined so as not to use
  682     loops, in order to improve performance in some environments. At the same
  683     time, I abstracted some of the common code into auxiliary macros to save
  684     repetition (this should not affect the compiled code).
  685 
  686 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
  687     compile-time error is now given if \c is not followed by an ASCII
  688     character, that is, a byte less than 128. (In EBCDIC mode, the code is
  689     different, and any byte value is allowed.)
  690 
  691 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
  692     START_OPTIMIZE option, which is now allowed at compile time - but just
  693     passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
  694     to pcregrep and other applications that have no direct access to PCRE
  695     options. The new /Y option in pcretest sets this option when calling
  696     pcre_compile().
  697 
  698 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
  699     back references. Groups containing recursive back references were forced to
  700     be atomic by that change, but in the case of named groups, the amount of
  701     memory required was incorrectly computed, leading to "Failed: internal
  702     error: code overflow". This has been fixed.
  703 
  704 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
  705     pcretest.c, to avoid build problems in some Borland environments.
  706 
  707 
  708 Version 8.10 25-Jun-2010
  709 ------------------------
  710 
  711 1.  Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
  712     THEN.
  713 
  714 2.  (*ACCEPT) was not working when inside an atomic group.
  715 
  716 3.  Inside a character class, \B is treated as a literal by default, but
  717     faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
  718     causes the error). The code is unchanged, but I tidied the documentation.
  719 
  720 4.  Inside a character class, PCRE always treated \R and \X as literals,
  721     whereas Perl faults them if its -w option is set. I have changed PCRE so
  722     that it faults them when PCRE_EXTRA is set.
  723 
  724 5.  Added support for \N, which always matches any character other than
  725     newline. (It is the same as "." when PCRE_DOTALL is not set.)
  726 
  727 6.  When compiling pcregrep with newer versions of gcc which may have
  728     FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
  729     declared with attribute warn_unused_result" were given. Just casting the
  730     result to (void) does not stop the warnings; a more elaborate fudge is
  731     needed. I've used a macro to implement this.
  732 
  733 7.  Minor change to pcretest.c to avoid a compiler warning.
  734 
  735 8.  Added four artifical Unicode properties to help with an option to make
  736     \s etc use properties (see next item). The new properties are: Xan
  737     (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
  738 
  739 9.  Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
  740     use Unicode properties. (*UCP) at the start of a pattern can be used to set
  741     this option. Modified pcretest to add /W to test this facility. Added
  742     REG_UCP to make it available via the POSIX interface.
  743 
  744 10. Added --line-buffered to pcregrep.
  745 
  746 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
  747     studied, and the match started with a letter with a code point greater than
  748     127 whose first byte was different to the first byte of the other case of
  749     the letter, the other case of this starting letter was not recognized
  750     (#976).
  751 
  752 12. If a pattern that was studied started with a repeated Unicode property
  753     test, for example, \p{Nd}+, there was the theoretical possibility of
  754     setting up an incorrect bitmap of starting bytes, but fortunately it could
  755     not have actually happened in practice until change 8 above was made (it
  756     added property types that matched character-matching opcodes).
  757 
  758 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
  759     possible starting bytes for non-anchored patterns.
  760 
  761 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
  762     \R, and also a number of cases that involve Unicode properties, both
  763     explicit and implicit when PCRE_UCP is set.
  764 
  765 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
  766     input, it could crash or give wrong results if characters with values
  767     greater than 0xc0 were present in the subject string. (Detail: it assumed
  768     UTF-8 input when processing these items.)
  769 
  770 16. Added a lot of (int) casts to avoid compiler warnings in systems where
  771     size_t is 64-bit (#991).
  772 
  773 17. Added a check for running out of memory when PCRE is compiled with
  774     --disable-stack-for-recursion (#990).
  775 
  776 18. If the last data line in a file for pcretest does not have a newline on
  777     the end, a newline was missing in the output.
  778 
  779 19. The default pcre_chartables.c file recognizes only ASCII characters (values
  780     less than 128) in its various bitmaps. However, there is a facility for
  781     generating tables according to the current locale when PCRE is compiled. It
  782     turns out that in some environments, 0x85 and 0xa0, which are Unicode space
  783     characters, are recognized by isspace() and therefore were getting set in
  784     these tables, and indeed these tables seem to approximate to ISO 8859. This
  785     caused a problem in UTF-8 mode when pcre_study() was used to create a list
  786     of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
  787     which of course cannot start UTF-8 characters. I have changed the code so
  788     that only real ASCII characters (less than 128) and the correct starting
  789     bytes for UTF-8 encodings are set for characters greater than 127 when in
  790     UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
  791     altogether.)
  792 
  793 20. Added the /T option to pcretest so as to be able to run tests with non-
  794     standard character tables, thus making it possible to include the tests
  795     used for 19 above in the standard set of tests.
  796 
  797 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
  798     reference to a subpattern the other side of a comment that contains an
  799     opening parenthesis caused either an internal compiling error, or a
  800     reference to the wrong subpattern.
  801 
  802 
  803 Version 8.02 19-Mar-2010
  804 ------------------------
  805 
  806 1.  The Unicode data tables have been updated to Unicode 5.2.0.
  807 
  808 2.  Added the option --libs-cpp to pcre-config, but only when C++ support is
  809     configured.
  810 
  811 3.  Updated the licensing terms in the pcregexp.pas file, as agreed with the
  812     original author of that file, following a query about its status.
  813 
  814 4.  On systems that do not have stdint.h (e.g. Solaris), check for and include
  815     inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
  816 
  817 5.  A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
  818     quantifier applied to a forward-referencing subroutine call, could compile
  819     incorrect code or give the error "internal error: previously-checked
  820     referenced subpattern not found".
  821 
  822 6.  Both MS Visual Studio and Symbian OS have problems with initializing
  823     variables to point to external functions. For these systems, therefore,
  824     pcre_malloc etc. are now initialized to local functions that call the
  825     relevant global functions.
  826 
  827 7.  There were two entries missing in the vectors called coptable and poptable
  828     in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
  829     I've fixed the data, and added a kludgy way of testing at compile time that
  830     the lengths are correct (equal to the number of opcodes).
  831 
  832 8.  Following on from 7, I added a similar kludge to check the length of the
  833     eint vector in pcreposix.c.
  834 
  835 9.  Error texts for pcre_compile() are held as one long string to avoid too
  836     much relocation at load time. To find a text, the string is searched,
  837     counting zeros. There was no check for running off the end of the string,
  838     which could happen if a new error number was added without updating the
  839     string.
  840 
  841 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
  842 
  843 11. \K was not working if it appeared in an atomic group or in a group that
  844     was called as a "subroutine", or in an assertion. Perl 5.11 documents that
  845     \K is "not well defined" if used in an assertion. PCRE now accepts it if
  846     the assertion is positive, but not if it is negative.
  847 
  848 12. Change 11 fortuitously reduced the size of the stack frame used in the
  849     "match()" function of pcre_exec.c by one pointer. Forthcoming
  850     implementation of support for (*MARK) will need an extra pointer on the
  851     stack; I have reserved it now, so that the stack frame size does not
  852     decrease.
  853 
  854 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
  855     item in branch that calls a recursion is a subroutine call - as in the
  856     second branch in the above example - was incorrectly given the compile-
  857     time error "recursive call could loop indefinitely" because pcre_compile()
  858     was not correctly checking the subroutine for matching a non-empty string.
  859 
  860 14. The checks for overrunning compiling workspace could trigger after an
  861     overrun had occurred. This is a "should never occur" error, but it can be
  862     triggered by pathological patterns such as hundreds of nested parentheses.
  863     The checks now trigger 100 bytes before the end of the workspace.
  864 
  865 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
  866 
  867 
  868 Version 8.01 19-Jan-2010
  869 ------------------------
  870 
  871 1.  If a pattern contained a conditional subpattern with only one branch (in
  872     particular, this includes all (*DEFINE) patterns), a call to pcre_study()
  873     computed the wrong minimum data length (which is of course zero for such
  874     subpatterns). This could cause incorrect "no match" results.
  875 
  876 2.  For patterns such as (?i)a(?-i)b|c where an option setting at the start of
  877     the pattern is reset in the first branch, pcre_compile() failed with
  878     "internal error: code overflow at offset...". This happened only when
  879     the reset was to the original external option setting. (An optimization
  880     abstracts leading options settings into an external setting, which was the
  881     cause of this.)
  882 
  883 3.  A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
  884     of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
  885     assertion pattern did not match (meaning that the assertion was true), it
  886     was incorrectly treated as false if the SKIP had been reached during the
  887     matching. This also applied to assertions used as conditions.
  888 
  889 4.  If an item that is not supported by pcre_dfa_exec() was encountered in an
  890     assertion subpattern, including such a pattern used as a condition,
  891     unpredictable results occurred, instead of the error return
  892     PCRE_ERROR_DFA_UITEM.
  893 
  894 5.  The C++ GlobalReplace function was not working like Perl for the special
  895     situation when an empty string is matched. It now does the fancy magic
  896     stuff that is necessary.
  897 
  898 6.  In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
  899     removed. (These were left over from very, very early versions of PCRE.)
  900 
  901 7.  Some cosmetic changes to the code to make life easier when compiling it
  902     as part of something else:
  903 
  904     (a) Change DEBUG to PCRE_DEBUG.
  905 
  906     (b) In pcre_compile(), rename the member of the "branch_chain" structure
  907         called "current" as "current_branch", to prevent a collision with the
  908         Linux macro when compiled as a kernel module.
  909 
  910     (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
  911         prevent a collision with the Linux macro when compiled as a kernel
  912         module.
  913 
  914 8.  In pcre_compile() there are some checks for integer overflows that used to
  915     cast potentially large values to (double). This has been changed to that
  916     when building, a check for int64_t is made, and if it is found, it is used
  917     instead, thus avoiding the use of floating point arithmetic. (There is no
  918     other use of FP in PCRE.) If int64_t is not found, the fallback is to
  919     double.
  920 
  921 9.  Added two casts to avoid signed/unsigned warnings from VS Studio Express
  922     2005 (difference between two addresses compared to an unsigned value).
  923 
  924 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
  925     custom one, because of the following reported problem in Windows:
  926 
  927       - libbz2 uses the Pascal calling convention (WINAPI) for the functions
  928           under Win32.
  929       - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
  930           therefore missing the function definition.
  931       - The compiler thus generates a "C" signature for the test function.
  932       - The linker fails to find the "C" function.
  933       - PCRE fails to configure if asked to do so against libbz2.
  934 
  935 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
  936     messages were output:
  937 
  938       Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
  939       rerunning libtoolize, to keep the correct libtool macros in-tree.
  940       Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
  941 
  942     I have done both of these things.
  943 
  944 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
  945     most of the time, it *can* run out if it is given a pattern that contains a
  946     runaway infinite recursion. I updated the discussion in the pcrestack man
  947     page.
  948 
  949 13. Now that we have gone to the x.xx style of version numbers, the minor
  950     version may start with zero. Using 08 or 09 is a bad idea because users
  951     might check the value of PCRE_MINOR in their code, and 08 or 09 may be
  952     interpreted as invalid octal numbers. I've updated the previous comment in
  953     configure.ac, and also added a check that gives an error if 08 or 09 are
  954     used.
  955 
  956 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
  957     causing partial matching to fail when the end of the subject matched \W
  958     in a UTF-8 pattern where \W was quantified with a minimum of 3.
  959 
  960 15. There were some discrepancies between the declarations in pcre_internal.h
  961     of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
  962     their definitions. The declarations used "const uschar *" and the
  963     definitions used USPTR. Even though USPTR is normally defined as "const
  964     unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
  965     reported that: "This difference in casting confuses some C++ compilers, for
  966     example, SunCC recognizes above declarations as different functions and
  967     generates broken code for hbpcre." I have changed the declarations to use
  968     USPTR.
  969 
  970 16. GNU libtool is named differently on some systems. The autogen.sh script now
  971     tries several variants such as glibtoolize (MacOSX) and libtoolize1x
  972     (FreeBSD).
  973 
  974 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
  975     (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
  976     comment: "Figure out how to create a longlong from a string: strtoll and
  977     equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
  978     instance, but it only takes 2 args instead of 3!"
  979 
  980 18. A subtle bug concerned with back references has been fixed by a change of
  981     specification, with a corresponding code fix. A pattern such as
  982     ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
  983     refers, was giving matches when it shouldn't. For example, xa=xaaa would
  984     match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
  985     same bug. Such groups have to be quantified to be useful, or contained
  986     inside another quantified group. (If there's no repetition, the reference
  987     can never match.) The problem arises because, having left the group and
  988     moved on to the rest of the pattern, a later failure that backtracks into
  989     the group uses the captured value from the final iteration of the group
  990     rather than the correct earlier one. I have fixed this in PCRE by forcing
  991     any group that contains a reference to itself to be an atomic group; that
  992     is, there cannot be any backtracking into it once it has completed. This is
  993     similar to recursive and subroutine calls.
  994 
  995 
  996 Version 8.00 19-Oct-09
  997 ----------------------
  998 
  999 1.  The table for translating pcre_compile() error codes into POSIX error codes
 1000     was out-of-date, and there was no check on the pcre_compile() error code
 1001     being within the table. This could lead to an OK return being given in
 1002     error.
 1003 
 1004 2.  Changed the call to open a subject file in pcregrep from fopen(pathname,
 1005     "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
 1006     in a Windows environment.
 1007 
 1008 3.  The pcregrep --count option prints the count for each file even when it is
 1009     zero, as does GNU grep. However, pcregrep was also printing all files when
 1010     --files-with-matches was added. Now, when both options are given, it prints
 1011     counts only for those files that have at least one match. (GNU grep just
 1012     prints the file name in this circumstance, but including the count seems
 1013     more useful - otherwise, why use --count?) Also ensured that the
 1014     combination -clh just lists non-zero counts, with no names.
 1015 
 1016 4.  The long form of the pcregrep -F option was incorrectly implemented as
 1017     --fixed_strings instead of --fixed-strings. This is an incompatible change,
 1018     but it seems right to fix it, and I didn't think it was worth preserving
 1019     the old behaviour.
 1020 
 1021 5.  The command line items --regex=pattern and --regexp=pattern were not
 1022     recognized by pcregrep, which required --regex pattern or --regexp pattern
 1023     (with a space rather than an '='). The man page documented the '=' forms,
 1024     which are compatible with GNU grep; these now work.
 1025 
 1026 6.  No libpcreposix.pc file was created for pkg-config; there was just
 1027     libpcre.pc and libpcrecpp.pc. The omission has been rectified.
 1028 
 1029 7.  Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
 1030     when UCP support is not needed, by modifying the Python script that
 1031     generates it from Unicode data files. This should not matter if the module
 1032     is correctly used as a library, but I received one complaint about 50K of
 1033     unwanted data. My guess is that the person linked everything into his
 1034     program rather than using a library. Anyway, it does no harm.
 1035 
 1036 8.  A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
 1037     was a minimum greater than 1 for a wide character in a possessive
 1038     repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
 1039     which had an unlimited repeat of a nested, fixed maximum repeat of a wide
 1040     character. Chaos in the form of incorrect output or a compiling loop could
 1041     result.
 1042 
 1043 9.  The restrictions on what a pattern can contain when partial matching is
 1044     requested for pcre_exec() have been removed. All patterns can now be
 1045     partially matched by this function. In addition, if there are at least two
 1046     slots in the offset vector, the offset of the earliest inspected character
 1047     for the match and the offset of the end of the subject are set in them when
 1048     PCRE_ERROR_PARTIAL is returned.
 1049 
 1050 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
 1051     synonymous with PCRE_PARTIAL, for backwards compatibility, and
 1052     PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
 1053     and may be more useful for multi-segment matching.
 1054 
 1055 11. Partial matching with pcre_exec() is now more intuitive. A partial match
 1056     used to be given if ever the end of the subject was reached; now it is
 1057     given only if matching could not proceed because another character was
 1058     needed. This makes a difference in some odd cases such as Z(*FAIL) with the
 1059     string "Z", which now yields "no match" instead of "partial match". In the
 1060     case of pcre_dfa_exec(), "no match" is given if every matching path for the
 1061     final character ended with (*FAIL).
 1062 
 1063 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
 1064     if the pattern had a "must contain" character that was already found in the
 1065     earlier partial match, unless partial matching was again requested. For
 1066     example, with the pattern /dog.(body)?/, the "must contain" character is
 1067     "g". If the first part-match was for the string "dog", restarting with
 1068     "sbody" failed. This bug has been fixed.
 1069 
 1070 13. The string returned by pcre_dfa_exec() after a partial match has been
 1071     changed so that it starts at the first inspected character rather than the
 1072     first character of the match. This makes a difference only if the pattern
 1073     starts with a lookbehind assertion or \b or \B (\K is not supported by
 1074     pcre_dfa_exec()). It's an incompatible change, but it makes the two
 1075     matching functions compatible, and I think it's the right thing to do.
 1076 
 1077 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
 1078     so that the demonstration program is easily available in environments where
 1079     PCRE has not been installed from source.
 1080 
 1081 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
 1082     libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
 1083     library.
 1084 
 1085 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
 1086     It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
 1087     is not the first non-POSIX option to be added. Clearly some people find
 1088     these options useful.
 1089 
 1090 17. If a caller to the POSIX matching function regexec() passes a non-zero
 1091     value for nmatch with a NULL value for pmatch, the value of
 1092     nmatch is forced to zero.
 1093 
 1094 18. RunGrepTest did not have a test for the availability of the -u option of
 1095     the diff command, as RunTest does. It now checks in the same way as
 1096     RunTest, and also checks for the -b option.
 1097 
 1098 19. If an odd number of negated classes containing just a single character
 1099     interposed, within parentheses, between a forward reference to a named
 1100     subpattern and the definition of the subpattern, compilation crashed with
 1101     an internal error, complaining that it could not find the referenced
 1102     subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
 1103     [The bug was that it was starting one character too far in when skipping
 1104     over the character class, thus treating the ] as data rather than
 1105     terminating the class. This meant it could skip too much.]
 1106 
 1107 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
 1108     /g option in pcretest when the pattern contains \K, which makes it possible
 1109     to have an empty string match not at the start, even when the pattern is
 1110     anchored. Updated pcretest and pcredemo to use this option.
 1111 
 1112 21. If the maximum number of capturing subpatterns in a recursion was greater
 1113     than the maximum at the outer level, the higher number was returned, but
 1114     with unset values at the outer level. The correct (outer level) value is
 1115     now given.
 1116 
 1117 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
 1118     PCRE did not set those parentheses (unlike Perl). I have now found a way to
 1119     make it do so. The string so far is captured, making this feature
 1120     compatible with Perl.
 1121 
 1122 23. The tests have been re-organized, adding tests 11 and 12, to make it
 1123     possible to check the Perl 5.10 features against Perl 5.10.
 1124 
 1125 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
 1126     pattern matches a fixed length string. PCRE did not allow this; now it
 1127     does. Neither allows recursion.
 1128 
 1129 25. I finally figured out how to implement a request to provide the minimum
 1130     length of subject string that was needed in order to match a given pattern.
 1131     (It was back references and recursion that I had previously got hung up
 1132     on.) This code has now been added to pcre_study(); it finds a lower bound
 1133     to the length of subject needed. It is not necessarily the greatest lower
 1134     bound, but using it to avoid searching strings that are too short does give
 1135     some useful speed-ups. The value is available to calling programs via
 1136     pcre_fullinfo().
 1137 
 1138 26. While implementing 25, I discovered to my embarrassment that pcretest had
 1139     not been passing the result of pcre_study() to pcre_dfa_exec(), so the
 1140     study optimizations had never been tested with that matching function.
 1141     Oops. What is worse, even when it was passed study data, there was a bug in
 1142     pcre_dfa_exec() that meant it never actually used it. Double oops. There
 1143     were also very few tests of studied patterns with pcre_dfa_exec().
 1144 
 1145 27. If (?| is used to create subpatterns with duplicate numbers, they are now
 1146     allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
 1147     on the other side of the coin, they are no longer allowed to have different
 1148     names, because these cannot be distinguished in PCRE, and this has caused
 1149     confusion. (This is a difference from Perl.)
 1150 
 1151 28. When duplicate subpattern names are present (necessarily with different
 1152     numbers, as required by 27 above), and a test is made by name in a
 1153     conditional pattern, either for a subpattern having been matched, or for
 1154     recursion in such a pattern, all the associated numbered subpatterns are
 1155     tested, and the overall condition is true if the condition is true for any
 1156     one of them. This is the way Perl works, and is also more like the way
 1157     testing by number works.
 1158 
 1159 
 1160 Version 7.9 11-Apr-09
 1161 ---------------------
 1162 
 1163 1.  When building with support for bzlib/zlib (pcregrep) and/or readline
 1164     (pcretest), all targets were linked against these libraries. This included
 1165     libpcre, libpcreposix, and libpcrecpp, even though they do not use these
 1166     libraries. This caused unwanted dependencies to be created. This problem
 1167     has been fixed, and now only pcregrep is linked with bzlib/zlib and only
 1168     pcretest is linked with readline.
 1169 
 1170 2.  The "typedef int BOOL" in pcre_internal.h that was included inside the
 1171     "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
 1172     moved outside it again, because FALSE and TRUE are already defined in AIX,
 1173     but BOOL is not.
 1174 
 1175 3.  The pcre_config() function was treating the PCRE_MATCH_LIMIT and
 1176     PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
 1177 
 1178 4.  The pcregrep documentation said spaces were inserted as well as colons (or
 1179     hyphens) following file names and line numbers when outputting matching
 1180     lines. This is not true; no spaces are inserted. I have also clarified the
 1181     wording for the --colour (or --color) option.
 1182 
 1183 5.  In pcregrep, when --colour was used with -o, the list of matching strings
 1184     was not coloured; this is different to GNU grep, so I have changed it to be
 1185     the same.
 1186 
 1187 6.  When --colo(u)r was used in pcregrep, only the first matching substring in
 1188     each matching line was coloured. Now it goes on to look for further matches
 1189     of any of the test patterns, which is the same behaviour as GNU grep.
 1190 
 1191 7.  A pattern that could match an empty string could cause pcregrep to loop; it
 1192     doesn't make sense to accept an empty string match in pcregrep, so I have
 1193     locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
 1194     seems to be how GNU grep behaves.
 1195 
 1196 8.  The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
 1197     start or after a newline", because the conditional assertion was not being
 1198     correctly handled. The rule now is that both the assertion and what follows
 1199     in the first alternative must satisfy the test.
 1200 
 1201 9.  If auto-callout was enabled in a pattern with a conditional group whose
 1202     condition was an assertion, PCRE could crash during matching, both with
 1203     pcre_exec() and pcre_dfa_exec().
 1204 
 1205 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
 1206     used for matching.
 1207 
 1208 11. Unicode property support in character classes was not working for
 1209     characters (bytes) greater than 127 when not in UTF-8 mode.
 1210 
 1211 12. Added the -M command line option to pcretest.
 1212 
 1213 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
 1214 
 1215 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
 1216 
 1217 16. Added comments and documentation about mis-use of no_arg in the C++
 1218     wrapper.
 1219 
 1220 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
 1221     from Martin Jerabek that uses macro names for all relevant character and
 1222     string constants.
 1223 
 1224 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
 1225     SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
 1226     SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
 1227     these, but not everybody uses configure.
 1228 
 1229 19. A conditional group that had only one branch was not being correctly
 1230     recognized as an item that could match an empty string. This meant that an
 1231     enclosing group might also not be so recognized, causing infinite looping
 1232     (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
 1233     with the subject "ab", where knowledge that the repeated group can match
 1234     nothing is needed in order to break the loop.
 1235 
 1236 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
 1237     exec(), but without supplying a callout function, matching went wrong.
 1238 
 1239 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
 1240     leak if the size of the offset vector was greater than 30. When the vector
 1241     is smaller, the saved offsets during recursion go onto a local stack
 1242     vector, but for larger vectors malloc() is used. It was failing to free
 1243     when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
 1244     error, in fact).
 1245 
 1246 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
 1247     heapframe that is used only when UTF-8 support is enabled. This caused no
 1248     problem, but was untidy.
 1249 
 1250 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
 1251     CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
 1252     included within another project.
 1253 
 1254 24. Steven Van Ingelgem's patches to add more options to the CMake support,
 1255     slightly modified by me:
 1256 
 1257       (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
 1258           not building pcregrep.
 1259 
 1260       (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
 1261           if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
 1262 
 1263 25. Forward references, both numeric and by name, in patterns that made use of
 1264     duplicate group numbers, could behave incorrectly or give incorrect errors,
 1265     because when scanning forward to find the reference group, PCRE was not
 1266     taking into account the duplicate group numbers. A pattern such as
 1267     ^X(?3)(a)(?|(b)|(q))(Y) is an example.
 1268 
 1269 26. Changed a few more instances of "const unsigned char *" to USPTR, making
 1270     the feature of a custom pointer more persuasive (as requested by a user).
 1271 
 1272 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
 1273     pcretest.c, inside #ifndefs, because it seems they are sometimes already
 1274     pre-defined.
 1275 
 1276 28. Added support for (*UTF8) at the start of a pattern.
 1277 
 1278 29. Arrange for flags added by the "release type" setting in CMake to be shown
 1279     in the configuration summary.
 1280 
 1281 
 1282 Version 7.8 05-Sep-08
 1283 ---------------------
 1284 
 1285 1.  Replaced UCP searching code with optimized version as implemented for Ad
 1286     Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
 1287     stage table and inline lookup instead of a function, giving speed ups of 2
 1288     to 5 times on some simple patterns that I tested. Permission was given to
 1289     distribute the MultiStage2.py script that generates the tables (it's not in
 1290     the tarball, but is in the Subversion repository).
 1291 
 1292 2.  Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
 1293     scripts.
 1294 
 1295 3.  Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
 1296     a group with a zero qualifier. The result of the study could be incorrect,
 1297     or the function might crash, depending on the pattern.
 1298 
 1299 4.  Caseless matching was not working for non-ASCII characters in back
 1300     references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
 1301     It now works when Unicode Property Support is available.
 1302 
 1303 5.  In pcretest, an escape such as \x{de} in the data was always generating
 1304     a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
 1305     non-UTF-8 mode. If the value is greater than 255, it gives a warning about
 1306     truncation.
 1307 
 1308 6.  Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
 1309 
 1310 7.  Added two (int) casts to pcregrep when printing the difference of two
 1311     pointers, in case they are 64-bit values.
 1312 
 1313 8.  Added comments about Mac OS X stack usage to the pcrestack man page and to
 1314     test 2 if it fails.
 1315 
 1316 9.  Added PCRE_CALL_CONVENTION just before the names of all exported functions,
 1317     and a #define of that name to empty if it is not externally set. This is to
 1318     allow users of MSVC to set it if necessary.
 1319 
 1320 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
 1321     the convenience functions in the pcre_get.c source file.
 1322 
 1323 11. An option change at the start of a pattern that had top-level alternatives
 1324     could cause overwriting and/or a crash. This command provoked a crash in
 1325     some environments:
 1326 
 1327       printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
 1328 
 1329     This potential security problem was recorded as CVE-2008-2371.
 1330 
 1331 12. For a pattern where the match had to start at the beginning or immediately
 1332     after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
 1333     pcre_dfa_exec() could read past the end of the passed subject if there was
 1334     no match. To help with detecting such bugs (e.g. with valgrind), I modified
 1335     pcretest so that it places the subject at the end of its malloc-ed buffer.
 1336 
 1337 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
 1338     exec() might read past the end of the data buffer in UTF-8 mode.
 1339 
 1340 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
 1341     the data contained the byte 0x85 as part of a UTF-8 character within its
 1342     first line. This applied both to normal and DFA matching.
 1343 
 1344 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
 1345     /^[^d]*?$/8 failed to match "abc".
 1346 
 1347 16. Added a missing copyright notice to pcrecpp_internal.h.
 1348 
 1349 17. Make it more clear in the documentation that values returned from
 1350     pcre_exec() in ovector are byte offsets, not character counts.
 1351 
 1352 18. Tidied a few places to stop certain compilers from issuing warnings.
 1353 
 1354 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
 1355     supplied by Stefan Weber. I made a further small update for 7.8 because
 1356     there is a change of source arrangements: the pcre_searchfuncs.c module is
 1357     replaced by pcre_ucd.c.
 1358 
 1359 
 1360 Version 7.7 07-May-08
 1361 ---------------------
 1362 
 1363 1.  Applied Craig's patch to sort out a long long problem: "If we can't convert
 1364     a string to a long long, pretend we don't even have a long long." This is
 1365     done by checking for the strtoq, strtoll, and _strtoi64 functions.
 1366 
 1367 2.  Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
 1368     pre-7.6 versions, which defined a global no_arg variable instead of putting
 1369     it in the RE class. (See also #8 below.)
 1370 
 1371 3.  Remove a line of dead code, identified by coverity and reported by Nuno
 1372     Lopes.
 1373 
 1374 4.  Fixed two related pcregrep bugs involving -r with --include or --exclude:
 1375 
 1376     (1) The include/exclude patterns were being applied to the whole pathnames
 1377         of files, instead of just to the final components.
 1378 
 1379     (2) If there was more than one level of directory, the subdirectories were
 1380         skipped unless they satisfied the include/exclude conditions. This is
 1381         inconsistent with GNU grep (and could even be seen as contrary to the
 1382         pcregrep specification - which I improved to make it absolutely clear).
 1383         The action now is always to scan all levels of directory, and just
 1384         apply the include/exclude patterns to regular files.
 1385 
 1386 5.  Added the --include_dir and --exclude_dir patterns to pcregrep, and used
 1387     --exclude_dir in the tests to avoid scanning .svn directories.
 1388 
 1389 6.  Applied Craig's patch to the QuoteMeta function so that it escapes the
 1390     NUL character as backslash + 0 rather than backslash + NUL, because PCRE
 1391     doesn't support NULs in patterns.
 1392 
 1393 7.  Added some missing "const"s to declarations of static tables in
 1394     pcre_compile.c and pcre_dfa_exec.c.
 1395 
 1396 8.  Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
 1397     caused by fix #2  above. (Subsequently also a second patch to fix the
 1398     first patch. And a third patch - this was a messy problem.)
 1399 
 1400 9.  Applied Craig's patch to remove the use of push_back().
 1401 
 1402 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
 1403     matching function regexec().
 1404 
 1405 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
 1406     which, however, unlike Perl's \g{...}, are subroutine calls, not back
 1407     references. PCRE supports relative numbers with this syntax (I don't think
 1408     Oniguruma does).
 1409 
 1410 12. Previously, a group with a zero repeat such as (...){0} was completely
 1411     omitted from the compiled regex. However, this means that if the group
 1412     was called as a subroutine from elsewhere in the pattern, things went wrong
 1413     (an internal error was given). Such groups are now left in the compiled
 1414     pattern, with a new opcode that causes them to be skipped at execution
 1415     time.
 1416 
 1417 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
 1418     to the way PCRE behaves:
 1419 
 1420     (a) A lone ] character is dis-allowed (Perl treats it as data).
 1421 
 1422     (b) A back reference to an unmatched subpattern matches an empty string
 1423         (Perl fails the current match path).
 1424 
 1425     (c) A data ] in a character class must be notated as \] because if the
 1426         first data character in a class is ], it defines an empty class. (In
 1427         Perl it is not possible to have an empty class.) The empty class []
 1428         never matches; it forces failure and is equivalent to (*FAIL) or (?!).
 1429         The negative empty class [^] matches any one character, independently
 1430         of the DOTALL setting.
 1431 
 1432 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
 1433     non-existent subpattern following a character class starting with ']' and
 1434     containing () gave an internal compiling error instead of "reference to
 1435     non-existent subpattern". Fortunately, when the pattern did exist, the
 1436     compiled code was correct. (When scanning forwards to check for the
 1437     existencd of the subpattern, it was treating the data ']' as terminating
 1438     the class, so got the count wrong. When actually compiling, the reference
 1439     was subsequently set up correctly.)
 1440 
 1441 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
 1442     it was being rejected as not supported by pcre_dfa_exec(), even though
 1443     other assertions are supported. I have made pcre_dfa_exec() support
 1444     (*FAIL).
 1445 
 1446 16. The implementation of 13c above involved the invention of a new opcode,
 1447     OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
 1448     cannot be changed at match time, I realized I could make a small
 1449     improvement to matching performance by compiling OP_ALLANY instead of
 1450     OP_ANY for "." when DOTALL was set, and then removing the runtime tests
 1451     on the OP_ANY path.
 1452 
 1453 17. Compiling pcretest on Windows with readline support failed without the
 1454     following two fixes: (1) Make the unistd.h include conditional on
 1455     HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
 1456 
 1457 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
 1458     ncurses library to be included for pcretest when ReadLine support is
 1459     requested, but also to allow for it to be overridden. This patch came from
 1460     Daniel Bergström.
 1461 
 1462 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
 1463     as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
 1464     any errors with the current Unicode tables. Thanks to Peter Kankowski for
 1465     spotting this.
 1466 
 1467 
 1468 Version 7.6 28-Jan-08
 1469 ---------------------
 1470 
 1471 1.  A character class containing a very large number of characters with
 1472     codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
 1473     overflow.
 1474 
 1475 2.  Patch to cut out the "long long" test in pcrecpp_unittest when
 1476     HAVE_LONG_LONG is not defined.
 1477 
 1478 3.  Applied Christian Ehrlicher's patch to update the CMake build files to
 1479     bring them up to date and include new features. This patch includes:
 1480 
 1481     - Fixed PH's badly added libz and libbz2 support.
 1482     - Fixed a problem with static linking.
 1483     - Added pcredemo. [But later removed - see 7 below.]
 1484     - Fixed dftables problem and added an option.
 1485     - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
 1486         HAVE_LONG_LONG.
 1487     - Added readline support for pcretest.
 1488     - Added an listing of the option settings after cmake has run.
 1489 
 1490 4.  A user submitted a patch to Makefile that makes it easy to create
 1491     "pcre.dll" under mingw when using Configure/Make. I added stuff to
 1492     Makefile.am that cause it to include this special target, without
 1493     affecting anything else. Note that the same mingw target plus all
 1494     the other distribution libraries and programs are now supported
 1495     when configuring with CMake (see 6 below) instead of with
 1496     Configure/Make.
 1497 
 1498 5.  Applied Craig's patch that moves no_arg into the RE class in the C++ code.
 1499     This is an attempt to solve the reported problem "pcrecpp::no_arg is not
 1500     exported in the Windows port". It has not yet been confirmed that the patch
 1501     solves the problem, but it does no harm.
 1502 
 1503 6.  Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
 1504     NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
 1505     with CMake, and also correct the comment about stack recursion.
 1506 
 1507 7.  Remove the automatic building of pcredemo from the ./configure system and
 1508     from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
 1509     of a program that users should build themselves after PCRE is installed, so
 1510     building it automatically is not really right. What is more, it gave
 1511     trouble in some build environments.
 1512 
 1513 8.  Further tidies to CMakeLists.txt from Sheri and Christian.
 1514 
 1515 
 1516 Version 7.5 10-Jan-08
 1517 ---------------------
 1518 
 1519 1.  Applied a patch from Craig: "This patch makes it possible to 'ignore'
 1520     values in parens when parsing an RE using the C++ wrapper."
 1521 
 1522 2.  Negative specials like \S did not work in character classes in UTF-8 mode.
 1523     Characters greater than 255 were excluded from the class instead of being
 1524     included.
 1525 
 1526 3.  The same bug as (2) above applied to negated POSIX classes such as
 1527     [:^space:].
 1528 
 1529 4.  PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
 1530     defined or documented. It seems to have been a typo for PCRE_STATIC, so
 1531     I have changed it.
 1532 
 1533 5.  The construct (?&) was not diagnosed as a syntax error (it referenced the
 1534     first named subpattern) and a construct such as (?&a) would reference the
 1535     first named subpattern whose name started with "a" (in other words, the
 1536     length check was missing). Both these problems are fixed. "Subpattern name
 1537     expected" is now given for (?&) (a zero-length name), and this patch also
 1538     makes it give the same error for \k'' (previously it complained that that
 1539     was a reference to a non-existent subpattern).
 1540 
 1541 6.  The erroneous patterns (?+-a) and (?-+a) give different error messages;
 1542     this is right because (?- can be followed by option settings as well as by
 1543     digits. I have, however, made the messages clearer.
 1544 
 1545 7.  Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
 1546     than the number used in the conditional) now cause a compile-time error.
 1547     This is actually not compatible with Perl, which accepts such patterns, but
 1548     treats the conditional as always being FALSE (as PCRE used to), but it
 1549     seems to me that giving a diagnostic is better.
 1550 
 1551 8.  Change "alphameric" to the more common word "alphanumeric" in comments
 1552     and messages.
 1553 
 1554 9.  Fix two occurrences of "backslash" in comments that should have been
 1555     "backspace".
 1556 
 1557 10. Remove two redundant lines of code that can never be obeyed (their function
 1558     was moved elsewhere).
 1559 
 1560 11. The program that makes PCRE's Unicode character property table had a bug
 1561     which caused it to generate incorrect table entries for sequences of
 1562     characters that have the same character type, but are in different scripts.
 1563     It amalgamated them into a single range, with the script of the first of
 1564     them. In other words, some characters were in the wrong script. There were
 1565     thirteen such cases, affecting characters in the following ranges:
 1566 
 1567       U+002b0 - U+002c1
 1568       U+0060c - U+0060d
 1569       U+0061e - U+00612
 1570       U+0064b - U+0065e
 1571       U+0074d - U+0076d
 1572       U+01800 - U+01805
 1573       U+01d00 - U+01d77
 1574       U+01d9b - U+01dbf
 1575       U+0200b - U+0200f
 1576       U+030fc - U+030fe
 1577       U+03260 - U+0327f
 1578       U+0fb46 - U+0fbb1
 1579       U+10450 - U+1049d
 1580 
 1581 12. The -o option (show only the matching part of a line) for pcregrep was not
 1582     compatible with GNU grep in that, if there was more than one match in a
 1583     line, it showed only the first of them. It now behaves in the same way as
 1584     GNU grep.
 1585 
 1586 13. If the -o and -v options were combined for pcregrep, it printed a blank
 1587     line for every non-matching line. GNU grep prints nothing, and pcregrep now
 1588     does the same. The return code can be used to tell if there were any
 1589     non-matching lines.
 1590 
 1591 14. Added --file-offsets and --line-offsets to pcregrep.
 1592 
 1593 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
 1594     infinitely looping recursion. The bug was that positive lookaheads were not
 1595     being skipped when checking for a possible empty match (negative lookaheads
 1596     and both kinds of lookbehind were skipped).
 1597 
 1598 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
 1599     inclusion of <windows.h> to before rather than after the definition of
 1600     INVALID_FILE_ATTRIBUTES (patch from David Byron).
 1601 
 1602 17. Specifying a possessive quantifier with a specific limit for a Unicode
 1603     character property caused pcre_compile() to compile bad code, which led at
 1604     runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
 1605     are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
 1606     caused the error; without that there was no problem.
 1607 
 1608 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
 1609 
 1610 19. Added --enable-pcretest-libreadline.
 1611 
 1612 20. In pcrecpp.cc, the variable 'count' was incremented twice in
 1613     RE::GlobalReplace(). As a result, the number of replacements returned was
 1614     double what it should be. I removed one of the increments, but Craig sent a
 1615     later patch that removed the other one (the right fix) and added unit tests
 1616     that check the return values (which was not done before).
 1617 
 1618 21. Several CMake things:
 1619 
 1620     (1) Arranged that, when cmake is used on Unix, the libraries end up with
 1621         the names libpcre and libpcreposix, not just pcre and pcreposix.
 1622 
 1623     (2) The above change means that pcretest and pcregrep are now correctly
 1624         linked with the newly-built libraries, not previously installed ones.
 1625 
 1626     (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
 1627 
 1628 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
 1629     crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
 1630     UTF-8 newline character). The key issue is that the pattern starts .*;
 1631     this means that the match must be either at the beginning, or after a
 1632     newline. The bug was in the code for advancing after a failed match and
 1633     checking that the new position followed a newline. It was not taking
 1634     account of UTF-8 characters correctly.
 1635 
 1636 23. PCRE was behaving differently from Perl in the way it recognized POSIX
 1637     character classes. PCRE was not treating the sequence [:...:] as a
 1638     character class unless the ... were all letters. Perl, however, seems to
 1639     allow any characters between [: and :], though of course it rejects as
 1640     unknown any "names" that contain non-letters, because all the known class
 1641     names consist only of letters. Thus, Perl gives an error for [[:1234:]],
 1642     for example, whereas PCRE did not - it did not recognize a POSIX character
 1643     class. This seemed a bit dangerous, so the code has been changed to be
 1644     closer to Perl. The behaviour is not identical to Perl, because PCRE will
 1645     diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
 1646     treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
 1647     Perl does, and where it didn't before.
 1648 
 1649 24. Rewrite so as to remove the single use of %n from pcregrep because in some
 1650     Windows environments %n is disabled by default.
 1651 
 1652 
 1653 Version 7.4 21-Sep-07
 1654 ---------------------
 1655 
 1656 1.  Change 7.3/28 was implemented for classes by looking at the bitmap. This
 1657     means that a class such as [\s] counted as "explicit reference to CR or
 1658     LF". That isn't really right - the whole point of the change was to try to
 1659     help when there was an actual mention of one of the two characters. So now
 1660     the change happens only if \r or \n (or a literal CR or LF) character is
 1661     encountered.
 1662 
 1663 2.  The 32-bit options word was also used for 6 internal flags, but the numbers
 1664     of both had grown to the point where there were only 3 bits left.
 1665     Fortunately, there was spare space in the data structure, and so I have
 1666     moved the internal flags into a new 16-bit field to free up more option
 1667     bits.
 1668 
 1669 3.  The appearance of (?J) at the start of a pattern set the DUPNAMES option,
 1670     but did not set the internal JCHANGED flag - either of these is enough to
 1671     control the way the "get" function works - but the PCRE_INFO_JCHANGED
 1672     facility is supposed to tell if (?J) was ever used, so now (?J) at the
 1673     start sets both bits.
 1674 
 1675 4.  Added options (at build time, compile time, exec time) to change \R from
 1676     matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
 1677 
 1678 5.  doc/pcresyntax.html was missing from the distribution.
 1679 
 1680 6.  Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
 1681     compatibility, even though it is no longer used.
 1682 
 1683 7.  Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
 1684     strtoull to pcrecpp.cc to select the available functions in WIN32 when the
 1685     windows.h file is present (where different names are used). [This was
 1686     reversed later after testing - see 16 below.]
 1687 
 1688 8.  Changed all #include <config.h> to #include "config.h". There were also
 1689     some further <pcre.h> cases that I changed to "pcre.h".
 1690 
 1691 9.  When pcregrep was used with the --colour option, it missed the line ending
 1692     sequence off the lines that it output.
 1693 
 1694 10. It was pointed out to me that arrays of string pointers cause lots of
 1695     relocations when a shared library is dynamically loaded. A technique of
 1696     using a single long string with a table of offsets can drastically reduce
 1697     these. I have refactored PCRE in four places to do this. The result is
 1698     dramatic:
 1699 
 1700       Originally:                          290
 1701       After changing UCP table:            187
 1702       After changing error message table:   43
 1703       After changing table of "verbs"       36
 1704       After changing table of Posix names   22
 1705 
 1706     Thanks to the folks working on Gregex for glib for this insight.
 1707 
 1708 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
 1709     unicode-properties was also set.
 1710 
 1711 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
 1712 
 1713 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
 1714     checked only for CRLF.
 1715 
 1716 14. Added casts to pcretest.c to avoid compiler warnings.
 1717 
 1718 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
 1719 
 1720 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
 1721     and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
 1722     entirely. This removes changes made in 7 above.
 1723 
 1724 17. The CMake files have been updated, and there is now more information about
 1725     building with CMake in the NON-UNIX-USE document.
 1726 
 1727 
 1728 Version 7.3 28-Aug-07
 1729 ---------------------
 1730 
 1731  1. In the rejigging of the build system that eventually resulted in 7.1, the
 1732     line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
 1733     brackets there is not right, since it causes compilers to look for an
 1734     installed pcre.h, not the version that is in the source that is being
 1735     compiled (which of course may be different). I have changed it back to:
 1736 
 1737       #include "pcre.h"
 1738 
 1739     I have a vague recollection that the change was concerned with compiling in
 1740     different directories, but in the new build system, that is taken care of
 1741     by the VPATH setting the Makefile.
 1742 
 1743  2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
 1744     when the subject happened to end in the byte 0x85 (e.g. if the last
 1745     character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
 1746     characters but of course it shouldn't be taken as a newline when it is part
 1747     of another character. The bug was that, for an unlimited repeat of . in
 1748     not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
 1749     characters when looking for a newline.
 1750 
 1751  3. A small performance improvement in the DOTALL UTF-8 mode .* case.
 1752 
 1753  4. Debugging: adjusted the names of opcodes for different kinds of parentheses
 1754     in debug output.
 1755 
 1756  5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
 1757     long printing in the pcrecpp unittest when running under MinGW.
 1758 
 1759  6. ESC_K was left out of the EBCDIC table.
 1760 
 1761  7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
 1762     parentheses; I made it 1000, which seemed large enough. Unfortunately, the
 1763     limit also applies to "virtual nesting" when a pattern is recursive, and in
 1764     this case 1000 isn't so big. I have been able to remove this limit at the
 1765     expense of backing off one optimization in certain circumstances. Normally,
 1766     when pcre_exec() would call its internal match() function recursively and
 1767     immediately return the result unconditionally, it uses a "tail recursion"
 1768     feature to save stack. However, when a subpattern that can match an empty
 1769     string has an unlimited repetition quantifier, it no longer makes this
 1770     optimization. That gives it a stack frame in which to save the data for
 1771     checking that an empty string has been matched. Previously this was taken
 1772     from the 1000-entry workspace that had been reserved. So now there is no
 1773     explicit limit, but more stack is used.
 1774 
 1775  8. Applied Daniel's patches to solve problems with the import/export magic
 1776     syntax that is required for Windows, and which was going wrong for the
 1777     pcreposix and pcrecpp parts of the library. These were overlooked when this
 1778     problem was solved for the main library.
 1779 
 1780  9. There were some crude static tests to avoid integer overflow when computing
 1781     the size of patterns that contain repeated groups with explicit upper
 1782     limits. As the maximum quantifier is 65535, the maximum group length was
 1783     set at 30,000 so that the product of these two numbers did not overflow a
 1784     32-bit integer. However, it turns out that people want to use groups that
 1785     are longer than 30,000 bytes (though not repeat them that many times).
 1786     Change 7.0/17 (the refactoring of the way the pattern size is computed) has
 1787     made it possible to implement the integer overflow checks in a much more
 1788     dynamic way, which I have now done. The artificial limitation on group
 1789     length has been removed - we now have only the limit on the total length of
 1790     the compiled pattern, which depends on the LINK_SIZE setting.
 1791 
 1792 10. Fixed a bug in the documentation for get/copy named substring when
 1793     duplicate names are permitted. If none of the named substrings are set, the
 1794     functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
 1795     empty string.
 1796 
 1797 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
 1798     instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
 1799     because the ] is interpreted as the first data character and the
 1800     terminating ] is not found. PCRE has been made compatible with Perl in this
 1801     regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
 1802     cause memory overwriting.
 1803 
 1804 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
 1805     string has been matched (to stop an infinite loop). It was not recognizing
 1806     a conditional subpattern that could match an empty string if that
 1807     subpattern was within another subpattern. For example, it looped when
 1808     trying to match  (((?(1)X|))*)  but it was OK with  ((?(1)X|)*)  where the
 1809     condition was not nested. This bug has been fixed.
 1810 
 1811 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
 1812     past the start of the subject in the presence of bytes with the top bit
 1813     set, for example "\x8aBCD".
 1814 
 1815 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
 1816     (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
 1817 
 1818 14. Optimized (?!) to (*FAIL).
 1819 
 1820 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
 1821     This restricts code points to be within the range 0 to 0x10FFFF, excluding
 1822     the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
 1823     full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
 1824     does: it's just the validity check that is more restrictive.
 1825 
 1826 16. Inserted checks for integer overflows during escape sequence (backslash)
 1827     processing, and also fixed erroneous offset values for syntax errors during
 1828     backslash processing.
 1829 
 1830 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
 1831     for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
 1832 
 1833 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
 1834     caused an overrun.
 1835 
 1836 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
 1837     something other than just ASCII characters) inside a group that had an
 1838     unlimited repeat caused a loop at compile time (while checking to see
 1839     whether the group could match an empty string).
 1840 
 1841 20. Debugging a pattern containing \p or \P could cause a crash. For example,
 1842     [\P{Any}] did so. (Error in the code for printing property names.)
 1843 
 1844 21. An orphan \E inside a character class could cause a crash.
 1845 
 1846 22. A repeated capturing bracket such as (A)? could cause a wild memory
 1847     reference during compilation.
 1848 
 1849 23. There are several functions in pcre_compile() that scan along a compiled
 1850     expression for various reasons (e.g. to see if it's fixed length for look
 1851     behind). There were bugs in these functions when a repeated \p or \P was
 1852     present in the pattern. These operators have additional parameters compared
 1853     with \d, etc, and these were not being taken into account when moving along
 1854     the compiled data. Specifically:
 1855 
 1856     (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
 1857         length.
 1858 
 1859     (b) An item such as \pL+ within a repeated group could cause crashes or
 1860         loops.
 1861 
 1862     (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
 1863         "reference to non-existent subpattern" error.
 1864 
 1865     (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
 1866 
 1867 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
 1868     characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
 1869 
 1870 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
 1871 
 1872 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
 1873     character were causing crashes (broken optimization).
 1874 
 1875 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
 1876     \p or \P) caused a compile-time loop.
 1877 
 1878 28. More problems have arisen in unanchored patterns when CRLF is a valid line
 1879     break. For example, the unstudied pattern [\r\n]A does not match the string
 1880     "\r\nA" because change 7.0/46 below moves the current point on by two
 1881     characters after failing to match at the start. However, the pattern \nA
 1882     *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
 1883     the same is true. There doesn't seem any very clean way out of this, but
 1884     what I have chosen to do makes the common cases work: PCRE now takes note
 1885     of whether there can be an explicit match for \r or \n anywhere in the
 1886     pattern, and if so, 7.0/46 no longer applies. As part of this change,
 1887     there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
 1888     pattern has explicit CR or LF references.
 1889 
 1890 29. Added (*CR) etc for changing newline setting at start of pattern.
 1891 
 1892 
 1893 Version 7.2 19-Jun-07
 1894 ---------------------
 1895 
 1896  1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
 1897     which is apparently normally available under Windows.
 1898 
 1899  2. Re-jig the pcregrep tests with different newline settings in an attempt
 1900     to make them independent of the local environment's newline setting.
 1901 
 1902  3. Add code to configure.ac to remove -g from the CFLAGS default settings.
 1903 
 1904  4. Some of the "internals" tests were previously cut out when the link size
 1905     was not 2, because the output contained actual offsets. The recent new
 1906     "Z" feature of pcretest means that these can be cut out, making the tests
 1907     usable with all link sizes.
 1908 
 1909  5. Implemented Stan Switzer's goto replacement for longjmp() when not using
 1910     stack recursion. This gives a massive performance boost under BSD, but just
 1911     a small improvement under Linux. However, it saves one field in the frame
 1912     in all cases.
 1913 
 1914  6. Added more features from the forthcoming Perl 5.10:
 1915 
 1916     (a) (?-n) (where n is a string of digits) is a relative subroutine or
 1917         recursion call. It refers to the nth most recently opened parentheses.
 1918 
 1919     (b) (?+n) is also a relative subroutine call; it refers to the nth next
 1920         to be opened parentheses.
 1921 
 1922     (c) Conditions that refer to capturing parentheses can be specified
 1923         relatively, for example, (?(-2)... or (?(+3)...
 1924 
 1925     (d) \K resets the start of the current match so that everything before
 1926         is not part of it.
 1927 
 1928     (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
 1929 
 1930     (f) \g{name} is another synonym - part of Perl 5.10's unification of
 1931         reference syntax.
 1932 
 1933     (g) (?| introduces a group in which the numbering of parentheses in each
 1934         alternative starts with the same number.
 1935 
 1936     (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
 1937 
 1938  7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
 1939     PCRE_INFO_JCHANGED.
 1940 
 1941  8. A pattern such as  (.*(.)?)*  caused pcre_exec() to fail by either not
 1942     terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
 1943     for detecting groups that can match an empty string.
 1944 
 1945  9. A pattern with a very large number of alternatives (more than several
 1946     hundred) was running out of internal workspace during the pre-compile
 1947     phase, where pcre_compile() figures out how much memory will be needed. A
 1948     bit of new cunning has reduced the workspace needed for groups with
 1949     alternatives. The 1000-alternative test pattern now uses 12 bytes of
 1950     workspace instead of running out of the 4096 that are available.
 1951 
 1952 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
 1953 
 1954 11. Applied patch from Google to remove an optimization that didn't quite work.
 1955     The report of the bug said:
 1956 
 1957       pcrecpp::RE("a*").FullMatch("aaa") matches, while
 1958       pcrecpp::RE("a*?").FullMatch("aaa") does not, and
 1959       pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
 1960 
 1961 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
 1962     it matched the wrong number of bytes.
 1963 
 1964 
 1965 Version 7.1 24-Apr-07
 1966 ---------------------
 1967 
 1968  1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
 1969     that is more "standard", making use of automake and other Autotools. There
 1970     is some re-arrangement of the files and adjustment of comments consequent
 1971     on this.
 1972 
 1973  2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
 1974     for recursive directory scanning broke on some systems because the files
 1975     are not scanned in any specific order and on different systems the order
 1976     was different. A call to "sort" has been inserted into RunGrepTest for the
 1977     approprate test as a short-term fix. In the longer term there may be an
 1978     alternative.
 1979 
 1980  3. I had an email from Eric Raymond about problems translating some of PCRE's
 1981     man pages to HTML (despite the fact that I distribute HTML pages, some
 1982     people do their own conversions for various reasons). The problems
 1983     concerned the use of low-level troff macros .br and .in. I have therefore
 1984     removed all such uses from the man pages (some were redundant, some could
 1985     be replaced by .nf/.fi pairs). The 132html script that I use to generate
 1986     HTML has been updated to handle .nf/.fi and to complain if it encounters
 1987     .br or .in.
 1988 
 1989  4. Updated comments in configure.ac that get placed in config.h.in and also
 1990     arranged for config.h to be included in the distribution, with the name
 1991     config.h.generic, for the benefit of those who have to compile without
 1992     Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
 1993 
 1994  5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
 1995     Weber: (1) pcre_internal.h was missing some function renames; (2) updated
 1996     makevp.bat for the current PCRE, using the additional files
 1997     makevp_c.txt, makevp_l.txt, and pcregexp.pas.
 1998 
 1999  6. A Windows user reported a minor discrepancy with test 2, which turned out
 2000     to be caused by a trailing space on an input line that had got lost in his
 2001     copy. The trailing space was an accident, so I've just removed it.
 2002 
 2003  7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
 2004     that is needed.
 2005 
 2006  8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
 2007     as "const" (a) because they are and (b) because it helps the PHP
 2008     maintainers who have recently made a script to detect big data structures
 2009     in the php code that should be moved to the .rodata section. I remembered
 2010     to update Builducptable as well, so it won't revert if ucptable.h is ever
 2011     re-created.
 2012 
 2013  9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
 2014     pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
 2015     order to be able to cut out the UTF-8 tables in the latter when UTF-8
 2016     support is not required. This saves 1.5-2K of code, which is important in
 2017     some applications.
 2018 
 2019     Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
 2020     so as not to refer to the tables, even though these functions will never be
 2021     called when UTF-8 support is disabled. Otherwise there are problems with a
 2022     shared library.
 2023 
 2024 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
 2025 
 2026     (a) It was defining its arguments as char * instead of void *.
 2027 
 2028     (b) It was assuming that all moves were upwards in memory; this was true
 2029         a long time ago when I wrote it, but is no longer the case.
 2030 
 2031     The emulated memove() is provided for those environments that have neither
 2032     memmove() nor bcopy(). I didn't think anyone used it these days, but that
 2033     is clearly not the case, as these two bugs were recently reported.
 2034 
 2035 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
 2036     and Detrail to create the HTML documentation, the .txt form of the man
 2037     pages, and it removes trailing spaces from listed files. It also creates
 2038     pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
 2039     case, it wraps all the #defines with #ifndefs. This script should be run
 2040     before "make dist".
 2041 
 2042 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
 2043     with Unicode property support.
 2044 
 2045     (a) For a maximizing quantifier, if the two different cases of the
 2046         character were of different lengths in their UTF-8 codings (there are
 2047         some cases like this - I found 11), and the matching function had to
 2048         back up over a mixture of the two cases, it incorrectly assumed they
 2049         were both the same length.
 2050 
 2051     (b) When PCRE was configured to use the heap rather than the stack for
 2052         recursion during matching, it was not correctly preserving the data for
 2053         the other case of a UTF-8 character when checking ahead for a match
 2054         while processing a minimizing repeat. If the check also involved
 2055         matching a wide character, but failed, corruption could cause an
 2056         erroneous result when trying to check for a repeat of the original
 2057         character.
 2058 
 2059 13. Some tidying changes to the testing mechanism:
 2060 
 2061     (a) The RunTest script now detects the internal link size and whether there
 2062         is UTF-8 and UCP support by running ./pcretest -C instead of relying on
 2063         values substituted by "configure". (The RunGrepTest script already did
 2064         this for UTF-8.) The configure.ac script no longer substitutes the
 2065         relevant variables.
 2066 
 2067     (b) The debugging options /B and /D in pcretest show the compiled bytecode
 2068         with length and offset values. This means that the output is different
 2069         for different internal link sizes. Test 2 is skipped for link sizes
 2070         other than 2 because of this, bypassing the problem. Unfortunately,
 2071         there was also a test in test 3 (the locale tests) that used /B and
 2072         failed for link sizes other than 2. Rather than cut the whole test out,
 2073         I have added a new /Z option to pcretest that replaces the length and
 2074         offset values with spaces. This is now used to make test 3 independent
 2075         of link size. (Test 2 will be tidied up later.)
 2076 
 2077 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
 2078     segmentation fault instead of returning the appropriate error message.
 2079 
 2080 15. In multiline mode when the newline sequence was set to "any", the pattern
 2081     ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
 2082     This doesn't seem right; it now treats the CRLF combination as the line
 2083     ending, and so does not match in that case. It's only a pattern such as ^$
 2084     that would hit this one: something like ^ABC$ would have failed after \r
 2085     and then tried again after \r\n.
 2086 
 2087 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
 2088     in an attempt to make files that differ only in their line terminators
 2089     compare equal. This works on Linux.
 2090 
 2091 17. Under certain error circumstances pcregrep might try to free random memory
 2092     as it exited. This is now fixed, thanks to valgrind.
 2093 
 2094 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
 2095     "abc\r\n\r\n", it found an unwanted second match after the second \r. This
 2096     was because its rules for how to advance for /g after matching an empty
 2097     string at the end of a line did not allow for this case. They now check for
 2098     it specially.
 2099 
 2100 20. pcretest is supposed to handle patterns and data of any length, by
 2101     extending its buffers when necessary. It was getting this wrong when the
 2102     buffer for a data line had to be extended.
 2103 
 2104 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
 2105     CRLF as a newline sequence.
 2106 
 2107 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
 2108     out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
 2109     I have nevertheless tidied it up.
 2110 
 2111 23. Added some casts to kill warnings from HP-UX ia64 compiler.
 2112 
 2113 24. Added a man page for pcre-config.
 2114 
 2115 
 2116 Version 7.0 19-Dec-06
 2117 ---------------------
 2118 
 2119  1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
 2120     moving to gcc 4.1.1.
 2121 
 2122  2. The -S option for pcretest uses setrlimit(); I had omitted to #include
 2123     sys/time.h, which is documented as needed for this function. It doesn't
 2124     seem to matter on Linux, but it showed up on some releases of OS X.
 2125 
 2126  3. It seems that there are systems where bytes whose values are greater than
 2127     127 match isprint() in the "C" locale. The "C" locale should be the
 2128     default when a C program starts up. In most systems, only ASCII printing
 2129     characters match isprint(). This difference caused the output from pcretest
 2130     to vary, making some of the tests fail. I have changed pcretest so that:
 2131 
 2132     (a) When it is outputting text in the compiled version of a pattern, bytes
 2133         other than 32-126 are always shown as hex escapes.
 2134 
 2135     (b) When it is outputting text that is a matched part of a subject string,
 2136         it does the same, unless a different locale has been set for the match
 2137         (using the /L modifier). In this case, it uses isprint() to decide.
 2138 
 2139  4. Fixed a major bug that caused incorrect computation of the amount of memory
 2140     required for a compiled pattern when options that changed within the
 2141     pattern affected the logic of the preliminary scan that determines the
 2142     length. The relevant options are -x, and -i in UTF-8 mode. The result was
 2143     that the computed length was too small. The symptoms of this bug were
 2144     either the PCRE error "internal error: code overflow" from pcre_compile(),
 2145     or a glibc crash with a message such as "pcretest: free(): invalid next
 2146     size (fast)". Examples of patterns that provoked this bug (shown in
 2147     pcretest format) are:
 2148 
 2149       /(?-x: )/x
 2150       /(?x)(?-x: \s*#\s*)/
 2151       /((?i)[\x{c0}])/8
 2152       /(?i:[\x{c0}])/8
 2153 
 2154     HOWEVER: Change 17 below makes this fix obsolete as the memory computation
 2155     is now done differently.
 2156 
 2157  5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
 2158     wrapper classes; (b) implement a new function in the C++ scanner that is
 2159     more efficient than the old way of doing things because it avoids levels of
 2160     recursion in the regex matching; (c) add a paragraph to the documentation
 2161     for the FullMatch() function.
 2162 
 2163  6. The escape sequence \n was being treated as whatever was defined as
 2164     "newline". Not only was this contrary to the documentation, which states
 2165     that \n is character 10 (hex 0A), but it also went horribly wrong when
 2166     "newline" was defined as CRLF. This has been fixed.
 2167 
 2168  7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
 2169     was being set to -1 for the "end of line" case (supposedly a value that no
 2170     character can have). Though this value is never used (the check for end of
 2171     line is "zero bytes in current character"), it caused compiler complaints.
 2172     I've changed it to 0xffffffff.
 2173 
 2174  8. In pcre_version.c, the version string was being built by a sequence of
 2175     C macros that, in the event of PCRE_PRERELEASE being defined as an empty
 2176     string (as it is for production releases) called a macro with an empty
 2177     argument. The C standard says the result of this is undefined. The gcc
 2178     compiler treats it as an empty string (which was what was wanted) but it is
 2179     reported that Visual C gives an error. The source has been hacked around to
 2180     avoid this problem.
 2181 
 2182  9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
 2183     builds of pcretest, and changed the call to _setmode() to use _O_BINARY
 2184     instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
 2185     of them did).
 2186 
 2187 10. Originally, pcretest opened its input and output without "b"; then I was
 2188     told that "b" was needed in some environments, so it was added for release
 2189     5.0 to both the input and output. (It makes no difference on Unix-like
 2190     systems.) Later I was told that it is wrong for the input on Windows. I've
 2191     now abstracted the modes into two macros, to make it easier to fiddle with
 2192     them, and removed "b" from the input mode under Windows.
 2193 
 2194 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
 2195 
 2196 12. Added -help and --help to pcretest as an official way of being reminded
 2197     of the options.
 2198 
 2199 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
 2200     and pcrecpp.cc because they annoy compilers at high warning levels.
 2201 
 2202 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
 2203 
 2204 15. Fixed an occurrence of == in configure.ac that should have been = (shell
 2205     scripts are not C programs :-) and which was not noticed because it works
 2206     on Linux.
 2207 
 2208 16. pcretest is supposed to handle any length of pattern and data line (as one
 2209     line or as a continued sequence of lines) by extending its input buffer if
 2210     necessary. This feature was broken for very long pattern lines, leading to
 2211     a string of junk being passed to pcre_compile() if the pattern was longer
 2212     than about 50K.
 2213 
 2214 17. I have done a major re-factoring of the way pcre_compile() computes the
 2215     amount of memory needed for a compiled pattern. Previously, there was code
 2216     that made a preliminary scan of the pattern in order to do this. That was
 2217     OK when PCRE was new, but as the facilities have expanded, it has become
 2218     harder and harder to keep it in step with the real compile phase, and there
 2219     have been a number of bugs (see for example, 4 above). I have now found a
 2220     cunning way of running the real compile function in a "fake" mode that
 2221     enables it to compute how much memory it would need, while actually only
 2222     ever using a few hundred bytes of working memory and without too many
 2223     tests of the mode. This should make future maintenance and development
 2224     easier. A side effect of this work is that the limit of 200 on the nesting
 2225     depth of parentheses has been removed (though this was never a serious
 2226     limitation, I suspect). However, there is a downside: pcre_compile() now
 2227     runs more slowly than before (30% or more, depending on the pattern). I
 2228     hope this isn't a big issue. There is no effect on runtime performance.
 2229 
 2230 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
 2231     newline (only possible for the last line of a file) and it was a
 2232     pattern that set a locale (followed by /Lsomething), pcretest crashed.
 2233 
 2234 19. Added additional timing features to pcretest. (1) The -tm option now times
 2235     matching only, not compiling. (2) Both -t and -tm can be followed, as a
 2236     separate command line item, by a number that specifies the number of
 2237     repeats to use when timing. The default is 50000; this gives better
 2238     precision, but takes uncomfortably long for very large patterns.
 2239 
 2240 20. Extended pcre_study() to be more clever in cases where a branch of a
 2241     subpattern has no definite first character. For example, (a*|b*)[cd] would
 2242     previously give no result from pcre_study(). Now it recognizes that the
 2243     first character must be a, b, c, or d.
 2244 
 2245 21. There was an incorrect error "recursive call could loop indefinitely" if
 2246     a subpattern (or the entire pattern) that was being tested for matching an
 2247     empty string contained only one non-empty item after a nested subpattern.
 2248     For example, the pattern (?>\x{100}*)\d(?R) provoked this error
 2249     incorrectly, because the \d was being skipped in the check.
 2250 
 2251 22. The pcretest program now has a new pattern option /B and a command line
 2252     option -b, which is equivalent to adding /B to every pattern. This causes
 2253     it to show the compiled bytecode, without the additional information that
 2254     -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
 2255     is the same as /B/I).
 2256 
 2257 23. A new optimization is now able automatically to treat some sequences such
 2258     as a*b as a*+b. More specifically, if something simple (such as a character
 2259     or a simple class like \d) has an unlimited quantifier, and is followed by
 2260     something that cannot possibly match the quantified thing, the quantifier
 2261     is automatically "possessified".
 2262 
 2263 24. A recursive reference to a subpattern whose number was greater than 39
 2264     went wrong under certain circumstances in UTF-8 mode. This bug could also
 2265     have affected the operation of pcre_study().
 2266 
 2267 25. Realized that a little bit of performance could be had by replacing
 2268     (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
 2269 
 2270 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
 2271 
 2272 27. Possessive quantifiers such as a++ were previously implemented by turning
 2273     them into atomic groups such as ($>a+). Now they have their own opcodes,
 2274     which improves performance. This includes the automatically created ones
 2275     from 23 above.
 2276 
 2277 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
 2278     lookahead was broken if it was not anchored. PCRE was mistakenly expecting
 2279     the first matched character to be a colon. This applied both to named and
 2280     numbered groups.
 2281 
 2282 29. The ucpinternal.h header file was missing its idempotency #ifdef.
 2283 
 2284 30. I was sent a "project" file called libpcre.a.dev which I understand makes
 2285     building PCRE on Windows easier, so I have included it in the distribution.
 2286 
 2287 31. There is now a check in pcretest against a ridiculously large number being
 2288     returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
 2289     loop, the loop is abandoned.
 2290 
 2291 32. Forward references to subpatterns in conditions such as (?(2)...) where
 2292     subpattern 2 is defined later cause pcre_compile() to search forwards in
 2293     the pattern for the relevant set of parentheses. This search went wrong
 2294     when there were unescaped parentheses in a character class, parentheses
 2295     escaped with \Q...\E, or parentheses in a #-comment in /x mode.
 2296 
 2297 33. "Subroutine" calls and backreferences were previously restricted to
 2298     referencing subpatterns earlier in the regex. This restriction has now
 2299     been removed.
 2300 
 2301 34. Added a number of extra features that are going to be in Perl 5.10. On the
 2302     whole, these are just syntactic alternatives for features that PCRE had
 2303     previously implemented using the Python syntax or my own invention. The
 2304     other formats are all retained for compatibility.
 2305 
 2306     (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
 2307         as (?P<name>...). The new forms, as well as being in Perl 5.10, are
 2308         also .NET compatible.
 2309 
 2310     (b) A recursion or subroutine call to a named group can now be defined as
 2311         (?&name) as well as (?P>name).
 2312 
 2313     (c) A backreference to a named group can now be defined as \k<name> or
 2314         \k'name' as well as (?P=name). The new forms, as well as being in Perl
 2315         5.10, are also .NET compatible.
 2316 
 2317     (d) A conditional reference to a named group can now use the syntax
 2318         (?(<name>) or (?('name') as well as (?(name).
 2319 
 2320     (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
 2321         groups (named and numbered) that are never evaluated inline, but can be
 2322         called as "subroutines" from elsewhere. In effect, the DEFINE condition
 2323         is always false. There may be only one alternative in such a group.
 2324 
 2325     (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
 2326         as the simple (?(R). The condition is true only if the most recent
 2327         recursion is that of the given number or name. It does not search out
 2328         through the entire recursion stack.
 2329 
 2330     (g) The escape \gN or \g{N} has been added, where N is a positive or
 2331         negative number, specifying an absolute or relative reference.
 2332 
 2333 35. Tidied to get rid of some further signed/unsigned compiler warnings and
 2334     some "unreachable code" warnings.
 2335 
 2336 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
 2337     things, this adds five new scripts.
 2338 
 2339 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
 2340     There were also incompatibilities regarding the handling of \Q..\E inside
 2341     character classes, for example with patterns like [\Qa\E-\Qz\E] where the
 2342     hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
 2343 
 2344 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
 2345     matches an empty string, and forcibly breaks the loop. There were bugs in
 2346     this code in non-simple cases. For a pattern such as  ^(a()*)*  matched
 2347     against  aaaa  the result was just "a" rather than "aaaa", for example. Two
 2348     separate and independent bugs (that affected different cases) have been
 2349     fixed.
 2350 
 2351 39. Refactored the code to abolish the use of different opcodes for small
 2352     capturing bracket numbers. This is a tidy that I avoided doing when I
 2353     removed the limit on the number of capturing brackets for 3.5 back in 2001.
 2354     The new approach is not only tidier, it makes it possible to reduce the
 2355     memory needed to fix the previous bug (38).
 2356 
 2357 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
 2358     sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
 2359     processing dot, circumflex, or dollar metacharacters, or #-comments in /x
 2360     mode.
 2361 
 2362 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
 2363     report.
 2364 
 2365 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
 2366     copy construction and assignment in the C++ wrapper.
 2367 
 2368 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
 2369     couple of bugs that could have given wrong results in the "--newline=crlf"
 2370     case.
 2371 
 2372 44. Added a number of casts and did some reorganization of signed/unsigned int
 2373     variables following suggestions from Dair Grant. Also renamed the variable
 2374     "this" as "item" because it is a C++ keyword.
 2375 
 2376 45. Arranged for dftables to add
 2377 
 2378       #include "pcre_internal.h"
 2379 
 2380     to pcre_chartables.c because without it, gcc 4.x may remove the array
 2381     definition from the final binary if PCRE is built into a static library and
 2382     dead code stripping is activated.
 2383 
 2384 46. For an unanchored pattern, if a match attempt fails at the start of a
 2385     newline sequence, and the newline setting is CRLF or ANY, and the next two
 2386     characters are CRLF, advance by two characters instead of one.
 2387 
 2388 
 2389 Version 6.7 04-Jul-06
 2390 ---------------------
 2391 
 2392  1. In order to handle tests when input lines are enormously long, pcretest has
 2393     been re-factored so that it automatically extends its buffers when
 2394     necessary. The code is crude, but this _is_ just a test program. The
 2395     default size has been increased from 32K to 50K.
 2396 
 2397  2. The code in pcre_study() was using the value of the re argument before
 2398     testing it for NULL. (Of course, in any sensible call of the function, it
 2399     won't be NULL.)
 2400 
 2401  3. The memmove() emulation function in pcre_internal.h, which is used on
 2402     systems that lack both memmove() and bcopy() - that is, hardly ever -
 2403     was missing a "static" storage class specifier.
 2404 
 2405  4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
 2406     containing an extended class (one that cannot be represented by a bitmap
 2407     because it contains high-valued characters or Unicode property items, e.g.
 2408     [\pZ]). Almost always one would set UTF-8 mode when processing such a
 2409     pattern, but PCRE should not loop if you do not (it no longer does).
 2410     [Detail: two cases were found: (a) a repeated subpattern containing an
 2411     extended class; (b) a recursive reference to a subpattern that followed a
 2412     previous extended class. It wasn't skipping over the extended class
 2413     correctly when UTF-8 mode was not set.]
 2414 
 2415  5. A negated single-character class was not being recognized as fixed-length
 2416     in lookbehind assertions such as (?<=[^f]), leading to an incorrect
 2417     compile error "lookbehind assertion is not fixed length".
 2418 
 2419  6. The RunPerlTest auxiliary script was showing an unexpected difference
 2420     between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
 2421     write a Perl script that can interpret lines of an input file either as
 2422     byte characters or as UTF-8, which is what "perltest" was being required to
 2423     do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
 2424     can't do is switch easily at run time between having the "use utf8;" pragma
 2425     or not. In the end, I fudged it by using the RunPerlTest script to insert
 2426     "use utf8;" explicitly for the UTF-8 tests.
 2427 
 2428  7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
 2429     the end of the subject string, contrary to the documentation and to what
 2430     Perl does. This was true of both matching functions. Now it matches only at
 2431     the start of the subject and immediately after *internal* newlines.
 2432 
 2433  8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
 2434     a pointer to an int instead of a pointer to an unsigned long int. This
 2435     caused problems on 64-bit systems.
 2436 
 2437  9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
 2438     instance of the 'standard' template library not being so standard".
 2439 
 2440 10. There was no check on the number of named subpatterns nor the maximum
 2441     length of a subpattern name. The product of these values is used to compute
 2442     the size of the memory block for a compiled pattern. By supplying a very
 2443     long subpattern name and a large number of named subpatterns, the size
 2444     computation could be caused to overflow. This is now prevented by limiting
 2445     the length of names to 32 characters, and the number of named subpatterns
 2446     to 10,000.
 2447 
 2448 11. Subpatterns that are repeated with specific counts have to be replicated in
 2449     the compiled pattern. The size of memory for this was computed from the
 2450     length of the subpattern and the repeat count. The latter is limited to
 2451     65535, but there was no limit on the former, meaning that integer overflow
 2452     could in principle occur. The compiled length of a repeated subpattern is
 2453     now limited to 30,000 bytes in order to prevent this.
 2454 
 2455 12. Added the optional facility to have named substrings with the same name.
 2456 
 2457 13. Added the ability to use a named substring as a condition, using the
 2458     Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
 2459     are numbers (not recommended). Forward references are permitted.
 2460 
 2461 14. Added forward references in named backreferences (if you see what I mean).
 2462 
 2463 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
 2464     pattern could run off the end of the subject. For example, the pattern
 2465     "(?s)(.{1,5})"8 did this with the subject "ab".
 2466 
 2467 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
 2468     PCRE_CASELESS was set when matching characters that were quantified with ?
 2469     or *.
 2470 
 2471 17. A character class other than a single negated character that had a minimum
 2472     but no maximum quantifier - for example [ab]{6,} - was not handled
 2473     correctly by pce_dfa_exec(). It would match only one character.
 2474 
 2475 18. A valid (though odd) pattern that looked like a POSIX character
 2476     class but used an invalid character after [ (for example [[,abc,]]) caused
 2477     pcre_compile() to give the error "Failed: internal error: code overflow" or
 2478     in some cases to crash with a glibc free() error. This could even happen if
 2479     the pattern terminated after [[ but there just happened to be a sequence of
 2480     letters, a binary zero, and a closing ] in the memory that followed.
 2481 
 2482 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
 2483     over the years. Originally (before any Unicode support), just the bottom 8
 2484     bits were taken. Thus, for example, \500 really meant \100. Nowadays the
 2485     output from "man perlunicode" includes this:
 2486 
 2487       The regular expression compiler produces polymorphic opcodes.  That
 2488       is, the pattern adapts to the data and automatically switches to
 2489       the Unicode character scheme when presented with Unicode data--or
 2490       instead uses a traditional byte scheme when presented with byte
 2491       data.
 2492 
 2493     Sadly, a wide octal escape does not cause a switch, and in a string with
 2494     no other multibyte characters, these octal escapes are treated as before.
 2495     Thus, in Perl, the pattern  /\500/ actually matches \100 but the pattern
 2496     /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
 2497     Unicode string.
 2498 
 2499     I have not perpetrated such confusion in PCRE. Up till now, it took just
 2500     the bottom 8 bits, as in old Perl. I have now made octal escapes with
 2501     values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
 2502     translate to the appropriate multibyte character.
 2503 
 2504 29. Applied some refactoring to reduce the number of warnings from Microsoft
 2505     and Borland compilers. This has included removing the fudge introduced
 2506     seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
 2507     a warning about an unused variable.
 2508 
 2509 21. PCRE has not included VT (character 0x0b) in the set of whitespace
 2510     characters since release 4.0, because Perl (from release 5.004) does not.
 2511     [Or at least, is documented not to: some releases seem to be in conflict
 2512     with the documentation.] However, when a pattern was studied with
 2513     pcre_study() and all its branches started with \s, PCRE still included VT
 2514     as a possible starting character. Of course, this did no harm; it just
 2515     caused an unnecessary match attempt.
 2516 
 2517 22. Removed a now-redundant internal flag bit that recorded the fact that case
 2518     dependency changed within the pattern. This was once needed for "required
 2519     byte" processing, but is no longer used. This recovers a now-scarce options
 2520     bit. Also moved the least significant internal flag bit to the most-
 2521     significant bit of the word, which was not previously used (hangover from
 2522     the days when it was an int rather than a uint) to free up another bit for
 2523     the future.
 2524 
 2525 23. Added support for CRLF line endings as well as CR and LF. As well as the
 2526     default being selectable at build time, it can now be changed at runtime
 2527     via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
 2528     specify that it is scanning data with non-default line endings.
 2529 
 2530 24. Changed the definition of CXXLINK to make it agree with the definition of
 2531     LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
 2532 
 2533 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
 2534     recursions. This makes a big different to stack usage for some patterns.
 2535 
 2536 26. If a subpattern containing a named recursion or subroutine reference such
 2537     as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
 2538     the space required for the compiled pattern went wrong and gave too small a
 2539     value. Depending on the environment, this could lead to "Failed: internal
 2540     error: code overflow at offset 49" or "glibc detected double free or
 2541     corruption" errors.
 2542 
 2543 27. Applied patches from Google (a) to support the new newline modes and (b) to
 2544     advance over multibyte UTF-8 characters in GlobalReplace.
 2545 
 2546 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
 2547     difference for some implementation of PCRE in some Windows version.
 2548 
 2549 29. Added some extra testing facilities to pcretest:
 2550 
 2551     \q<number>   in a data line sets the "match limit" value
 2552     \Q<number>   in a data line sets the "match recursion limt" value
 2553     -S <number>  sets the stack size, where <number> is in megabytes
 2554 
 2555     The -S option isn't available for Windows.
 2556 
 2557 
 2558 Version 6.6 06-Feb-06
 2559 ---------------------
 2560 
 2561  1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
 2562     in pcreposix.h. I have copied the definition from pcre.h.
 2563 
 2564  2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
 2565     because pcre.h is no longer a built file.
 2566 
 2567  3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
 2568     not normally included in the compiled code.
 2569 
 2570 
 2571 Version 6.5 01-Feb-06
 2572 ---------------------
 2573 
 2574  1. When using the partial match feature with pcre_dfa_exec(), it was not
 2575     anchoring the second and subsequent partial matches at the new starting
 2576     point. This could lead to incorrect results. For example, with the pattern
 2577     /1234/, partially matching against "123" and then "a4" gave a match.
 2578 
 2579  2. Changes to pcregrep:
 2580 
 2581     (a) All non-match returns from pcre_exec() were being treated as failures
 2582         to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
 2583         error message is output. Some extra information is given for the
 2584         PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
 2585         probably the only errors that are likely to be caused by users (by
 2586         specifying a regex that has nested indefinite repeats, for instance).
 2587         If there are more than 20 of these errors, pcregrep is abandoned.
 2588 
 2589     (b) A binary zero was treated as data while matching, but terminated the
 2590         output line if it was written out. This has been fixed: binary zeroes
 2591         are now no different to any other data bytes.
 2592 
 2593     (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
 2594         used to set a locale for matching. The --locale=xxxx long option has
 2595         been added (no short equivalent) to specify a locale explicitly on the
 2596         pcregrep command, overriding the environment variables.
 2597 
 2598     (d) When -B was used with -n, some line numbers in the output were one less
 2599         than they should have been.
 2600 
 2601     (e) Added the -o (--only-matching) option.
 2602 
 2603     (f) If -A or -C was used with -c (count only), some lines of context were
 2604         accidentally printed for the final match.
 2605 
 2606     (g) Added the -H (--with-filename) option.
 2607 
 2608     (h) The combination of options -rh failed to suppress file names for files
 2609         that were found from directory arguments.
 2610 
 2611     (i) Added the -D (--devices) and -d (--directories) options.
 2612 
 2613     (j) Added the -F (--fixed-strings) option.
 2614 
 2615     (k) Allow "-" to be used as a file name for -f as well as for a data file.
 2616 
 2617     (l) Added the --colo(u)r option.
 2618 
 2619     (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
 2620         is not present by default.
 2621 
 2622  3. A nasty bug was discovered in the handling of recursive patterns, that is,
 2623     items such as (?R) or (?1), when the recursion could match a number of
 2624     alternatives. If it matched one of the alternatives, but subsequently,
 2625     outside the recursion, there was a failure, the code tried to back up into
 2626     the recursion. However, because of the way PCRE is implemented, this is not
 2627     possible, and the result was an incorrect result from the match.
 2628 
 2629     In order to prevent this happening, the specification of recursion has
 2630     been changed so that all such subpatterns are automatically treated as
 2631     atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
 2632 
 2633  4. I had overlooked the fact that, in some locales, there are characters for
 2634     which isalpha() is true but neither isupper() nor islower() are true. In
 2635     the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
 2636     and ordfeminine) are like this. This affected the treatment of \w and \W
 2637     when they appeared in character classes, but not when they appeared outside
 2638     a character class. The bit map for "word" characters is now created
 2639     separately from the results of isalnum() instead of just taking it from the
 2640     upper, lower, and digit maps. (Plus the underscore character, of course.)
 2641 
 2642  5. The above bug also affected the handling of POSIX character classes such as
 2643     [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
 2644     permanent tables. Instead, the bit maps for such a class were previously
 2645     created as the appropriate unions of the upper, lower, and digit bitmaps.
 2646     Now they are created by subtraction from the [[:word:]] class, which has
 2647     its own bitmap.
 2648 
 2649  6. The [[:blank:]] character class matches horizontal, but not vertical space.
 2650     It is created by subtracting the vertical space characters (\x09, \x0a,
 2651     \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
 2652     subtraction was done in the overall bitmap for a character class, meaning
 2653     that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
 2654     be recognized. This bug has been fixed.
 2655 
 2656  7. Patches from the folks at Google:
 2657 
 2658       (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
 2659       real life, but is still worth protecting against".
 2660 
 2661       (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
 2662       regular expressions".
 2663 
 2664       (c) pcre_scanner.cc: avoid use of std::count() because not all systems
 2665       have it.
 2666 
 2667       (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
 2668       "configure" and the latter not, in order to fix a problem somebody had
 2669       with compiling the Arg class on HP-UX.
 2670 
 2671       (e) Improve the error-handling of the C++ wrapper a little bit.
 2672 
 2673       (f) New tests for checking recursion limiting.
 2674 
 2675  8. The pcre_memmove() function, which is used only if the environment does not
 2676     have a standard memmove() function (and is therefore rarely compiled),
 2677     contained two bugs: (a) use of int instead of size_t, and (b) it was not
 2678     returning a result (though PCRE never actually uses the result).
 2679 
 2680  9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
 2681     large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
 2682     returned instead of calling malloc() with an overflowing number that would
 2683     most likely cause subsequent chaos.
 2684 
 2685 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
 2686 
 2687 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
 2688     with this option is matched, the nmatch and pmatch options of regexec() are
 2689     ignored.
 2690 
 2691 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
 2692     provided in case anyone wants to the the POSIX interface with UTF-8
 2693     strings.
 2694 
 2695 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
 2696     C++ linking (needed for some HP-UX environments).
 2697 
 2698 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
 2699     (unused parameter) and in the pcre_printint() function (omitted "default"
 2700     switch label when the default is to do nothing).
 2701 
 2702 15. Added some code to make it possible, when PCRE is compiled as a C++
 2703     library, to replace subject pointers for pcre_exec() with a smart pointer
 2704     class, thus making it possible to process discontinuous strings.
 2705 
 2706 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
 2707     much the same function. They were added by different people who were trying
 2708     to make PCRE easy to compile on non-Unix systems. It has been suggested
 2709     that PCRE_EXPORT be abolished now that there is more automatic apparatus
 2710     for compiling on Windows systems. I have therefore replaced it with
 2711     PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
 2712     defaults to "extern" for C or "extern C" for C++, which works fine on
 2713     Unix-like systems. It is now possible to override the value of PCRE_DATA_
 2714     SCOPE with something explicit in config.h. In addition:
 2715 
 2716     (a) pcreposix.h still had just "extern" instead of either of these macros;
 2717         I have replaced it with PCRE_DATA_SCOPE.
 2718 
 2719     (b) Functions such as _pcre_xclass(), which are internal to the library,
 2720         but external in the C sense, all had PCRE_EXPORT in their definitions.
 2721         This is apparently wrong for the Windows case, so I have removed it.
 2722         (It makes no difference on Unix-like systems.)
 2723 
 2724 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
 2725     of recursive calls to match(). This is different to MATCH_LIMIT because
 2726     that limits the total number of calls to match(), not all of which increase
 2727     the depth of recursion. Limiting the recursion depth limits the amount of
 2728     stack (or heap if NO_RECURSE is set) that is used. The default can be set
 2729     when PCRE is compiled, and changed at run time. A patch from Google adds
 2730     this functionality to the C++ interface.
 2731 
 2732 18. Changes to the handling of Unicode character properties:
 2733 
 2734     (a) Updated the table to Unicode 4.1.0.
 2735 
 2736     (b) Recognize characters that are not in the table as "Cn" (undefined).
 2737 
 2738     (c) I revised the way the table is implemented to a much improved format
 2739         which includes recognition of ranges. It now supports the ranges that
 2740         are defined in UnicodeData.txt, and it also amalgamates other
 2741         characters into ranges. This has reduced the number of entries in the
 2742         table from around 16,000 to around 3,000, thus reducing its size
 2743         considerably. I realized I did not need to use a tree structure after
 2744         all - a binary chop search is just as efficient. Having reduced the
 2745         number of entries, I extended their size from 6 bytes to 8 bytes to
 2746         allow for more data.
 2747 
 2748     (d) Added support for Unicode script names via properties such as \p{Han}.
 2749 
 2750 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
 2751     matching that character.
 2752 
 2753 20. When matching a repeated Unicode property with a minimum greater than zero,
 2754     (for example \pL{2,}), PCRE could look past the end of the subject if it
 2755     reached it while seeking the minimum number of characters. This could
 2756     happen only if some of the characters were more than one byte long, because
 2757     there is a check for at least the minimum number of bytes.
 2758 
 2759 21. Refactored the implementation of \p and \P so as to be more general, to
 2760     allow for more different types of property in future. This has changed the
 2761     compiled form incompatibly. Anybody with saved compiled patterns that use
 2762     \p or \P will have to recompile them.
 2763 
 2764 22. Added "Any" and "L&" to the supported property types.
 2765 
 2766 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
 2767     but give a compile time error if the value is greater than 0xff.
 2768 
 2769 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
 2770     accidentally not being installed or uninstalled.
 2771 
 2772 25. The pcre.h file was built from pcre.h.in, but the only changes that were
 2773     made were to insert the current release number. This seemed silly, because
 2774     it made things harder for people building PCRE on systems that don't run
 2775     "configure". I have turned pcre.h into a distributed file, no longer built
 2776     by "configure", with the version identification directly included. There is
 2777     no longer a pcre.h.in file.
 2778 
 2779     However, this change necessitated a change to the pcre-config script as
 2780     well. It is built from pcre-config.in, and one of the substitutions was the
 2781     release number. I have updated configure.ac so that ./configure now finds
 2782     the release number by grepping pcre.h.
 2783 
 2784 26. Added the ability to run the tests under valgrind.
 2785 
 2786 
 2787 Version 6.4 05-Sep-05
 2788 ---------------------
 2789 
 2790  1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
 2791     "--" to be printed when multiple files were scanned, even when none of the
 2792     -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
 2793     consider it to be a bug, and have restored the previous behaviour.
 2794 
 2795  2. A couple of code tidies to get rid of compiler warnings.
 2796 
 2797  3. The pcretest program used to cheat by referring to symbols in the library
 2798     whose names begin with _pcre_. These are internal symbols that are not
 2799     really supposed to be visible externally, and in some environments it is
 2800     possible to suppress them. The cheating is now confined to including
 2801     certain files from the library's source, which is a bit cleaner.
 2802 
 2803  4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
 2804     file's purpose clearer.
 2805 
 2806  5. Reorganized pcre_ucp_findchar().
 2807 
 2808 
 2809 Version 6.3 15-Aug-05
 2810 ---------------------
 2811 
 2812  1. The file libpcre.pc.in did not have general read permission in the tarball.
 2813 
 2814  2. There were some problems when building without C++ support:
 2815 
 2816     (a) If C++ support was not built, "make install" and "make test" still
 2817         tried to test it.
 2818 
 2819     (b) There were problems when the value of CXX was explicitly set. Some
 2820         changes have been made to try to fix these, and ...
 2821 
 2822     (c) --disable-cpp can now be used to explicitly disable C++ support.
 2823 
 2824     (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
 2825         backslash in a target when C++ was disabled. This confuses some
 2826         versions of "make", apparently. Using an intermediate variable solves
 2827         this. (Same for CPP_LOBJ.)
 2828 
 2829  3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
 2830     (non-Windows) now includes $(CFLAGS) because these flags are sometimes
 2831     necessary on certain architectures.
 2832 
 2833  4. Added a setting of -export-symbols-regex to the link command to remove
 2834     those symbols that are exported in the C sense, but actually are local
 2835     within the library, and not documented. Their names all begin with
 2836     "_pcre_". This is not a perfect job, because (a) we have to except some
 2837     symbols that pcretest ("illegally") uses, and (b) the facility isn't always
 2838     available (and never for static libraries). I have made a note to try to
 2839     find a way round (a) in the future.
 2840 
 2841 
 2842 Version 6.2 01-Aug-05
 2843 ---------------------
 2844 
 2845  1. There was no test for integer overflow of quantifier values. A construction
 2846     such as {1111111111111111} would give undefined results. What is worse, if
 2847     a minimum quantifier for a parenthesized subpattern overflowed and became
 2848     negative, the calculation of the memory size went wrong. This could have
 2849     led to memory overwriting.
 2850 
 2851  2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
 2852 
 2853  3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
 2854     operating environments where this matters.
 2855 
 2856  4. Applied Giuseppe Maxia's patch to add additional features for controlling
 2857     PCRE options from within the C++ wrapper.
 2858 
 2859  5. Named capturing subpatterns were not being correctly counted when a pattern
 2860     was compiled. This caused two problems: (a) If there were more than 100
 2861     such subpatterns, the calculation of the memory needed for the whole
 2862     compiled pattern went wrong, leading to an overflow error. (b) Numerical
 2863     back references of the form \12, where the number was greater than 9, were
 2864     not recognized as back references, even though there were sufficient
 2865     previous subpatterns.
 2866 
 2867  6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
 2868     versions of gcc, e.g. 2.95.4.
 2869 
 2870 
 2871 Version 6.1 21-Jun-05
 2872 ---------------------
 2873 
 2874  1. There was one reference to the variable "posix" in pcretest.c that was not
 2875     surrounded by "#if !defined NOPOSIX".
 2876 
 2877  2. Make it possible to compile pcretest without DFA support, UTF8 support, or
 2878     the cross-check on the old pcre_info() function, for the benefit of the
 2879     cut-down version of PCRE that is currently imported into Exim.
 2880 
 2881  3. A (silly) pattern starting with (?i)(?-i) caused an internal space
 2882     allocation error. I've done the easy fix, which wastes 2 bytes for sensible
 2883     patterns that start (?i) but I don't think that matters. The use of (?i) is
 2884     just an example; this all applies to the other options as well.
 2885 
 2886  4. Since libtool seems to echo the compile commands it is issuing, the output
 2887     from "make" can be reduced a bit by putting "@" in front of each libtool
 2888     compile command.
 2889 
 2890  5. Patch from the folks at Google for configure.in to be a bit more thorough
 2891     in checking for a suitable C++ installation before trying to compile the
 2892     C++ stuff. This should fix a reported problem when a compiler was present,
 2893     but no suitable headers.
 2894 
 2895  6. The man pages all had just "PCRE" as their title. I have changed them to
 2896     be the relevant file name. I have also arranged that these names are
 2897     retained in the file doc/pcre.txt, which is a concatenation in text format
 2898     of all the man pages except the little individual ones for each function.
 2899 
 2900  7. The NON-UNIX-USE file had not been updated for the different set of source
 2901     files that come with release 6. I also added a few comments about the C++
 2902     wrapper.
 2903 
 2904 
 2905 Version 6.0 07-Jun-05
 2906 ---------------------
 2907 
 2908  1. Some minor internal re-organization to help with my DFA experiments.
 2909 
 2910  2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
 2911     didn't matter for the library itself when fully configured, but did matter
 2912     when compiling without UCP support, or within Exim, where the ucp files are
 2913     not imported.
 2914 
 2915  3. Refactoring of the library code to split up the various functions into
 2916     different source modules. The addition of the new DFA matching code (see
 2917     below) to a single monolithic source would have made it really too
 2918     unwieldy, quite apart from causing all the code to be include in a
 2919     statically linked application, when only some functions are used. This is
 2920     relevant even without the DFA addition now that patterns can be compiled in
 2921     one application and matched in another.
 2922 
 2923     The downside of splitting up is that there have to be some external
 2924     functions and data tables that are used internally in different modules of
 2925     the library but which are not part of the API. These have all had their
 2926     names changed to start with "_pcre_" so that they are unlikely to clash
 2927     with other external names.
 2928 
 2929  4. Added an alternate matching function, pcre_dfa_exec(), which matches using
 2930     a different (DFA) algorithm. Although it is slower than the original
 2931     function, it does have some advantages for certain types of matching
 2932     problem.
 2933 
 2934  5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
 2935     including restarting after a partial match.
 2936 
 2937  6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
 2938     defined when compiling for Windows was sent to me. I have put it into the
 2939     code, though I have no means of testing or verifying it.
 2940 
 2941  7. Added the pcre_refcount() auxiliary function.
 2942 
 2943  8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
 2944     match before or at the first newline in the subject string. In pcretest,
 2945     the /f option on a pattern can be used to set this.
 2946 
 2947  9. A repeated \w when used in UTF-8 mode with characters greater than 256
 2948     would behave wrongly. This has been present in PCRE since release 4.0.
 2949 
 2950 10. A number of changes to the pcregrep command:
 2951 
 2952     (a) Refactored how -x works; insert ^(...)$ instead of setting
 2953         PCRE_ANCHORED and checking the length, in preparation for adding
 2954         something similar for -w.
 2955 
 2956     (b) Added the -w (match as a word) option.
 2957 
 2958     (c) Refactored the way lines are read and buffered so as to have more
 2959         than one at a time available.
 2960 
 2961     (d) Implemented a pcregrep test script.
 2962 
 2963     (e) Added the -M (multiline match) option. This allows patterns to match
 2964         over several lines of the subject. The buffering ensures that at least
 2965         8K, or the rest of the document (whichever is the shorter) is available
 2966         for matching (and similarly the previous 8K for lookbehind assertions).
 2967 
 2968     (f) Changed the --help output so that it now says
 2969 
 2970           -w, --word-regex(p)
 2971 
 2972         instead of two lines, one with "regex" and the other with "regexp"
 2973         because that confused at least one person since the short forms are the
 2974         same. (This required a bit of code, as the output is generated
 2975         automatically from a table. It wasn't just a text change.)
 2976 
 2977     (g) -- can be used to terminate pcregrep options if the next thing isn't an
 2978         option but starts with a hyphen. Could be a pattern or a path name
 2979         starting with a hyphen, for instance.
 2980 
 2981     (h) "-" can be given as a file name to represent stdin.
 2982 
 2983     (i) When file names are being printed, "(standard input)" is used for
 2984         the standard input, for compatibility with GNU grep. Previously
 2985         "<stdin>" was used.
 2986 
 2987     (j) The option --label=xxx can be used to supply a name to be used for
 2988         stdin when file names are being printed. There is no short form.
 2989 
 2990     (k) Re-factored the options decoding logic because we are going to add
 2991         two more options that take data. Such options can now be given in four
 2992         different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
 2993 
 2994     (l) Added the -A, -B, and -C options for requesting that lines of context
 2995         around matches be printed.
 2996 
 2997     (m) Added the -L option to print the names of files that do not contain
 2998         any matching lines, that is, the complement of -l.
 2999 
 3000     (n) The return code is 2 if any file cannot be opened, but pcregrep does
 3001         continue to scan other files.
 3002 
 3003     (o) The -s option was incorrectly implemented. For compatibility with other
 3004         greps, it now suppresses the error message for a non-existent or non-
 3005         accessible file (but not the return code). There is a new option called
 3006         -q that suppresses the output of matching lines, which was what -s was
 3007         previously doing.
 3008 
 3009     (p) Added --include and --exclude options to specify files for inclusion
 3010         and exclusion when recursing.
 3011 
 3012 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
 3013     Hopefully, it now does.
 3014 
 3015 12. Missing cast in pcre_study().
 3016 
 3017 13. Added an "uninstall" target to the makefile.
 3018 
 3019 14. Replaced "extern" in the function prototypes in Makefile.in with
 3020     "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
 3021     world, but is set differently for Windows.
 3022 
 3023 15. Added a second compiling function called pcre_compile2(). The only
 3024     difference is that it has an extra argument, which is a pointer to an
 3025     integer error code. When there is a compile-time failure, this is set
 3026     non-zero, in addition to the error test pointer being set to point to an
 3027     error message. The new argument may be NULL if no error number is required
 3028     (but then you may as well call pcre_compile(), which is now just a
 3029     wrapper). This facility is provided because some applications need a
 3030     numeric error indication, but it has also enabled me to tidy up the way
 3031     compile-time errors are handled in the POSIX wrapper.
 3032 
 3033 16. Added VPATH=.libs to the makefile; this should help when building with one
 3034     prefix path and installing with another. (Or so I'm told by someone who
 3035     knows more about this stuff than I do.)
 3036 
 3037 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
 3038     passes PCRE_DOTALL to the pcre_compile() function, making the "." character
 3039     match everything, including newlines. This is not POSIX-compatible, but
 3040     somebody wanted the feature. From pcretest it can be activated by using
 3041     both the P and the s flags.
 3042 
 3043 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
 3044 
 3045 19. libpcre.pc was being incorrectly installed as executable.
 3046 
 3047 20. A couple of places in pcretest check for end-of-line by looking for '\n';
 3048     it now also looks for '\r' so that it will work unmodified on Windows.
 3049 
 3050 21. Added Google's contributed C++ wrapper to the distribution.
 3051 
 3052 22. Added some untidy missing memory free() calls in pcretest, to keep
 3053     Electric Fence happy when testing.
 3054 
 3055 
 3056 
 3057 Version 5.0 13-Sep-04
 3058 ---------------------
 3059 
 3060  1. Internal change: literal characters are no longer packed up into items
 3061     containing multiple characters in a single byte-string. Each character
 3062     is now matched using a separate opcode. However, there may be more than one
 3063     byte in the character in UTF-8 mode.
 3064 
 3065  2. The pcre_callout_block structure has two new fields: pattern_position and
 3066     next_item_length. These contain the offset in the pattern to the next match
 3067     item, and its length, respectively.
 3068 
 3069  3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
 3070     insertion of callouts before each pattern item. Added the /C option to
 3071     pcretest to make use of this.
 3072 
 3073  4. On the advice of a Windows user, the lines
 3074 
 3075       #if defined(_WIN32) || defined(WIN32)
 3076       _setmode( _fileno( stdout ), 0x8000 );
 3077       #endif  /* defined(_WIN32) || defined(WIN32) */
 3078 
 3079     have been added to the source of pcretest. This apparently does useful
 3080     magic in relation to line terminators.
 3081 
 3082  5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
 3083     for the benefit of those environments where the "b" makes a difference.
 3084 
 3085  6. The icc compiler has the same options as gcc, but "configure" doesn't seem
 3086     to know about it. I have put a hack into configure.in that adds in code
 3087     to set GCC=yes if CC=icc. This seems to end up at a point in the
 3088     generated configure script that is early enough to affect the setting of
 3089     compiler options, which is what is needed, but I have no means of testing
 3090     whether it really works. (The user who reported this had patched the
 3091     generated configure script, which of course I cannot do.)
 3092 
 3093     LATER: After change 22 below (new libtool files), the configure script
 3094     seems to know about icc (and also ecc). Therefore, I have commented out
 3095     this hack in configure.in.
 3096 
 3097  7. Added support for pkg-config (2 patches were sent in).
 3098 
 3099  8. Negated POSIX character classes that used a combination of internal tables
 3100     were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
 3101     [[:^ascii]]. Typically, they would match almost any characters. The other
 3102     POSIX classes were not broken in this way.
 3103 
 3104  9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
 3105     to find the match, as PCRE was deluded into thinking that the match had to
 3106     start at the start point or following a newline. The same bug applied to
 3107     patterns with negative forward assertions or any backward assertions
 3108     preceding ".*" at the start, unless the pattern required a fixed first
 3109     character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
 3110 
 3111 10. In UTF-8 mode, when moving forwards in the subject after a failed match
 3112     starting at the last subject character, bytes beyond the end of the subject
 3113     string were read.
 3114 
 3115 11. Renamed the variable "class" as "classbits" to make life easier for C++
 3116     users. (Previously there was a macro definition, but it apparently wasn't
 3117     enough.)
 3118 
 3119 12. Added the new field "tables" to the extra data so that tables can be passed
 3120     in at exec time, or the internal tables can be re-selected. This allows
 3121     a compiled regex to be saved and re-used at a later time by a different
 3122     program that might have everything at different addresses.
 3123 
 3124 13. Modified the pcre-config script so that, when run on Solaris, it shows a
 3125     -R library as well as a -L library.
 3126 
 3127 14. The debugging options of pcretest (-d on the command line or D on a
 3128     pattern) showed incorrect output for anything following an extended class
 3129     that contained multibyte characters and which was followed by a quantifier.
 3130 
 3131 15. Added optional support for general category Unicode character properties
 3132     via the \p, \P, and \X escapes. Unicode property support implies UTF-8
 3133     support. It adds about 90K to the size of the library. The meanings of the
 3134     inbuilt class escapes such as \d and \s have NOT been changed.
 3135 
 3136 16. Updated pcredemo.c to include calls to free() to release the memory for the
 3137     compiled pattern.
 3138 
 3139 17. The generated file chartables.c was being created in the source directory
 3140     instead of in the building directory. This caused the build to fail if the
 3141     source directory was different from the building directory, and was
 3142     read-only.
 3143 
 3144 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
 3145     file. No doubt somebody will tell me if they don't make sense... Also added
 3146     Dan Mooney's comments about building on OpenVMS.
 3147 
 3148 19. Added support for partial matching via the PCRE_PARTIAL option for
 3149     pcre_exec() and the \P data escape in pcretest.
 3150 
 3151 20. Extended pcretest with 3 new pattern features:
 3152 
 3153     (i)   A pattern option of the form ">rest-of-line" causes pcretest to
 3154           write the compiled pattern to the file whose name is "rest-of-line".
 3155           This is a straight binary dump of the data, with the saved pointer to
 3156           the character tables forced to be NULL. The study data, if any, is
 3157           written too. After writing, pcretest reads a new pattern.
 3158 
 3159     (ii)  If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
 3160           compiled pattern from the given file. There must not be any
 3161           occurrences of "<" in the file name (pretty unlikely); if there are,
 3162           pcretest will instead treat the initial "<" as a pattern delimiter.
 3163           After reading in the pattern, pcretest goes on to read data lines as
 3164           usual.
 3165 
 3166     (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
 3167           and 16-bit fields in a compiled pattern, to simulate a pattern that
 3168           was compiled on a host of opposite endianness.
 3169 
 3170 21. The pcre-exec() function can now cope with patterns that were compiled on
 3171     hosts of opposite endianness, with this restriction:
 3172 
 3173       As for any compiled expression that is saved and used later, the tables
 3174       pointer field cannot be preserved; the extra_data field in the arguments
 3175       to pcre_exec() should be used to pass in a tables address if a value
 3176       other than the default internal tables were used at compile time.
 3177 
 3178 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
 3179     now diagnosed as an error. Previously, most of the time, a negative number
 3180     would have been treated as zero, but if in addition "ovector" was passed as
 3181     NULL, a crash could occur.
 3182 
 3183 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
 3184     new versions from the libtool 1.5 distribution (the last one is a copy of
 3185     a file called libtool.m4). This seems to have fixed the need to patch
 3186     "configure" to support Darwin 1.3 (which I used to do). However, I still
 3187     had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
 3188     workstation).
 3189 
 3190 24. Changed the PCRE licence to be the more standard "BSD" licence.
 3191 
 3192 
 3193 Version 4.5 01-Dec-03
 3194 ---------------------
 3195 
 3196  1. There has been some re-arrangement of the code for the match() function so
 3197     that it can be compiled in a version that does not call itself recursively.
 3198     Instead, it keeps those local variables that need separate instances for
 3199     each "recursion" in a frame on the heap, and gets/frees frames whenever it
 3200     needs to "recurse". Keeping track of where control must go is done by means
 3201     of setjmp/longjmp. The whole thing is implemented by a set of macros that
 3202     hide most of the details from the main code, and operates only if
 3203     NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
 3204     "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
 3205     operating.
 3206 
 3207     To make it easier for callers to provide specially tailored get/free
 3208     functions for this usage, two new functions, pcre_stack_malloc, and
 3209     pcre_stack_free, are used. They are always called in strict stacking order,
 3210     and the size of block requested is always the same.
 3211 
 3212     The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
 3213     PCRE has been compiled to use the stack or the heap for recursion. The
 3214     -C option of pcretest uses this to show which version is compiled.
 3215 
 3216     A new data escape \S, is added to pcretest; it causes the amounts of store
 3217     obtained and freed by both kinds of malloc/free at match time to be added
 3218     to the output.
 3219 
 3220  2. Changed the locale test to use "fr_FR" instead of "fr" because that's
 3221     what's available on my current Linux desktop machine.
 3222 
 3223  3. When matching a UTF-8 string, the test for a valid string at the start has
 3224     been extended. If start_offset is not zero, PCRE now checks that it points
 3225     to a byte that is the start of a UTF-8 character. If not, it returns
 3226     PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
 3227     this is necessary because there may be backward assertions in the pattern.
 3228     When matching the same subject several times, it may save resources to use
 3229     PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
 3230 
 3231  4. The code for checking the validity of UTF-8 strings has been tightened so
 3232     that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
 3233     containing "overlong sequences".
 3234 
 3235  5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
 3236     I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
 3237     should have been "&", but it just so happened that all the cases this let
 3238     through by mistake were picked up later in the function.
 3239 
 3240  6. I had used a variable called "isblank" - this is a C99 function, causing
 3241     some compilers to warn. To avoid this, I renamed it (as "blankclass").
 3242 
 3243  7. Cosmetic: (a) only output another newline at the end of pcretest if it is
 3244     prompting; (b) run "./pcretest /dev/null" at the start of the test script
 3245     so the version is shown; (c) stop "make test" echoing "./RunTest".
 3246 
 3247  8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
 3248 
 3249  9. The prototype for memmove() for systems that don't have it was using
 3250     size_t, but the inclusion of the header that defines size_t was later. I've
 3251     moved the #includes for the C headers earlier to avoid this.
 3252 
 3253 10. Added some adjustments to the code to make it easier to compiler on certain
 3254     special systems:
 3255 
 3256       (a) Some "const" qualifiers were missing.
 3257       (b) Added the macro EXPORT before all exported functions; by default this
 3258           is defined to be empty.
 3259       (c) Changed the dftables auxiliary program (that builds chartables.c) so
 3260           that it reads its output file name as an argument instead of writing
 3261           to the standard output and assuming this can be redirected.
 3262 
 3263 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
 3264     class containing characters with values greater than 255, PCRE compilation
 3265     went into a loop.
 3266 
 3267 12. A recursive reference to a subpattern that was within another subpattern
 3268     that had a minimum quantifier of zero caused PCRE to crash. For example,
 3269     (x(y(?2))z)? provoked this bug with a subject that got as far as the
 3270     recursion. If the recursively-called subpattern itself had a zero repeat,
 3271     that was OK.
 3272 
 3273 13. In pcretest, the buffer for reading a data line was set at 30K, but the
 3274     buffer into which it was copied (for escape processing) was still set at
 3275     1024, so long lines caused crashes.
 3276 
 3277 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
 3278     "internal error: code overflow...". This applied to any character class
 3279     that was followed by a possessive quantifier.
 3280 
 3281 15. Modified the Makefile to add libpcre.la as a prerequisite for
 3282     libpcreposix.la because I was told this is needed for a parallel build to
 3283     work.
 3284 
 3285 16. If a pattern that contained .* following optional items at the start was
 3286     studied, the wrong optimizing data was generated, leading to matching
 3287     errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
 3288     matching string must start with a or b or c. The correct conclusion for
 3289     this pattern is that a match can start with any character.
 3290 
 3291 
 3292 Version 4.4 13-Aug-03
 3293 ---------------------
 3294 
 3295  1. In UTF-8 mode, a character class containing characters with values between
 3296     127 and 255 was not handled correctly if the compiled pattern was studied.
 3297     In fixing this, I have also improved the studying algorithm for such
 3298     classes (slightly).
 3299 
 3300  2. Three internal functions had redundant arguments passed to them. Removal
 3301     might give a very teeny performance improvement.
 3302 
 3303  3. Documentation bug: the value of the capture_top field in a callout is *one
 3304     more than* the number of the hightest numbered captured substring.
 3305 
 3306  4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
 3307     in incorrectly linking with a previously installed version. They now link
 3308     explicitly with libpcre.la.
 3309 
 3310  5. configure.in no longer needs to recognize Cygwin specially.
 3311 
 3312  6. A problem in pcre.in for Windows platforms is fixed.
 3313 
 3314  7. If a pattern was successfully studied, and the -d (or /D) flag was given to
 3315     pcretest, it used to include the size of the study block as part of its
 3316     output. Unfortunately, the structure contains a field that has a different
 3317     size on different hardware architectures. This meant that the tests that
 3318     showed this size failed. As the block is currently always of a fixed size,
 3319     this information isn't actually particularly useful in pcretest output, so
 3320     I have just removed it.
 3321 
 3322  8. Three pre-processor statements accidentally did not start in column 1.
 3323     Sadly, there are *still* compilers around that complain, even though
 3324     standard C has not required this for well over a decade. Sigh.
 3325 
 3326  9. In pcretest, the code for checking callouts passed small integers in the
 3327     callout_data field, which is a void * field. However, some picky compilers
 3328     complained about the casts involved for this on 64-bit systems. Now
 3329     pcretest passes the address of the small integer instead, which should get
 3330     rid of the warnings.
 3331 
 3332 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
 3333     both compile and run time, and gives an error if an invalid UTF-8 sequence
 3334     is found. There is a option for disabling this check in cases where the
 3335     string is known to be correct and/or the maximum performance is wanted.
 3336 
 3337 11. In response to a bug report, I changed one line in Makefile.in from
 3338 
 3339         -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
 3340     to
 3341         -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
 3342 
 3343     to look similar to other lines, but I have no way of telling whether this
 3344     is the right thing to do, as I do not use Windows. No doubt I'll get told
 3345     if it's wrong...
 3346 
 3347 
 3348 Version 4.3 21-May-03
 3349 ---------------------
 3350 
 3351 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
 3352    Makefile.
 3353 
 3354 2. Some refactoring to improve the quality of the code:
 3355 
 3356    (i)   The utf8_table... variables are now declared "const".
 3357 
 3358    (ii)  The code for \cx, which used the "case flipping" table to upper case
 3359          lower case letters, now just substracts 32. This is ASCII-specific,
 3360          but the whole concept of \cx is ASCII-specific, so it seems
 3361          reasonable.
 3362 
 3363    (iii) PCRE was using its character types table to recognize decimal and
 3364          hexadecimal digits in the pattern. This is silly, because it handles
 3365          only 0-9, a-f, and A-F, but the character types table is locale-
 3366          specific, which means strange things might happen. A private
 3367          table is now used for this - though it costs 256 bytes, a table is
 3368          much faster than multiple explicit tests. Of course, the standard
 3369          character types table is still used for matching digits in subject
 3370          strings against \d.
 3371 
 3372    (iv)  Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
 3373          ending in _t are). So I've renamed it as ESC_tee.
 3374 
 3375 3. The first argument for regexec() in the POSIX wrapper should have been
 3376    defined as "const".
 3377 
 3378 4. Changed pcretest to use malloc() for its buffers so that they can be
 3379    Electric Fenced for debugging.
 3380 
 3381 5. There were several places in the code where, in UTF-8 mode, PCRE would try
 3382    to read one or more bytes before the start of the subject string. Often this
 3383    had no effect on PCRE's behaviour, but in some circumstances it could
 3384    provoke a segmentation fault.
 3385 
 3386 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
 3387    to try to read one or more bytes before the start of the subject string.
 3388 
 3389 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
 3390    UTF-8 support could misbehave in various ways if the subject string
 3391    contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
 3392    area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
 3393    back over UTF-8 characters.)
 3394 
 3395 
 3396 Version 4.2 14-Apr-03
 3397 ---------------------
 3398 
 3399 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
 3400 
 3401 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
 3402      [ON_WINDOWS]: new variable, "#" on non-Windows platforms
 3403      [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
 3404      [WIN_PREFIX]: new variable, "cyg" for Cygwin
 3405      * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
 3406        and BUILD_EXEEXT
 3407      Note: automatic setting of the BUILD variables is not yet working
 3408      set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
 3409        compile-time but not at link-time
 3410      [LINK]: use for linking executables only
 3411      make different versions for Windows and non-Windows
 3412      [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
 3413        libraries
 3414      [LINK_FOR_BUILD]: new variable
 3415      [OBJEXT]: use throughout
 3416      [EXEEXT]: use throughout
 3417      <winshared>: new target
 3418      <wininstall>: new target
 3419      <dftables.o>: use native compiler
 3420      <dftables>: use native linker
 3421      <install>: handle Windows platform correctly
 3422      <clean>: ditto
 3423      <check>: ditto
 3424      copy DLL to top builddir before testing
 3425 
 3426    As part of these changes, -no-undefined was removed again. This was reported
 3427    to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
 3428    in any case.
 3429 
 3430 3. Some tidies to get rid of compiler warnings:
 3431 
 3432    . In the match_data structure, match_limit was an unsigned long int, whereas
 3433      match_call_count was an int. I've made them both unsigned long ints.
 3434 
 3435    . In pcretest the fact that a const uschar * doesn't automatically cast to
 3436      a void * provoked a warning.
 3437 
 3438    . Turning on some more compiler warnings threw up some "shadow" variables
 3439      and a few more missing casts.
 3440 
 3441 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
 3442    option, a class that contained a single character with a value between 128
 3443    and 255 (e.g. /[\xFF]/) caused PCRE to crash.
 3444 
 3445 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
 3446    option, a class that contained several characters, but with at least one
 3447    whose value was between 128 and 255 caused PCRE to crash.
 3448 
 3449 
 3450 Version 4.1 12-Mar-03
 3451 ---------------------
 3452 
 3453 1. Compiling with gcc -pedantic found a couple of places where casts were
 3454 needed, and a string in dftables.c that was longer than standard compilers are
 3455 required to support.
 3456 
 3457 2. Compiling with Sun's compiler found a few more places where the code could
 3458 be tidied up in order to avoid warnings.
 3459 
 3460 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
 3461 first of these names is deprecated in the latest Autoconf in favour of the name
 3462 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
 3463 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
 3464 analogy I have changed it to CFLAGS_FOR_BUILD.
 3465 
 3466 4. Added -no-undefined to the linking command in the Makefile, because this is
 3467 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
 3468 linking step for the pcreposix library.
 3469 
 3470 5. PCRE was failing to diagnose the case of two named groups with the same
 3471 name.
 3472 
 3473 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
 3474 literal character that is needed in the subject for a match, and scans along to
 3475 ensure that it is present before embarking on the full matching process. This
 3476 saves time in cases of nested unlimited repeats that are never going to match.
 3477 Problem: the scan can take a lot of time if the subject is very long (e.g.
 3478 megabytes), thus penalizing straightforward matches. It is now done only if the
 3479 amount of subject to be scanned is less than 1000 bytes.
 3480 
 3481 7. A lesser problem with the same optimization is that it was recording the
 3482 first character of an anchored pattern as "needed", thus provoking a search
 3483 right along the subject, even when the first match of the pattern was going to
 3484 fail. The "needed" character is now not set for anchored patterns, unless it
 3485 follows something in the pattern that is of non-fixed length. Thus, it still
 3486 fulfils its original purpose of finding quick non-matches in cases of nested
 3487 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
 3488 
 3489 
 3490 Version 4.0 17-Feb-03
 3491 ---------------------
 3492 
 3493 1. If a comment in an extended regex that started immediately after a meta-item
 3494 extended to the end of string, PCRE compiled incorrect data. This could lead to
 3495 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
 3496 
 3497 2. Moved to autoconf 2.53 and libtool 1.4.2.
 3498 
 3499 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
 3500 the special perltest8 script is no longer needed - all the tests can be run
 3501 from a single perltest script.
 3502 
 3503 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
 3504 by \s. It has now been removed in PCRE. This means it isn't recognized as
 3505 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
 3506 class [:space:] *does* include VT, thereby creating a mess.
 3507 
 3508 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
 3509 space and tab.
 3510 
 3511 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
 3512 its new features into the main test script, reducing the number of scripts.
 3513 
 3514 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
 3515 were backward compatible, and made the (?i) apply to the whole pattern, as if
 3516 /i were given. Now it behaves more logically, and applies the option setting
 3517 only to what follows. PCRE has been changed to follow suit. However, if it
 3518 finds options settings right at the start of the pattern, it extracts them into
 3519 the global options, as before. Thus, they show up in the info data.
 3520 
 3521 8. Added support for the \Q...\E escape sequence. Characters in between are
 3522 treated as literals. This is slightly different from Perl in that $ and @ are
 3523 also handled as literals inside the quotes. In Perl, they will cause variable
 3524 interpolation. Note the following examples:
 3525 
 3526     Pattern            PCRE matches      Perl matches
 3527 
 3528     \Qabc$xyz\E        abc$xyz           abc followed by the contents of $xyz
 3529     \Qabc\$xyz\E       abc\$xyz          abc\$xyz
 3530     \Qabc\E\$\Qxyz\E   abc$xyz           abc$xyz
 3531 
 3532 For compatibility with Perl, \Q...\E sequences are recognized inside character
 3533 classes as well as outside them.
 3534 
 3535 9. Re-organized 3 code statements in pcretest to avoid "overflow in
 3536 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
 3537 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
 3538 signed/unsigned warnings.
 3539 
 3540 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
 3541 option for pcretest, so I've replaced it by a simple function that does just
 3542 that job.
 3543 
 3544 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
 3545 "pcregrep -".
 3546 
 3547 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
 3548 Java package. This provides some syntactic sugar for simple cases of what my
 3549 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
 3550 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
 3551 item, you can use this simplified notation. Note that only makes sense with
 3552 greedy quantifiers. Consequently, the use of the possessive quantifier forces
 3553 greediness, whatever the setting of the PCRE_UNGREEDY option.
 3554 
 3555 13. A change of greediness default within a pattern was not taking effect at
 3556 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
 3557 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
 3558 was abstracted outside.
 3559 
 3560 14. PCRE now supports the \G assertion. It is true when the current matching
 3561 position is at the start point of the match. This differs from \A when the
 3562 starting offset is non-zero. Used with the /g option of pcretest (or similar
 3563 code), it works in the same way as it does for Perl's /g option. If all
 3564 alternatives of a regex begin with \G, the expression is anchored to the start
 3565 match position, and the "anchored" flag is set in the compiled expression.
 3566 
 3567 15. Some bugs concerning the handling of certain option changes within patterns
 3568 have been fixed. These applied to options other than (?ims). For example,
 3569 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
 3570 been the other way round. Some of this was related to change 7 above.
 3571 
 3572 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
 3573 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
 3574 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
 3575 POSIX classes only within a class (e.g. /[[:alpha:]]/).
 3576 
 3577 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
 3578 mode. Unlike ".", it always matches newline, whatever the setting of
 3579 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
 3580 assertions. Perl allows it, but it doesn't (in general) work because it can't
 3581 calculate the length of the lookbehind. At least, that's the case for Perl
 3582 5.8.0 - I've been told they are going to document that it doesn't work in
 3583 future.
 3584 
 3585 18. Added an error diagnosis for escapes that PCRE does not support: these are
 3586 \L, \l, \N, \P, \p, \U, \u, and \X.
 3587 
 3588 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
 3589 reading past the end of the pattern in cases such as /[abcd/.
 3590 
 3591 20. PCRE was getting more memory than necessary for patterns with classes that
 3592 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
 3593 
 3594 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
 3595 compiling PCRE for use with Virtual Pascal.
 3596 
 3597 22. Small fix to the Makefile to make it work properly if the build is done
 3598 outside the source tree.
 3599 
 3600 23. Added a new extension: a condition to go with recursion. If a conditional
 3601 subpattern starts with (?(R) the "true" branch is used if recursion has
 3602 happened, whereas the "false" branch is used only at the top level.
 3603 
 3604 24. When there was a very long string of literal characters (over 255 bytes
 3605 without UTF support, over 250 bytes with UTF support), the computation of how
 3606 much memory was required could be incorrect, leading to segfaults or other
 3607 strange effects.
 3608 
 3609 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
 3610 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
 3611 there was a subsequent back reference to those brackets. This meant that, for
 3612 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
 3613 possible to check for precisely this case. All we can do is abandon the
 3614 optimization if .* occurs inside capturing brackets when there are any back
 3615 references whatsoever. (See below for a better fix that came later.)
 3616 
 3617 26. The handling of the optimization for finding the first character of a
 3618 non-anchored pattern, and for finding a character that is required later in the
 3619 match were failing in some cases. This didn't break the matching; it just
 3620 failed to optimize when it could. The way this is done has been re-implemented.
 3621 
 3622 27. Fixed typo in error message for invalid (?R item (it said "(?p").
 3623 
 3624 28. Added a new feature that provides some of the functionality that Perl
 3625 provides with (?{...}). The facility is termed a "callout". The way it is done
 3626 in PCRE is for the caller to provide an optional function, by setting
 3627 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
 3628 global variable. By default it is unset, which disables all calling out. To get
 3629 the function called, the regex must include (?C) at appropriate points. This
 3630 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
 3631 This provides a means of identifying different callout points. When PCRE
 3632 reaches such a point in the regex, if pcre_callout has been set, the external
 3633 function is called. It is provided with data in a structure called
 3634 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
 3635 matching continues; if it returns a non-zero value, the match at the current
 3636 point fails. However, backtracking will occur if possible. [This was changed
 3637 later and other features added - see item 49 below.]
 3638 
 3639 29. pcretest is upgraded to test the callout functionality. It provides a
 3640 callout function that displays information. By default, it shows the start of
 3641 the match and the current position in the text. There are some new data escapes
 3642 to vary what happens:
 3643 
 3644     \C+         in addition, show current contents of captured substrings
 3645     \C-         do not supply a callout function
 3646     \C!n        return 1 when callout number n is reached
 3647     \C!n!m      return 1 when callout number n is reached for the mth time
 3648 
 3649 30. If pcregrep was called with the -l option and just a single file name, it
 3650 output "<stdin>" if a match was found, instead of the file name.
 3651 
 3652 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
 3653 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
 3654 pcre_exec(). This saves a malloc/free per call. The default value of
 3655 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
 3656 when configuring.
 3657 
 3658 32. The default maximum size of a compiled pattern is 64K. There have been a
 3659 few cases of people hitting this limit. The code now uses macros to handle the
 3660 storing of links as offsets within the compiled pattern. It defaults to 2-byte
 3661 links, but this can be changed to 3 or 4 bytes by --with-link-size when
 3662 configuring. Tests 2 and 5 work only with 2-byte links because they output
 3663 debugging information about compiled patterns.
 3664 
 3665 33. Internal code re-arrangements:
 3666 
 3667 (a) Moved the debugging function for printing out a compiled regex into
 3668     its own source file (printint.c) and used #include to pull it into
 3669     pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
 3670     separate copies.
 3671 
 3672 (b) Defined the list of op-code names for debugging as a macro in
 3673     internal.h so that it is next to the definition of the opcodes.
 3674 
 3675 (c) Defined a table of op-code lengths for simpler skipping along compiled
 3676     code. This is again a macro in internal.h so that it is next to the
 3677     definition of the opcodes.
 3678 
 3679 34. Added support for recursive calls to individual subpatterns, along the
 3680 lines of Robin Houston's patch (but implemented somewhat differently).
 3681 
 3682 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
 3683 allow it to read and process whole directories in Win32. This code was
 3684 contributed by Lionel Fourquaux; it has not been tested by me.
 3685 
 3686 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
 3687 used to name a group. Names consist of alphanumerics and underscores, and must
 3688 be unique. Back references use the syntax (?P=name) and recursive calls use
 3689 (?P>name) which is a PCRE extension to the Python extension. Groups still have
 3690 numbers. The function pcre_fullinfo() can be used after compilation to extract
 3691 a name/number map. There are three relevant calls:
 3692 
 3693   PCRE_INFO_NAMEENTRYSIZE        yields the size of each entry in the map
 3694   PCRE_INFO_NAMECOUNT            yields the number of entries
 3695   PCRE_INFO_NAMETABLE            yields a pointer to the map.
 3696 
 3697 The map is a vector of fixed-size entries. The size of each entry depends on
 3698 the length of the longest name used. The first two bytes of each entry are the
 3699 group number, most significant byte first. There follows the corresponding
 3700 name, zero terminated. The names are in alphabetical order.
 3701 
 3702 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
 3703 case instead of 255. Making it the same both with and without UTF-8 support
 3704 means that the same test output works with both.
 3705 
 3706 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
 3707 calling malloc() with a zero argument.
 3708 
 3709 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
 3710 optimization. I've improved things by keeping a bitmap of backreferences with
 3711 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
 3712 fact referenced, the optimization can be applied. It is unlikely that a
 3713 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
 3714 the match to follow \n) will appear inside brackets with a number greater than
 3715 31, but if it does, any back reference > 31 suppresses the optimization.
 3716 
 3717 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
 3718 of disabling numbered capturing parentheses. Any opening parenthesis that is
 3719 not followed by ? behaves as if it were followed by ?: but named parentheses
 3720 can still be used for capturing (and they will acquire numbers in the usual
 3721 way).
 3722 
 3723 41. Redesigned the return codes from the match() function into yes/no/error so
 3724 that errors can be passed back from deep inside the nested calls. A malloc
 3725 failure while inside a recursive subpattern call now causes the
 3726 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
 3727 
 3728 42. It is now possible to set a limit on the number of times the match()
 3729 function is called in a call to pcre_exec(). This facility makes it possible to
 3730 limit the amount of recursion and backtracking, though not in a directly
 3731 obvious way, because the match() function is used in a number of different
 3732 circumstances. The count starts from zero for each position in the subject
 3733 string (for non-anchored patterns). The default limit is, for compatibility, a
 3734 large number, namely 10 000 000. You can change this in two ways:
 3735 
 3736 (a) When configuring PCRE before making, you can use --with-match-limit=n
 3737     to set a default value for the compiled library.
 3738 
 3739 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
 3740     a different value is set. See 45 below.
 3741 
 3742 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
 3743 
 3744 43. Added a new function pcre_config(int, void *) to enable run-time extraction
 3745 of things that can be changed at compile time. The first argument specifies
 3746 what is wanted and the second points to where the information is to be placed.
 3747 The current list of available information is:
 3748 
 3749   PCRE_CONFIG_UTF8
 3750 
 3751 The output is an integer that is set to one if UTF-8 support is available;
 3752 otherwise it is set to zero.
 3753 
 3754   PCRE_CONFIG_NEWLINE
 3755 
 3756 The output is an integer that it set to the value of the code that is used for
 3757 newline. It is either LF (10) or CR (13).
 3758 
 3759   PCRE_CONFIG_LINK_SIZE
 3760 
 3761 The output is an integer that contains the number of bytes used for internal
 3762 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
 3763 
 3764   PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
 3765 
 3766 The output is an integer that contains the threshold above which the POSIX
 3767 interface uses malloc() for output vectors. See item 31 above.
 3768 
 3769   PCRE_CONFIG_MATCH_LIMIT
 3770 
 3771 The output is an unsigned integer that contains the default limit of the number
 3772 of match() calls in a pcre_exec() execution. See 42 above.
 3773 
 3774 44. pcretest has been upgraded by the addition of the -C option. This causes it
 3775 to extract all the available output from the new pcre_config() function, and to
 3776 output it. The program then exits immediately.
 3777 
 3778 45. A need has arisen to pass over additional data with calls to pcre_exec() in
 3779 order to support additional features. One way would have been to define
 3780 pcre_exec2() (for example) with extra arguments, but this would not have been
 3781 extensible, and would also have required all calls to the original function to
 3782 be mapped to the new one. Instead, I have chosen to extend the mechanism that
 3783 is used for passing in "extra" data from pcre_study().
 3784 
 3785 The pcre_extra structure is now exposed and defined in pcre.h. It currently
 3786 contains the following fields:
 3787 
 3788   flags         a bitmap indicating which of the following fields are set
 3789   study_data    opaque data from pcre_study()
 3790   match_limit   a way of specifying a limit on match() calls for a specific
 3791                   call to pcre_exec()
 3792   callout_data  data for callouts (see 49 below)
 3793 
 3794 The flag bits are also defined in pcre.h, and are
 3795 
 3796   PCRE_EXTRA_STUDY_DATA
 3797   PCRE_EXTRA_MATCH_LIMIT
 3798   PCRE_EXTRA_CALLOUT_DATA
 3799 
 3800 The pcre_study() function now returns one of these new pcre_extra blocks, with
 3801 the actual study data pointed to by the study_data field, and the
 3802 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
 3803 before. That is, this change is entirely upwards-compatible and requires no
 3804 change to existing code.
 3805 
 3806 If you want to pass in additional data to pcre_exec(), you can either place it
 3807 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
 3808 block.
 3809 
 3810 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
 3811 data string contains the escape sequence \M, pcretest calls pcre_exec() several
 3812 times with different match limits, until it finds the minimum value needed for
 3813 pcre_exec() to complete. The value is then output. This can be instructive; for
 3814 most simple matches the number is quite small, but for pathological cases it
 3815 gets very large very quickly.
 3816 
 3817 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
 3818 returns the size of the data block pointed to by the study_data field in a
 3819 pcre_extra block, that is, the value that was passed as the argument to
 3820 pcre_malloc() when PCRE was getting memory in which to place the information
 3821 created by pcre_study(). The fourth argument should point to a size_t variable.
 3822 pcretest has been extended so that this information is shown after a successful
 3823 pcre_study() call when information about the compiled regex is being displayed.
 3824 
 3825 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
 3826 because what follows is always an absolute path. (Later: it turns out that this
 3827 is more than cosmetic for MinGW, because it doesn't like empty path
 3828 components.)
 3829 
 3830 49. Some changes have been made to the callout feature (see 28 above):
 3831 
 3832 (i)  A callout function now has three choices for what it returns:
 3833 
 3834        0  =>  success, carry on matching
 3835      > 0  =>  failure at this point, but backtrack if possible
 3836      < 0  =>  serious error, return this value from pcre_exec()
 3837 
 3838      Negative values should normally be chosen from the set of PCRE_ERROR_xxx
 3839      values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
 3840      "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
 3841      use by callout functions. It will never be used by PCRE itself.
 3842 
 3843 (ii) The pcre_extra structure (see 45 above) has a void * field called
 3844      callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
 3845      pcre_callout_block structure has a field of the same name. The contents of
 3846      the field passed in the pcre_extra structure are passed to the callout
 3847      function in the corresponding field in the callout block. This makes it
 3848      easier to use the same callout-containing regex from multiple threads. For
 3849      testing, the pcretest program has a new data escape
 3850 
 3851        \C*n        pass the number n (may be negative) as callout_data
 3852 
 3853      If the callout function in pcretest receives a non-zero value as
 3854      callout_data, it returns that value.
 3855 
 3856 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
 3857 there were some redundant $(CFLAGS) in commands that are now specified as
 3858 $(LINK), which already includes $(CFLAGS).
 3859 
 3860 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
 3861 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
 3862 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
 3863 one-byte characters throughout. Note that case-insensitive matching applies
 3864 only to characters whose values are less than 256. PCRE doesn't support the
 3865 notion of cases for higher-valued characters.
 3866 
 3867 (i)   A character class whose characters are all within 0-255 is handled as
 3868       a bit map, and the map is inverted for negative classes. Previously, a
 3869       character > 255 always failed to match such a class; however it should
 3870       match if the class was a negative one (e.g. [^ab]). This has been fixed.
 3871 
 3872 (ii)  A negated character class with a single character < 255 is coded as
 3873       "not this character" (OP_NOT). This wasn't working properly when the test
 3874       character was multibyte, either singly or repeated.
 3875 
 3876 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
 3877       mode, for example: \x{100}{2,3}.
 3878 
 3879 (iv)  The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
 3880       singly or repeated) now correctly test multibyte characters. However,
 3881       PCRE doesn't recognize any characters with values greater than 255 as
 3882       digits, spaces, or word characters. Such characters always match \D, \S,
 3883       and \W, and never match \d, \s, or \w.
 3884 
 3885 (v)   Classes may now contain characters and character ranges with values
 3886       greater than 255. For example: [ab\x{100}-\x{400}].
 3887 
 3888 (vi)  pcregrep now has a --utf-8 option (synonym -u) which makes it call
 3889       PCRE in UTF-8 mode.
 3890 
 3891 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
 3892 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
 3893 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
 3894 value.)
 3895 
 3896 53. The single man page has become too large. I have therefore split it up into
 3897 a number of separate man pages. These also give rise to individual HTML pages;
 3898 these are now put in a separate directory, and there is an index.html page that
 3899 lists them all. Some hyperlinking between the pages has been installed.
 3900 
 3901 54. Added convenience functions for handling named capturing parentheses.
 3902 
 3903 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
 3904 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
 3905 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
 3906 are faulted.
 3907 
 3908 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
 3909 calling configure. These values are used when compiling the dftables.c program
 3910 which is run to generate the source of the default character tables. They
 3911 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
 3912 you will need to set these values.
 3913 
 3914 57. Updated the building process for Windows DLL, as provided by Fred Cox.
 3915 
 3916 
 3917 Version 3.9 02-Jan-02
 3918 ---------------------
 3919 
 3920 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
 3921 
 3922 2. If --disable-static was given, the building process failed when trying to
 3923 build pcretest and pcregrep. (For some reason it was using libtool to compile
 3924 them, which is not right, as they aren't part of the library.)
 3925 
 3926 
 3927 Version 3.8 18-Dec-01
 3928 ---------------------
 3929 
 3930 1. The experimental UTF-8 code was completely screwed up. It was packing the
 3931 bytes in the wrong order. How dumb can you get?
 3932 
 3933 
 3934 Version 3.7 29-Oct-01
 3935 ---------------------
 3936 
 3937 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
 3938 This caused pcretest, when used on the test data, to segfault. Unfortunately,
 3939 this didn't happen under Solaris 8, where I normally test things.
 3940 
 3941 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
 3942 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
 3943 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
 3944 this fix an hour or so after the initial 3.7 release.)
 3945 
 3946 
 3947 Version 3.6 23-Oct-01
 3948 ---------------------
 3949 
 3950 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
 3951 offsets passed as NULL with zero offset count.
 3952 
 3953 2. The config.guess and config.sub files had not been updated when I moved to
 3954 the latest autoconf.
 3955 
 3956 
 3957 Version 3.5 15-Aug-01
 3958 ---------------------
 3959 
 3960 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
 3961 had been forgotten.
 3962 
 3963 2. By using declared but undefined structures, we can avoid using "void"
 3964 definitions in pcre.h while keeping the internal definitions of the structures
 3965 private.
 3966 
 3967 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
 3968 user point of view, this means that both static and shared libraries are built
 3969 by default, but this can be individually controlled. More of the work of
 3970 handling this static/shared cases is now inside libtool instead of PCRE's make
 3971 file.
 3972 
 3973 4. The pcretest utility is now installed along with pcregrep because it is
 3974 useful for users (to test regexs) and by doing this, it automatically gets
 3975 relinked by libtool. The documentation has been turned into a man page, so
 3976 there are now .1, .txt, and .html versions in /doc.
 3977 
 3978 5. Upgrades to pcregrep:
 3979    (i)   Added long-form option names like gnu grep.
 3980    (ii)  Added --help to list all options with an explanatory phrase.
 3981    (iii) Added -r, --recursive to recurse into sub-directories.
 3982    (iv)  Added -f, --file to read patterns from a file.
 3983 
 3984 6. pcre_exec() was referring to its "code" argument before testing that
 3985 argument for NULL (and giving an error if it was NULL).
 3986 
 3987 7. Upgraded Makefile.in to allow for compiling in a different directory from
 3988 the source directory.
 3989 
 3990 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
 3991 options bits, the pointer it was passed was to an int instead of to an unsigned
 3992 long int. This mattered only on 64-bit systems.
 3993 
 3994 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
 3995 generated) instead of pcre.in, which it its source. Also made the same change
 3996 in several of the .c files.
 3997 
 3998 10. A new release of gcc defines printf() as a macro, which broke pcretest
 3999 because it had an ifdef in the middle of a string argument for printf(). Fixed
 4000 by using separate calls to printf().
 4001 
 4002 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
 4003 script, to force use of CR or LF instead of \n in the source. On non-Unix
 4004 systems, the value can be set in config.h.
 4005 
 4006 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
 4007 absolute limit. Changed the text of the error message to make this clear, and
 4008 likewise updated the man page.
 4009 
 4010 13. The limit of 99 on the number of capturing subpatterns has been removed.
 4011 The new limit is 65535, which I hope will not be a "real" limit.
 4012 
 4013 
 4014 Version 3.4 22-Aug-00
 4015 ---------------------
 4016 
 4017 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
 4018 
 4019 2. Diagnose condition (?(0) as an error instead of crashing on matching.
 4020 
 4021 
 4022 Version 3.3 01-Aug-00
 4023 ---------------------
 4024 
 4025 1. If an octal character was given, but the value was greater than \377, it
 4026 was not getting masked to the least significant bits, as documented. This could
 4027 lead to crashes in some systems.
 4028 
 4029 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
 4030 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
 4031 
 4032 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
 4033 These just pass their arguments on to (pcre_free)(), but they are provided
 4034 because some uses of PCRE bind it to non-C systems that can call its functions,
 4035 but cannot call free() or pcre_free() directly.
 4036 
 4037 4. Add "make test" as a synonym for "make check". Corrected some comments in
 4038 the Makefile.
 4039 
 4040 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
 4041 Makefile.
 4042 
 4043 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
 4044 command called pgrep for grepping around the active processes.
 4045 
 4046 7. Added the beginnings of support for UTF-8 character strings.
 4047 
 4048 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
 4049 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
 4050 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
 4051 out for the ar command.)
 4052 
 4053 
 4054 Version 3.2 12-May-00
 4055 ---------------------
 4056 
 4057 This is purely a bug fixing release.
 4058 
 4059 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
 4060 of ZA. This was just one example of several cases that could provoke this bug,
 4061 which was introduced by change 9 of version 2.00. The code for breaking
 4062 infinite loops after an iteration that matches an empty string was't working
 4063 correctly.
 4064 
 4065 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
 4066 when matched against abbab (for example). After matching an empty string, it
 4067 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
 4068 caused it to match further down the string than it should.
 4069 
 4070 3. The code contained an inclusion of sys/types.h. It isn't clear why this
 4071 was there because it doesn't seem to be needed, and it causes trouble on some
 4072 systems, as it is not a Standard C header. It has been removed.
 4073 
 4074 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
 4075 were reported on the Macintosh. The changes were from
 4076 
 4077   while ((c = *(++ptr)) != 0 && c != '\n');
 4078 to
 4079   while ((c = *(++ptr)) != 0 && c != '\n') ;
 4080 
 4081 Totally extraordinary, but if that's what it takes...
 4082 
 4083 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
 4084 available. Added HAVE_BCOPY and an autoconf test for it; if neither
 4085 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
 4086 assumes the way PCRE uses memmove() (always moving upwards).
 4087 
 4088 6. PCRE is being used in one environment where strchr() is not available. There
 4089 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
 4090 faster code anyway.
 4091 
 4092 
 4093 Version 3.1 09-Feb-00
 4094 ---------------------
 4095 
 4096 The only change in this release is the fixing of some bugs in Makefile.in for
 4097 the "install" target:
 4098 
 4099 (1) It was failing to install pcreposix.h.
 4100 
 4101 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
 4102 
 4103 
 4104 Version 3.0 01-Feb-00
 4105 ---------------------
 4106 
 4107 1. Add support for the /+ modifier to perltest (to output $` like it does in
 4108 pcretest).
 4109 
 4110 2. Add support for the /g modifier to perltest.
 4111 
 4112 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
 4113 matches null strings.
 4114 
 4115 4. Fix perltest so that it doesn't do unwanted things when fed an empty
 4116 pattern. Perl treats empty patterns specially - it reuses the most recent
 4117 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
 4118 effect.
 4119 
 4120 5. The POSIX interface was broken in that it was just handing over the POSIX
 4121 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
 4122 required a bigger vector, with some working space on the end. This means that
 4123 the POSIX wrapper now has to get and free some memory, and copy the results.
 4124 
 4125 6. Added some simple autoconf support, placing the test data and the
 4126 documentation in separate directories, re-organizing some of the
 4127 information files, and making it build pcre-config (a GNU standard). Also added
 4128 libtool support for building PCRE as a shared library, which is now the
 4129 default.
 4130 
 4131 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
 4132 09 are not valid octal constants. Single digits will be used for minor values
 4133 less than 10.
 4134 
 4135 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
 4136 existing programs that set these in the POSIX interface can use PCRE without
 4137 modification.
 4138 
 4139 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
 4140 return all that pcre_info() returns, plus additional data. The pcre_info()
 4141 function is retained for compatibility, but is considered to be obsolete.
 4142 
 4143 10. Added experimental recursion feature (?R) to handle one common case that
 4144 Perl 5.6 will be able to do with (?p{...}).
 4145 
 4146 11. Added support for POSIX character classes like [:alpha:], which Perl is
 4147 adopting.
 4148 
 4149 
 4150 Version 2.08 31-Aug-99
 4151 ----------------------
 4152 
 4153 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
 4154 trying to match at the startoffset position, but instead was moving forward to
 4155 the next newline as if a previous match had failed.
 4156 
 4157 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
 4158 and could get into a loop if a null string was matched other than at the start
 4159 of the subject.
 4160 
 4161 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
 4162 be distinguished at compile time, and for completeness also added PCRE_DATE.
 4163 
 4164 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
 4165 in GnuWin32 environments.
 4166 
 4167 
 4168 Version 2.07 29-Jul-99
 4169 ----------------------
 4170 
 4171 1. The documentation is now supplied in plain text form and HTML as well as in
 4172 the form of man page sources.
 4173 
 4174 2. C++ compilers don't like assigning (void *) values to other pointer types.
 4175 In particular this affects malloc(). Although there is no problem in Standard
 4176 C, I've put in casts to keep C++ compilers happy.
 4177 
 4178 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
 4179 should be (const char *).
 4180 
 4181 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
 4182 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
 4183 However, I haven't made this a standard facility. The documentation doesn't
 4184 mention it, and the Makefile doesn't support it.
 4185 
 4186 5. The Makefile now contains an "install" target, with editable destinations at
 4187 the top of the file. The pcretest program is not installed.
 4188 
 4189 6. pgrep -V now gives the PCRE version number and date.
 4190 
 4191 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
 4192 causing the entire string to be ignored, instead of just the last character.
 4193 
 4194 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
 4195 non-matching string, it can take a very, very long time, even for strings of
 4196 quite modest length, because of the nested recursion. PCRE now does better in
 4197 some of these cases. It does this by remembering the last required literal
 4198 character in the pattern, and pre-searching the subject to ensure it is present
 4199 before running the real match. In other words, it applies a heuristic to detect
 4200 some types of certain failure quickly, and in the above example, if presented
 4201 with a string that has no trailing " it gives "no match" very quickly.
 4202 
 4203 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
 4204 other alternatives are tried instead.
 4205 
 4206 
 4207 Version 2.06 09-Jun-99
 4208 ----------------------
 4209 
 4210 1. Change pcretest's output for amount of store used to show just the code
 4211 space, because the remainder (the data block) varies in size between 32-bit and
 4212 64-bit systems.
 4213 
 4214 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
 4215 start matching at. This allows lookbehinds to work when searching for multiple
 4216 occurrences in a string.
 4217 
 4218 3. Added additional options to pcretest for testing multiple occurrences:
 4219 
 4220    /+   outputs the rest of the string that follows a match
 4221    /g   loops for multiple occurrences, using the new startoffset argument
 4222    /G   loops for multiple occurrences by passing an incremented pointer
 4223 
 4224 4. PCRE wasn't doing the "first character" optimization for patterns starting
 4225 with \b or \B, though it was doing it for other lookbehind assertions. That is,
 4226 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
 4227 the letter 'x'. On long subject strings, this gives a significant speed-up.
 4228 
 4229 
 4230 Version 2.05 21-Apr-99
 4231 ----------------------
 4232 
 4233 1. Changed the type of magic_number from int to long int so that it works
 4234 properly on 16-bit systems.
 4235 
 4236 2. Fixed a bug which caused patterns starting with .* not to work correctly
 4237 when the subject string contained newline characters. PCRE was assuming
 4238 anchoring for such patterns in all cases, which is not correct because .* will
 4239 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
 4240 DOTALL is set at top level; otherwise it knows that patterns starting with .*
 4241 must be retried after every newline in the subject.
 4242 
 4243 
 4244 Version 2.04 18-Feb-99
 4245 ----------------------
 4246 
 4247 1. For parenthesized subpatterns with repeats whose minimum was zero, the
 4248 computation of the store needed to hold the pattern was incorrect (too large).
 4249 If such patterns were nested a few deep, this could multiply and become a real
 4250 problem.
 4251 
 4252 2. Added /M option to pcretest to show the memory requirement of a specific
 4253 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
 4254 
 4255 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
 4256 compiled in such a way that the backtracking after subsequent failure was
 4257 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
 4258 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
 4259 
 4260 
 4261 Version 2.03 02-Feb-99
 4262 ----------------------
 4263 
 4264 1. Fixed typo and small mistake in man page.
 4265 
 4266 2. Added 4th condition (GPL supersedes if conflict) and created separate
 4267 LICENCE file containing the conditions.
 4268 
 4269 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
 4270 Perl, that is the internal \ allows the delimiter to be included in the
 4271 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
 4272 the final delimiter, add \ to the end of the pattern (to test the error).
 4273 
 4274 4. Added the convenience functions for extracting substrings after a successful
 4275 match. Updated pcretest to make it able to test these functions.
 4276 
 4277 
 4278 Version 2.02 14-Jan-99
 4279 ----------------------
 4280 
 4281 1. Initialized the working variables associated with each extraction so that
 4282 their saving and restoring doesn't refer to uninitialized store.
 4283 
 4284 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
 4285 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
 4286 fix the problem.
 4287 
 4288 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
 4289 calls, and wasn't printing the correct value for compiling calls. Increased the
 4290 default value of LOOPREPEAT, and the number of significant figures in the
 4291 times.
 4292 
 4293 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
 4294 
 4295 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
 4296 a building problem on Windows NT with a FAT file system.
 4297 
 4298 
 4299 Version 2.01 21-Oct-98
 4300 ----------------------
 4301 
 4302 1. Changed the API for pcre_compile() to allow for the provision of a pointer
 4303 to character tables built by pcre_maketables() in the current locale. If NULL
 4304 is passed, the default tables are used.
 4305 
 4306 
 4307 Version 2.00 24-Sep-98
 4308 ----------------------
 4309 
 4310 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
 4311 it any more.
 4312 
 4313 2. Allow quantification of (?>) groups, and make it work correctly.
 4314 
 4315 3. The first character computation wasn't working for (?>) groups.
 4316 
 4317 4. Correct the implementation of \Z (it is permitted to match on the \n at the
 4318 end of the subject) and add 5.005's \z, which really does match only at the
 4319 very end of the subject.
 4320 
 4321 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
 4322 
 4323 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
 4324 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
 4325 localized options. All options to pcre_study() were also removed.
 4326 
 4327 7. Add other new features from 5.005:
 4328 
 4329    $(?<=           positive lookbehind
 4330    $(?<!           negative lookbehind
 4331    (?imsx-imsx)    added the unsetting capability
 4332                    such a setting is global if at outer level; local otherwise
 4333    (?imsx-imsx:)   non-capturing groups with option setting
 4334    (?(cond)re|re)  conditional pattern matching
 4335 
 4336    A backreference to itself in a repeated group matches the previous
 4337    captured string.
 4338 
 4339 8. General tidying up of studying (both automatic and via "study")
 4340 consequential on the addition of new assertions.
 4341 
 4342 9. As in 5.005, unlimited repeated groups that could match an empty substring
 4343 are no longer faulted at compile time. Instead, the loop is forcibly broken at
 4344 runtime if any iteration does actually match an empty substring.
 4345 
 4346 10. Include the RunTest script in the distribution.
 4347 
 4348 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
 4349 discrepancies, some of which were old and were also with respect to 5.004. They
 4350 have now been fixed.
 4351 
 4352 
 4353 Version 1.09 28-Apr-98
 4354 ----------------------
 4355 
 4356 1. A negated single character class followed by a quantifier with a minimum
 4357 value of one (e.g.  [^x]{1,6}  ) was not compiled correctly. This could lead to
 4358 program crashes, or just wrong answers. This did not apply to negated classes
 4359 containing more than one character, or to minima other than one.
 4360 
 4361 
 4362 Version 1.08 27-Mar-98
 4363 ----------------------
 4364 
 4365 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
 4366 
 4367 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
 4368 latter must appear before anything that relies on it in the pattern.
 4369 
 4370 
 4371 Version 1.07 16-Feb-98
 4372 ----------------------
 4373 
 4374 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
 4375 repeat of a potentially empty string).
 4376 
 4377 
 4378 Version 1.06 23-Jan-98
 4379 ----------------------
 4380 
 4381 1. Added Markus Oberhumer's little patches for C++.
 4382 
 4383 2. Literal strings longer than 255 characters were broken.
 4384 
 4385 
 4386 Version 1.05 23-Dec-97
 4387 ----------------------
 4388 
 4389 1. Negated character classes containing more than one character were failing if
 4390 PCRE_CASELESS was set at run time.
 4391 
 4392 
 4393 Version 1.04 19-Dec-97
 4394 ----------------------
 4395 
 4396 1. Corrected the man page, where some "const" qualifiers had been omitted.
 4397 
 4398 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
 4399 input syntax.
 4400 
 4401 3. Fixed memory leak which occurred when a regex with back references was
 4402 matched with an offsets vector that wasn't big enough. The temporary memory
 4403 that is used in this case wasn't being freed if the match failed.
 4404 
 4405 4. Tidied pcretest to ensure it frees memory that it gets.
 4406 
 4407 5. Temporary memory was being obtained in the case where the passed offsets
 4408 vector was exactly big enough.
 4409 
 4410 6. Corrected definition of offsetof() from change 5 below.
 4411 
 4412 7. I had screwed up change 6 below and broken the rules for the use of
 4413 setjmp(). Now fixed.
 4414 
 4415 
 4416 Version 1.03 18-Dec-97
 4417 ----------------------
 4418 
 4419 1. A erroneous regex with a missing opening parenthesis was correctly
 4420 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
 4421 on some systems.
 4422 
 4423 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
 4424 it was reported that one broken compiler failed on the former because "code" is
 4425 also an independent variable.
 4426 
 4427 3. The erroneous regex a[]b caused an array overrun reference.
 4428 
 4429 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
 4430 fail on data ending with that character. (It was going on too far, and checking
 4431 the next character, typically a binary zero.) This was specific to the
 4432 optimized code for single-character negative classes.
 4433 
 4434 5. Added a contributed patch from the TIN world which does the following:
 4435 
 4436   + Add an undef for memmove, in case the the system defines a macro for it.
 4437 
 4438   + Add a definition of offsetof(), in case there isn't one. (I don't know
 4439     the reason behind this - offsetof() is part of the ANSI standard - but
 4440     it does no harm).
 4441 
 4442   + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
 4443     most of the places where whitespace preceded '#'. I have given up and
 4444     allowed the remaining 2 cases to be at the margin.
 4445 
 4446   + Rename some variables in pcre to eliminate shadowing. This seems very
 4447     pedantic, but does no harm, of course.
 4448 
 4449 6. Moved the call to setjmp() into its own function, to get rid of warnings
 4450 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
 4451 
 4452 7. Constructs such as \d{8,} were compiling into the equivalent of
 4453 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
 4454 outcome, but in this particular case used more store than had been allocated,
 4455 which caused the bug to be discovered because it threw up an internal error.
 4456 
 4457 8. The debugging code in both pcre and pcretest for outputting the compiled
 4458 form of a regex was going wrong in the case of back references followed by
 4459 curly-bracketed repeats.
 4460 
 4461 
 4462 Version 1.02 12-Dec-97
 4463 ----------------------
 4464 
 4465 1. Typos in pcre.3 and comments in the source fixed.
 4466 
 4467 2. Applied a contributed patch to get rid of places where it used to remove
 4468 'const' from variables, and fixed some signed/unsigned and uninitialized
 4469 variable warnings.
 4470 
 4471 3. Added the "runtest" target to Makefile.
 4472 
 4473 4. Set default compiler flag to -O2 rather than just -O.
 4474 
 4475 
 4476 Version 1.01 19-Nov-97
 4477 ----------------------
 4478 
 4479 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
 4480 like /([ab]*)*/, that is, for classes with more than one character in them.
 4481 
 4482 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
 4483 as /((?>a*))*/ (a PCRE_EXTRA facility).
 4484 
 4485 
 4486 Version 1.00 18-Nov-97
 4487 ----------------------
 4488 
 4489 1. Added compile-time macros to support systems such as SunOS4 which don't have
 4490 memmove() or strerror() but have other things that can be used instead.
 4491 
 4492 2. Arranged that "make clean" removes the executables.
 4493 
 4494 
 4495 Version 0.99 27-Oct-97
 4496 ----------------------
 4497 
 4498 1. Fixed bug in code for optimizing classes with only one character. It was
 4499 initializing a 32-byte map regardless, which could cause it to run off the end
 4500 of the memory it had got.
 4501 
 4502 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
 4503 
 4504 
 4505 Version 0.98 22-Oct-97
 4506 ----------------------
 4507 
 4508 1. Fixed bug in code for handling temporary memory usage when there are more
 4509 back references than supplied space in the ovector. This could cause segfaults.
 4510 
 4511 
 4512 Version 0.97 21-Oct-97
 4513 ----------------------
 4514 
 4515 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
 4516 
 4517 2. Optimized negated single characters not to use a bit map.
 4518 
 4519 3. Brought error texts together as macro definitions; clarified some of them;
 4520 fixed one that was wrong - it said "range out of order" when it meant "invalid
 4521 escape sequence".
 4522 
 4523 4. Changed some char * arguments to const char *.
 4524 
 4525 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
 4526 
 4527 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
 4528 pcretest.
 4529 
 4530 
 4531 Version 0.96 16-Oct-97
 4532 ----------------------
 4533 
 4534 1. Added a simple "pgrep" utility to the distribution.
 4535 
 4536 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
 4537 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
 4538 where "ddd" means "one or more decimal digits".
 4539 
 4540 3. Fixed serious bug. If a pattern had a back reference, but the call to
 4541 pcre_exec() didn't supply a large enough ovector to record the related
 4542 identifying subpattern, the match always failed. PCRE now remembers the number
 4543 of the largest back reference, and gets some temporary memory in which to save
 4544 the offsets during matching if necessary, in order to ensure that
 4545 backreferences always work.
 4546 
 4547 4. Increased the compatibility with Perl in a number of ways:
 4548 
 4549   (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
 4550       to request this handling. The option can be set at compile or exec time.
 4551 
 4552   (b) $ matches before a terminating newline by default; an option
 4553       PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
 4554       mode). The option can be set at compile or exec time.
 4555 
 4556   (c) The handling of \ followed by a digit other than 0 is now supposed to be
 4557       the same as Perl's. If the decimal number it represents is less than 10
 4558       or there aren't that many previous left capturing parentheses, an octal
 4559       escape is read. Inside a character class, it's always an octal escape,
 4560       even if it is a single digit.
 4561 
 4562   (d) An escaped but undefined alphabetic character is taken as a literal,
 4563       unless PCRE_EXTRA is set. Currently this just reserves the remaining
 4564       escapes.
 4565 
 4566   (e) {0} is now permitted. (The previous item is removed from the compiled
 4567       pattern).
 4568 
 4569 5. Changed all the names of code files so that the basic parts are no longer
 4570 than 10 characters, and abolished the teeny "globals.c" file.
 4571 
 4572 6. Changed the handling of character classes; they are now done with a 32-byte
 4573 bit map always.
 4574 
 4575 7. Added the -d and /D options to pcretest to make it possible to look at the
 4576 internals of compilation without having to recompile pcre.
 4577 
 4578 
 4579 Version 0.95 23-Sep-97
 4580 ----------------------
 4581 
 4582 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
 4583 \x20 at the start of a run of normal characters. These were being treated as
 4584 real characters, instead of the source characters being re-checked.
 4585 
 4586 
 4587 Version 0.94 18-Sep-97
 4588 ----------------------
 4589 
 4590 1. The functions are now thread-safe, with the caveat that the global variables
 4591 containing pointers to malloc() and free() or alternative functions are the
 4592 same for all threads.
 4593 
 4594 2. Get pcre_study() to generate a bitmap of initial characters for non-
 4595 anchored patterns when this is possible, and use it if passed to pcre_exec().
 4596 
 4597 
 4598 Version 0.93 15-Sep-97
 4599 ----------------------
 4600 
 4601 1. /(b)|(:+)/ was computing an incorrect first character.
 4602 
 4603 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
 4604 but not actually doing anything yet.
 4605 
 4606 3. Treat "-" characters in classes that cannot be part of ranges as literals,
 4607 as Perl does (e.g. [-az] or [az-]).
 4608 
 4609 4. Set the anchored flag if a branch starts with .* or .*? because that tests
 4610 all possible positions.
 4611 
 4612 5. Split up into different modules to avoid including unneeded functions in a
 4613 compiled binary. However, compile and exec are still in one module. The "study"
 4614 function is split off.
 4615 
 4616 6. The character tables are now in a separate module whose source is generated
 4617 by an auxiliary program - but can then be edited by hand if required. There are
 4618 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
 4619 toupper() in the code.
 4620 
 4621 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
 4622 make them global. Abolish the function for setting them, as the caller can now
 4623 set them directly.
 4624 
 4625 
 4626 Version 0.92 11-Sep-97
 4627 ----------------------
 4628 
 4629 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
 4630 (e.g. /a{1,3}/) was broken (I mis-optimized it).
 4631 
 4632 2. Caseless matching was not working in character classes if the characters in
 4633 the pattern were in upper case.
 4634 
 4635 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
 4636 
 4637 4. Make PCRE_ANCHORED public and accept as a compile option.
 4638 
 4639 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
 4640 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
 4641 pass them.
 4642 
 4643 6. Give an error if bad option bits passed at compile or run time.
 4644 
 4645 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
 4646 pcretest to cause it to pass that flag.
 4647 
 4648 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
 4649 options, and the first character, if set.
 4650 
 4651 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
 4652 
 4653 
 4654 Version 0.91 10-Sep-97
 4655 ----------------------
 4656 
 4657 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
 4658 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
 4659 
 4660 2. PCRE was looping on encountering an indefinitely repeated back reference to
 4661 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
 4662 Perl does - treats the match as successful.
 4663 
 4664 ****