"Fossies" - the Fresh Open Source Software Archive

Member "pcre-8.42/ChangeLog" (20 Mar 2018, 296019 Bytes) of package /linux/misc/pcre-8.42.tar.bz2:


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. See also the latest Fossies "Diffs" side-by-side code changes report for "ChangeLog": 8.41_vs_8.42.

    1 ChangeLog for PCRE
    2 ------------------
    3 
    4 Note that the PCRE 8.xx series (PCRE1) is now in a bugfix-only state. All
    5 development is happening in the PCRE2 10.xx series.
    6 
    7 
    8 Version 8.42 20-March-2018
    9 --------------------------
   10 
   11 1.  Fixed a MIPS issue in the JIT compiler reported by Joshua Kinard.
   12 
   13 2.  Fixed outdated real_pcre definitions in pcre.h.in (patch by Evgeny Kotkov).
   14 
   15 3.  pcregrep was truncating components of file names to 128 characters when
   16 processing files with the -r option, and also (some very odd code) truncating
   17 path names to 512 characters. There is now a check on the absolute length of
   18 full path file names, which may be up to 2047 characters long.
   19 
   20 4.  Using pcre_dfa_exec(), in UTF mode when UCP support was not defined, there
   21 was the possibility of a false positive match when caselessly matching a "not
   22 this character" item such as [^\x{1234}] (with a code point greater than 127)
   23 because the "other case" variable was not being initialized.
   24 
   25 5. Although pcre_jit_exec checks whether the pattern is compiled
   26 in a given mode, it was also expected that at least one mode is available.
   27 This is fixed and pcre_jit_exec returns with PCRE_ERROR_JIT_BADOPTION
   28 when the pattern is not optimized by JIT at all.
   29 
   30 6. The line number and related variables such as match counts in pcregrep
   31 were all int variables, causing overflow when files with more than 2147483647
   32 lines were processed (assuming 32-bit ints). They have all been changed to
   33 unsigned long ints.
   34 
   35 7. If a backreference with a minimum repeat count of zero was first in a
   36 pattern, apart from assertions, an incorrect first matching character could be
   37 recorded. For example, for the pattern /(?=(a))\1?b/, "b" was incorrectly set
   38 as the first character of a match.
   39 
   40 8. Fix out-of-bounds read for partial matching of /./ against an empty string
   41 when the newline type is CRLF.
   42 
   43 9. When matching using the the REG_STARTEND feature of the POSIX API with a
   44 non-zero starting offset, unset capturing groups with lower numbers than a
   45 group that did capture something were not being correctly returned as "unset"
   46 (that is, with offset values of -1).
   47 
   48 10. Matching the pattern /(*UTF)\C[^\v]+\x80/ against an 8-bit string
   49 containing multi-code-unit characters caused bad behaviour and possibly a
   50 crash. This issue was fixed for other kinds of repeat in release 8.37 by change
   51 38, but repeating character classes were overlooked.
   52 
   53 11. A small fix to pcregrep to avoid compiler warnings for -Wformat-overflow=2.
   54 
   55 12. Added --enable-jit=auto support to configure.ac.
   56 
   57 13. Fix misleading error message in configure.ac.
   58 
   59 
   60 Version 8.41 05-July-2017
   61 -------------------------
   62 
   63 1.  Fixed typo in CMakeLists.txt (wrong number of arguments for
   64 PCRE_STATIC_RUNTIME (affects MSVC only).
   65 
   66 2.  Issue 1 for 8.40 below was not correctly fixed. If pcregrep in multiline
   67 mode with --only-matching matched several lines, it restarted scanning at the
   68 next line instead of moving on to the end of the matched string, which can be
   69 several lines after the start.
   70 
   71 3.  Fix a missing else in the JIT compiler reported by 'idaifish'.
   72 
   73 4.  A (?# style comment is now ignored between a basic quantifier and a
   74 following '+' or '?' (example: /X+(?#comment)?Y/.
   75 
   76 5.  Avoid use of a potentially overflowing buffer in pcregrep (patch by Petr
   77 Pisar).
   78 
   79 6.  Fuzzers have reported issues in pcretest. These are NOT serious (it is,
   80 after all, just a test program). However, to stop the reports, some easy ones
   81 are fixed:
   82 
   83     (a) Check for values < 256 when calling isprint() in pcretest.
   84     (b) Give an error for too big a number after \O.
   85 
   86 7.  In the 32-bit library in non-UTF mode, an attempt to find a Unicode
   87 property for a character with a code point greater than 0x10ffff (the Unicode
   88 maximum) caused a crash.
   89 
   90 8. The alternative matching function, pcre_dfa_exec() misbehaved if it
   91 encountered a character class with a possessive repeat, for example [a-f]{3}+.
   92 
   93 9. When pcretest called pcre_copy_substring() in 32-bit mode, it set the buffer
   94 length incorrectly, which could result in buffer overflow.
   95 
   96 10. Remove redundant line of code (accidentally left in ages ago).
   97 
   98 11. Applied C++ patch from Irfan Adilovic to guard 'using std::' directives
   99 with namespace pcrecpp (Bugzilla #2084).
  100 
  101 12. Remove a duplication typo in pcre_tables.c.
  102 
  103 13. Fix returned offsets from regexec() when REG_STARTEND is used with a
  104 starting offset greater than zero.
  105 
  106 
  107 Version 8.40 11-January-2017
  108 ----------------------------
  109 
  110 1.  Using -o with -M in pcregrep could cause unnecessary repeated output when
  111     the match extended over a line boundary.
  112 
  113 2.  Applied Chris Wilson's second patch (Bugzilla #1681) to CMakeLists.txt for
  114     MSVC static compilation, putting the first patch under a new option.
  115 
  116 3.  Fix register overwite in JIT when SSE2 acceleration is enabled.
  117 
  118 4.  Ignore "show all captures" (/=) for DFA matching.
  119 
  120 5.  Fix JIT unaligned accesses on x86. Patch by Marc Mutz.
  121 
  122 6.  In any wide-character mode (8-bit UTF or any 16-bit or 32-bit mode),
  123     without PCRE_UCP set, a negative character type such as \D in a positive
  124     class should cause all characters greater than 255 to match, whatever else
  125     is in the class. There was a bug that caused this not to happen if a
  126     Unicode property item was added to such a class, for example [\D\P{Nd}] or
  127     [\W\pL].
  128 
  129 7.  When pcretest was outputing information from a callout, the caret indicator
  130     for the current position in the subject line was incorrect if it was after
  131     an escape sequence for a character whose code point was greater than
  132     \x{ff}.
  133 
  134 8.  A pattern such as (?<RA>abc)(?(R)xyz) was incorrectly compiled such that
  135     the conditional was interpreted as a reference to capturing group 1 instead
  136     of a test for recursion. Any group whose name began with R was
  137     misinterpreted in this way. (The reference interpretation should only
  138     happen if the group's name is precisely "R".)
  139 
  140 9.  A number of bugs have been mended relating to match start-up optimizations
  141     when the first thing in a pattern is a positive lookahead. These all
  142     applied only when PCRE_NO_START_OPTIMIZE was *not* set:
  143 
  144     (a) A pattern such as (?=.*X)X$ was incorrectly optimized as if it needed
  145         both an initial 'X' and a following 'X'.
  146     (b) Some patterns starting with an assertion that started with .* were
  147         incorrectly optimized as having to match at the start of the subject or
  148         after a newline. There are cases where this is not true, for example,
  149         (?=.*[A-Z])(?=.{8,16})(?!.*[\s]) matches after the start in lines that
  150         start with spaces. Starting .* in an assertion is no longer taken as an
  151         indication of matching at the start (or after a newline).
  152 
  153 
  154 Version 8.39 14-June-2016
  155 -------------------------
  156 
  157 1.  If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
  158     an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
  159     misbehaved. This bug was found by the LLVM fuzzer.
  160 
  161 2.  Similar to the above, if an isolated \E was present between an item and its
  162     qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
  163     bug was found by the LLVM fuzzer.
  164 
  165 3.  Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
  166     working correctly in UCP mode.
  167 
  168 4.  The POSIX wrapper function regexec() crashed if the option REG_STARTEND
  169     was set when the pmatch argument was NULL. It now returns REG_INVARG.
  170 
  171 5.  Allow for up to 32-bit numbers in the ordin() function in pcregrep.
  172 
  173 6.  An empty \Q\E sequence between an item and its qualifier caused
  174     pcre_compile() to misbehave when auto callouts were enabled. This bug was
  175     found by the LLVM fuzzer.
  176 
  177 7.  If a pattern that was compiled with PCRE_EXTENDED started with white
  178     space or a #-type comment that was followed by (?-x), which turns off
  179     PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
  180     pcre_compile() assumed that (?-x) applied to the whole pattern and
  181     consequently mis-compiled it. This bug was found by the LLVM fuzzer.
  182 
  183 8.  A call of pcre_copy_named_substring() for a named substring whose number
  184     was greater than the space in the ovector could cause a crash.
  185 
  186 9.  Yet another buffer overflow bug involved duplicate named groups with a
  187     group that reset capture numbers (compare 8.38/7 below). Once again, I have
  188     just allowed for more memory, even if not needed. (A proper fix is
  189     implemented in PCRE2, but it involves a lot of refactoring.)
  190 
  191 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
  192     start of the match to be earlier than the end.
  193 
  194 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
  195 
  196 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
  197     assertion, caused pcretest to generate incorrect output, and also to read
  198     uninitialized memory (detected by ASAN or valgrind).
  199 
  200 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
  201     nested set of parentheses of sufficient size caused an overflow of the
  202     compiling workspace (which was diagnosed, but of course is not desirable).
  203 
  204 14. And yet another buffer overflow bug involving duplicate named groups, this
  205     time nested, with a nested back reference. Yet again, I have just allowed
  206     for more memory, because anything more needs all the refactoring that has
  207     been done for PCRE2. An example pattern that provoked this bug is:
  208     /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
  209     registered as CVE-2016-1283.
  210 
  211 15. pcretest went into a loop if global matching was requested with an ovector
  212     size less than 2. It now gives an error message. This bug was found by
  213     afl-fuzz.
  214 
  215 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
  216     ("assertion expected") when (?(?C) was not followed by an opening
  217     parenthesis.
  218 
  219 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
  220     actually affect anything, by sheer luck.
  221 
  222 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
  223     static compilation.
  224 
  225 19. Modified the RunTest script to incorporate a valgrind suppressions file so
  226     that certain errors, provoked by the SSE2 instruction set when JIT is used,
  227     are ignored.
  228 
  229 20. A racing condition is fixed in JIT reported by Mozilla.
  230 
  231 21. Minor code refactor to avoid "array subscript is below array bounds"
  232     compiler warning.
  233 
  234 22. Minor code refactor to avoid "left shift of negative number" warning.
  235 
  236 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
  237     UCP support.
  238 
  239 24. Refactor to avoid compiler warnings in pcrecpp.cc.
  240 
  241 25. Refactor to fix a typo in pcre_jit_test.c
  242 
  243 26. Patch to support compiling pcrecpp.cc with Intel compiler.
  244 
  245 
  246 Version 8.38 23-November-2015
  247 -----------------------------
  248 
  249 1.  If a group that contained a recursive back reference also contained a
  250     forward reference subroutine call followed by a non-forward-reference
  251     subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
  252     compile correct code, leading to undefined behaviour or an internally
  253     detected error. This bug was discovered by the LLVM fuzzer.
  254 
  255 2.  Quantification of certain items (e.g. atomic back references) could cause
  256     incorrect code to be compiled when recursive forward references were
  257     involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
  258     This bug was discovered by the LLVM fuzzer.
  259 
  260 3.  A repeated conditional group whose condition was a reference by name caused
  261     a buffer overflow if there was more than one group with the given name.
  262     This bug was discovered by the LLVM fuzzer.
  263 
  264 4.  A recursive back reference by name within a group that had the same name as
  265     another group caused a buffer overflow. For example:
  266     /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
  267 
  268 5.  A forward reference by name to a group whose number is the same as the
  269     current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
  270     a buffer overflow at compile time. This bug was discovered by the LLVM
  271     fuzzer.
  272 
  273 6.  A lookbehind assertion within a set of mutually recursive subpatterns could
  274     provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
  275 
  276 7.  Another buffer overflow bug involved duplicate named groups with a
  277     reference between their definition, with a group that reset capture
  278     numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
  279     fixed by always allowing for more memory, even if not needed. (A proper fix
  280     is implemented in PCRE2, but it involves more refactoring.)
  281 
  282 8.  There was no check for integer overflow in subroutine calls such as (?123).
  283 
  284 9.  The table entry for \l in EBCDIC environments was incorrect, leading to its
  285     being treated as a literal 'l' instead of causing an error.
  286 
  287 10. There was a buffer overflow if pcre_exec() was called with an ovector of
  288     size 1. This bug was found by american fuzzy lop.
  289 
  290 11. If a non-capturing group containing a conditional group that could match
  291     an empty string was repeated, it was not identified as matching an empty
  292     string itself. For example: /^(?:(?(1)x|)+)+$()/.
  293 
  294 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
  295     \a and \e in test subject lines.
  296 
  297 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
  298     instead of the EBCDIC value.
  299 
  300 14. The handling of \c in an EBCDIC environment has been revised so that it is
  301     now compatible with the specification in Perl's perlebcdic page.
  302 
  303 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
  304     ASCII/Unicode. This has now been added to the list of characters that are
  305     recognized as white space in EBCDIC.
  306 
  307 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
  308     error (correctly) when used outside a class, but did not give an error
  309     within a class.
  310 
  311 17. \h within a class was incorrectly compiled in EBCDIC environments.
  312 
  313 18. A pattern with an unmatched closing parenthesis that contained a backward
  314     assertion which itself contained a forward reference caused buffer
  315     overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
  316 
  317 19. JIT should return with error when the compiled pattern requires more stack
  318     space than the maximum.
  319 
  320 20. A possessively repeated conditional group that could match an empty string,
  321     for example, /(?(R))*+/, was incorrectly compiled.
  322 
  323 21. Fix infinite recursion in the JIT compiler when certain patterns such as
  324     /(?:|a|){100}x/ are analysed.
  325 
  326 22. Some patterns with character classes involving [: and \\ were incorrectly
  327     compiled and could cause reading from uninitialized memory or an incorrect
  328     error diagnosis.
  329 
  330 23. Pathological patterns containing many nested occurrences of [: caused
  331     pcre_compile() to run for a very long time.
  332 
  333 24. A conditional group with only one branch has an implicit empty alternative
  334     branch and must therefore be treated as potentially matching an empty
  335     string.
  336 
  337 25. If (?R was followed by - or + incorrect behaviour happened instead of a
  338     diagnostic.
  339 
  340 26. Arrange to give up on finding the minimum matching length for overly
  341     complex patterns.
  342 
  343 27. Similar to (4) above: in a pattern with duplicated named groups and an
  344     occurrence of (?| it is possible for an apparently non-recursive back
  345     reference to become recursive if a later named group with the relevant
  346     number is encountered. This could lead to a buffer overflow. Wen Guanxing
  347     from Venustech ADLAB discovered this bug.
  348 
  349 28. If pcregrep was given the -q option with -c or -l, or when handling a
  350     binary file, it incorrectly wrote output to stdout.
  351 
  352 29. The JIT compiler did not restore the control verb head in case of *THEN
  353     control verbs. This issue was found by Karl Skomski with a custom LLVM
  354     fuzzer.
  355 
  356 30. Error messages for syntax errors following \g and \k were giving inaccurate
  357     offsets in the pattern.
  358 
  359 31. Added a check for integer overflow in conditions (?(<digits>) and
  360     (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
  361     fuzzer.
  362 
  363 32. Handling recursive references such as (?2) when the reference is to a group
  364     later in the pattern uses code that is very hacked about and error-prone.
  365     It has been re-written for PCRE2. Here in PCRE1, a check has been added to
  366     give an internal error if it is obvious that compiling has gone wrong.
  367 
  368 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
  369     This issue was found by Karl Skomski with a custom LLVM fuzzer.
  370 
  371 34. The JIT compiler should restore the control chain for empty possessive
  372     repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
  373 
  374 35. Match limit check added to JIT recursion. This issue was found by Karl
  375     Skomski with a custom LLVM fuzzer.
  376 
  377 36. Yet another case similar to 27 above has been circumvented by an
  378     unconditional allocation of extra memory. This issue is fixed "properly" in
  379     PCRE2 by refactoring the way references are handled. Wen Guanxing
  380     from Venustech ADLAB discovered this bug.
  381 
  382 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
  383     with a custom LLVM fuzzer.
  384 
  385 38. Fixed a corner case of range optimization in JIT.
  386 
  387 39. An incorrect error "overran compiling workspace" was given if there were
  388     exactly enough group forward references such that the last one extended
  389     into the workspace safety margin. The next one would have expanded the
  390     workspace. The test for overflow was not including the safety margin.
  391 
  392 40. A match limit issue is fixed in JIT which was found by Karl Skomski
  393     with a custom LLVM fuzzer.
  394 
  395 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
  396     work under Windows. (Why has it taken so long for anyone to notice?)
  397 
  398 42. In a character class such as [\W\p{Any}] where both a negative-type escape
  399     ("not a word character") and a property escape were present, the property
  400     escape was being ignored.
  401 
  402 43. Fix crash caused by very long (*MARK) or (*THEN) names.
  403 
  404 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
  405     by a single ASCII character in a class item, was incorrectly compiled in
  406     UCP mode. The POSIX class got lost, but only if the single character
  407     followed it.
  408 
  409 45. [:punct:] in UCP mode was matching some characters in the range 128-255
  410     that should not have been matched.
  411 
  412 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
  413     class, all characters with code points greater than 255 are in the class.
  414     When a Unicode property was also in the class (if PCRE_UCP is set, escapes
  415     such as \w are turned into Unicode properties), wide characters were not
  416     correctly handled, and could fail to match.
  417 
  418 
  419 Version 8.37 28-April-2015
  420 --------------------------
  421 
  422 1.  When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
  423     for those parentheses to be closed with whatever has been captured so far.
  424     However, it was failing to mark any other groups between the hightest
  425     capture so far and the currrent group as "unset". Thus, the ovector for
  426     those groups contained whatever was previously there. An example is the
  427     pattern /(x)|((*ACCEPT))/ when matched against "abcd".
  428 
  429 2.  If an assertion condition was quantified with a minimum of zero (an odd
  430     thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
  431 
  432 3.  If a pattern in pcretest input had the P (POSIX) modifier followed by an
  433     unrecognized modifier, a crash could occur.
  434 
  435 4.  An attempt to do global matching in pcretest with a zero-length ovector
  436     caused a crash.
  437 
  438 5.  Fixed a memory leak during matching that could occur for a subpattern
  439     subroutine call (recursive or otherwise) if the number of captured groups
  440     that had to be saved was greater than ten.
  441 
  442 6.  Catch a bad opcode during auto-possessification after compiling a bad UTF
  443     string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
  444     UTF with NO_UTF_CHECK is documented as having an undefined outcome.
  445 
  446 7.  A UTF pattern containing a "not" match of a non-ASCII character and a
  447     subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
  448 
  449 8. When a pattern is compiled, it remembers the highest back reference so that
  450    when matching, if the ovector is too small, extra memory can be obtained to
  451    use instead. A conditional subpattern whose condition is a check on a
  452    capture having happened, such as, for example in the pattern
  453    /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
  454    setting the highest backreference number. This mattered only if pcre_exec()
  455    was called with an ovector that was too small to hold the capture, and there
  456    was no other kind of back reference (a situation which is probably quite
  457    rare). The effect of the bug was that the condition was always treated as
  458    FALSE when the capture could not be consulted, leading to a incorrect
  459    behaviour by pcre_exec(). This bug has been fixed.
  460 
  461 9. A reference to a duplicated named group (either a back reference or a test
  462    for being set in a conditional) that occurred in a part of the pattern where
  463    PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
  464    to be incorrectly calculated, leading to overwriting.
  465 
  466 10. A mutually recursive set of back references such as (\2)(\1) caused a
  467     segfault at study time (while trying to find the minimum matching length).
  468     The infinite loop is now broken (with the minimum length unset, that is,
  469     zero).
  470 
  471 11. If an assertion that was used as a condition was quantified with a minimum
  472     of zero, matching went wrong. In particular, if the whole group had
  473     unlimited repetition and could match an empty string, a segfault was
  474     likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
  475     assertions to be quantified, but not if they are being used as conditions,
  476     so the above pattern is faulted by Perl. PCRE has now been changed so that
  477     it also rejects such patterns.
  478 
  479 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
  480     failed to allow the zero-repeat case if pcre2_exec() was called with an
  481     ovector too small to capture the group.
  482 
  483 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
  484     Red Hat Product Security:
  485 
  486     (a) A crash if /K and /F were both set with the option to save the compiled
  487     pattern.
  488 
  489     (b) Another crash if the option to print captured substrings in a callout
  490     was combined with setting a null ovector, for example \O\C+ as a subject
  491     string.
  492 
  493 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
  494     forward reference repeated a large (but limited) number of times within a
  495     repeated outer group that has a zero minimum quantifier, caused incorrect
  496     code to be compiled, leading to the error "internal error:
  497     previously-checked referenced subpattern not found" when an incorrect
  498     memory address was read. This bug was reported as "heap overflow",
  499     discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
  500     CVE-2015-2325.
  501 
  502 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
  503     call within a group that also contained a recursive back reference caused
  504     incorrect code to be compiled. This bug was reported as "heap overflow",
  505     discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
  506     number CVE-2015-2326.
  507 
  508 24. Computing the size of the JIT read-only data in advance has been a source
  509     of various issues, and new ones are still appear unfortunately. To fix
  510     existing and future issues, size computation is eliminated from the code,
  511     and replaced by on-demand memory allocation.
  512 
  513 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
  514     adjacent to the end of the range, and the range contained characters with
  515     more than one other case, caused incorrect behaviour when compiled in UTF
  516     mode. In that example, the range a-j was left out of the class.
  517 
  518 26. Fix JIT compilation of conditional blocks, which assertion
  519     is converted to (*FAIL). E.g: /(?(?!))/.
  520 
  521 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
  522     discovered by the LLVM fuzzer.
  523 
  524 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
  525     when this assertion was used as a condition, for example (?(?!)a|b). In
  526     pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
  527     error about an unsupported item.
  528 
  529 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
  530     possessification code could take exponential time to complete. A recursion
  531     depth limit of 1000 has been imposed to limit the resources used by this
  532     optimization.
  533 
  534 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
  535     such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
  536     because \S ensures they are all in the class. The code for doing this was
  537     interacting badly with the code for computing the amount of space needed to
  538     compile the pattern, leading to a buffer overflow. This bug was discovered
  539     by the LLVM fuzzer.
  540 
  541 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
  542     other kinds of group caused stack overflow at compile time. This bug was
  543     discovered by the LLVM fuzzer.
  544 
  545 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
  546     between a subroutine call and its quantifier was incorrectly compiled,
  547     leading to buffer overflow or other errors. This bug was discovered by the
  548     LLVM fuzzer.
  549 
  550 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
  551     assertion after (?(. The code was failing to check the character after
  552     (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
  553     was discovered by the LLVM fuzzer.
  554 
  555 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
  556     a fixed maximum following a group that contains a subroutine reference was
  557     incorrectly compiled and could trigger buffer overflow. This bug was
  558     discovered by the LLVM fuzzer.
  559 
  560 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
  561     caused a stack overflow instead of the diagnosis of a non-fixed length
  562     lookbehind assertion. This bug was discovered by the LLVM fuzzer.
  563 
  564 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
  565     (e.g. /(?<=\Ka)/) could make pcregrep loop.
  566 
  567 37. There was a similar problem to 36 in pcretest for global matches.
  568 
  569 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
  570     and a subsequent item in the pattern caused a non-match, backtracking over
  571     the repeated \X did not stop, but carried on past the start of the subject,
  572     causing reference to random memory and/or a segfault. There were also some
  573     other cases where backtracking after \C could crash. This set of bugs was
  574     discovered by the LLVM fuzzer.
  575 
  576 39. The function for finding the minimum length of a matching string could take
  577     a very long time if mutual recursion was present many times in a pattern,
  578     for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
  579     method has been implemented. This infelicity was discovered by the LLVM
  580     fuzzer.
  581 
  582 40. Static linking against the PCRE library using the pkg-config module was
  583     failing on missing pthread symbols.
  584 
  585 
  586 Version 8.36 26-September-2014
  587 ------------------------------
  588 
  589 1.  Got rid of some compiler warnings in the C++ modules that were shown up by
  590     -Wmissing-field-initializers and -Wunused-parameter.
  591 
  592 2.  The tests for quantifiers being too big (greater than 65535) were being
  593     applied after reading the number, and stupidly assuming that integer
  594     overflow would give a negative number. The tests are now applied as the
  595     numbers are read.
  596 
  597 3.  Tidy code in pcre_exec.c where two branches that used to be different are
  598     now the same.
  599 
  600 4.  The JIT compiler did not generate match limit checks for certain
  601     bracketed expressions with quantifiers. This may lead to exponential
  602     backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
  603     issue should be resolved now.
  604 
  605 5.  Fixed an issue, which occures when nested alternatives are optimized
  606     with table jumps.
  607 
  608 6.  Inserted two casts and changed some ints to size_t in the light of some
  609     reported 64-bit compiler warnings (Bugzilla 1477).
  610 
  611 7.  Fixed a bug concerned with zero-minimum possessive groups that could match
  612     an empty string, which sometimes were behaving incorrectly in the
  613     interpreter (though correctly in the JIT matcher). This pcretest input is
  614     an example:
  615 
  616       '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
  617       NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
  618 
  619     the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
  620     JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
  621     for an empty string was breaking the inner loop and carrying on at a lower
  622     level, when possessive repeated groups should always return to a higher
  623     level as they have no backtrack points in them. The empty string test now
  624     occurs at the outer level.
  625 
  626 8.  Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
  627     ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
  628 
  629 9.  Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
  630     doesn't).
  631 
  632 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
  633     to hold a pattern to be incorrectly computed (too small) when there were
  634     named back references to duplicated names. This could cause "internal
  635     error: code overflow" or "double free or corruption" or other memory
  636     handling errors.
  637 
  638 11. When named subpatterns had the same prefixes, back references could be
  639     confused. For example, in this pattern:
  640 
  641       /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
  642 
  643     the reference to 'Name' was incorrectly treated as a reference to a
  644     duplicate name.
  645 
  646 12. A pattern such as /^s?c/mi8 where the optional character has more than
  647     one "other case" was incorrectly compiled such that it would only try to
  648     match starting at "c".
  649 
  650 13. When a pattern starting with \s was studied, VT was not included in the
  651     list of possible starting characters; this should have been part of the
  652     8.34/18 patch.
  653 
  654 14. If a character class started [\Qx]... where x is any character, the class
  655     was incorrectly terminated at the ].
  656 
  657 15. If a pattern that started with a caseless match for a character with more
  658     than one "other case" was studied, PCRE did not set up the starting code
  659     unit bit map for the list of possible characters. Now it does. This is an
  660     optimization improvement, not a bug fix.
  661 
  662 16. The Unicode data tables have been updated to Unicode 7.0.0.
  663 
  664 17. Fixed a number of memory leaks in pcregrep.
  665 
  666 18. Avoid a compiler warning (from some compilers) for a function call with
  667     a cast that removes "const" from an lvalue by using an intermediate
  668     variable (to which the compiler does not object).
  669 
  670 19. Incorrect code was compiled if a group that contained an internal recursive
  671     back reference was optional (had quantifier with a minimum of zero). This
  672     example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
  673     caused segmentation faults because of stack overflows at compile time.
  674 
  675 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
  676     group that is quantified with an indefinite repeat, caused a compile-time
  677     loop which used up all the system stack and provoked a segmentation fault.
  678     This was not the same bug as 19 above.
  679 
  680 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
  681     Patch by Mike Frysinger.
  682 
  683 
  684 Version 8.35 04-April-2014
  685 --------------------------
  686 
  687 1.  A new flag is set, when property checks are present in an XCLASS.
  688     When this flag is not set, PCRE can perform certain optimizations
  689     such as studying these XCLASS-es.
  690 
  691 2.  The auto-possessification of character sets were improved: a normal
  692     and an extended character set can be compared now. Furthermore
  693     the JIT compiler optimizes more character set checks.
  694 
  695 3.  Got rid of some compiler warnings for potentially uninitialized variables
  696     that show up only when compiled with -O2.
  697 
  698 4.  A pattern such as (?=ab\K) that uses \K in an assertion can set the start
  699     of a match later then the end of the match. The pcretest program was not
  700     handling the case sensibly - it was outputting from the start to the next
  701     binary zero. It now reports this situation in a message, and outputs the
  702     text from the end to the start.
  703 
  704 5.  Fast forward search is improved in JIT. Instead of the first three
  705     characters, any three characters with fixed position can be searched.
  706     Search order: first, last, middle.
  707 
  708 6.  Improve character range checks in JIT. Characters are read by an inprecise
  709     function now, which returns with an unknown value if the character code is
  710     above a certain threshold (e.g: 256). The only limitation is that the value
  711     must be bigger than the threshold as well. This function is useful when
  712     the characters above the threshold are handled in the same way.
  713 
  714 7.  The macros whose names start with RAWUCHAR are placeholders for a future
  715     mode in which only the bottom 21 bits of 32-bit data items are used. To
  716     make this more memorable for those maintaining the code, the names have
  717     been changed to start with UCHAR21, and an extensive comment has been added
  718     to their definition.
  719 
  720 8.  Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
  721     to the export list in Makefile.am (they were accidentally omitted from the
  722     8.34 tarball).
  723 
  724 9.  The informational output from pcretest used the phrase "starting byte set"
  725     which is inappropriate for the 16-bit and 32-bit libraries. As the output
  726     for "first char" and "need char" really means "non-UTF-char", I've changed
  727     "byte" to "char", and slightly reworded the output. The documentation about
  728     these values has also been (I hope) clarified.
  729 
  730 10. Another JIT related optimization: use table jumps for selecting the correct
  731     backtracking path, when more than four alternatives are present inside a
  732     bracket.
  733 
  734 11. Empty match is not possible, when the minimum length is greater than zero,
  735     and there is no \K in the pattern. JIT should avoid empty match checks in
  736     such cases.
  737 
  738 12. In a caseless character class with UCP support, when a character with more
  739     than one alternative case was not the first character of a range, not all
  740     the alternative cases were added to the class. For example, s and \x{17f}
  741     are both alternative cases for S: the class [RST] was handled correctly,
  742     but [R-T] was not.
  743 
  744 13. The configure.ac file always checked for pthread support when JIT was
  745     enabled. This is not used in Windows, so I have put this test inside a
  746     check for the presence of windows.h (which was already tested for).
  747 
  748 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
  749     The algorithm provides a way to skip certain starting offsets, and usually
  750     faster than linear prefix searches.
  751 
  752 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
  753     as for 'fr_FR' and 'french'. For some reason, however, it then used the
  754     Windows-specific input and output files, which have 'french' screwed in.
  755     So this could never have worked. One of the problems with locales is that
  756     they aren't always the same. I have now updated RunTest so that it checks
  757     the output of the locale test (test 3) against three different output
  758     files, and it allows the test to pass if any one of them matches. With luck
  759     this should make the test pass on some versions of Solaris where it was
  760     failing. Because of the uncertainty, the script did not used to stop if
  761     test 3 failed; it now does. If further versions of a French locale ever
  762     come to light, they can now easily be added.
  763 
  764 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
  765     there was a message during ./configure, but it did not stop. This now
  766     provokes an error. The invalid example in README has been corrected.
  767     If a value less than the minimum is given, the minimum value has always
  768     been used, but now a warning is given.
  769 
  770 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
  771     was a bug in the test system, which is now fixed. Also, the list of various
  772     configurations that are tested for each release did not have one with both
  773     16/32 bits and --enable-bar-anycrlf. It now does.
  774 
  775 18. pcretest was missing "-C bsr" for displaying the \R default setting.
  776 
  777 19. Little endian PowerPC systems are supported now by the JIT compiler.
  778 
  779 20. The fast forward newline mechanism could enter to an infinite loop on
  780     certain invalid UTF-8 input. Although we don't support these cases
  781     this issue can be fixed by a performance optimization.
  782 
  783 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
  784     not take account if existing stack usage. There is now a new global
  785     variable called pcre_stack_guard that can be set to point to an external
  786     function to check stack availability. It is called at the start of
  787     processing every parenthesized group.
  788 
  789 22. A typo in the code meant that in ungreedy mode the max/min qualifier
  790     behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
  791     not match "ab".
  792 
  793 23. When UTF was disabled, the JIT program reported some incorrect compile
  794     errors. These messages are silenced now.
  795 
  796 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
  797     compiler.
  798 
  799 25. Change all the temporary files used in RunGrepTest to be different to those
  800     used by RunTest so that the tests can be run simultaneously, for example by
  801     "make -j check".
  802 
  803 
  804 Version 8.34 15-December-2013
  805 -----------------------------
  806 
  807 1.  Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
  808     executable memory. Patch inspired by Carsten Klein.
  809 
  810 2.  ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
  811     this macro is never tested and has no effect, because the work to support
  812     coverage involves only compiling and linking options and special targets in
  813     the Makefile. The comment in config.h implied that defining the macro would
  814     enable coverage support, which is totally false. There was also support for
  815     setting this macro in the CMake files (my fault, I just copied it from
  816     configure). SUPPORT_GCOV has now been removed.
  817 
  818 3.  Make a small performance improvement in strlen16() and strlen32() in
  819     pcretest.
  820 
  821 4.  Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
  822     detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
  823     these cases). There was also one in pcretest.c.
  824 
  825 5.  Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
  826 
  827 6.  In UTF mode, the code for checking whether a group could match an empty
  828     string (which is used for indefinitely repeated groups to allow for
  829     breaking an infinite loop) was broken when the group contained a repeated
  830     negated single-character class with a character that occupied more than one
  831     data item and had a minimum repetition of zero (for example, [^\x{100}]* in
  832     UTF-8 mode). The effect was undefined: the group might or might not be
  833     deemed as matching an empty string, or the program might have crashed.
  834 
  835 7.  The code for checking whether a group could match an empty string was not
  836     recognizing that \h, \H, \v, \V, and \R must match a character.
  837 
  838 8.  Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
  839     an empty string. If it can, pcretest shows this in its information output.
  840 
  841 9.  Fixed two related bugs that applied to Unicode extended grapheme clusters
  842     that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
  843     matched by pcre_exec() without using JIT:
  844 
  845     (a) If the rest of the pattern did not match after a maximal run of
  846         grapheme clusters, the code for backing up to try with fewer of them
  847         did not always back up over a full grapheme when characters that do not
  848         have the modifier quality were involved, e.g. Hangul syllables.
  849 
  850     (b) If the match point in a subject started with modifier character, and
  851         there was no match, the code could incorrectly back up beyond the match
  852         point, and potentially beyond the first character in the subject,
  853         leading to a segfault or an incorrect match result.
  854 
  855 10. A conditional group with an assertion condition could lead to PCRE
  856     recording an incorrect first data item for a match if no other first data
  857     item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
  858     first data item, and therefore matched "ca" after "c" instead of at the
  859     start.
  860 
  861 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
  862     bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
  863 
  864 12. The source of pcregrep now includes z/OS-specific code so that it can be
  865     compiled for z/OS as part of the special z/OS distribution.
  866 
  867 13. Added the -T and -TM options to pcretest.
  868 
  869 14. The code in pcre_compile.c for creating the table of named capturing groups
  870     has been refactored. Instead of creating the table dynamically during the
  871     actual compiling pass, the information is remembered during the pre-compile
  872     pass (on the stack unless there are more than 20 named groups, in which
  873     case malloc() is used) and the whole table is created before the actual
  874     compile happens. This has simplified the code (it is now nearly 150 lines
  875     shorter) and prepared the way for better handling of references to groups
  876     with duplicate names.
  877 
  878 15. A back reference to a named subpattern when there is more than one of the
  879     same name now checks them in the order in which they appear in the pattern.
  880     The first one that is set is used for the reference. Previously only the
  881     first one was inspected. This change makes PCRE more compatible with Perl.
  882 
  883 16. Unicode character properties were updated from Unicode 6.3.0.
  884 
  885 17. The compile-time code for auto-possessification has been refactored, based
  886     on a patch by Zoltan Herczeg. It now happens after instead of during
  887     compilation. The code is cleaner, and more cases are handled. The option
  888     PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
  889     options in pcretest are provided to set it. It can also be set by
  890     (*NO_AUTO_POSSESS) at the start of a pattern.
  891 
  892 18. The character VT has been added to the default ("C" locale) set of
  893     characters that match \s and are generally treated as white space,
  894     following this same change in Perl 5.18. There is now no difference between
  895     "Perl space" and "POSIX space". Whether VT is treated as white space in
  896     other locales depends on the locale.
  897 
  898 19. The code for checking named groups as conditions, either for being set or
  899     for being recursed, has been refactored (this is related to 14 and 15
  900     above). Processing unduplicated named groups should now be as fast at
  901     numerical groups, and processing duplicated groups should be faster than
  902     before.
  903 
  904 20. Two patches to the CMake build system, by Alexander Barkov:
  905 
  906       (1) Replace the "source" command by "." in CMakeLists.txt because
  907           "source" is a bash-ism.
  908 
  909       (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
  910           without these the CMake build does not work on Solaris.
  911 
  912 21. Perl has changed its handling of \8 and \9. If there is no previously
  913     encountered capturing group of those numbers, they are treated as the
  914     literal characters 8 and 9 instead of a binary zero followed by the
  915     literals. PCRE now does the same.
  916 
  917 22. Following Perl, added \o{} to specify codepoints in octal, making it
  918     possible to specify values greater than 0777 and also making them
  919     unambiguous.
  920 
  921 23. Perl now gives an error for missing closing braces after \x{... instead of
  922     treating the string as literal. PCRE now does the same.
  923 
  924 24. RunTest used to grumble if an inappropriate test was selected explicitly,
  925     but just skip it when running all tests. This make it awkward to run ranges
  926     of tests when one of them was inappropriate. Now it just skips any
  927     inappropriate tests, as it always did when running all tests.
  928 
  929 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
  930     character types such as \d or \w, too many callouts were inserted, and the
  931     data that they returned was rubbish.
  932 
  933 26. In UCP mode, \s was not matching two of the characters that Perl matches,
  934     namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
  935     were matched by \h. The code has now been refactored so that the lists of
  936     the horizontal and vertical whitespace characters used for \h and \v (which
  937     are defined only in one place) are now also used for \s.
  938 
  939 27. Add JIT support for the 64 bit TileGX architecture.
  940     Patch by Jiong Wang (Tilera Corporation).
  941 
  942 28. Possessive quantifiers for classes (both explicit and automatically
  943     generated) now use special opcodes instead of wrapping in ONCE brackets.
  944 
  945 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
  946     (because it's meaningless), this was not happening when PCRE_CASELESS was
  947     set. Not wrong, but inefficient.
  948 
  949 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
  950     properties for \w, \d, etc) is present in a test regex. Otherwise if the
  951     test contains no characters greater than 255, Perl doesn't realise it
  952     should be using Unicode semantics.
  953 
  954 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
  955     [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
  956     does in Unicode mode.
  957 
  958 32. Added the "forbid" facility to pcretest so that putting tests into the
  959     wrong test files can sometimes be quickly detected.
  960 
  961 33. There is now a limit (default 250) on the depth of nesting of parentheses.
  962     This limit is imposed to control the amount of system stack used at compile
  963     time. It can be changed at build time by --with-parens-nest-limit=xxx or
  964     the equivalent in CMake.
  965 
  966 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
  967     errors. Perl warns for these when in warning mode, but PCRE has no facility
  968     for giving warnings.
  969 
  970 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
  971     However, this was not working for (?!) because it is optimized to (*FAIL),
  972     for which PCRE does not allow quantifiers. The optimization is now disabled
  973     when a quantifier follows (?!). I can't see any use for this, but it makes
  974     things uniform.
  975 
  976 36. Perl no longer allows group names to start with digits, so I have made this
  977     change also in PCRE. It simplifies the code a bit.
  978 
  979 37. In extended mode, Perl ignores spaces before a + that indicates a
  980     possessive quantifier. PCRE allowed a space before the quantifier, but not
  981     before the possessive +. It now does.
  982 
  983 38. The use of \K (reset reported match start) within a repeated possessive
  984     group such as (a\Kb)*+ was not working.
  985 
  986 40. Document that the same character tables must be used at compile time and
  987     run time, and that the facility to pass tables to pcre_exec() and
  988     pcre_dfa_exec() is for use only with saved/restored patterns.
  989 
  990 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
  991     features for Builds with MSVC:
  992 
  993     1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
  994        builds). The .rc files can be used to set FileDescription and many other
  995        attributes.
  996 
  997     2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
  998        This allows higher-level build scripts which want .pdb files to avoid
  999        hard-coding the exact files needed."
 1000 
 1001 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
 1002     mean "start of word" and "end of word", respectively, as a transition aid.
 1003 
 1004 43. A minimizing repeat of a class containing codepoints greater than 255 in
 1005     non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
 1006     compiled to use the heap for recursion.
 1007 
 1008 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
 1009     is configured.
 1010 
 1011 
 1012 Version 8.33 28-May-2013
 1013 ------------------------
 1014 
 1015 1.  Added 'U' to some constants that are compared to unsigned integers, to
 1016     avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
 1017     variables that are added to signed variables, to ensure the result is
 1018     signed and can be negated.
 1019 
 1020 2.  Applied patch by Daniel Richard G for quashing MSVC warnings to the
 1021     CMake config files.
 1022 
 1023 3.  Revise the creation of config.h.generic so that all boolean macros are
 1024     #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
 1025     overriding via -D on the command line possible.
 1026 
 1027 4.  Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
 1028     to unsigned int is reported to make a quite noticeable speed difference in
 1029     a specific Windows environment. Testing on Linux did also appear to show
 1030     some benefit (and it is clearly not harmful). Also fixed the definition of
 1031     Xop which should be unsigned.
 1032 
 1033 5.  Related to (4), changing the definition of the intermediate variable cc
 1034     in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
 1035     improvements.
 1036 
 1037 6.  Fix forward search in JIT when link size is 3 or greater. Also removed some
 1038     unnecessary spaces.
 1039 
 1040 7.  Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
 1041     and later.
 1042 
 1043 8.  Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
 1044 
 1045 9.  Optimizing fast_forward_start_bits in JIT.
 1046 
 1047 10. Adding support for callouts in JIT, and fixing some issues revealed
 1048     during this work. Namely:
 1049 
 1050     (a) Unoptimized capturing brackets incorrectly reset on backtrack.
 1051 
 1052     (b) Minimum length was not checked before the matching is started.
 1053 
 1054 11. The value of capture_last that is passed to callouts was incorrect in some
 1055     cases when there was a capture on one path that was subsequently abandoned
 1056     after a backtrack. Also, the capture_last value is now reset after a
 1057     recursion, since all captures are also reset in this case.
 1058 
 1059 12. The interpreter no longer returns the "too many substrings" error in the
 1060     case when an overflowing capture is in a branch that is subsequently
 1061     abandoned after a backtrack.
 1062 
 1063 13. In the pathological case when an offset vector of size 2 is used, pcretest
 1064     now prints out the matched string after a yield of 0 or 1.
 1065 
 1066 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
 1067     Only supported by the JIT compiler at the moment.
 1068 
 1069 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
 1070 
 1071 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
 1072     offset of the starting point of the matching process, provided the offsets
 1073     vector is large enough.
 1074 
 1075 17. The \A escape now records a lookbehind value of 1, though its execution
 1076     does not actually inspect the previous character. This is to ensure that,
 1077     in partial multi-segment matching, at least one character from the old
 1078     segment is retained when a new segment is processed. Otherwise, if there
 1079     are no lookbehinds in the pattern, \A might match incorrectly at the start
 1080     of a new segment.
 1081 
 1082 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
 1083 
 1084 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
 1085     gives some modest performance improvement in 8-bit mode.
 1086 
 1087 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
 1088     be expressed in certain programming languages using Universal Character
 1089     Names.
 1090 
 1091 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
 1092     which points out that "non characters" are not "characters that may not
 1093     appear in Unicode strings" but rather "characters that are reserved for
 1094     internal use and have only local meaning".
 1095 
 1096 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
 1097     there was a conditional group that depended on an assertion, if the
 1098     assertion was false, the callout that immediately followed the alternation
 1099     in the condition was skipped when pcre_exec() was used for matching.
 1100 
 1101 23. Allow an explicit callout to be inserted before an assertion that is the
 1102     condition for a conditional group, for compatibility with automatic
 1103     callouts, which always insert a callout at this point.
 1104 
 1105 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
 1106     confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
 1107 
 1108 25. (*PRUNE) is now supported by the JIT compiler.
 1109 
 1110 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
 1111 
 1112 27. Fix the case where there are two or more SKIPs with arguments that may be
 1113     ignored.
 1114 
 1115 28. (*SKIP) is now supported by the JIT compiler.
 1116 
 1117 29. (*THEN) is now supported by the JIT compiler.
 1118 
 1119 30. Update RunTest with additional test selector options.
 1120 
 1121 31. The way PCRE handles backtracking verbs has been changed in two ways.
 1122 
 1123     (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
 1124     SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
 1125     backtracking. In some cases this makes it more Perl-compatible, but Perl's
 1126     rather obscure rules do not always do the same thing.
 1127 
 1128     (2) Previously, backtracking verbs were confined within assertions. This is
 1129     no longer the case for positive assertions, except for (*ACCEPT). Again,
 1130     this sometimes improves Perl compatibility, and sometimes does not.
 1131 
 1132 32. A number of tests that were in test 2 because Perl did things differently
 1133     have been moved to test 1, because either Perl or PCRE has changed, and
 1134     these tests are now compatible.
 1135 
 1136 32. Backtracking control verbs are now handled in the same way in JIT and
 1137     interpreter.
 1138 
 1139 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
 1140     contained a forward subroutine reference caused a compile error.
 1141 
 1142 34. Auto-detect and optimize limited repetitions in JIT.
 1143 
 1144 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
 1145     blocking (*UTF) etc.
 1146 
 1147 36. In the interpreter, maximizing pattern repetitions for characters and
 1148     character types now use tail recursion, which reduces stack usage.
 1149 
 1150 37. The value of the max lookbehind was not correctly preserved if a compiled
 1151     and saved regex was reloaded on a host of different endianness.
 1152 
 1153 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
 1154     of the compiled pattern block, expand the flags field from 16 to 32 bits
 1155     because it was almost full.
 1156 
 1157 39. Try madvise first before posix_madvise.
 1158 
 1159 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
 1160     with a pattern such as ^$. It has taken 4 years for anybody to notice! The
 1161     original change locked out all matches of empty strings. This has been
 1162     changed so that one match of an empty string per line is recognized.
 1163     Subsequent searches on the same line (for colouring or for --only-matching,
 1164     for example) do not recognize empty strings.
 1165 
 1166 41. Applied a user patch to fix a number of spelling mistakes in comments.
 1167 
 1168 42. Data lines longer than 65536 caused pcretest to crash.
 1169 
 1170 43. Clarified the data type for length and startoffset arguments for pcre_exec
 1171     and pcre_dfa_exec in the function-specific man pages, where they were
 1172     explicitly stated to be in bytes, never having been updated. I also added
 1173     some clarification to the pcreapi man page.
 1174 
 1175 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
 1176     a segmentation fault.
 1177 
 1178 
 1179 Version 8.32 30-November-2012
 1180 -----------------------------
 1181 
 1182 1.  Improved JIT compiler optimizations for first character search and single
 1183     character iterators.
 1184 
 1185 2.  Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
 1186     Patch by Daniel Richard G.
 1187 
 1188 3.  Single character iterator optimizations in the JIT compiler.
 1189 
 1190 4.  Improved JIT compiler optimizations for character ranges.
 1191 
 1192 5.  Rename the "leave" variable names to "quit" to improve WinCE compatibility.
 1193     Reported by Giuseppe D'Angelo.
 1194 
 1195 6.  The PCRE_STARTLINE bit, indicating that a match can occur only at the start
 1196     of a line, was being set incorrectly in cases where .* appeared inside
 1197     atomic brackets at the start of a pattern, or where there was a subsequent
 1198     *PRUNE or *SKIP.
 1199 
 1200 7.  Improved instruction cache flush for POWER/PowerPC.
 1201     Patch by Daniel Richard G.
 1202 
 1203 8.  Fixed a number of issues in pcregrep, making it more compatible with GNU
 1204     grep:
 1205 
 1206     (a) There is now no limit to the number of patterns to be matched.
 1207 
 1208     (b) An error is given if a pattern is too long.
 1209 
 1210     (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
 1211         are now supported.
 1212 
 1213     (d) --exclude-from and --include-from (multiple use) have been added.
 1214 
 1215     (e) Exclusions and inclusions now apply to all files and directories, not
 1216         just to those obtained from scanning a directory recursively.
 1217 
 1218     (f) Multiple uses of -f and --file-list are now supported.
 1219 
 1220     (g) In a Windows environment, the default for -d has been changed from
 1221         "read" (the GNU grep default) to "skip", because otherwise the presence
 1222         of a directory in the file list provokes an error.
 1223 
 1224     (h) The documentation has been revised and clarified in places.
 1225 
 1226 9.  Improve the matching speed of capturing brackets.
 1227 
 1228 10. Changed the meaning of \X so that it now matches a Unicode extended
 1229     grapheme cluster.
 1230 
 1231 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
 1232     out POSIX threads when JIT support is configured.
 1233 
 1234 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
 1235 
 1236 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
 1237     structure could go wrong in environments where size_t is not the same size
 1238     as int.
 1239 
 1240 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
 1241     set.
 1242 
 1243 15. The EBCDIC support had decayed; later updates to the code had included
 1244     explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
 1245     general tidy up of EBCDIC-related issues, and the documentation was also
 1246     not quite right. There is now a test that can be run on ASCII systems to
 1247     check some of the EBCDIC-related things (but is it not a full test).
 1248 
 1249 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
 1250     in a small tidy to the code.
 1251 
 1252 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
 1253 
 1254 18. If the --only-matching (-o) option in pcregrep is specified multiple
 1255     times, each one causes appropriate output. For example, -o1 -o2 outputs the
 1256     substrings matched by the 1st and 2nd capturing parentheses. A separating
 1257     string can be specified by --om-separator (default empty).
 1258 
 1259 19. Improving the first n character searches.
 1260 
 1261 20. Turn case lists for horizontal and vertical white space into macros so that
 1262     they are defined only once.
 1263 
 1264 21. This set of changes together give more compatible Unicode case-folding
 1265     behaviour for characters that have more than one other case when UCP
 1266     support is available.
 1267 
 1268     (a) The Unicode property table now has offsets into a new table of sets of
 1269         three or more characters that are case-equivalent. The MultiStage2.py
 1270         script that generates these tables (the pcre_ucd.c file) now scans
 1271         CaseFolding.txt instead of UnicodeData.txt for character case
 1272         information.
 1273 
 1274     (b) The code for adding characters or ranges of characters to a character
 1275         class has been abstracted into a generalized function that also handles
 1276         case-independence. In UTF-mode with UCP support, this uses the new data
 1277         to handle characters with more than one other case.
 1278 
 1279     (c) A bug that is fixed as a result of (b) is that codepoints less than 256
 1280         whose other case is greater than 256 are now correctly matched
 1281         caselessly. Previously, the high codepoint matched the low one, but not
 1282         vice versa.
 1283 
 1284     (d) The processing of \h, \H, \v, and \ in character classes now makes use
 1285         of the new class addition function, using character lists defined as
 1286         macros alongside the case definitions of 20 above.
 1287 
 1288     (e) Caseless back references now work with characters that have more than
 1289         one other case.
 1290 
 1291     (f) General caseless matching of characters with more than one other case
 1292         is supported.
 1293 
 1294 22. Unicode character properties were updated from Unicode 6.2.0
 1295 
 1296 23. Improved CMake support under Windows. Patch by Daniel Richard G.
 1297 
 1298 24. Add support for 32-bit character strings, and UTF-32
 1299 
 1300 25. Major JIT compiler update (code refactoring and bugfixing).
 1301     Experimental Sparc 32 support is added.
 1302 
 1303 26. Applied a modified version of Daniel Richard G's patch to create
 1304     pcre.h.generic and config.h.generic by "make" instead of in the
 1305     PrepareRelease script.
 1306 
 1307 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
 1308     pcre_compile.c when checking for a zero character.
 1309 
 1310 28. Introducing a native interface for JIT. Through this interface, the compiled
 1311     machine code can be directly executed. The purpose of this interface is to
 1312     provide fast pattern matching, so several sanity checks are not performed.
 1313     However, feature tests are still performed. The new interface provides
 1314     1.4x speedup compared to the old one.
 1315 
 1316 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
 1317     the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
 1318     was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
 1319 
 1320 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
 1321     greater than 0x7fffffff (which cannot be represented in UTF-8, even under
 1322     the "old" RFC 2279). Instead, it ended up passing a negative length to
 1323     pcre_exec().
 1324 
 1325 31. Add support for GCC's visibility feature to hide internal functions.
 1326 
 1327 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
 1328     "unknown -C option" after outputting 0 or 1.
 1329 
 1330 33. There is now support for generating a code coverage report for the test
 1331     suite in environments where gcc is the compiler and lcov is installed. This
 1332     is mainly for the benefit of the developers.
 1333 
 1334 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
 1335     unaddressable using valgrind annotations, allowing valgrind to detect
 1336     invalid memory accesses. This is mainly for the benefit of the developers.
 1337 
 1338 25. (*UTF) can now be used to start a pattern in any of the three libraries.
 1339 
 1340 26. Give configure error if --enable-cpp but no C++ compiler found.
 1341 
 1342 
 1343 Version 8.31 06-July-2012
 1344 -------------------------
 1345 
 1346 1.  Fixing a wrong JIT test case and some compiler warnings.
 1347 
 1348 2.  Removed a bashism from the RunTest script.
 1349 
 1350 3.  Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
 1351     to unsigned type, result still unsigned" that was given by an MS compiler
 1352     on encountering the code "-sizeof(xxx)".
 1353 
 1354 4.  Partial matching support is added to the JIT compiler.
 1355 
 1356 5.  Fixed several bugs concerned with partial matching of items that consist
 1357     of more than one character:
 1358 
 1359     (a) /^(..)\1/ did not partially match "aba" because checking references was
 1360         done on an "all or nothing" basis. This also applied to repeated
 1361         references.
 1362 
 1363     (b) \R did not give a hard partial match if \r was found at the end of the
 1364         subject.
 1365 
 1366     (c) \X did not give a hard partial match after matching one or more
 1367         characters at the end of the subject.
 1368 
 1369     (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
 1370         a partial match for the string "\r".
 1371 
 1372     (e) When newline was set to CRLF, the metacharacter "." did not recognize
 1373         a partial match for a CR character at the end of the subject string.
 1374 
 1375 6.  If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
 1376     running pcretest, the text "(JIT)" added to the output whenever JIT is
 1377     actually used to run the match.
 1378 
 1379 7.  Individual JIT compile options can be set in pcretest by following -s+[+]
 1380     or /S+[+] with a digit between 1 and 7.
 1381 
 1382 8.  OP_NOT now supports any UTF character not just single-byte ones.
 1383 
 1384 9.  (*MARK) control verb is now supported by the JIT compiler.
 1385 
 1386 10. The command "./RunTest list" lists the available tests without actually
 1387     running any of them. (Because I keep forgetting what they all are.)
 1388 
 1389 11. Add PCRE_INFO_MAXLOOKBEHIND.
 1390 
 1391 12. Applied a (slightly modified) user-supplied patch that improves performance
 1392     when the heap is used for recursion (compiled with --disable-stack-for-
 1393     recursion). Instead of malloc and free for each heap frame each time a
 1394     logical recursion happens, frames are retained on a chain and re-used where
 1395     possible. This sometimes gives as much as 30% improvement.
 1396 
 1397 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
 1398     call.
 1399 
 1400 14. As documented, (*COMMIT) is now confined to within a positive assertion.
 1401 
 1402 15. It is now possible to link pcretest with libedit as an alternative to
 1403     libreadline.
 1404 
 1405 16. (*COMMIT) control verb is now supported by the JIT compiler.
 1406 
 1407 17. The Unicode data tables have been updated to Unicode 6.1.0.
 1408 
 1409 18. Added --file-list option to pcregrep.
 1410 
 1411 19. Added binary file support to pcregrep, including the -a, --binary-files,
 1412     -I, and --text options.
 1413 
 1414 20. The madvise function is renamed for posix_madvise for QNX compatibility
 1415     reasons. Fixed by Giuseppe D'Angelo.
 1416 
 1417 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
 1418 
 1419 22. Changed the option for creating symbolic links for 16-bit man pages from
 1420     -s to -sf so that re-installing does not cause issues.
 1421 
 1422 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
 1423 
 1424 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
 1425     matches in certain environments (the workspace was not being correctly
 1426     retained). Also added to pcre_dfa_exec() a simple plausibility check on
 1427     some of the workspace data at the beginning of a restart.
 1428 
 1429 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
 1430     was not doing so when it should - probably a typo introduced by SVN 528
 1431     (change 8.10/14).
 1432 
 1433 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
 1434     \w+ when the character tables indicated that \x{c4} was a word character.
 1435     There were several related cases, all because the tests for doing a table
 1436     lookup were testing for characters less than 127 instead of 255.
 1437 
 1438 27. If a pattern contains capturing parentheses that are not used in a match,
 1439     their slots in the ovector are set to -1. For those that are higher than
 1440     any matched groups, this happens at the end of processing. In the case when
 1441     there were back references that the ovector was too small to contain
 1442     (causing temporary malloc'd memory to be used during matching), and the
 1443     highest capturing number was not used, memory off the end of the ovector
 1444     was incorrectly being set to -1. (It was using the size of the temporary
 1445     memory instead of the true size.)
 1446 
 1447 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
 1448     ovector size, it uses memory at the end of the block that it has got.
 1449 
 1450 29. Check for an overlong MARK name and give an error at compile time. The
 1451     limit is 255 for the 8-bit library and 65535 for the 16-bit library.
 1452 
 1453 30. JIT compiler update.
 1454 
 1455 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
 1456     Rill for the patch.
 1457 
 1458 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
 1459 
 1460 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
 1461 
 1462 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
 1463     SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
 1464     was enabled.
 1465 
 1466 35. Improve JIT code generation for greedy plus quantifier.
 1467 
 1468 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
 1469     1 to "aa" instead of to an empty string. The bug affected repeated groups
 1470     that could potentially match an empty string.
 1471 
 1472 37. Optimizing single character iterators in JIT.
 1473 
 1474 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
 1475     the same checks as \x{...} characters in non-JavaScript mode. Specifically,
 1476     codepoints that are too big for the mode are faulted, and in a UTF mode,
 1477     disallowed codepoints are also faulted.
 1478 
 1479 39. If PCRE was compiled with UTF support, in three places in the DFA
 1480     matcher there was code that should only have been obeyed in UTF mode, but
 1481     was being obeyed unconditionally. In 8-bit mode this could cause incorrect
 1482     processing when bytes with values greater than 127 were present. In 16-bit
 1483     mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
 1484     both cases the values are those that cannot be the first data item in a UTF
 1485     character. The three items that might have provoked this were recursions,
 1486     possessively repeated groups, and atomic groups.
 1487 
 1488 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
 1489     and pcregrep, because some OS require shared objects to be explicitly
 1490     passed to ld, causing the link step to fail if they are not.
 1491 
 1492 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
 1493     mode, patterns that started with \h* or \R* might be incorrectly matched.
 1494 
 1495 
 1496 Version 8.30 04-February-2012
 1497 -----------------------------
 1498 
 1499 1.  Renamed "isnumber" as "is_a_number" because in some Mac environments this
 1500     name is defined in ctype.h.
 1501 
 1502 2.  Fixed a bug in fixed-length calculation for lookbehinds that would show up
 1503     only in quite long subpatterns.
 1504 
 1505 3.  Removed the function pcre_info(), which has been obsolete and deprecated
 1506     since it was replaced by pcre_fullinfo() in February 2000.
 1507 
 1508 4.  For a non-anchored pattern, if (*SKIP) was given with a name that did not
 1509     match a (*MARK), and the match failed at the start of the subject, a
 1510     reference to memory before the start of the subject could occur. This bug
 1511     was introduced by fix 17 of release 8.21.
 1512 
 1513 5.  A reference to an unset group with zero minimum repetition was giving
 1514     totally wrong answers (in non-JavaScript-compatibility mode). For example,
 1515     /(another)?(\1?)test/ matched against "hello world test". This bug was
 1516     introduced in release 8.13.
 1517 
 1518 6.  Add support for 16-bit character strings (a large amount of work involving
 1519     many changes and refactorings).
 1520 
 1521 7.  RunGrepTest failed on msys because \r\n was replaced by whitespace when the
 1522     command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
 1523     from a file.
 1524 
 1525 8.  Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
 1526     rounding is not applied in this particular case).
 1527 
 1528 9.  The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
 1529     if they appear, or are escaped, in patterns.
 1530 
 1531 10. Get rid of a number of -Wunused-but-set-variable warnings.
 1532 
 1533 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
 1534     "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
 1535     Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
 1536     also returns the mark "x". This bug applied to capturing parentheses,
 1537     non-capturing parentheses, and atomic parentheses. It also applied to some
 1538     assertions.
 1539 
 1540 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
 1541     information out of configure.ac instead of relying on pcre.h.generic, which
 1542     is not stored in the repository.
 1543 
 1544 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
 1545     -lreadline.
 1546 
 1547 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
 1548 
 1549 15. Applied Graycode's patch to put the top-level frame on the stack rather
 1550     than the heap when not using the stack for recursion. This gives a
 1551     performance improvement in many cases when recursion is not deep.
 1552 
 1553 16. Experimental code added to "pcretest -C" to output the stack frame size.
 1554 
 1555 
 1556 Version 8.21 12-Dec-2011
 1557 ------------------------
 1558 
 1559 1.  Updating the JIT compiler.
 1560 
 1561 2.  JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
 1562     are added as well.
 1563 
 1564 3.  Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
 1565     PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
 1566     calling _pcre_jit_exec. Some extra comments are added.
 1567 
 1568 4.  (*MARK) settings inside atomic groups that do not contain any capturing
 1569     parentheses, for example, (?>a(*:m)), were not being passed out. This bug
 1570     was introduced by change 18 for 8.20.
 1571 
 1572 5.  Supporting of \x, \U and \u in JavaScript compatibility mode based on the
 1573     ECMA-262 standard.
 1574 
 1575 6.  Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
 1576     erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
 1577     This bug was probably introduced by change 9 of 8.13.
 1578 
 1579 7.  While fixing 6 above, I noticed that a number of other items were being
 1580     incorrectly rejected as "not fixed length". This arose partly because newer
 1581     opcodes had not been added to the fixed-length checking code. I have (a)
 1582     corrected the bug and added tests for these items, and (b) arranged for an
 1583     error to occur if an unknown opcode is encountered while checking for fixed
 1584     length instead of just assuming "not fixed length". The items that were
 1585     rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
 1586     (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
 1587     repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
 1588 
 1589 8.  A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
 1590     being incorrectly compiled and would have given unpredicatble results.
 1591 
 1592 9.  A possessively repeated subpattern with minimum repeat count greater than
 1593     one behaved incorrectly. For example, (A){2,}+ behaved as if it was
 1594     (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
 1595     the first (A) could occur when it should not.
 1596 
 1597 10. Add a cast and remove a redundant test from the code.
 1598 
 1599 11. JIT should use pcre_malloc/pcre_free for allocation.
 1600 
 1601 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
 1602     best practice nowadays, and helps with cross-compiling. (If the exec_prefix
 1603     is anything other than /usr, -L is still shown).
 1604 
 1605 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
 1606 
 1607 14. Perl does not support \N without a following name in a [] class; PCRE now
 1608     also gives an error.
 1609 
 1610 15. If a forward reference was repeated with an upper limit of around 2000,
 1611     it caused the error "internal error: overran compiling workspace". The
 1612     maximum number of forward references (including repeats) was limited by the
 1613     internal workspace, and dependent on the LINK_SIZE. The code has been
 1614     rewritten so that the workspace expands (via pcre_malloc) if necessary, and
 1615     the default depends on LINK_SIZE. There is a new upper limit (for safety)
 1616     of around 200,000 forward references. While doing this, I also speeded up
 1617     the filling in of repeated forward references.
 1618 
 1619 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
 1620     incorrectly expecting the subject to contain another "a" after the start.
 1621 
 1622 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
 1623     in the match, the SKIP should be ignored. This was not happening; instead
 1624     the SKIP was being treated as NOMATCH. For patterns such as
 1625     /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
 1626     tested.
 1627 
 1628 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
 1629     now much more compatible with Perl, in particular in cases where the result
 1630     is a non-match for a non-anchored pattern. For example, if
 1631     /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
 1632     "m", where previously it did not return a name. A side effect of this
 1633     change is that for partial matches, the last encountered mark name is
 1634     returned, as for non matches. A number of tests that were previously not
 1635     Perl-compatible have been moved into the Perl-compatible test files. The
 1636     refactoring has had the pleasing side effect of removing one argument from
 1637     the match() function, thus reducing its stack requirements.
 1638 
 1639 19. If the /S+ option was used in pcretest to study a pattern using JIT,
 1640     subsequent uses of /S (without +) incorrectly behaved like /S+.
 1641 
 1642 21. Retrieve executable code size support for the JIT compiler and fixing
 1643     some warnings.
 1644 
 1645 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
 1646     not work when the shorter character appeared right at the end of the
 1647     subject string.
 1648 
 1649 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
 1650     systems.
 1651 
 1652 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
 1653     output it when the /M option is used in pcretest.
 1654 
 1655 25. The CheckMan script was not being included in the distribution. Also, added
 1656     an explicit "perl" to run Perl scripts from the PrepareRelease script
 1657     because this is reportedly needed in Windows.
 1658 
 1659 26. If study data was being save in a file and studying had not found a set of
 1660     "starts with" bytes for the pattern, the data written to the file (though
 1661     never used) was taken from uninitialized memory and so caused valgrind to
 1662     complain.
 1663 
 1664 27. Updated RunTest.bat as provided by Sheri Pierce.
 1665 
 1666 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
 1667 
 1668 29. Computation of memory usage for the table of capturing group names was
 1669     giving an unnecessarily large value.
 1670 
 1671 
 1672 Version 8.20 21-Oct-2011
 1673 ------------------------
 1674 
 1675 1.  Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
 1676     a POSIX class. After further experiments with Perl, which convinced me that
 1677     Perl has bugs and confusions, a closing square bracket is no longer allowed
 1678     in a POSIX name. This bug also affected patterns with classes that started
 1679     with full stops.
 1680 
 1681 2.  If a pattern such as /(a)b|ac/ is matched against "ac", there is no
 1682     captured substring, but while checking the failing first alternative,
 1683     substring 1 is temporarily captured. If the output vector supplied to
 1684     pcre_exec() was not big enough for this capture, the yield of the function
 1685     was still zero ("insufficient space for captured substrings"). This cannot
 1686     be totally fixed without adding another stack variable, which seems a lot
 1687     of expense for a edge case. However, I have improved the situation in cases
 1688     such as /(a)(b)x|abc/ matched against "abc", where the return code
 1689     indicates that fewer than the maximum number of slots in the ovector have
 1690     been set.
 1691 
 1692 3.  Related to (2) above: when there are more back references in a pattern than
 1693     slots in the output vector, pcre_exec() uses temporary memory during
 1694     matching, and copies in the captures as far as possible afterwards. It was
 1695     using the entire output vector, but this conflicts with the specification
 1696     that only 2/3 is used for passing back captured substrings. Now it uses
 1697     only the first 2/3, for compatibility. This is, of course, another edge
 1698     case.
 1699 
 1700 4.  Zoltan Herczeg's just-in-time compiler support has been integrated into the
 1701     main code base, and can be used by building with --enable-jit. When this is
 1702     done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
 1703     runtime --no-jit option is given.
 1704 
 1705 5.  When the number of matches in a pcre_dfa_exec() run exactly filled the
 1706     ovector, the return from the function was zero, implying that there were
 1707     other matches that did not fit. The correct "exactly full" value is now
 1708     returned.
 1709 
 1710 6.  If a subpattern that was called recursively or as a subroutine contained
 1711     (*PRUNE) or any other control that caused it to give a non-standard return,
 1712     invalid errors such as "Error -26 (nested recursion at the same subject
 1713     position)" or even infinite loops could occur.
 1714 
 1715 7.  If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
 1716     computing the minimum length on reaching *ACCEPT, and so ended up with the
 1717     wrong value of 1 rather than 0. Further investigation indicates that
 1718     computing a minimum subject length in the presence of *ACCEPT is difficult
 1719     (think back references, subroutine calls), and so I have changed the code
 1720     so that no minimum is registered for a pattern that contains *ACCEPT.
 1721 
 1722 8.  If (*THEN) was present in the first (true) branch of a conditional group,
 1723     it was not handled as intended. [But see 16 below.]
 1724 
 1725 9.  Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
 1726     Sheri Pierce.
 1727 
 1728 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
 1729     the first byte in a match must be "a".
 1730 
 1731 11. Change 17 for 8.13 increased the recursion depth for patterns like
 1732     /a(?:.)*?a/ drastically. I've improved things by remembering whether a
 1733     pattern contains any instances of (*THEN). If it does not, the old
 1734     optimizations are restored. It would be nice to do this on a per-group
 1735     basis, but at the moment that is not feasible.
 1736 
 1737 12. In some environments, the output of pcretest -C is CRLF terminated. This
 1738     broke RunTest's code that checks for the link size. A single white space
 1739     character after the value is now allowed for.
 1740 
 1741 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
 1742     For "fr", it uses the Windows-specific input and output files.
 1743 
 1744 14. If (*THEN) appeared in a group that was called recursively or as a
 1745     subroutine, it did not work as intended. [But see next item.]
 1746 
 1747 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
 1748     pattern fragments (but not containing any | characters). If A and B are
 1749     matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
 1750     was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
 1751     D. In other words, Perl considers parentheses that do not contain any |
 1752     characters to be part of a surrounding alternative, whereas PCRE was
 1753     treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
 1754     differently. PCRE now behaves in the same way as Perl, except in the case
 1755     of subroutine/recursion calls such as (?1) which have in any case always
 1756     been different (but PCRE had them first :-).
 1757 
 1758 16. Related to 15 above: Perl does not treat the | in a conditional group as
 1759     creating alternatives. Such a group is treated in the same way as an
 1760     ordinary group without any | characters when processing (*THEN). PCRE has
 1761     been changed to match Perl's behaviour.
 1762 
 1763 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
 1764     RunGrepTest script failed.
 1765 
 1766 18. Change 22 for version 13 caused atomic groups to use more stack. This is
 1767     inevitable for groups that contain captures, but it can lead to a lot of
 1768     stack use in large patterns. The old behaviour has been restored for atomic
 1769     groups that do not contain any capturing parentheses.
 1770 
 1771 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
 1772     suppress the check for a minimum subject length at run time. (If it was
 1773     given to pcre_exec() or pcre_dfa_exec() it did work.)
 1774 
 1775 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
 1776     fail to work when decoding hex characters in data strings in EBCDIC
 1777     environments.
 1778 
 1779 21. It appears that in at least one Mac OS environment, the isxdigit() function
 1780     is implemented as a macro that evaluates to its argument more than once,
 1781     contravening the C 90 Standard (I haven't checked a later standard). There
 1782     was an instance in pcretest which caused it to go wrong when processing
 1783     \x{...} escapes in subject strings. The has been rewritten to avoid using
 1784     things like p++ in the argument of isxdigit().
 1785 
 1786 
 1787 Version 8.13 16-Aug-2011
 1788 ------------------------
 1789 
 1790 1.  The Unicode data tables have been updated to Unicode 6.0.0.
 1791 
 1792 2.  Two minor typos in pcre_internal.h have been fixed.
 1793 
 1794 3.  Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
 1795     pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
 1796     in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
 1797 
 1798 4.  There were a number of related bugs in the code for matching backrefences
 1799     caselessly in UTF-8 mode when codes for the characters concerned were
 1800     different numbers of bytes. For example, U+023A and U+2C65 are an upper
 1801     and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
 1802     (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
 1803     code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
 1804     2-byte code at the end of the subject (it thought there wasn't enough data
 1805     left).
 1806 
 1807 5.  Comprehensive information about what went wrong is now returned by
 1808     pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
 1809     as the output vector has at least 2 elements. The offset of the start of
 1810     the failing character and a reason code are placed in the vector.
 1811 
 1812 6.  When the UTF-8 string check fails for pcre_compile(), the offset that is
 1813     now returned is for the first byte of the failing character, instead of the
 1814     last byte inspected. This is an incompatible change, but I hope it is small
 1815     enough not to be a problem. It makes the returned offset consistent with
 1816     pcre_exec() and pcre_dfa_exec().
 1817 
 1818 7.  pcretest now gives a text phrase as well as the error number when
 1819     pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
 1820     failure, the offset and reason code are output.
 1821 
 1822 8.  When \R was used with a maximizing quantifier it failed to skip backwards
 1823     over a \r\n pair if the subsequent match failed. Instead, it just skipped
 1824     back over a single character (\n). This seems wrong (because it treated the
 1825     two characters as a single entity when going forwards), conflicts with the
 1826     documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
 1827     behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
 1828     has been changed.
 1829 
 1830 9.  Some internal refactoring has changed the processing so that the handling
 1831     of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
 1832     time (the PCRE_DOTALL option was changed this way some time ago: version
 1833     7.7 change 16). This has made it possible to abolish the OP_OPT op code,
 1834     which was always a bit of a fudge. It also means that there is one less
 1835     argument for the match() function, which reduces its stack requirements
 1836     slightly. This change also fixes an incompatibility with Perl: the pattern
 1837     (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
 1838 
 1839 10. More internal refactoring has drastically reduced the number of recursive
 1840     calls to match() for possessively repeated groups such as (abc)++ when
 1841     using pcre_exec().
 1842 
 1843 11. While implementing 10, a number of bugs in the handling of groups were
 1844     discovered and fixed:
 1845 
 1846     (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
 1847     (a|)*(?1) gave a compile-time internal error.
 1848     ((a|)+)+  did not notice that the outer group could match an empty string.
 1849     (^a|^)+   was not marked as anchored.
 1850     (.*a|.*)+ was not marked as matching at start or after a newline.
 1851 
 1852 12. Yet more internal refactoring has removed another argument from the match()
 1853     function. Special calls to this function are now indicated by setting a
 1854     value in a variable in the "match data" data block.
 1855 
 1856 13. Be more explicit in pcre_study() instead of relying on "default" for
 1857     opcodes that mean there is no starting character; this means that when new
 1858     ones are added and accidentally left out of pcre_study(), testing should
 1859     pick them up.
 1860 
 1861 14. The -s option of pcretest has been documented for ages as being an old
 1862     synonym of -m (show memory usage). I have changed it to mean "force study
 1863     for every regex", that is, assume /S for every regex. This is similar to -i
 1864     and -d etc. It's slightly incompatible, but I'm hoping nobody is still
 1865     using it. It makes it easier to run collections of tests with and without
 1866     study enabled, and thereby test pcre_study() more easily. All the standard
 1867     tests are now run with and without -s (but some patterns can be marked as
 1868     "never study" - see 20 below).
 1869 
 1870 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
 1871     restoration of the capturing data to the outer values was not happening
 1872     correctly.
 1873 
 1874 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
 1875     empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
 1876     pattern had matched an empty string, and so incorrectly returned a no
 1877     match.
 1878 
 1879 17. There was optimizing code for the last branch of non-capturing parentheses,
 1880     and also for the obeyed branch of a conditional subexpression, which used
 1881     tail recursion to cut down on stack usage. Unfortunately, now that there is
 1882     the possibility of (*THEN) occurring in these branches, tail recursion is
 1883     no longer possible because the return has to be checked for (*THEN). These
 1884     two optimizations have therefore been removed. [But see 8.20/11 above.]
 1885 
 1886 18. If a pattern containing \R was studied, it was assumed that \R always
 1887     matched two bytes, thus causing the minimum subject length to be
 1888     incorrectly computed because \R can also match just one byte.
 1889 
 1890 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
 1891     was incorrectly computed.
 1892 
 1893 20. If /S is present twice on a test pattern in pcretest input, it now
 1894     *disables* studying, thereby overriding the use of -s on the command line
 1895     (see 14 above). This is necessary for one or two tests to keep the output
 1896     identical in both cases.
 1897 
 1898 21. When (*ACCEPT) was used in an assertion that matched an empty string and
 1899     PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
 1900 
 1901 22. When an atomic group that contained a capturing parenthesis was
 1902     successfully matched, but the branch in which it appeared failed, the
 1903     capturing was not being forgotten if a higher numbered group was later
 1904     captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
 1905     group 1 to "a", when in fact it should be unset. This applied to multi-
 1906     branched capturing and non-capturing groups, repeated or not, and also to
 1907     positive assertions (capturing in negative assertions does not happen
 1908     in PCRE) and also to nested atomic groups.
 1909 
 1910 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
 1911     subject after a captured substring, to make it easier to tell which of a
 1912     number of identical substrings has been captured.
 1913 
 1914 24. The way atomic groups are processed by pcre_exec() has been changed so that
 1915     if they are repeated, backtracking one repetition now resets captured
 1916     values correctly. For example, if ((?>(a+)b)+aabab) is matched against
 1917     "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
 1918     "aaa". Previously, it would have been "a". As part of this code
 1919     refactoring, the way recursive calls are handled has also been changed.
 1920 
 1921 25. If an assertion condition captured any substrings, they were not passed
 1922     back unless some other capturing happened later. For example, if
 1923     (?(?=(a))a) was matched against "a", no capturing was returned.
 1924 
 1925 26. When studying a pattern that contained subroutine calls or assertions,
 1926     the code for finding the minimum length of a possible match was handling
 1927     direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
 1928     group 1 called group 2 while simultaneously a separate group 2 called group
 1929     1). A stack overflow occurred in this case. I have fixed this by limiting
 1930     the recursion depth to 10.
 1931 
 1932 27. Updated RunTest.bat in the distribution to the version supplied by Tom
 1933     Fortmann. This supports explicit test numbers on the command line, and has
 1934     argument validation and error reporting.
 1935 
 1936 28. An instance of \X with an unlimited repeat could fail if at any point the
 1937     first character it looked at was a mark character.
 1938 
 1939 29. Some minor code refactoring concerning Unicode properties and scripts
 1940     should reduce the stack requirement of match() slightly.
 1941 
 1942 30. Added the '=' option to pcretest to check the setting of unused capturing
 1943     slots at the end of the pattern, which are documented as being -1, but are
 1944     not included in the return count.
 1945 
 1946 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
 1947     compiled something random. Now it gives a compile-time error (as does
 1948     Perl).
 1949 
 1950 32. A *MARK encountered during the processing of a positive assertion is now
 1951     recorded and passed back (compatible with Perl).
 1952 
 1953 33. If --only-matching or --colour was set on a pcregrep call whose pattern
 1954     had alternative anchored branches, the search for a second match in a line
 1955     was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
 1956     matched the line "0102" twice. The same bug affected patterns that started
 1957     with a backwards assertion. For example /\b01|\b02/ also matched "0102"
 1958     twice.
 1959 
 1960 34. Previously, PCRE did not allow quantification of assertions. However, Perl
 1961     does, and because of capturing effects, quantifying parenthesized
 1962     assertions may at times be useful. Quantifiers are now allowed for
 1963     parenthesized assertions.
 1964 
 1965 35. A minor code tidy in pcre_compile() when checking options for \R usage.
 1966 
 1967 36. \g was being checked for fancy things in a character class, when it should
 1968     just be a literal "g".
 1969 
 1970 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
 1971     appearance of a nested POSIX class supersedes an apparent external class.
 1972     For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
 1973     unescaped square brackets may also appear as part of class names. For
 1974     example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
 1975     more like Perl. (But see 8.20/1 above.)
 1976 
 1977 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
 1978     was because it thought it was \N{name}, which is not supported).
 1979 
 1980 39. Add minix to OS list not supporting the -S option in pcretest.
 1981 
 1982 40. PCRE tries to detect cases of infinite recursion at compile time, but it
 1983     cannot analyze patterns in sufficient detail to catch mutual recursions
 1984     such as ((?1))((?2)). There is now a runtime test that gives an error if a
 1985     subgroup is called recursively as a subpattern for a second time at the
 1986     same position in the subject string. In previous releases this might have
 1987     been caught by the recursion limit, or it might have run out of stack.
 1988 
 1989 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
 1990     happen only once. PCRE was, however incorrectly giving a compile time error
 1991     "recursive call could loop indefinitely" because it cannot analyze the
 1992     pattern in sufficient detail. The compile time test no longer happens when
 1993     PCRE is compiling a conditional subpattern, but actual runaway loops are
 1994     now caught at runtime (see 40 above).
 1995 
 1996 42. It seems that Perl allows any characters other than a closing parenthesis
 1997     to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
 1998     has been changed to be the same.
 1999 
 2000 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
 2001     as not to get warnings when autogen.sh is called. Also changed
 2002     AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
 2003 
 2004 44. To help people who use pcregrep to scan files containing exceedingly long
 2005     lines, the following changes have been made:
 2006 
 2007     (a) The default value of the buffer size parameter has been increased from
 2008         8K to 20K. (The actual buffer used is three times this size.)
 2009 
 2010     (b) The default can be changed by ./configure --with-pcregrep-bufsize when
 2011         PCRE is built.
 2012 
 2013     (c) A --buffer-size=n option has been added to pcregrep, to allow the size
 2014         to be set at run time.
 2015 
 2016     (d) Numerical values in pcregrep options can be followed by K or M, for
 2017         example --buffer-size=50K.
 2018 
 2019     (e) If a line being scanned overflows pcregrep's buffer, an error is now
 2020         given and the return code is set to 2.
 2021 
 2022 45. Add a pointer to the latest mark to the callout data block.
 2023 
 2024 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
 2025     partial match of an empty string instead of no match. This was specific to
 2026     the use of ".".
 2027 
 2028 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
 2029     complete match instead of a partial match. This bug was dependent on both
 2030     the PCRE_UTF8 and PCRE_DOTALL options being set.
 2031 
 2032 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
 2033     starting byte set, because \b was not being ignored.
 2034 
 2035 
 2036 Version 8.12 15-Jan-2011
 2037 ------------------------
 2038 
 2039 1.  Fixed some typos in the markup of the man pages, and wrote a script that
 2040     checks for such things as part of the documentation building process.
 2041 
 2042 2.  On a big-endian 64-bit system, pcregrep did not correctly process the
 2043     --match-limit and --recursion-limit options (added for 8.11). In
 2044     particular, this made one of the standard tests fail. (The integer value
 2045     went into the wrong half of a long int.)
 2046 
 2047 3.  If the --colour option was given to pcregrep with -v (invert match), it
 2048     did strange things, either producing crazy output, or crashing. It should,
 2049     of course, ignore a request for colour when reporting lines that do not
 2050     match.
 2051 
 2052 4.  Another pcregrep bug caused similar problems if --colour was specified with
 2053     -M (multiline) and the pattern match finished with a line ending.
 2054 
 2055 5.  In pcregrep, when a pattern that ended with a literal newline sequence was
 2056     matched in multiline mode, the following line was shown as part of the
 2057     match. This seems wrong, so I have changed it.
 2058 
 2059 6.  Another pcregrep bug in multiline mode, when --colour was specified, caused
 2060     the check for further matches in the same line (so they could be coloured)
 2061     to overrun the end of the current line. If another match was found, it was
 2062     incorrectly shown (and then shown again when found in the next line).
 2063 
 2064 7.  If pcregrep was compiled under Windows, there was a reference to the
 2065     function pcregrep_exit() before it was defined. I am assuming this was
 2066     the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
 2067     reported by a user. I've moved the definition above the reference.
 2068 
 2069 
 2070 Version 8.11 10-Dec-2010
 2071 ------------------------
 2072 
 2073 1.  (*THEN) was not working properly if there were untried alternatives prior
 2074     to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
 2075     backtracked to try for "b" instead of moving to the next alternative branch
 2076     at the same level (in this case, to look for "c"). The Perl documentation
 2077     is clear that when (*THEN) is backtracked onto, it goes to the "next
 2078     alternative in the innermost enclosing group".
 2079 
 2080 2.  (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
 2081     such as   (A(*COMMIT)B(*THEN)C|D)  any failure after matching A should
 2082     result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
 2083     (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
 2084     (*THEN).
 2085 
 2086 3.  If \s appeared in a character class, it removed the VT character from
 2087     the class, even if it had been included by some previous item, for example
 2088     in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
 2089     of \s, but is part of the POSIX "space" class.)
 2090 
 2091 4.  A partial match never returns an empty string (because you can always
 2092     match an empty string at the end of the subject); however the checking for
 2093     an empty string was starting at the "start of match" point. This has been
 2094     changed to the "earliest inspected character" point, because the returned
 2095     data for a partial match starts at this character. This means that, for
 2096     example, /(?<=abc)def/ gives a partial match for the subject "abc"
 2097     (previously it gave "no match").
 2098 
 2099 5.  Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
 2100     of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
 2101     previously a full match would be given. However, setting PCRE_PARTIAL_HARD
 2102     has an implication that the given string is incomplete (because a partial
 2103     match is preferred over a full match). For this reason, these items now
 2104     give a partial match in this situation. [Aside: previously, the one case
 2105     /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
 2106     match rather than a full match, which was wrong by the old rules, but is
 2107     now correct.]
 2108 
 2109 6.  There was a bug in the handling of #-introduced comments, recognized when
 2110     PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
 2111     If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
 2112     UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
 2113     scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
 2114     but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
 2115     places in pcre_compile().
 2116 
 2117 7.  Related to (6) above, when pcre_compile() was skipping #-introduced
 2118     comments when looking ahead for named forward references to subpatterns,
 2119     the only newline sequence it recognized was NL. It now handles newlines
 2120     according to the set newline convention.
 2121 
 2122 8.  SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
 2123     former, but used strtoul(), whereas pcretest avoided strtoul() but did not
 2124     cater for a lack of strerror(). These oversights have been fixed.
 2125 
 2126 9.  Added --match-limit and --recursion-limit to pcregrep.
 2127 
 2128 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
 2129 
 2130 11. When the -o option was used, pcregrep was setting a return code of 1, even
 2131     when matches were found, and --line-buffered was not being honoured.
 2132 
 2133 12. Added an optional parentheses number to the -o and --only-matching options
 2134     of pcregrep.
 2135 
 2136 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
 2137     can match an empty string. The code to do it in pcretest and pcredemo
 2138     needed fixing:
 2139 
 2140     (a) When the newline convention was "crlf", pcretest got it wrong, skipping
 2141         only one byte after an empty string match just before CRLF (this case
 2142         just got forgotten; "any" and "anycrlf" were OK).
 2143 
 2144     (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
 2145         mode when an empty string match preceded an ASCII character followed by
 2146         a non-ASCII character. (The code for advancing by one character rather
 2147         than one byte was nonsense.)
 2148 
 2149     (c) The pcredemo.c sample program did not have any code at all to handle
 2150         the cases when CRLF is a valid newline sequence.
 2151 
 2152 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
 2153     as a starting offset was within the subject string. There is now a new
 2154     error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
 2155     negative or greater than the length of the string. In order to test this,
 2156     pcretest is extended to allow the setting of negative starting offsets.
 2157 
 2158 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
 2159     starting offset points to the beginning of a UTF-8 character was
 2160     unnecessarily clumsy. I tidied it up.
 2161 
 2162 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
 2163     bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
 2164 
 2165 17. Nobody had reported that the --include_dir option, which was added in
 2166     release 7.7 should have been called --include-dir (hyphen, not underscore)
 2167     for compatibility with GNU grep. I have changed it to --include-dir, but
 2168     left --include_dir as an undocumented synonym, and the same for
 2169     --exclude-dir, though that is not available in GNU grep, at least as of
 2170     release 2.5.4.
 2171 
 2172 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
 2173     characters from a string of bytes) have been redefined so as not to use
 2174     loops, in order to improve performance in some environments. At the same
 2175     time, I abstracted some of the common code into auxiliary macros to save
 2176     repetition (this should not affect the compiled code).
 2177 
 2178 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
 2179     compile-time error is now given if \c is not followed by an ASCII
 2180     character, that is, a byte less than 128. (In EBCDIC mode, the code is
 2181     different, and any byte value is allowed.)
 2182 
 2183 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
 2184     START_OPTIMIZE option, which is now allowed at compile time - but just
 2185     passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
 2186     to pcregrep and other applications that have no direct access to PCRE
 2187     options. The new /Y option in pcretest sets this option when calling
 2188     pcre_compile().
 2189 
 2190 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
 2191     back references. Groups containing recursive back references were forced to
 2192     be atomic by that change, but in the case of named groups, the amount of
 2193     memory required was incorrectly computed, leading to "Failed: internal
 2194     error: code overflow". This has been fixed.
 2195 
 2196 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
 2197     pcretest.c, to avoid build problems in some Borland environments.
 2198 
 2199 
 2200 Version 8.10 25-Jun-2010
 2201 ------------------------
 2202 
 2203 1.  Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
 2204     THEN.
 2205 
 2206 2.  (*ACCEPT) was not working when inside an atomic group.
 2207 
 2208 3.  Inside a character class, \B is treated as a literal by default, but
 2209     faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
 2210     causes the error). The code is unchanged, but I tidied the documentation.
 2211 
 2212 4.  Inside a character class, PCRE always treated \R and \X as literals,
 2213     whereas Perl faults them if its -w option is set. I have changed PCRE so
 2214     that it faults them when PCRE_EXTRA is set.
 2215 
 2216 5.  Added support for \N, which always matches any character other than
 2217     newline. (It is the same as "." when PCRE_DOTALL is not set.)
 2218 
 2219 6.  When compiling pcregrep with newer versions of gcc which may have
 2220     FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
 2221     declared with attribute warn_unused_result" were given. Just casting the
 2222     result to (void) does not stop the warnings; a more elaborate fudge is
 2223     needed. I've used a macro to implement this.
 2224 
 2225 7.  Minor change to pcretest.c to avoid a compiler warning.
 2226 
 2227 8.  Added four artifical Unicode properties to help with an option to make
 2228     \s etc use properties (see next item). The new properties are: Xan
 2229     (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
 2230 
 2231 9.  Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
 2232     use Unicode properties. (*UCP) at the start of a pattern can be used to set
 2233     this option. Modified pcretest to add /W to test this facility. Added
 2234     REG_UCP to make it available via the POSIX interface.
 2235 
 2236 10. Added --line-buffered to pcregrep.
 2237 
 2238 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
 2239     studied, and the match started with a letter with a code point greater than
 2240     127 whose first byte was different to the first byte of the other case of
 2241     the letter, the other case of this starting letter was not recognized
 2242     (#976).
 2243 
 2244 12. If a pattern that was studied started with a repeated Unicode property
 2245     test, for example, \p{Nd}+, there was the theoretical possibility of
 2246     setting up an incorrect bitmap of starting bytes, but fortunately it could
 2247     not have actually happened in practice until change 8 above was made (it
 2248     added property types that matched character-matching opcodes).
 2249 
 2250 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
 2251     possible starting bytes for non-anchored patterns.
 2252 
 2253 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
 2254     \R, and also a number of cases that involve Unicode properties, both
 2255     explicit and implicit when PCRE_UCP is set.
 2256 
 2257 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
 2258     input, it could crash or give wrong results if characters with values
 2259     greater than 0xc0 were present in the subject string. (Detail: it assumed
 2260     UTF-8 input when processing these items.)
 2261 
 2262 16. Added a lot of (int) casts to avoid compiler warnings in systems where
 2263     size_t is 64-bit (#991).
 2264 
 2265 17. Added a check for running out of memory when PCRE is compiled with
 2266     --disable-stack-for-recursion (#990).
 2267 
 2268 18. If the last data line in a file for pcretest does not have a newline on
 2269     the end, a newline was missing in the output.
 2270 
 2271 19. The default pcre_chartables.c file recognizes only ASCII characters (values
 2272     less than 128) in its various bitmaps. However, there is a facility for
 2273     generating tables according to the current locale when PCRE is compiled. It
 2274     turns out that in some environments, 0x85 and 0xa0, which are Unicode space
 2275     characters, are recognized by isspace() and therefore were getting set in
 2276     these tables, and indeed these tables seem to approximate to ISO 8859. This
 2277     caused a problem in UTF-8 mode when pcre_study() was used to create a list
 2278     of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
 2279     which of course cannot start UTF-8 characters. I have changed the code so
 2280     that only real ASCII characters (less than 128) and the correct starting
 2281     bytes for UTF-8 encodings are set for characters greater than 127 when in
 2282     UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
 2283     altogether.)
 2284 
 2285 20. Added the /T option to pcretest so as to be able to run tests with non-
 2286     standard character tables, thus making it possible to include the tests
 2287     used for 19 above in the standard set of tests.
 2288 
 2289 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
 2290     reference to a subpattern the other side of a comment that contains an
 2291     opening parenthesis caused either an internal compiling error, or a
 2292     reference to the wrong subpattern.
 2293 
 2294 
 2295 Version 8.02 19-Mar-2010
 2296 ------------------------
 2297 
 2298 1.  The Unicode data tables have been updated to Unicode 5.2.0.
 2299 
 2300 2.  Added the option --libs-cpp to pcre-config, but only when C++ support is
 2301     configured.
 2302 
 2303 3.  Updated the licensing terms in the pcregexp.pas file, as agreed with the
 2304     original author of that file, following a query about its status.
 2305 
 2306 4.  On systems that do not have stdint.h (e.g. Solaris), check for and include
 2307     inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
 2308 
 2309 5.  A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
 2310     quantifier applied to a forward-referencing subroutine call, could compile
 2311     incorrect code or give the error "internal error: previously-checked
 2312     referenced subpattern not found".
 2313 
 2314 6.  Both MS Visual Studio and Symbian OS have problems with initializing
 2315     variables to point to external functions. For these systems, therefore,
 2316     pcre_malloc etc. are now initialized to local functions that call the
 2317     relevant global functions.
 2318 
 2319 7.  There were two entries missing in the vectors called coptable and poptable
 2320     in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
 2321     I've fixed the data, and added a kludgy way of testing at compile time that
 2322     the lengths are correct (equal to the number of opcodes).
 2323 
 2324 8.  Following on from 7, I added a similar kludge to check the length of the
 2325     eint vector in pcreposix.c.
 2326 
 2327 9.  Error texts for pcre_compile() are held as one long string to avoid too
 2328     much relocation at load time. To find a text, the string is searched,
 2329     counting zeros. There was no check for running off the end of the string,
 2330     which could happen if a new error number was added without updating the
 2331     string.
 2332 
 2333 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
 2334 
 2335 11. \K was not working if it appeared in an atomic group or in a group that
 2336     was called as a "subroutine", or in an assertion. Perl 5.11 documents that
 2337     \K is "not well defined" if used in an assertion. PCRE now accepts it if
 2338     the assertion is positive, but not if it is negative.
 2339 
 2340 12. Change 11 fortuitously reduced the size of the stack frame used in the
 2341     "match()" function of pcre_exec.c by one pointer. Forthcoming
 2342     implementation of support for (*MARK) will need an extra pointer on the
 2343     stack; I have reserved it now, so that the stack frame size does not
 2344     decrease.
 2345 
 2346 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
 2347     item in branch that calls a recursion is a subroutine call - as in the
 2348     second branch in the above example - was incorrectly given the compile-
 2349     time error "recursive call could loop indefinitely" because pcre_compile()
 2350     was not correctly checking the subroutine for matching a non-empty string.
 2351 
 2352 14. The checks for overrunning compiling workspace could trigger after an
 2353     overrun had occurred. This is a "should never occur" error, but it can be
 2354     triggered by pathological patterns such as hundreds of nested parentheses.
 2355     The checks now trigger 100 bytes before the end of the workspace.
 2356 
 2357 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
 2358 
 2359 
 2360 Version 8.01 19-Jan-2010
 2361 ------------------------
 2362 
 2363 1.  If a pattern contained a conditional subpattern with only one branch (in
 2364     particular, this includes all (*DEFINE) patterns), a call to pcre_study()
 2365     computed the wrong minimum data length (which is of course zero for such
 2366     subpatterns). This could cause incorrect "no match" results.
 2367 
 2368 2.  For patterns such as (?i)a(?-i)b|c where an option setting at the start of
 2369     the pattern is reset in the first branch, pcre_compile() failed with
 2370     "internal error: code overflow at offset...". This happened only when
 2371     the reset was to the original external option setting. (An optimization
 2372     abstracts leading options settings into an external setting, which was the
 2373     cause of this.)
 2374 
 2375 3.  A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
 2376     of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
 2377     assertion pattern did not match (meaning that the assertion was true), it
 2378     was incorrectly treated as false if the SKIP had been reached during the
 2379     matching. This also applied to assertions used as conditions.
 2380 
 2381 4.  If an item that is not supported by pcre_dfa_exec() was encountered in an
 2382     assertion subpattern, including such a pattern used as a condition,
 2383     unpredictable results occurred, instead of the error return
 2384     PCRE_ERROR_DFA_UITEM.
 2385 
 2386 5.  The C++ GlobalReplace function was not working like Perl for the special
 2387     situation when an empty string is matched. It now does the fancy magic
 2388     stuff that is necessary.
 2389 
 2390 6.  In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
 2391     removed. (These were left over from very, very early versions of PCRE.)
 2392 
 2393 7.  Some cosmetic changes to the code to make life easier when compiling it
 2394     as part of something else:
 2395 
 2396     (a) Change DEBUG to PCRE_DEBUG.
 2397 
 2398     (b) In pcre_compile(), rename the member of the "branch_chain" structure
 2399         called "current" as "current_branch", to prevent a collision with the
 2400         Linux macro when compiled as a kernel module.
 2401 
 2402     (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
 2403         prevent a collision with the Linux macro when compiled as a kernel
 2404         module.
 2405 
 2406 8.  In pcre_compile() there are some checks for integer overflows that used to
 2407     cast potentially large values to (double). This has been changed to that
 2408     when building, a check for int64_t is made, and if it is found, it is used
 2409     instead, thus avoiding the use of floating point arithmetic. (There is no
 2410     other use of FP in PCRE.) If int64_t is not found, the fallback is to
 2411     double.
 2412 
 2413 9.  Added two casts to avoid signed/unsigned warnings from VS Studio Express
 2414     2005 (difference between two addresses compared to an unsigned value).
 2415 
 2416 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
 2417     custom one, because of the following reported problem in Windows:
 2418 
 2419       - libbz2 uses the Pascal calling convention (WINAPI) for the functions
 2420           under Win32.
 2421       - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
 2422           therefore missing the function definition.
 2423       - The compiler thus generates a "C" signature for the test function.
 2424       - The linker fails to find the "C" function.
 2425       - PCRE fails to configure if asked to do so against libbz2.
 2426 
 2427 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
 2428     messages were output:
 2429 
 2430       Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
 2431       rerunning libtoolize, to keep the correct libtool macros in-tree.
 2432       Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
 2433 
 2434     I have done both of these things.
 2435 
 2436 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
 2437     most of the time, it *can* run out if it is given a pattern that contains a
 2438     runaway infinite recursion. I updated the discussion in the pcrestack man
 2439     page.
 2440 
 2441 13. Now that we have gone to the x.xx style of version numbers, the minor
 2442     version may start with zero. Using 08 or 09 is a bad idea because users
 2443     might check the value of PCRE_MINOR in their code, and 08 or 09 may be
 2444     interpreted as invalid octal numbers. I've updated the previous comment in
 2445     configure.ac, and also added a check that gives an error if 08 or 09 are
 2446     used.
 2447 
 2448 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
 2449     causing partial matching to fail when the end of the subject matched \W
 2450     in a UTF-8 pattern where \W was quantified with a minimum of 3.
 2451 
 2452 15. There were some discrepancies between the declarations in pcre_internal.h
 2453     of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
 2454     their definitions. The declarations used "const uschar *" and the
 2455     definitions used USPTR. Even though USPTR is normally defined as "const
 2456     unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
 2457     reported that: "This difference in casting confuses some C++ compilers, for
 2458     example, SunCC recognizes above declarations as different functions and
 2459     generates broken code for hbpcre." I have changed the declarations to use
 2460     USPTR.
 2461 
 2462 16. GNU libtool is named differently on some systems. The autogen.sh script now
 2463     tries several variants such as glibtoolize (MacOSX) and libtoolize1x
 2464     (FreeBSD).
 2465 
 2466 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
 2467     (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
 2468     comment: "Figure out how to create a longlong from a string: strtoll and
 2469     equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
 2470     instance, but it only takes 2 args instead of 3!"
 2471 
 2472 18. A subtle bug concerned with back references has been fixed by a change of
 2473     specification, with a corresponding code fix. A pattern such as
 2474     ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
 2475     refers, was giving matches when it shouldn't. For example, xa=xaaa would
 2476     match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
 2477     same bug. Such groups have to be quantified to be useful, or contained
 2478     inside another quantified group. (If there's no repetition, the reference
 2479     can never match.) The problem arises because, having left the group and
 2480     moved on to the rest of the pattern, a later failure that backtracks into
 2481     the group uses the captured value from the final iteration of the group
 2482     rather than the correct earlier one. I have fixed this in PCRE by forcing
 2483     any group that contains a reference to itself to be an atomic group; that
 2484     is, there cannot be any backtracking into it once it has completed. This is
 2485     similar to recursive and subroutine calls.
 2486 
 2487 
 2488 Version 8.00 19-Oct-09
 2489 ----------------------
 2490 
 2491 1.  The table for translating pcre_compile() error codes into POSIX error codes
 2492     was out-of-date, and there was no check on the pcre_compile() error code
 2493     being within the table. This could lead to an OK return being given in
 2494     error.
 2495 
 2496 2.  Changed the call to open a subject file in pcregrep from fopen(pathname,
 2497     "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
 2498     in a Windows environment.
 2499 
 2500 3.  The pcregrep --count option prints the count for each file even when it is
 2501     zero, as does GNU grep. However, pcregrep was also printing all files when
 2502     --files-with-matches was added. Now, when both options are given, it prints
 2503     counts only for those files that have at least one match. (GNU grep just
 2504     prints the file name in this circumstance, but including the count seems
 2505     more useful - otherwise, why use --count?) Also ensured that the
 2506     combination -clh just lists non-zero counts, with no names.
 2507 
 2508 4.  The long form of the pcregrep -F option was incorrectly implemented as
 2509     --fixed_strings instead of --fixed-strings. This is an incompatible change,
 2510     but it seems right to fix it, and I didn't think it was worth preserving
 2511     the old behaviour.
 2512 
 2513 5.  The command line items --regex=pattern and --regexp=pattern were not
 2514     recognized by pcregrep, which required --regex pattern or --regexp pattern
 2515     (with a space rather than an '='). The man page documented the '=' forms,
 2516     which are compatible with GNU grep; these now work.
 2517 
 2518 6.  No libpcreposix.pc file was created for pkg-config; there was just
 2519     libpcre.pc and libpcrecpp.pc. The omission has been rectified.
 2520 
 2521 7.  Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
 2522     when UCP support is not needed, by modifying the Python script that
 2523     generates it from Unicode data files. This should not matter if the module
 2524     is correctly used as a library, but I received one complaint about 50K of
 2525     unwanted data. My guess is that the person linked everything into his
 2526     program rather than using a library. Anyway, it does no harm.
 2527 
 2528 8.  A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
 2529     was a minimum greater than 1 for a wide character in a possessive
 2530     repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
 2531     which had an unlimited repeat of a nested, fixed maximum repeat of a wide
 2532     character. Chaos in the form of incorrect output or a compiling loop could
 2533     result.
 2534 
 2535 9.  The restrictions on what a pattern can contain when partial matching is
 2536     requested for pcre_exec() have been removed. All patterns can now be
 2537     partially matched by this function. In addition, if there are at least two
 2538     slots in the offset vector, the offset of the earliest inspected character
 2539     for the match and the offset of the end of the subject are set in them when
 2540     PCRE_ERROR_PARTIAL is returned.
 2541 
 2542 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
 2543     synonymous with PCRE_PARTIAL, for backwards compatibility, and
 2544     PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
 2545     and may be more useful for multi-segment matching.
 2546 
 2547 11. Partial matching with pcre_exec() is now more intuitive. A partial match
 2548     used to be given if ever the end of the subject was reached; now it is
 2549     given only if matching could not proceed because another character was
 2550     needed. This makes a difference in some odd cases such as Z(*FAIL) with the
 2551     string "Z", which now yields "no match" instead of "partial match". In the
 2552     case of pcre_dfa_exec(), "no match" is given if every matching path for the
 2553     final character ended with (*FAIL).
 2554 
 2555 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
 2556     if the pattern had a "must contain" character that was already found in the
 2557     earlier partial match, unless partial matching was again requested. For
 2558     example, with the pattern /dog.(body)?/, the "must contain" character is
 2559     "g". If the first part-match was for the string "dog", restarting with
 2560     "sbody" failed. This bug has been fixed.
 2561 
 2562 13. The string returned by pcre_dfa_exec() after a partial match has been
 2563     changed so that it starts at the first inspected character rather than the
 2564     first character of the match. This makes a difference only if the pattern
 2565     starts with a lookbehind assertion or \b or \B (\K is not supported by
 2566     pcre_dfa_exec()). It's an incompatible change, but it makes the two
 2567     matching functions compatible, and I think it's the right thing to do.
 2568 
 2569 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
 2570     so that the demonstration program is easily available in environments where
 2571     PCRE has not been installed from source.
 2572 
 2573 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
 2574     libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
 2575     library.
 2576 
 2577 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
 2578     It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
 2579     is not the first non-POSIX option to be added. Clearly some people find
 2580     these options useful.
 2581 
 2582 17. If a caller to the POSIX matching function regexec() passes a non-zero
 2583     value for nmatch with a NULL value for pmatch, the value of
 2584     nmatch is forced to zero.
 2585 
 2586 18. RunGrepTest did not have a test for the availability of the -u option of
 2587     the diff command, as RunTest does. It now checks in the same way as
 2588     RunTest, and also checks for the -b option.
 2589 
 2590 19. If an odd number of negated classes containing just a single character
 2591     interposed, within parentheses, between a forward reference to a named
 2592     subpattern and the definition of the subpattern, compilation crashed with
 2593     an internal error, complaining that it could not find the referenced
 2594     subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
 2595     [The bug was that it was starting one character too far in when skipping
 2596     over the character class, thus treating the ] as data rather than
 2597     terminating the class. This meant it could skip too much.]
 2598 
 2599 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
 2600     /g option in pcretest when the pattern contains \K, which makes it possible
 2601     to have an empty string match not at the start, even when the pattern is
 2602     anchored. Updated pcretest and pcredemo to use this option.
 2603 
 2604 21. If the maximum number of capturing subpatterns in a recursion was greater
 2605     than the maximum at the outer level, the higher number was returned, but
 2606     with unset values at the outer level. The correct (outer level) value is
 2607     now given.
 2608 
 2609 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
 2610     PCRE did not set those parentheses (unlike Perl). I have now found a way to
 2611     make it do so. The string so far is captured, making this feature
 2612     compatible with Perl.
 2613 
 2614 23. The tests have been re-organized, adding tests 11 and 12, to make it
 2615     possible to check the Perl 5.10 features against Perl 5.10.
 2616 
 2617 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
 2618     pattern matches a fixed length string. PCRE did not allow this; now it
 2619     does. Neither allows recursion.
 2620 
 2621 25. I finally figured out how to implement a request to provide the minimum
 2622     length of subject string that was needed in order to match a given pattern.
 2623     (It was back references and recursion that I had previously got hung up
 2624     on.) This code has now been added to pcre_study(); it finds a lower bound
 2625     to the length of subject needed. It is not necessarily the greatest lower
 2626     bound, but using it to avoid searching strings that are too short does give
 2627     some useful speed-ups. The value is available to calling programs via
 2628     pcre_fullinfo().
 2629 
 2630 26. While implementing 25, I discovered to my embarrassment that pcretest had
 2631     not been passing the result of pcre_study() to pcre_dfa_exec(), so the
 2632     study optimizations had never been tested with that matching function.
 2633     Oops. What is worse, even when it was passed study data, there was a bug in
 2634     pcre_dfa_exec() that meant it never actually used it. Double oops. There
 2635     were also very few tests of studied patterns with pcre_dfa_exec().
 2636 
 2637 27. If (?| is used to create subpatterns with duplicate numbers, they are now
 2638     allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
 2639     on the other side of the coin, they are no longer allowed to have different
 2640     names, because these cannot be distinguished in PCRE, and this has caused
 2641     confusion. (This is a difference from Perl.)
 2642 
 2643 28. When duplicate subpattern names are present (necessarily with different
 2644     numbers, as required by 27 above), and a test is made by name in a
 2645     conditional pattern, either for a subpattern having been matched, or for
 2646     recursion in such a pattern, all the associated numbered subpatterns are
 2647     tested, and the overall condition is true if the condition is true for any
 2648     one of them. This is the way Perl works, and is also more like the way
 2649     testing by number works.
 2650 
 2651 
 2652 Version 7.9 11-Apr-09
 2653 ---------------------
 2654 
 2655 1.  When building with support for bzlib/zlib (pcregrep) and/or readline
 2656     (pcretest), all targets were linked against these libraries. This included
 2657     libpcre, libpcreposix, and libpcrecpp, even though they do not use these
 2658     libraries. This caused unwanted dependencies to be created. This problem
 2659     has been fixed, and now only pcregrep is linked with bzlib/zlib and only
 2660     pcretest is linked with readline.
 2661 
 2662 2.  The "typedef int BOOL" in pcre_internal.h that was included inside the
 2663     "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
 2664     moved outside it again, because FALSE and TRUE are already defined in AIX,
 2665     but BOOL is not.
 2666 
 2667 3.  The pcre_config() function was treating the PCRE_MATCH_LIMIT and
 2668     PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
 2669 
 2670 4.  The pcregrep documentation said spaces were inserted as well as colons (or
 2671     hyphens) following file names and line numbers when outputting matching
 2672     lines. This is not true; no spaces are inserted. I have also clarified the
 2673     wording for the --colour (or --color) option.
 2674 
 2675 5.  In pcregrep, when --colour was used with -o, the list of matching strings
 2676     was not coloured; this is different to GNU grep, so I have changed it to be
 2677     the same.
 2678 
 2679 6.  When --colo(u)r was used in pcregrep, only the first matching substring in
 2680     each matching line was coloured. Now it goes on to look for further matches
 2681     of any of the test patterns, which is the same behaviour as GNU grep.
 2682 
 2683 7.  A pattern that could match an empty string could cause pcregrep to loop; it
 2684     doesn't make sense to accept an empty string match in pcregrep, so I have
 2685     locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
 2686     seems to be how GNU grep behaves. [But see later change 40 for release
 2687     8.33.]
 2688 
 2689 8.  The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
 2690     start or after a newline", because the conditional assertion was not being
 2691     correctly handled. The rule now is that both the assertion and what follows
 2692     in the first alternative must satisfy the test.
 2693 
 2694 9.  If auto-callout was enabled in a pattern with a conditional group whose
 2695     condition was an assertion, PCRE could crash during matching, both with
 2696     pcre_exec() and pcre_dfa_exec().
 2697 
 2698 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
 2699     used for matching.
 2700 
 2701 11. Unicode property support in character classes was not working for
 2702     characters (bytes) greater than 127 when not in UTF-8 mode.
 2703 
 2704 12. Added the -M command line option to pcretest.
 2705 
 2706 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
 2707 
 2708 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
 2709 
 2710 16. Added comments and documentation about mis-use of no_arg in the C++
 2711     wrapper.
 2712 
 2713 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
 2714     from Martin Jerabek that uses macro names for all relevant character and
 2715     string constants.
 2716 
 2717 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
 2718     SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
 2719     SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
 2720     these, but not everybody uses configure.
 2721 
 2722 19. A conditional group that had only one branch was not being correctly
 2723     recognized as an item that could match an empty string. This meant that an
 2724     enclosing group might also not be so recognized, causing infinite looping
 2725     (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
 2726     with the subject "ab", where knowledge that the repeated group can match
 2727     nothing is needed in order to break the loop.
 2728 
 2729 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
 2730     exec(), but without supplying a callout function, matching went wrong.
 2731 
 2732 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
 2733     leak if the size of the offset vector was greater than 30. When the vector
 2734     is smaller, the saved offsets during recursion go onto a local stack
 2735     vector, but for larger vectors malloc() is used. It was failing to free
 2736     when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
 2737     error, in fact).
 2738 
 2739 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
 2740     heapframe that is used only when UTF-8 support is enabled. This caused no
 2741     problem, but was untidy.
 2742 
 2743 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
 2744     CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
 2745     included within another project.
 2746 
 2747 24. Steven Van Ingelgem's patches to add more options to the CMake support,
 2748     slightly modified by me:
 2749 
 2750       (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
 2751           not building pcregrep.
 2752 
 2753       (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
 2754           if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
 2755 
 2756 25. Forward references, both numeric and by name, in patterns that made use of
 2757     duplicate group numbers, could behave incorrectly or give incorrect errors,
 2758     because when scanning forward to find the reference group, PCRE was not
 2759     taking into account the duplicate group numbers. A pattern such as
 2760     ^X(?3)(a)(?|(b)|(q))(Y) is an example.
 2761 
 2762 26. Changed a few more instances of "const unsigned char *" to USPTR, making
 2763     the feature of a custom pointer more persuasive (as requested by a user).
 2764 
 2765 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
 2766     pcretest.c, inside #ifndefs, because it seems they are sometimes already
 2767     pre-defined.
 2768 
 2769 28. Added support for (*UTF8) at the start of a pattern.
 2770 
 2771 29. Arrange for flags added by the "release type" setting in CMake to be shown
 2772     in the configuration summary.
 2773 
 2774 
 2775 Version 7.8 05-Sep-08
 2776 ---------------------
 2777 
 2778 1.  Replaced UCP searching code with optimized version as implemented for Ad
 2779     Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
 2780     stage table and inline lookup instead of a function, giving speed ups of 2
 2781     to 5 times on some simple patterns that I tested. Permission was given to
 2782     distribute the MultiStage2.py script that generates the tables (it's not in
 2783     the tarball, but is in the Subversion repository).
 2784 
 2785 2.  Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
 2786     scripts.
 2787 
 2788 3.  Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
 2789     a group with a zero qualifier. The result of the study could be incorrect,
 2790     or the function might crash, depending on the pattern.
 2791 
 2792 4.  Caseless matching was not working for non-ASCII characters in back
 2793     references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
 2794     It now works when Unicode Property Support is available.
 2795 
 2796 5.  In pcretest, an escape such as \x{de} in the data was always generating
 2797     a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
 2798     non-UTF-8 mode. If the value is greater than 255, it gives a warning about
 2799     truncation.
 2800 
 2801 6.  Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
 2802 
 2803 7.  Added two (int) casts to pcregrep when printing the difference of two
 2804     pointers, in case they are 64-bit values.
 2805 
 2806 8.  Added comments about Mac OS X stack usage to the pcrestack man page and to
 2807     test 2 if it fails.
 2808 
 2809 9.  Added PCRE_CALL_CONVENTION just before the names of all exported functions,
 2810     and a #define of that name to empty if it is not externally set. This is to
 2811     allow users of MSVC to set it if necessary.
 2812 
 2813 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
 2814     the convenience functions in the pcre_get.c source file.
 2815 
 2816 11. An option change at the start of a pattern that had top-level alternatives
 2817     could cause overwriting and/or a crash. This command provoked a crash in
 2818     some environments:
 2819 
 2820       printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
 2821 
 2822     This potential security problem was recorded as CVE-2008-2371.
 2823 
 2824 12. For a pattern where the match had to start at the beginning or immediately
 2825     after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
 2826     pcre_dfa_exec() could read past the end of the passed subject if there was
 2827     no match. To help with detecting such bugs (e.g. with valgrind), I modified
 2828     pcretest so that it places the subject at the end of its malloc-ed buffer.
 2829 
 2830 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
 2831     exec() might read past the end of the data buffer in UTF-8 mode.
 2832 
 2833 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
 2834     the data contained the byte 0x85 as part of a UTF-8 character within its
 2835     first line. This applied both to normal and DFA matching.
 2836 
 2837 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
 2838     /^[^d]*?$/8 failed to match "abc".
 2839 
 2840 16. Added a missing copyright notice to pcrecpp_internal.h.
 2841 
 2842 17. Make it more clear in the documentation that values returned from
 2843     pcre_exec() in ovector are byte offsets, not character counts.
 2844 
 2845 18. Tidied a few places to stop certain compilers from issuing warnings.
 2846 
 2847 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
 2848     supplied by Stefan Weber. I made a further small update for 7.8 because
 2849     there is a change of source arrangements: the pcre_searchfuncs.c module is
 2850     replaced by pcre_ucd.c.
 2851 
 2852 
 2853 Version 7.7 07-May-08
 2854 ---------------------
 2855 
 2856 1.  Applied Craig's patch to sort out a long long problem: "If we can't convert
 2857     a string to a long long, pretend we don't even have a long long." This is
 2858     done by checking for the strtoq, strtoll, and _strtoi64 functions.
 2859 
 2860 2.  Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
 2861     pre-7.6 versions, which defined a global no_arg variable instead of putting
 2862     it in the RE class. (See also #8 below.)
 2863 
 2864 3.  Remove a line of dead code, identified by coverity and reported by Nuno
 2865     Lopes.
 2866 
 2867 4.  Fixed two related pcregrep bugs involving -r with --include or --exclude:
 2868 
 2869     (1) The include/exclude patterns were being applied to the whole pathnames
 2870         of files, instead of just to the final components.
 2871 
 2872     (2) If there was more than one level of directory, the subdirectories were
 2873         skipped unless they satisfied the include/exclude conditions. This is
 2874         inconsistent with GNU grep (and could even be seen as contrary to the
 2875         pcregrep specification - which I improved to make it absolutely clear).
 2876         The action now is always to scan all levels of directory, and just
 2877         apply the include/exclude patterns to regular files.
 2878 
 2879 5.  Added the --include_dir and --exclude_dir patterns to pcregrep, and used
 2880     --exclude_dir in the tests to avoid scanning .svn directories.
 2881 
 2882 6.  Applied Craig's patch to the QuoteMeta function so that it escapes the
 2883     NUL character as backslash + 0 rather than backslash + NUL, because PCRE
 2884     doesn't support NULs in patterns.
 2885 
 2886 7.  Added some missing "const"s to declarations of static tables in
 2887     pcre_compile.c and pcre_dfa_exec.c.
 2888 
 2889 8.  Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
 2890     caused by fix #2  above. (Subsequently also a second patch to fix the
 2891     first patch. And a third patch - this was a messy problem.)
 2892 
 2893 9.  Applied Craig's patch to remove the use of push_back().
 2894 
 2895 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
 2896     matching function regexec().
 2897 
 2898 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
 2899     which, however, unlike Perl's \g{...}, are subroutine calls, not back
 2900     references. PCRE supports relative numbers with this syntax (I don't think
 2901     Oniguruma does).
 2902 
 2903 12. Previously, a group with a zero repeat such as (...){0} was completely
 2904     omitted from the compiled regex. However, this means that if the group
 2905     was called as a subroutine from elsewhere in the pattern, things went wrong
 2906     (an internal error was given). Such groups are now left in the compiled
 2907     pattern, with a new opcode that causes them to be skipped at execution
 2908     time.
 2909 
 2910 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
 2911     to the way PCRE behaves:
 2912 
 2913     (a) A lone ] character is dis-allowed (Perl treats it as data).
 2914 
 2915     (b) A back reference to an unmatched subpattern matches an empty string
 2916         (Perl fails the current match path).
 2917 
 2918     (c) A data ] in a character class must be notated as \] because if the
 2919         first data character in a class is ], it defines an empty class. (In
 2920         Perl it is not possible to have an empty class.) The empty class []
 2921         never matches; it forces failure and is equivalent to (*FAIL) or (?!).
 2922         The negative empty class [^] matches any one character, independently
 2923         of the DOTALL setting.
 2924 
 2925 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
 2926     non-existent subpattern following a character class starting with ']' and
 2927     containing () gave an internal compiling error instead of "reference to
 2928     non-existent subpattern". Fortunately, when the pattern did exist, the
 2929     compiled code was correct. (When scanning forwards to check for the
 2930     existence of the subpattern, it was treating the data ']' as terminating
 2931     the class, so got the count wrong. When actually compiling, the reference
 2932     was subsequently set up correctly.)
 2933 
 2934 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
 2935     it was being rejected as not supported by pcre_dfa_exec(), even though
 2936     other assertions are supported. I have made pcre_dfa_exec() support
 2937     (*FAIL).
 2938 
 2939 16. The implementation of 13c above involved the invention of a new opcode,
 2940     OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
 2941     cannot be changed at match time, I realized I could make a small
 2942     improvement to matching performance by compiling OP_ALLANY instead of
 2943     OP_ANY for "." when DOTALL was set, and then removing the runtime tests
 2944     on the OP_ANY path.
 2945 
 2946 17. Compiling pcretest on Windows with readline support failed without the
 2947     following two fixes: (1) Make the unistd.h include conditional on
 2948     HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
 2949 
 2950 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
 2951     ncurses library to be included for pcretest when ReadLine support is
 2952     requested, but also to allow for it to be overridden. This patch came from
 2953     Daniel Bergström.
 2954 
 2955 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
 2956     as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
 2957     any errors with the current Unicode tables. Thanks to Peter Kankowski for
 2958     spotting this.
 2959 
 2960 
 2961 Version 7.6 28-Jan-08
 2962 ---------------------
 2963 
 2964 1.  A character class containing a very large number of characters with
 2965     codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
 2966     overflow.
 2967 
 2968 2.  Patch to cut out the "long long" test in pcrecpp_unittest when
 2969     HAVE_LONG_LONG is not defined.
 2970 
 2971 3.  Applied Christian Ehrlicher's patch to update the CMake build files to
 2972     bring them up to date and include new features. This patch includes:
 2973 
 2974     - Fixed PH's badly added libz and libbz2 support.
 2975     - Fixed a problem with static linking.
 2976     - Added pcredemo. [But later removed - see 7 below.]
 2977     - Fixed dftables problem and added an option.
 2978     - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
 2979         HAVE_LONG_LONG.
 2980     - Added readline support for pcretest.
 2981     - Added an listing of the option settings after cmake has run.
 2982 
 2983 4.  A user submitted a patch to Makefile that makes it easy to create
 2984     "pcre.dll" under mingw when using Configure/Make. I added stuff to
 2985     Makefile.am that cause it to include this special target, without
 2986     affecting anything else. Note that the same mingw target plus all
 2987     the other distribution libraries and programs are now supported
 2988     when configuring with CMake (see 6 below) instead of with
 2989     Configure/Make.
 2990 
 2991 5.  Applied Craig's patch that moves no_arg into the RE class in the C++ code.
 2992     This is an attempt to solve the reported problem "pcrecpp::no_arg is not
 2993     exported in the Windows port". It has not yet been confirmed that the patch
 2994     solves the problem, but it does no harm.
 2995 
 2996 6.  Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
 2997     NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
 2998     with CMake, and also correct the comment about stack recursion.
 2999 
 3000 7.  Remove the automatic building of pcredemo from the ./configure system and
 3001     from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
 3002     of a program that users should build themselves after PCRE is installed, so
 3003     building it automatically is not really right. What is more, it gave
 3004     trouble in some build environments.
 3005 
 3006 8.  Further tidies to CMakeLists.txt from Sheri and Christian.
 3007 
 3008 
 3009 Version 7.5 10-Jan-08
 3010 ---------------------
 3011 
 3012 1.  Applied a patch from Craig: "This patch makes it possible to 'ignore'
 3013     values in parens when parsing an RE using the C++ wrapper."
 3014 
 3015 2.  Negative specials like \S did not work in character classes in UTF-8 mode.
 3016     Characters greater than 255 were excluded from the class instead of being
 3017     included.
 3018 
 3019 3.  The same bug as (2) above applied to negated POSIX classes such as
 3020     [:^space:].
 3021 
 3022 4.  PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
 3023     defined or documented. It seems to have been a typo for PCRE_STATIC, so
 3024     I have changed it.
 3025 
 3026 5.  The construct (?&) was not diagnosed as a syntax error (it referenced the
 3027     first named subpattern) and a construct such as (?&a) would reference the
 3028     first named subpattern whose name started with "a" (in other words, the
 3029     length check was missing). Both these problems are fixed. "Subpattern name
 3030     expected" is now given for (?&) (a zero-length name), and this patch also
 3031     makes it give the same error for \k'' (previously it complained that that
 3032     was a reference to a non-existent subpattern).
 3033 
 3034 6.  The erroneous patterns (?+-a) and (?-+a) give different error messages;
 3035     this is right because (?- can be followed by option settings as well as by
 3036     digits. I have, however, made the messages clearer.
 3037 
 3038 7.  Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
 3039     than the number used in the conditional) now cause a compile-time error.
 3040     This is actually not compatible with Perl, which accepts such patterns, but
 3041     treats the conditional as always being FALSE (as PCRE used to), but it
 3042     seems to me that giving a diagnostic is better.
 3043 
 3044 8.  Change "alphameric" to the more common word "alphanumeric" in comments
 3045     and messages.
 3046 
 3047 9.  Fix two occurrences of "backslash" in comments that should have been
 3048     "backspace".
 3049 
 3050 10. Remove two redundant lines of code that can never be obeyed (their function
 3051     was moved elsewhere).
 3052 
 3053 11. The program that makes PCRE's Unicode character property table had a bug
 3054     which caused it to generate incorrect table entries for sequences of
 3055     characters that have the same character type, but are in different scripts.
 3056     It amalgamated them into a single range, with the script of the first of
 3057     them. In other words, some characters were in the wrong script. There were
 3058     thirteen such cases, affecting characters in the following ranges:
 3059 
 3060       U+002b0 - U+002c1
 3061       U+0060c - U+0060d
 3062       U+0061e - U+00612
 3063       U+0064b - U+0065e
 3064       U+0074d - U+0076d
 3065       U+01800 - U+01805
 3066       U+01d00 - U+01d77
 3067       U+01d9b - U+01dbf
 3068       U+0200b - U+0200f
 3069       U+030fc - U+030fe
 3070       U+03260 - U+0327f
 3071       U+0fb46 - U+0fbb1
 3072       U+10450 - U+1049d
 3073 
 3074 12. The -o option (show only the matching part of a line) for pcregrep was not
 3075     compatible with GNU grep in that, if there was more than one match in a
 3076     line, it showed only the first of them. It now behaves in the same way as
 3077     GNU grep.
 3078 
 3079 13. If the -o and -v options were combined for pcregrep, it printed a blank
 3080     line for every non-matching line. GNU grep prints nothing, and pcregrep now
 3081     does the same. The return code can be used to tell if there were any
 3082     non-matching lines.
 3083 
 3084 14. Added --file-offsets and --line-offsets to pcregrep.
 3085 
 3086 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
 3087     infinitely looping recursion. The bug was that positive lookaheads were not
 3088     being skipped when checking for a possible empty match (negative lookaheads
 3089     and both kinds of lookbehind were skipped).
 3090 
 3091 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
 3092     inclusion of <windows.h> to before rather than after the definition of
 3093     INVALID_FILE_ATTRIBUTES (patch from David Byron).
 3094 
 3095 17. Specifying a possessive quantifier with a specific limit for a Unicode
 3096     character property caused pcre_compile() to compile bad code, which led at
 3097     runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
 3098     are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
 3099     caused the error; without that there was no problem.
 3100 
 3101 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
 3102 
 3103 19. Added --enable-pcretest-libreadline.
 3104 
 3105 20. In pcrecpp.cc, the variable 'count' was incremented twice in
 3106     RE::GlobalReplace(). As a result, the number of replacements returned was
 3107     double what it should be. I removed one of the increments, but Craig sent a
 3108     later patch that removed the other one (the right fix) and added unit tests
 3109     that check the return values (which was not done before).
 3110 
 3111 21. Several CMake things:
 3112 
 3113     (1) Arranged that, when cmake is used on Unix, the libraries end up with
 3114         the names libpcre and libpcreposix, not just pcre and pcreposix.
 3115 
 3116     (2) The above change means that pcretest and pcregrep are now correctly
 3117         linked with the newly-built libraries, not previously installed ones.
 3118 
 3119     (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
 3120 
 3121 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
 3122     crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
 3123     UTF-8 newline character). The key issue is that the pattern starts .*;
 3124     this means that the match must be either at the beginning, or after a
 3125     newline. The bug was in the code for advancing after a failed match and
 3126     checking that the new position followed a newline. It was not taking
 3127     account of UTF-8 characters correctly.
 3128 
 3129 23. PCRE was behaving differently from Perl in the way it recognized POSIX
 3130     character classes. PCRE was not treating the sequence [:...:] as a
 3131     character class unless the ... were all letters. Perl, however, seems to
 3132     allow any characters between [: and :], though of course it rejects as
 3133     unknown any "names" that contain non-letters, because all the known class
 3134     names consist only of letters. Thus, Perl gives an error for [[:1234:]],
 3135     for example, whereas PCRE did not - it did not recognize a POSIX character
 3136     class. This seemed a bit dangerous, so the code has been changed to be
 3137     closer to Perl. The behaviour is not identical to Perl, because PCRE will
 3138     diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
 3139     treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
 3140     Perl does, and where it didn't before.
 3141 
 3142 24. Rewrite so as to remove the single use of %n from pcregrep because in some
 3143     Windows environments %n is disabled by default.
 3144 
 3145 
 3146 Version 7.4 21-Sep-07
 3147 ---------------------
 3148 
 3149 1.  Change 7.3/28 was implemented for classes by looking at the bitmap. This
 3150     means that a class such as [\s] counted as "explicit reference to CR or
 3151     LF". That isn't really right - the whole point of the change was to try to
 3152     help when there was an actual mention of one of the two characters. So now
 3153     the change happens only if \r or \n (or a literal CR or LF) character is
 3154     encountered.
 3155 
 3156 2.  The 32-bit options word was also used for 6 internal flags, but the numbers
 3157     of both had grown to the point where there were only 3 bits left.
 3158     Fortunately, there was spare space in the data structure, and so I have
 3159     moved the internal flags into a new 16-bit field to free up more option
 3160     bits.
 3161 
 3162 3.  The appearance of (?J) at the start of a pattern set the DUPNAMES option,
 3163     but did not set the internal JCHANGED flag - either of these is enough to
 3164     control the way the "get" function works - but the PCRE_INFO_JCHANGED
 3165     facility is supposed to tell if (?J) was ever used, so now (?J) at the
 3166     start sets both bits.
 3167 
 3168 4.  Added options (at build time, compile time, exec time) to change \R from
 3169     matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
 3170 
 3171 5.  doc/pcresyntax.html was missing from the distribution.
 3172 
 3173 6.  Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
 3174     compatibility, even though it is no longer used.
 3175 
 3176 7.  Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
 3177     strtoull to pcrecpp.cc to select the available functions in WIN32 when the
 3178     windows.h file is present (where different names are used). [This was
 3179     reversed later after testing - see 16 below.]
 3180 
 3181 8.  Changed all #include <config.h> to #include "config.h". There were also
 3182     some further <pcre.h> cases that I changed to "pcre.h".
 3183 
 3184 9.  When pcregrep was used with the --colour option, it missed the line ending
 3185     sequence off the lines that it output.
 3186 
 3187 10. It was pointed out to me that arrays of string pointers cause lots of
 3188     relocations when a shared library is dynamically loaded. A technique of
 3189     using a single long string with a table of offsets can drastically reduce
 3190     these. I have refactored PCRE in four places to do this. The result is
 3191     dramatic:
 3192 
 3193       Originally:                          290
 3194       After changing UCP table:            187
 3195       After changing error message table:   43
 3196       After changing table of "verbs"       36
 3197       After changing table of Posix names   22
 3198 
 3199     Thanks to the folks working on Gregex for glib for this insight.
 3200 
 3201 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
 3202     unicode-properties was also set.
 3203 
 3204 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
 3205 
 3206 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
 3207     checked only for CRLF.
 3208 
 3209 14. Added casts to pcretest.c to avoid compiler warnings.
 3210 
 3211 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
 3212 
 3213 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
 3214     and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
 3215     entirely. This removes changes made in 7 above.
 3216 
 3217 17. The CMake files have been updated, and there is now more information about
 3218     building with CMake in the NON-UNIX-USE document.
 3219 
 3220 
 3221 Version 7.3 28-Aug-07
 3222 ---------------------
 3223 
 3224  1. In the rejigging of the build system that eventually resulted in 7.1, the
 3225     line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
 3226     brackets there is not right, since it causes compilers to look for an
 3227     installed pcre.h, not the version that is in the source that is being
 3228     compiled (which of course may be different). I have changed it back to:
 3229 
 3230       #include "pcre.h"
 3231 
 3232     I have a vague recollection that the change was concerned with compiling in
 3233     different directories, but in the new build system, that is taken care of
 3234     by the VPATH setting the Makefile.
 3235 
 3236  2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
 3237     when the subject happened to end in the byte 0x85 (e.g. if the last
 3238     character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
 3239     characters but of course it shouldn't be taken as a newline when it is part
 3240     of another character. The bug was that, for an unlimited repeat of . in
 3241     not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
 3242     characters when looking for a newline.
 3243 
 3244  3. A small performance improvement in the DOTALL UTF-8 mode .* case.
 3245 
 3246  4. Debugging: adjusted the names of opcodes for different kinds of parentheses
 3247     in debug output.
 3248 
 3249  5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
 3250     long printing in the pcrecpp unittest when running under MinGW.
 3251 
 3252  6. ESC_K was left out of the EBCDIC table.
 3253 
 3254  7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
 3255     parentheses; I made it 1000, which seemed large enough. Unfortunately, the
 3256     limit also applies to "virtual nesting" when a pattern is recursive, and in
 3257     this case 1000 isn't so big. I have been able to remove this limit at the
 3258     expense of backing off one optimization in certain circumstances. Normally,
 3259     when pcre_exec() would call its internal match() function recursively and
 3260     immediately return the result unconditionally, it uses a "tail recursion"
 3261     feature to save stack. However, when a subpattern that can match an empty
 3262     string has an unlimited repetition quantifier, it no longer makes this
 3263     optimization. That gives it a stack frame in which to save the data for
 3264     checking that an empty string has been matched. Previously this was taken
 3265     from the 1000-entry workspace that had been reserved. So now there is no
 3266     explicit limit, but more stack is used.
 3267 
 3268  8. Applied Daniel's patches to solve problems with the import/export magic
 3269     syntax that is required for Windows, and which was going wrong for the
 3270     pcreposix and pcrecpp parts of the library. These were overlooked when this
 3271     problem was solved for the main library.
 3272 
 3273  9. There were some crude static tests to avoid integer overflow when computing
 3274     the size of patterns that contain repeated groups with explicit upper
 3275     limits. As the maximum quantifier is 65535, the maximum group length was
 3276     set at 30,000 so that the product of these two numbers did not overflow a
 3277     32-bit integer. However, it turns out that people want to use groups that
 3278     are longer than 30,000 bytes (though not repeat them that many times).
 3279     Change 7.0/17 (the refactoring of the way the pattern size is computed) has
 3280     made it possible to implement the integer overflow checks in a much more
 3281     dynamic way, which I have now done. The artificial limitation on group
 3282     length has been removed - we now have only the limit on the total length of
 3283     the compiled pattern, which depends on the LINK_SIZE setting.
 3284 
 3285 10. Fixed a bug in the documentation for get/copy named substring when
 3286     duplicate names are permitted. If none of the named substrings are set, the
 3287     functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
 3288     empty string.
 3289 
 3290 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
 3291     instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
 3292     because the ] is interpreted as the first data character and the
 3293     terminating ] is not found. PCRE has been made compatible with Perl in this
 3294     regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
 3295     cause memory overwriting.
 3296 
 3297 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
 3298     string has been matched (to stop an infinite loop). It was not recognizing
 3299     a conditional subpattern that could match an empty string if that
 3300     subpattern was within another subpattern. For example, it looped when
 3301     trying to match  (((?(1)X|))*)  but it was OK with  ((?(1)X|)*)  where the
 3302     condition was not nested. This bug has been fixed.
 3303 
 3304 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
 3305     past the start of the subject in the presence of bytes with the top bit
 3306     set, for example "\x8aBCD".
 3307 
 3308 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
 3309     (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
 3310 
 3311 14. Optimized (?!) to (*FAIL).
 3312 
 3313 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
 3314     This restricts code points to be within the range 0 to 0x10FFFF, excluding
 3315     the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
 3316     full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
 3317     does: it's just the validity check that is more restrictive.
 3318 
 3319 16. Inserted checks for integer overflows during escape sequence (backslash)
 3320     processing, and also fixed erroneous offset values for syntax errors during
 3321     backslash processing.
 3322 
 3323 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
 3324     for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
 3325 
 3326 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
 3327     caused an overrun.
 3328 
 3329 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
 3330     something other than just ASCII characters) inside a group that had an
 3331     unlimited repeat caused a loop at compile time (while checking to see
 3332     whether the group could match an empty string).
 3333 
 3334 20. Debugging a pattern containing \p or \P could cause a crash. For example,
 3335     [\P{Any}] did so. (Error in the code for printing property names.)
 3336 
 3337 21. An orphan \E inside a character class could cause a crash.
 3338 
 3339 22. A repeated capturing bracket such as (A)? could cause a wild memory
 3340     reference during compilation.
 3341 
 3342 23. There are several functions in pcre_compile() that scan along a compiled
 3343     expression for various reasons (e.g. to see if it's fixed length for look
 3344     behind). There were bugs in these functions when a repeated \p or \P was
 3345     present in the pattern. These operators have additional parameters compared
 3346     with \d, etc, and these were not being taken into account when moving along
 3347     the compiled data. Specifically:
 3348 
 3349     (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
 3350         length.
 3351 
 3352     (b) An item such as \pL+ within a repeated group could cause crashes or
 3353         loops.
 3354 
 3355     (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
 3356         "reference to non-existent subpattern" error.
 3357 
 3358     (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
 3359 
 3360 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
 3361     characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
 3362 
 3363 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
 3364 
 3365 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
 3366     character were causing crashes (broken optimization).
 3367 
 3368 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
 3369     \p or \P) caused a compile-time loop.
 3370 
 3371 28. More problems have arisen in unanchored patterns when CRLF is a valid line
 3372     break. For example, the unstudied pattern [\r\n]A does not match the string
 3373     "\r\nA" because change 7.0/46 below moves the current point on by two
 3374     characters after failing to match at the start. However, the pattern \nA
 3375     *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
 3376     the same is true. There doesn't seem any very clean way out of this, but
 3377     what I have chosen to do makes the common cases work: PCRE now takes note
 3378     of whether there can be an explicit match for \r or \n anywhere in the
 3379     pattern, and if so, 7.0/46 no longer applies. As part of this change,
 3380     there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
 3381     pattern has explicit CR or LF references.
 3382 
 3383 29. Added (*CR) etc for changing newline setting at start of pattern.
 3384 
 3385 
 3386 Version 7.2 19-Jun-07
 3387 ---------------------
 3388 
 3389  1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
 3390     which is apparently normally available under Windows.
 3391 
 3392  2. Re-jig the pcregrep tests with different newline settings in an attempt
 3393     to make them independent of the local environment's newline setting.
 3394 
 3395  3. Add code to configure.ac to remove -g from the CFLAGS default settings.
 3396 
 3397  4. Some of the "internals" tests were previously cut out when the link size
 3398     was not 2, because the output contained actual offsets. The recent new
 3399     "Z" feature of pcretest means that these can be cut out, making the tests
 3400     usable with all link sizes.
 3401 
 3402  5. Implemented Stan Switzer's goto replacement for longjmp() when not using
 3403     stack recursion. This gives a massive performance boost under BSD, but just
 3404     a small improvement under Linux. However, it saves one field in the frame
 3405     in all cases.
 3406 
 3407  6. Added more features from the forthcoming Perl 5.10:
 3408 
 3409     (a) (?-n) (where n is a string of digits) is a relative subroutine or
 3410         recursion call. It refers to the nth most recently opened parentheses.
 3411 
 3412     (b) (?+n) is also a relative subroutine call; it refers to the nth next
 3413         to be opened parentheses.
 3414 
 3415     (c) Conditions that refer to capturing parentheses can be specified
 3416         relatively, for example, (?(-2)... or (?(+3)...
 3417 
 3418     (d) \K resets the start of the current match so that everything before
 3419         is not part of it.
 3420 
 3421     (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
 3422 
 3423     (f) \g{name} is another synonym - part of Perl 5.10's unification of
 3424         reference syntax.
 3425 
 3426     (g) (?| introduces a group in which the numbering of parentheses in each
 3427         alternative starts with the same number.
 3428 
 3429     (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
 3430 
 3431  7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
 3432     PCRE_INFO_JCHANGED.
 3433 
 3434  8. A pattern such as  (.*(.)?)*  caused pcre_exec() to fail by either not
 3435     terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
 3436     for detecting groups that can match an empty string.
 3437 
 3438  9. A pattern with a very large number of alternatives (more than several
 3439     hundred) was running out of internal workspace during the pre-compile
 3440     phase, where pcre_compile() figures out how much memory will be needed. A
 3441     bit of new cunning has reduced the workspace needed for groups with
 3442     alternatives. The 1000-alternative test pattern now uses 12 bytes of
 3443     workspace instead of running out of the 4096 that are available.
 3444 
 3445 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
 3446 
 3447 11. Applied patch from Google to remove an optimization that didn't quite work.
 3448     The report of the bug said:
 3449 
 3450       pcrecpp::RE("a*").FullMatch("aaa") matches, while
 3451       pcrecpp::RE("a*?").FullMatch("aaa") does not, and
 3452       pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
 3453 
 3454 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
 3455     it matched the wrong number of bytes.
 3456 
 3457 
 3458 Version 7.1 24-Apr-07
 3459 ---------------------
 3460 
 3461  1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
 3462     that is more "standard", making use of automake and other Autotools. There
 3463     is some re-arrangement of the files and adjustment of comments consequent
 3464     on this.
 3465 
 3466  2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
 3467     for recursive directory scanning broke on some systems because the files
 3468     are not scanned in any specific order and on different systems the order
 3469     was different. A call to "sort" has been inserted into RunGrepTest for the
 3470     approprate test as a short-term fix. In the longer term there may be an
 3471     alternative.
 3472 
 3473  3. I had an email from Eric Raymond about problems translating some of PCRE's
 3474     man pages to HTML (despite the fact that I distribute HTML pages, some
 3475     people do their own conversions for various reasons). The problems
 3476     concerned the use of low-level troff macros .br and .in. I have therefore
 3477     removed all such uses from the man pages (some were redundant, some could
 3478     be replaced by .nf/.fi pairs). The 132html script that I use to generate
 3479     HTML has been updated to handle .nf/.fi and to complain if it encounters
 3480     .br or .in.
 3481 
 3482  4. Updated comments in configure.ac that get placed in config.h.in and also
 3483     arranged for config.h to be included in the distribution, with the name
 3484     config.h.generic, for the benefit of those who have to compile without
 3485     Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
 3486 
 3487  5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
 3488     Weber: (1) pcre_internal.h was missing some function renames; (2) updated
 3489     makevp.bat for the current PCRE, using the additional files
 3490     makevp_c.txt, makevp_l.txt, and pcregexp.pas.
 3491 
 3492  6. A Windows user reported a minor discrepancy with test 2, which turned out
 3493     to be caused by a trailing space on an input line that had got lost in his
 3494     copy. The trailing space was an accident, so I've just removed it.
 3495 
 3496  7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
 3497     that is needed.
 3498 
 3499  8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
 3500     as "const" (a) because they are and (b) because it helps the PHP
 3501     maintainers who have recently made a script to detect big data structures
 3502     in the php code that should be moved to the .rodata section. I remembered
 3503     to update Builducptable as well, so it won't revert if ucptable.h is ever
 3504     re-created.
 3505 
 3506  9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
 3507     pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
 3508     order to be able to cut out the UTF-8 tables in the latter when UTF-8
 3509     support is not required. This saves 1.5-2K of code, which is important in
 3510     some applications.
 3511 
 3512     Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
 3513     so as not to refer to the tables, even though these functions will never be
 3514     called when UTF-8 support is disabled. Otherwise there are problems with a
 3515     shared library.
 3516 
 3517 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
 3518 
 3519     (a) It was defining its arguments as char * instead of void *.
 3520 
 3521     (b) It was assuming that all moves were upwards in memory; this was true
 3522         a long time ago when I wrote it, but is no longer the case.
 3523 
 3524     The emulated memove() is provided for those environments that have neither
 3525     memmove() nor bcopy(). I didn't think anyone used it these days, but that
 3526     is clearly not the case, as these two bugs were recently reported.
 3527 
 3528 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
 3529     and Detrail to create the HTML documentation, the .txt form of the man
 3530     pages, and it removes trailing spaces from listed files. It also creates
 3531     pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
 3532     case, it wraps all the #defines with #ifndefs. This script should be run
 3533     before "make dist".
 3534 
 3535 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
 3536     with Unicode property support.
 3537 
 3538     (a) For a maximizing quantifier, if the two different cases of the
 3539         character were of different lengths in their UTF-8 codings (there are
 3540         some cases like this - I found 11), and the matching function had to
 3541         back up over a mixture of the two cases, it incorrectly assumed they
 3542         were both the same length.
 3543 
 3544     (b) When PCRE was configured to use the heap rather than the stack for
 3545         recursion during matching, it was not correctly preserving the data for
 3546         the other case of a UTF-8 character when checking ahead for a match
 3547         while processing a minimizing repeat. If the check also involved
 3548         matching a wide character, but failed, corruption could cause an
 3549         erroneous result when trying to check for a repeat of the original
 3550         character.
 3551 
 3552 13. Some tidying changes to the testing mechanism:
 3553 
 3554     (a) The RunTest script now detects the internal link size and whether there
 3555         is UTF-8 and UCP support by running ./pcretest -C instead of relying on
 3556         values substituted by "configure". (The RunGrepTest script already did
 3557         this for UTF-8.) The configure.ac script no longer substitutes the
 3558         relevant variables.
 3559 
 3560     (b) The debugging options /B and /D in pcretest show the compiled bytecode
 3561         with length and offset values. This means that the output is different
 3562         for different internal link sizes. Test 2 is skipped for link sizes
 3563         other than 2 because of this, bypassing the problem. Unfortunately,
 3564         there was also a test in test 3 (the locale tests) that used /B and
 3565         failed for link sizes other than 2. Rather than cut the whole test out,
 3566         I have added a new /Z option to pcretest that replaces the length and
 3567         offset values with spaces. This is now used to make test 3 independent
 3568         of link size. (Test 2 will be tidied up later.)
 3569 
 3570 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
 3571     segmentation fault instead of returning the appropriate error message.
 3572 
 3573 15. In multiline mode when the newline sequence was set to "any", the pattern
 3574     ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
 3575     This doesn't seem right; it now treats the CRLF combination as the line
 3576     ending, and so does not match in that case. It's only a pattern such as ^$
 3577     that would hit this one: something like ^ABC$ would have failed after \r
 3578     and then tried again after \r\n.
 3579 
 3580 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
 3581     in an attempt to make files that differ only in their line terminators
 3582     compare equal. This works on Linux.
 3583 
 3584 17. Under certain error circumstances pcregrep might try to free random memory
 3585     as it exited. This is now fixed, thanks to valgrind.
 3586 
 3587 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
 3588     "abc\r\n\r\n", it found an unwanted second match after the second \r. This
 3589     was because its rules for how to advance for /g after matching an empty
 3590     string at the end of a line did not allow for this case. They now check for
 3591     it specially.
 3592 
 3593 20. pcretest is supposed to handle patterns and data of any length, by
 3594     extending its buffers when necessary. It was getting this wrong when the
 3595     buffer for a data line had to be extended.
 3596 
 3597 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
 3598     CRLF as a newline sequence.
 3599 
 3600 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
 3601     out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
 3602     I have nevertheless tidied it up.
 3603 
 3604 23. Added some casts to kill warnings from HP-UX ia64 compiler.
 3605 
 3606 24. Added a man page for pcre-config.
 3607 
 3608 
 3609 Version 7.0 19-Dec-06
 3610 ---------------------
 3611 
 3612  1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
 3613     moving to gcc 4.1.1.
 3614 
 3615  2. The -S option for pcretest uses setrlimit(); I had omitted to #include
 3616     sys/time.h, which is documented as needed for this function. It doesn't
 3617     seem to matter on Linux, but it showed up on some releases of OS X.
 3618 
 3619  3. It seems that there are systems where bytes whose values are greater than
 3620     127 match isprint() in the "C" locale. The "C" locale should be the
 3621     default when a C program starts up. In most systems, only ASCII printing
 3622     characters match isprint(). This difference caused the output from pcretest
 3623     to vary, making some of the tests fail. I have changed pcretest so that:
 3624 
 3625     (a) When it is outputting text in the compiled version of a pattern, bytes
 3626         other than 32-126 are always shown as hex escapes.
 3627 
 3628     (b) When it is outputting text that is a matched part of a subject string,
 3629         it does the same, unless a different locale has been set for the match
 3630         (using the /L modifier). In this case, it uses isprint() to decide.
 3631 
 3632  4. Fixed a major bug that caused incorrect computation of the amount of memory
 3633     required for a compiled pattern when options that changed within the
 3634     pattern affected the logic of the preliminary scan that determines the
 3635     length. The relevant options are -x, and -i in UTF-8 mode. The result was
 3636     that the computed length was too small. The symptoms of this bug were
 3637     either the PCRE error "internal error: code overflow" from pcre_compile(),
 3638     or a glibc crash with a message such as "pcretest: free(): invalid next
 3639     size (fast)". Examples of patterns that provoked this bug (shown in
 3640     pcretest format) are:
 3641 
 3642       /(?-x: )/x
 3643       /(?x)(?-x: \s*#\s*)/
 3644       /((?i)[\x{c0}])/8
 3645       /(?i:[\x{c0}])/8
 3646 
 3647     HOWEVER: Change 17 below makes this fix obsolete as the memory computation
 3648     is now done differently.
 3649 
 3650  5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
 3651     wrapper classes; (b) implement a new function in the C++ scanner that is
 3652     more efficient than the old way of doing things because it avoids levels of
 3653     recursion in the regex matching; (c) add a paragraph to the documentation
 3654     for the FullMatch() function.
 3655 
 3656  6. The escape sequence \n was being treated as whatever was defined as
 3657     "newline". Not only was this contrary to the documentation, which states
 3658     that \n is character 10 (hex 0A), but it also went horribly wrong when
 3659     "newline" was defined as CRLF. This has been fixed.
 3660 
 3661  7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
 3662     was being set to -1 for the "end of line" case (supposedly a value that no
 3663     character can have). Though this value is never used (the check for end of
 3664     line is "zero bytes in current character"), it caused compiler complaints.
 3665     I've changed it to 0xffffffff.
 3666 
 3667  8. In pcre_version.c, the version string was being built by a sequence of
 3668     C macros that, in the event of PCRE_PRERELEASE being defined as an empty
 3669     string (as it is for production releases) called a macro with an empty
 3670     argument. The C standard says the result of this is undefined. The gcc
 3671     compiler treats it as an empty string (which was what was wanted) but it is
 3672     reported that Visual C gives an error. The source has been hacked around to
 3673     avoid this problem.
 3674 
 3675  9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
 3676     builds of pcretest, and changed the call to _setmode() to use _O_BINARY
 3677     instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
 3678     of them did).
 3679 
 3680 10. Originally, pcretest opened its input and output without "b"; then I was
 3681     told that "b" was needed in some environments, so it was added for release
 3682     5.0 to both the input and output. (It makes no difference on Unix-like
 3683     systems.) Later I was told that it is wrong for the input on Windows. I've
 3684     now abstracted the modes into two macros, to make it easier to fiddle with
 3685     them, and removed "b" from the input mode under Windows.
 3686 
 3687 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
 3688 
 3689 12. Added -help and --help to pcretest as an official way of being reminded
 3690     of the options.
 3691 
 3692 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
 3693     and pcrecpp.cc because they annoy compilers at high warning levels.
 3694 
 3695 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
 3696 
 3697 15. Fixed an occurrence of == in configure.ac that should have been = (shell
 3698     scripts are not C programs :-) and which was not noticed because it works
 3699     on Linux.
 3700 
 3701 16. pcretest is supposed to handle any length of pattern and data line (as one
 3702     line or as a continued sequence of lines) by extending its input buffer if
 3703     necessary. This feature was broken for very long pattern lines, leading to
 3704     a string of junk being passed to pcre_compile() if the pattern was longer
 3705     than about 50K.
 3706 
 3707 17. I have done a major re-factoring of the way pcre_compile() computes the
 3708     amount of memory needed for a compiled pattern. Previously, there was code
 3709     that made a preliminary scan of the pattern in order to do this. That was
 3710     OK when PCRE was new, but as the facilities have expanded, it has become
 3711     harder and harder to keep it in step with the real compile phase, and there
 3712     have been a number of bugs (see for example, 4 above). I have now found a
 3713     cunning way of running the real compile function in a "fake" mode that
 3714     enables it to compute how much memory it would need, while actually only
 3715     ever using a few hundred bytes of working memory and without too many
 3716     tests of the mode. This should make future maintenance and development
 3717     easier. A side effect of this work is that the limit of 200 on the nesting
 3718     depth of parentheses has been removed (though this was never a serious
 3719     limitation, I suspect). However, there is a downside: pcre_compile() now
 3720     runs more slowly than before (30% or more, depending on the pattern). I
 3721     hope this isn't a big issue. There is no effect on runtime performance.
 3722 
 3723 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
 3724     newline (only possible for the last line of a file) and it was a
 3725     pattern that set a locale (followed by /Lsomething), pcretest crashed.
 3726 
 3727 19. Added additional timing features to pcretest. (1) The -tm option now times
 3728     matching only, not compiling. (2) Both -t and -tm can be followed, as a
 3729     separate command line item, by a number that specifies the number of
 3730     repeats to use when timing. The default is 50000; this gives better
 3731     precision, but takes uncomfortably long for very large patterns.
 3732 
 3733 20. Extended pcre_study() to be more clever in cases where a branch of a
 3734     subpattern has no definite first character. For example, (a*|b*)[cd] would
 3735     previously give no result from pcre_study(). Now it recognizes that the
 3736     first character must be a, b, c, or d.
 3737 
 3738 21. There was an incorrect error "recursive call could loop indefinitely" if
 3739     a subpattern (or the entire pattern) that was being tested for matching an
 3740     empty string contained only one non-empty item after a nested subpattern.
 3741     For example, the pattern (?>\x{100}*)\d(?R) provoked this error
 3742     incorrectly, because the \d was being skipped in the check.
 3743 
 3744 22. The pcretest program now has a new pattern option /B and a command line
 3745     option -b, which is equivalent to adding /B to every pattern. This causes
 3746     it to show the compiled bytecode, without the additional information that
 3747     -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
 3748     is the same as /B/I).
 3749 
 3750 23. A new optimization is now able automatically to treat some sequences such
 3751     as a*b as a*+b. More specifically, if something simple (such as a character
 3752     or a simple class like \d) has an unlimited quantifier, and is followed by
 3753     something that cannot possibly match the quantified thing, the quantifier
 3754     is automatically "possessified".
 3755 
 3756 24. A recursive reference to a subpattern whose number was greater than 39
 3757     went wrong under certain circumstances in UTF-8 mode. This bug could also
 3758     have affected the operation of pcre_study().
 3759 
 3760 25. Realized that a little bit of performance could be had by replacing
 3761     (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
 3762 
 3763 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
 3764 
 3765 27. Possessive quantifiers such as a++ were previously implemented by turning
 3766     them into atomic groups such as ($>a+). Now they have their own opcodes,
 3767     which improves performance. This includes the automatically created ones
 3768     from 23 above.
 3769 
 3770 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
 3771     lookahead was broken if it was not anchored. PCRE was mistakenly expecting
 3772     the first matched character to be a colon. This applied both to named and
 3773     numbered groups.
 3774 
 3775 29. The ucpinternal.h header file was missing its idempotency #ifdef.
 3776 
 3777 30. I was sent a "project" file called libpcre.a.dev which I understand makes
 3778     building PCRE on Windows easier, so I have included it in the distribution.
 3779 
 3780 31. There is now a check in pcretest against a ridiculously large number being
 3781     returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
 3782     loop, the loop is abandoned.
 3783 
 3784 32. Forward references to subpatterns in conditions such as (?(2)...) where
 3785     subpattern 2 is defined later cause pcre_compile() to search forwards in
 3786     the pattern for the relevant set of parentheses. This search went wrong
 3787     when there were unescaped parentheses in a character class, parentheses
 3788     escaped with \Q...\E, or parentheses in a #-comment in /x mode.
 3789 
 3790 33. "Subroutine" calls and backreferences were previously restricted to
 3791     referencing subpatterns earlier in the regex. This restriction has now
 3792     been removed.
 3793 
 3794 34. Added a number of extra features that are going to be in Perl 5.10. On the
 3795     whole, these are just syntactic alternatives for features that PCRE had
 3796     previously implemented using the Python syntax or my own invention. The
 3797     other formats are all retained for compatibility.
 3798 
 3799     (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
 3800         as (?P<name>...). The new forms, as well as being in Perl 5.10, are
 3801         also .NET compatible.
 3802 
 3803     (b) A recursion or subroutine call to a named group can now be defined as
 3804         (?&name) as well as (?P>name).
 3805 
 3806     (c) A backreference to a named group can now be defined as \k<name> or
 3807         \k'name' as well as (?P=name). The new forms, as well as being in Perl
 3808         5.10, are also .NET compatible.
 3809 
 3810     (d) A conditional reference to a named group can now use the syntax
 3811         (?(<name>) or (?('name') as well as (?(name).
 3812 
 3813     (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
 3814         groups (named and numbered) that are never evaluated inline, but can be
 3815         called as "subroutines" from elsewhere. In effect, the DEFINE condition
 3816         is always false. There may be only one alternative in such a group.
 3817 
 3818     (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
 3819         as the simple (?(R). The condition is true only if the most recent
 3820         recursion is that of the given number or name. It does not search out
 3821         through the entire recursion stack.
 3822 
 3823     (g) The escape \gN or \g{N} has been added, where N is a positive or
 3824         negative number, specifying an absolute or relative reference.
 3825 
 3826 35. Tidied to get rid of some further signed/unsigned compiler warnings and
 3827     some "unreachable code" warnings.
 3828 
 3829 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
 3830     things, this adds five new scripts.
 3831 
 3832 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
 3833     There were also incompatibilities regarding the handling of \Q..\E inside
 3834     character classes, for example with patterns like [\Qa\E-\Qz\E] where the
 3835     hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
 3836 
 3837 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
 3838     matches an empty string, and forcibly breaks the loop. There were bugs in
 3839     this code in non-simple cases. For a pattern such as  ^(a()*)*  matched
 3840     against  aaaa  the result was just "a" rather than "aaaa", for example. Two
 3841     separate and independent bugs (that affected different cases) have been
 3842     fixed.
 3843 
 3844 39. Refactored the code to abolish the use of different opcodes for small
 3845     capturing bracket numbers. This is a tidy that I avoided doing when I
 3846     removed the limit on the number of capturing brackets for 3.5 back in 2001.
 3847     The new approach is not only tidier, it makes it possible to reduce the
 3848     memory needed to fix the previous bug (38).
 3849 
 3850 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
 3851     sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
 3852     processing dot, circumflex, or dollar metacharacters, or #-comments in /x
 3853     mode.
 3854 
 3855 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
 3856     report.
 3857 
 3858 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
 3859     copy construction and assignment in the C++ wrapper.
 3860 
 3861 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
 3862     couple of bugs that could have given wrong results in the "--newline=crlf"
 3863     case.
 3864 
 3865 44. Added a number of casts and did some reorganization of signed/unsigned int
 3866     variables following suggestions from Dair Grant. Also renamed the variable
 3867     "this" as "item" because it is a C++ keyword.
 3868 
 3869 45. Arranged for dftables to add
 3870 
 3871       #include "pcre_internal.h"
 3872 
 3873     to pcre_chartables.c because without it, gcc 4.x may remove the array
 3874     definition from the final binary if PCRE is built into a static library and
 3875     dead code stripping is activated.
 3876 
 3877 46. For an unanchored pattern, if a match attempt fails at the start of a
 3878     newline sequence, and the newline setting is CRLF or ANY, and the next two
 3879     characters are CRLF, advance by two characters instead of one.
 3880 
 3881 
 3882 Version 6.7 04-Jul-06
 3883 ---------------------
 3884 
 3885  1. In order to handle tests when input lines are enormously long, pcretest has
 3886     been re-factored so that it automatically extends its buffers when
 3887     necessary. The code is crude, but this _is_ just a test program. The
 3888     default size has been increased from 32K to 50K.
 3889 
 3890  2. The code in pcre_study() was using the value of the re argument before
 3891     testing it for NULL. (Of course, in any sensible call of the function, it
 3892     won't be NULL.)
 3893 
 3894  3. The memmove() emulation function in pcre_internal.h, which is used on
 3895     systems that lack both memmove() and bcopy() - that is, hardly ever -
 3896     was missing a "static" storage class specifier.
 3897 
 3898  4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
 3899     containing an extended class (one that cannot be represented by a bitmap
 3900     because it contains high-valued characters or Unicode property items, e.g.
 3901     [\pZ]). Almost always one would set UTF-8 mode when processing such a
 3902     pattern, but PCRE should not loop if you do not (it no longer does).
 3903     [Detail: two cases were found: (a) a repeated subpattern containing an
 3904     extended class; (b) a recursive reference to a subpattern that followed a
 3905     previous extended class. It wasn't skipping over the extended class
 3906     correctly when UTF-8 mode was not set.]
 3907 
 3908  5. A negated single-character class was not being recognized as fixed-length
 3909     in lookbehind assertions such as (?<=[^f]), leading to an incorrect
 3910     compile error "lookbehind assertion is not fixed length".
 3911 
 3912  6. The RunPerlTest auxiliary script was showing an unexpected difference
 3913     between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
 3914     write a Perl script that can interpret lines of an input file either as
 3915     byte characters or as UTF-8, which is what "perltest" was being required to
 3916     do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
 3917     can't do is switch easily at run time between having the "use utf8;" pragma
 3918     or not. In the end, I fudged it by using the RunPerlTest script to insert
 3919     "use utf8;" explicitly for the UTF-8 tests.
 3920 
 3921  7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
 3922     the end of the subject string, contrary to the documentation and to what
 3923     Perl does. This was true of both matching functions. Now it matches only at
 3924     the start of the subject and immediately after *internal* newlines.
 3925 
 3926  8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
 3927     a pointer to an int instead of a pointer to an unsigned long int. This
 3928     caused problems on 64-bit systems.
 3929 
 3930  9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
 3931     instance of the 'standard' template library not being so standard".
 3932 
 3933 10. There was no check on the number of named subpatterns nor the maximum
 3934     length of a subpattern name. The product of these values is used to compute
 3935     the size of the memory block for a compiled pattern. By supplying a very
 3936     long subpattern name and a large number of named subpatterns, the size
 3937     computation could be caused to overflow. This is now prevented by limiting
 3938     the length of names to 32 characters, and the number of named subpatterns
 3939     to 10,000.
 3940 
 3941 11. Subpatterns that are repeated with specific counts have to be replicated in
 3942     the compiled pattern. The size of memory for this was computed from the
 3943     length of the subpattern and the repeat count. The latter is limited to
 3944     65535, but there was no limit on the former, meaning that integer overflow
 3945     could in principle occur. The compiled length of a repeated subpattern is
 3946     now limited to 30,000 bytes in order to prevent this.
 3947 
 3948 12. Added the optional facility to have named substrings with the same name.
 3949 
 3950 13. Added the ability to use a named substring as a condition, using the
 3951     Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
 3952     are numbers (not recommended). Forward references are permitted.
 3953 
 3954 14. Added forward references in named backreferences (if you see what I mean).
 3955 
 3956 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
 3957     pattern could run off the end of the subject. For example, the pattern
 3958     "(?s)(.{1,5})"8 did this with the subject "ab".
 3959 
 3960 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
 3961     PCRE_CASELESS was set when matching characters that were quantified with ?
 3962     or *.
 3963 
 3964 17. A character class other than a single negated character that had a minimum
 3965     but no maximum quantifier - for example [ab]{6,} - was not handled
 3966     correctly by pce_dfa_exec(). It would match only one character.
 3967 
 3968 18. A valid (though odd) pattern that looked like a POSIX character
 3969     class but used an invalid character after [ (for example [[,abc,]]) caused
 3970     pcre_compile() to give the error "Failed: internal error: code overflow" or
 3971     in some cases to crash with a glibc free() error. This could even happen if
 3972     the pattern terminated after [[ but there just happened to be a sequence of
 3973     letters, a binary zero, and a closing ] in the memory that followed.
 3974 
 3975 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
 3976     over the years. Originally (before any Unicode support), just the bottom 8
 3977     bits were taken. Thus, for example, \500 really meant \100. Nowadays the
 3978     output from "man perlunicode" includes this:
 3979 
 3980       The regular expression compiler produces polymorphic opcodes.  That
 3981       is, the pattern adapts to the data and automatically switches to
 3982       the Unicode character scheme when presented with Unicode data--or
 3983       instead uses a traditional byte scheme when presented with byte
 3984       data.
 3985 
 3986     Sadly, a wide octal escape does not cause a switch, and in a string with
 3987     no other multibyte characters, these octal escapes are treated as before.
 3988     Thus, in Perl, the pattern  /\500/ actually matches \100 but the pattern
 3989     /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
 3990     Unicode string.
 3991 
 3992     I have not perpetrated such confusion in PCRE. Up till now, it took just
 3993     the bottom 8 bits, as in old Perl. I have now made octal escapes with
 3994     values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
 3995     translate to the appropriate multibyte character.
 3996 
 3997 29. Applied some refactoring to reduce the number of warnings from Microsoft
 3998     and Borland compilers. This has included removing the fudge introduced
 3999     seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
 4000     a warning about an unused variable.
 4001 
 4002 21. PCRE has not included VT (character 0x0b) in the set of whitespace
 4003     characters since release 4.0, because Perl (from release 5.004) does not.
 4004     [Or at least, is documented not to: some releases seem to be in conflict
 4005     with the documentation.] However, when a pattern was studied with
 4006     pcre_study() and all its branches started with \s, PCRE still included VT
 4007     as a possible starting character. Of course, this did no harm; it just
 4008     caused an unnecessary match attempt.
 4009 
 4010 22. Removed a now-redundant internal flag bit that recorded the fact that case
 4011     dependency changed within the pattern. This was once needed for "required
 4012     byte" processing, but is no longer used. This recovers a now-scarce options
 4013     bit. Also moved the least significant internal flag bit to the most-
 4014     significant bit of the word, which was not previously used (hangover from
 4015     the days when it was an int rather than a uint) to free up another bit for
 4016     the future.
 4017 
 4018 23. Added support for CRLF line endings as well as CR and LF. As well as the
 4019     default being selectable at build time, it can now be changed at runtime
 4020     via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
 4021     specify that it is scanning data with non-default line endings.
 4022 
 4023 24. Changed the definition of CXXLINK to make it agree with the definition of
 4024     LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
 4025 
 4026 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
 4027     recursions. This makes a big different to stack usage for some patterns.
 4028 
 4029 26. If a subpattern containing a named recursion or subroutine reference such
 4030     as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
 4031     the space required for the compiled pattern went wrong and gave too small a
 4032     value. Depending on the environment, this could lead to "Failed: internal
 4033     error: code overflow at offset 49" or "glibc detected double free or
 4034     corruption" errors.
 4035 
 4036 27. Applied patches from Google (a) to support the new newline modes and (b) to
 4037     advance over multibyte UTF-8 characters in GlobalReplace.
 4038 
 4039 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
 4040     difference for some implementation of PCRE in some Windows version.
 4041 
 4042 29. Added some extra testing facilities to pcretest:
 4043 
 4044     \q<number>   in a data line sets the "match limit" value
 4045     \Q<number>   in a data line sets the "match recursion limt" value
 4046     -S <number>  sets the stack size, where <number> is in megabytes
 4047 
 4048     The -S option isn't available for Windows.
 4049 
 4050 
 4051 Version 6.6 06-Feb-06
 4052 ---------------------
 4053 
 4054  1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
 4055     in pcreposix.h. I have copied the definition from pcre.h.
 4056 
 4057  2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
 4058     because pcre.h is no longer a built file.
 4059 
 4060  3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
 4061     not normally included in the compiled code.
 4062 
 4063 
 4064 Version 6.5 01-Feb-06
 4065 ---------------------
 4066 
 4067  1. When using the partial match feature with pcre_dfa_exec(), it was not
 4068     anchoring the second and subsequent partial matches at the new starting
 4069     point. This could lead to incorrect results. For example, with the pattern
 4070     /1234/, partially matching against "123" and then "a4" gave a match.
 4071 
 4072  2. Changes to pcregrep:
 4073 
 4074     (a) All non-match returns from pcre_exec() were being treated as failures
 4075         to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
 4076         error message is output. Some extra information is given for the
 4077         PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
 4078         probably the only errors that are likely to be caused by users (by
 4079         specifying a regex that has nested indefinite repeats, for instance).
 4080         If there are more than 20 of these errors, pcregrep is abandoned.
 4081 
 4082     (b) A binary zero was treated as data while matching, but terminated the
 4083         output line if it was written out. This has been fixed: binary zeroes
 4084         are now no different to any other data bytes.
 4085 
 4086     (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
 4087         used to set a locale for matching. The --locale=xxxx long option has
 4088         been added (no short equivalent) to specify a locale explicitly on the
 4089         pcregrep command, overriding the environment variables.
 4090 
 4091     (d) When -B was used with -n, some line numbers in the output were one less
 4092         than they should have been.
 4093 
 4094     (e) Added the -o (--only-matching) option.
 4095 
 4096     (f) If -A or -C was used with -c (count only), some lines of context were
 4097         accidentally printed for the final match.
 4098 
 4099     (g) Added the -H (--with-filename) option.
 4100 
 4101     (h) The combination of options -rh failed to suppress file names for files
 4102         that were found from directory arguments.
 4103 
 4104     (i) Added the -D (--devices) and -d (--directories) options.
 4105 
 4106     (j) Added the -F (--fixed-strings) option.
 4107 
 4108     (k) Allow "-" to be used as a file name for -f as well as for a data file.
 4109 
 4110     (l) Added the --colo(u)r option.
 4111 
 4112     (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
 4113         is not present by default.
 4114 
 4115  3. A nasty bug was discovered in the handling of recursive patterns, that is,
 4116     items such as (?R) or (?1), when the recursion could match a number of
 4117     alternatives. If it matched one of the alternatives, but subsequently,
 4118     outside the recursion, there was a failure, the code tried to back up into
 4119     the recursion. However, because of the way PCRE is implemented, this is not
 4120     possible, and the result was an incorrect result from the match.
 4121 
 4122     In order to prevent this happening, the specification of recursion has
 4123     been changed so that all such subpatterns are automatically treated as
 4124     atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
 4125 
 4126  4. I had overlooked the fact that, in some locales, there are characters for
 4127     which isalpha() is true but neither isupper() nor islower() are true. In
 4128     the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
 4129     and ordfeminine) are like this. This affected the treatment of \w and \W
 4130     when they appeared in character classes, but not when they appeared outside
 4131     a character class. The bit map for "word" characters is now created
 4132     separately from the results of isalnum() instead of just taking it from the
 4133     upper, lower, and digit maps. (Plus the underscore character, of course.)
 4134 
 4135  5. The above bug also affected the handling of POSIX character classes such as
 4136     [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
 4137     permanent tables. Instead, the bit maps for such a class were previously
 4138     created as the appropriate unions of the upper, lower, and digit bitmaps.
 4139     Now they are created by subtraction from the [[:word:]] class, which has
 4140     its own bitmap.
 4141 
 4142  6. The [[:blank:]] character class matches horizontal, but not vertical space.
 4143     It is created by subtracting the vertical space characters (\x09, \x0a,
 4144     \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
 4145     subtraction was done in the overall bitmap for a character class, meaning
 4146     that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
 4147     be recognized. This bug has been fixed.
 4148 
 4149  7. Patches from the folks at Google:
 4150 
 4151       (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
 4152       real life, but is still worth protecting against".
 4153 
 4154       (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
 4155       regular expressions".
 4156 
 4157       (c) pcre_scanner.cc: avoid use of std::count() because not all systems
 4158       have it.
 4159 
 4160       (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
 4161       "configure" and the latter not, in order to fix a problem somebody had
 4162       with compiling the Arg class on HP-UX.
 4163 
 4164       (e) Improve the error-handling of the C++ wrapper a little bit.
 4165 
 4166       (f) New tests for checking recursion limiting.
 4167 
 4168  8. The pcre_memmove() function, which is used only if the environment does not
 4169     have a standard memmove() function (and is therefore rarely compiled),
 4170     contained two bugs: (a) use of int instead of size_t, and (b) it was not
 4171     returning a result (though PCRE never actually uses the result).
 4172 
 4173  9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
 4174     large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
 4175     returned instead of calling malloc() with an overflowing number that would
 4176     most likely cause subsequent chaos.
 4177 
 4178 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
 4179 
 4180 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
 4181     with this option is matched, the nmatch and pmatch options of regexec() are
 4182     ignored.
 4183 
 4184 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
 4185     provided in case anyone wants to the the POSIX interface with UTF-8
 4186     strings.
 4187 
 4188 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
 4189     C++ linking (needed for some HP-UX environments).
 4190 
 4191 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
 4192     (unused parameter) and in the pcre_printint() function (omitted "default"
 4193     switch label when the default is to do nothing).
 4194 
 4195 15. Added some code to make it possible, when PCRE is compiled as a C++
 4196     library, to replace subject pointers for pcre_exec() with a smart pointer
 4197     class, thus making it possible to process discontinuous strings.
 4198 
 4199 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
 4200     much the same function. They were added by different people who were trying
 4201     to make PCRE easy to compile on non-Unix systems. It has been suggested
 4202     that PCRE_EXPORT be abolished now that there is more automatic apparatus
 4203     for compiling on Windows systems. I have therefore replaced it with
 4204     PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
 4205     defaults to "extern" for C or "extern C" for C++, which works fine on
 4206     Unix-like systems. It is now possible to override the value of PCRE_DATA_
 4207     SCOPE with something explicit in config.h. In addition:
 4208 
 4209     (a) pcreposix.h still had just "extern" instead of either of these macros;
 4210         I have replaced it with PCRE_DATA_SCOPE.
 4211 
 4212     (b) Functions such as _pcre_xclass(), which are internal to the library,
 4213         but external in the C sense, all had PCRE_EXPORT in their definitions.
 4214         This is apparently wrong for the Windows case, so I have removed it.
 4215         (It makes no difference on Unix-like systems.)
 4216 
 4217 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
 4218     of recursive calls to match(). This is different to MATCH_LIMIT because
 4219     that limits the total number of calls to match(), not all of which increase
 4220     the depth of recursion. Limiting the recursion depth limits the amount of
 4221     stack (or heap if NO_RECURSE is set) that is used. The default can be set
 4222     when PCRE is compiled, and changed at run time. A patch from Google adds
 4223     this functionality to the C++ interface.
 4224 
 4225 18. Changes to the handling of Unicode character properties:
 4226 
 4227     (a) Updated the table to Unicode 4.1.0.
 4228 
 4229     (b) Recognize characters that are not in the table as "Cn" (undefined).
 4230 
 4231     (c) I revised the way the table is implemented to a much improved format
 4232         which includes recognition of ranges. It now supports the ranges that
 4233         are defined in UnicodeData.txt, and it also amalgamates other
 4234         characters into ranges. This has reduced the number of entries in the
 4235         table from around 16,000 to around 3,000, thus reducing its size
 4236         considerably. I realized I did not need to use a tree structure after
 4237         all - a binary chop search is just as efficient. Having reduced the
 4238         number of entries, I extended their size from 6 bytes to 8 bytes to
 4239         allow for more data.
 4240 
 4241     (d) Added support for Unicode script names via properties such as \p{Han}.
 4242 
 4243 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
 4244     matching that character.
 4245 
 4246 20. When matching a repeated Unicode property with a minimum greater than zero,
 4247     (for example \pL{2,}), PCRE could look past the end of the subject if it
 4248     reached it while seeking the minimum number of characters. This could
 4249     happen only if some of the characters were more than one byte long, because
 4250     there is a check for at least the minimum number of bytes.
 4251 
 4252 21. Refactored the implementation of \p and \P so as to be more general, to
 4253     allow for more different types of property in future. This has changed the
 4254     compiled form incompatibly. Anybody with saved compiled patterns that use
 4255     \p or \P will have to recompile them.
 4256 
 4257 22. Added "Any" and "L&" to the supported property types.
 4258 
 4259 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
 4260     but give a compile time error if the value is greater than 0xff.
 4261 
 4262 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
 4263     accidentally not being installed or uninstalled.
 4264 
 4265 25. The pcre.h file was built from pcre.h.in, but the only changes that were
 4266     made were to insert the current release number. This seemed silly, because
 4267     it made things harder for people building PCRE on systems that don't run
 4268     "configure". I have turned pcre.h into a distributed file, no longer built
 4269     by "configure", with the version identification directly included. There is
 4270     no longer a pcre.h.in file.
 4271 
 4272     However, this change necessitated a change to the pcre-config script as
 4273     well. It is built from pcre-config.in, and one of the substitutions was the
 4274     release number. I have updated configure.ac so that ./configure now finds
 4275     the release number by grepping pcre.h.
 4276 
 4277 26. Added the ability to run the tests under valgrind.
 4278 
 4279 
 4280 Version 6.4 05-Sep-05
 4281 ---------------------
 4282 
 4283  1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
 4284     "--" to be printed when multiple files were scanned, even when none of the
 4285     -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
 4286     consider it to be a bug, and have restored the previous behaviour.
 4287 
 4288  2. A couple of code tidies to get rid of compiler warnings.
 4289 
 4290  3. The pcretest program used to cheat by referring to symbols in the library
 4291     whose names begin with _pcre_. These are internal symbols that are not
 4292     really supposed to be visible externally, and in some environments it is
 4293     possible to suppress them. The cheating is now confined to including
 4294     certain files from the library's source, which is a bit cleaner.
 4295 
 4296  4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
 4297     file's purpose clearer.
 4298 
 4299  5. Reorganized pcre_ucp_findchar().
 4300 
 4301 
 4302 Version 6.3 15-Aug-05
 4303 ---------------------
 4304 
 4305  1. The file libpcre.pc.in did not have general read permission in the tarball.
 4306 
 4307  2. There were some problems when building without C++ support:
 4308 
 4309     (a) If C++ support was not built, "make install" and "make test" still
 4310         tried to test it.
 4311 
 4312     (b) There were problems when the value of CXX was explicitly set. Some
 4313         changes have been made to try to fix these, and ...
 4314 
 4315     (c) --disable-cpp can now be used to explicitly disable C++ support.
 4316 
 4317     (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
 4318         backslash in a target when C++ was disabled. This confuses some
 4319         versions of "make", apparently. Using an intermediate variable solves
 4320         this. (Same for CPP_LOBJ.)
 4321 
 4322  3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
 4323     (non-Windows) now includes $(CFLAGS) because these flags are sometimes
 4324     necessary on certain architectures.
 4325 
 4326  4. Added a setting of -export-symbols-regex to the link command to remove
 4327     those symbols that are exported in the C sense, but actually are local
 4328     within the library, and not documented. Their names all begin with
 4329     "_pcre_". This is not a perfect job, because (a) we have to except some
 4330     symbols that pcretest ("illegally") uses, and (b) the facility isn't always
 4331     available (and never for static libraries). I have made a note to try to
 4332     find a way round (a) in the future.
 4333 
 4334 
 4335 Version 6.2 01-Aug-05
 4336 ---------------------
 4337 
 4338  1. There was no test for integer overflow of quantifier values. A construction
 4339     such as {1111111111111111} would give undefined results. What is worse, if
 4340     a minimum quantifier for a parenthesized subpattern overflowed and became
 4341     negative, the calculation of the memory size went wrong. This could have
 4342     led to memory overwriting.
 4343 
 4344  2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
 4345 
 4346  3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
 4347     operating environments where this matters.
 4348 
 4349  4. Applied Giuseppe Maxia's patch to add additional features for controlling
 4350     PCRE options from within the C++ wrapper.
 4351 
 4352  5. Named capturing subpatterns were not being correctly counted when a pattern
 4353     was compiled. This caused two problems: (a) If there were more than 100
 4354     such subpatterns, the calculation of the memory needed for the whole
 4355     compiled pattern went wrong, leading to an overflow error. (b) Numerical
 4356     back references of the form \12, where the number was greater than 9, were
 4357     not recognized as back references, even though there were sufficient
 4358     previous subpatterns.
 4359 
 4360  6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
 4361     versions of gcc, e.g. 2.95.4.
 4362 
 4363 
 4364 Version 6.1 21-Jun-05
 4365 ---------------------
 4366 
 4367  1. There was one reference to the variable "posix" in pcretest.c that was not
 4368     surrounded by "#if !defined NOPOSIX".
 4369 
 4370  2. Make it possible to compile pcretest without DFA support, UTF8 support, or
 4371     the cross-check on the old pcre_info() function, for the benefit of the
 4372     cut-down version of PCRE that is currently imported into Exim.
 4373 
 4374  3. A (silly) pattern starting with (?i)(?-i) caused an internal space
 4375     allocation error. I've done the easy fix, which wastes 2 bytes for sensible
 4376     patterns that start (?i) but I don't think that matters. The use of (?i) is
 4377     just an example; this all applies to the other options as well.
 4378 
 4379  4. Since libtool seems to echo the compile commands it is issuing, the output
 4380     from "make" can be reduced a bit by putting "@" in front of each libtool
 4381     compile command.
 4382 
 4383  5. Patch from the folks at Google for configure.in to be a bit more thorough
 4384     in checking for a suitable C++ installation before trying to compile the
 4385     C++ stuff. This should fix a reported problem when a compiler was present,
 4386     but no suitable headers.
 4387 
 4388  6. The man pages all had just "PCRE" as their title. I have changed them to
 4389     be the relevant file name. I have also arranged that these names are
 4390     retained in the file doc/pcre.txt, which is a concatenation in text format
 4391     of all the man pages except the little individual ones for each function.
 4392 
 4393  7. The NON-UNIX-USE file had not been updated for the different set of source
 4394     files that come with release 6. I also added a few comments about the C++
 4395     wrapper.
 4396 
 4397 
 4398 Version 6.0 07-Jun-05
 4399 ---------------------
 4400 
 4401  1. Some minor internal re-organization to help with my DFA experiments.
 4402 
 4403  2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
 4404     didn't matter for the library itself when fully configured, but did matter
 4405     when compiling without UCP support, or within Exim, where the ucp files are
 4406     not imported.
 4407 
 4408  3. Refactoring of the library code to split up the various functions into
 4409     different source modules. The addition of the new DFA matching code (see
 4410     below) to a single monolithic source would have made it really too
 4411     unwieldy, quite apart from causing all the code to be include in a
 4412     statically linked application, when only some functions are used. This is
 4413     relevant even without the DFA addition now that patterns can be compiled in
 4414     one application and matched in another.
 4415 
 4416     The downside of splitting up is that there have to be some external
 4417     functions and data tables that are used internally in different modules of
 4418     the library but which are not part of the API. These have all had their
 4419     names changed to start with "_pcre_" so that they are unlikely to clash
 4420     with other external names.
 4421 
 4422  4. Added an alternate matching function, pcre_dfa_exec(), which matches using
 4423     a different (DFA) algorithm. Although it is slower than the original
 4424     function, it does have some advantages for certain types of matching
 4425     problem.
 4426 
 4427  5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
 4428     including restarting after a partial match.
 4429 
 4430  6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
 4431     defined when compiling for Windows was sent to me. I have put it into the
 4432     code, though I have no means of testing or verifying it.
 4433 
 4434  7. Added the pcre_refcount() auxiliary function.
 4435 
 4436  8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
 4437     match before or at the first newline in the subject string. In pcretest,
 4438     the /f option on a pattern can be used to set this.
 4439 
 4440  9. A repeated \w when used in UTF-8 mode with characters greater than 256
 4441     would behave wrongly. This has been present in PCRE since release 4.0.
 4442 
 4443 10. A number of changes to the pcregrep command:
 4444 
 4445     (a) Refactored how -x works; insert ^(...)$ instead of setting
 4446         PCRE_ANCHORED and checking the length, in preparation for adding
 4447         something similar for -w.
 4448 
 4449     (b) Added the -w (match as a word) option.
 4450 
 4451     (c) Refactored the way lines are read and buffered so as to have more
 4452         than one at a time available.
 4453 
 4454     (d) Implemented a pcregrep test script.
 4455 
 4456     (e) Added the -M (multiline match) option. This allows patterns to match
 4457         over several lines of the subject. The buffering ensures that at least
 4458         8K, or the rest of the document (whichever is the shorter) is available
 4459         for matching (and similarly the previous 8K for lookbehind assertions).
 4460 
 4461     (f) Changed the --help output so that it now says
 4462 
 4463           -w, --word-regex(p)
 4464 
 4465         instead of two lines, one with "regex" and the other with "regexp"
 4466         because that confused at least one person since the short forms are the
 4467         same. (This required a bit of code, as the output is generated
 4468         automatically from a table. It wasn't just a text change.)
 4469 
 4470     (g) -- can be used to terminate pcregrep options if the next thing isn't an
 4471         option but starts with a hyphen. Could be a pattern or a path name
 4472         starting with a hyphen, for instance.
 4473 
 4474     (h) "-" can be given as a file name to represent stdin.
 4475 
 4476     (i) When file names are being printed, "(standard input)" is used for
 4477         the standard input, for compatibility with GNU grep. Previously
 4478         "<stdin>" was used.
 4479 
 4480     (j) The option --label=xxx can be used to supply a name to be used for
 4481         stdin when file names are being printed. There is no short form.
 4482 
 4483     (k) Re-factored the options decoding logic because we are going to add
 4484         two more options that take data. Such options can now be given in four
 4485         different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
 4486 
 4487     (l) Added the -A, -B, and -C options for requesting that lines of context
 4488         around matches be printed.
 4489 
 4490     (m) Added the -L option to print the names of files that do not contain
 4491         any matching lines, that is, the complement of -l.
 4492 
 4493     (n) The return code is 2 if any file cannot be opened, but pcregrep does
 4494         continue to scan other files.
 4495 
 4496     (o) The -s option was incorrectly implemented. For compatibility with other
 4497         greps, it now suppresses the error message for a non-existent or non-
 4498         accessible file (but not the return code). There is a new option called
 4499         -q that suppresses the output of matching lines, which was what -s was
 4500         previously doing.
 4501 
 4502     (p) Added --include and --exclude options to specify files for inclusion
 4503         and exclusion when recursing.
 4504 
 4505 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
 4506     Hopefully, it now does.
 4507 
 4508 12. Missing cast in pcre_study().
 4509 
 4510 13. Added an "uninstall" target to the makefile.
 4511 
 4512 14. Replaced "extern" in the function prototypes in Makefile.in with
 4513     "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
 4514     world, but is set differently for Windows.
 4515 
 4516 15. Added a second compiling function called pcre_compile2(). The only
 4517     difference is that it has an extra argument, which is a pointer to an
 4518     integer error code. When there is a compile-time failure, this is set
 4519     non-zero, in addition to the error test pointer being set to point to an
 4520     error message. The new argument may be NULL if no error number is required
 4521     (but then you may as well call pcre_compile(), which is now just a
 4522     wrapper). This facility is provided because some applications need a
 4523     numeric error indication, but it has also enabled me to tidy up the way
 4524     compile-time errors are handled in the POSIX wrapper.
 4525 
 4526 16. Added VPATH=.libs to the makefile; this should help when building with one
 4527     prefix path and installing with another. (Or so I'm told by someone who
 4528     knows more about this stuff than I do.)
 4529 
 4530 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
 4531     passes PCRE_DOTALL to the pcre_compile() function, making the "." character
 4532     match everything, including newlines. This is not POSIX-compatible, but
 4533     somebody wanted the feature. From pcretest it can be activated by using
 4534     both the P and the s flags.
 4535 
 4536 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
 4537 
 4538 19. libpcre.pc was being incorrectly installed as executable.
 4539 
 4540 20. A couple of places in pcretest check for end-of-line by looking for '\n';
 4541     it now also looks for '\r' so that it will work unmodified on Windows.
 4542 
 4543 21. Added Google's contributed C++ wrapper to the distribution.
 4544 
 4545 22. Added some untidy missing memory free() calls in pcretest, to keep
 4546     Electric Fence happy when testing.
 4547 
 4548 
 4549 
 4550 Version 5.0 13-Sep-04
 4551 ---------------------
 4552 
 4553  1. Internal change: literal characters are no longer packed up into items
 4554     containing multiple characters in a single byte-string. Each character
 4555     is now matched using a separate opcode. However, there may be more than one
 4556     byte in the character in UTF-8 mode.
 4557 
 4558  2. The pcre_callout_block structure has two new fields: pattern_position and
 4559     next_item_length. These contain the offset in the pattern to the next match
 4560     item, and its length, respectively.
 4561 
 4562  3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
 4563     insertion of callouts before each pattern item. Added the /C option to
 4564     pcretest to make use of this.
 4565 
 4566  4. On the advice of a Windows user, the lines
 4567 
 4568       #if defined(_WIN32) || defined(WIN32)
 4569       _setmode( _fileno( stdout ), 0x8000 );
 4570       #endif  /* defined(_WIN32) || defined(WIN32) */
 4571 
 4572     have been added to the source of pcretest. This apparently does useful
 4573     magic in relation to line terminators.
 4574 
 4575  5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
 4576     for the benefit of those environments where the "b" makes a difference.
 4577 
 4578  6. The icc compiler has the same options as gcc, but "configure" doesn't seem
 4579     to know about it. I have put a hack into configure.in that adds in code
 4580     to set GCC=yes if CC=icc. This seems to end up at a point in the
 4581     generated configure script that is early enough to affect the setting of
 4582     compiler options, which is what is needed, but I have no means of testing
 4583     whether it really works. (The user who reported this had patched the
 4584     generated configure script, which of course I cannot do.)
 4585 
 4586     LATER: After change 22 below (new libtool files), the configure script
 4587     seems to know about icc (and also ecc). Therefore, I have commented out
 4588     this hack in configure.in.
 4589 
 4590  7. Added support for pkg-config (2 patches were sent in).
 4591 
 4592  8. Negated POSIX character classes that used a combination of internal tables
 4593     were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
 4594     [[:^ascii]]. Typically, they would match almost any characters. The other
 4595     POSIX classes were not broken in this way.
 4596 
 4597  9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
 4598     to find the match, as PCRE was deluded into thinking that the match had to
 4599     start at the start point or following a newline. The same bug applied to
 4600     patterns with negative forward assertions or any backward assertions
 4601     preceding ".*" at the start, unless the pattern required a fixed first
 4602     character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
 4603 
 4604 10. In UTF-8 mode, when moving forwards in the subject after a failed match
 4605     starting at the last subject character, bytes beyond the end of the subject
 4606     string were read.
 4607 
 4608 11. Renamed the variable "class" as "classbits" to make life easier for C++
 4609     users. (Previously there was a macro definition, but it apparently wasn't
 4610     enough.)
 4611 
 4612 12. Added the new field "tables" to the extra data so that tables can be passed
 4613     in at exec time, or the internal tables can be re-selected. This allows
 4614     a compiled regex to be saved and re-used at a later time by a different
 4615     program that might have everything at different addresses.
 4616 
 4617 13. Modified the pcre-config script so that, when run on Solaris, it shows a
 4618     -R library as well as a -L library.
 4619 
 4620 14. The debugging options of pcretest (-d on the command line or D on a
 4621     pattern) showed incorrect output for anything following an extended class
 4622     that contained multibyte characters and which was followed by a quantifier.
 4623 
 4624 15. Added optional support for general category Unicode character properties
 4625     via the \p, \P, and \X escapes. Unicode property support implies UTF-8
 4626     support. It adds about 90K to the size of the library. The meanings of the
 4627     inbuilt class escapes such as \d and \s have NOT been changed.
 4628 
 4629 16. Updated pcredemo.c to include calls to free() to release the memory for the
 4630     compiled pattern.
 4631 
 4632 17. The generated file chartables.c was being created in the source directory
 4633     instead of in the building directory. This caused the build to fail if the
 4634     source directory was different from the building directory, and was
 4635     read-only.
 4636 
 4637 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
 4638     file. No doubt somebody will tell me if they don't make sense... Also added
 4639     Dan Mooney's comments about building on OpenVMS.
 4640 
 4641 19. Added support for partial matching via the PCRE_PARTIAL option for
 4642     pcre_exec() and the \P data escape in pcretest.
 4643 
 4644 20. Extended pcretest with 3 new pattern features:
 4645 
 4646     (i)   A pattern option of the form ">rest-of-line" causes pcretest to
 4647           write the compiled pattern to the file whose name is "rest-of-line".
 4648           This is a straight binary dump of the data, with the saved pointer to
 4649           the character tables forced to be NULL. The study data, if any, is
 4650           written too. After writing, pcretest reads a new pattern.
 4651 
 4652     (ii)  If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
 4653           compiled pattern from the given file. There must not be any
 4654           occurrences of "<" in the file name (pretty unlikely); if there are,
 4655           pcretest will instead treat the initial "<" as a pattern delimiter.
 4656           After reading in the pattern, pcretest goes on to read data lines as
 4657           usual.
 4658 
 4659     (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
 4660           and 16-bit fields in a compiled pattern, to simulate a pattern that
 4661           was compiled on a host of opposite endianness.
 4662 
 4663 21. The pcre-exec() function can now cope with patterns that were compiled on
 4664     hosts of opposite endianness, with this restriction:
 4665 
 4666       As for any compiled expression that is saved and used later, the tables
 4667       pointer field cannot be preserved; the extra_data field in the arguments
 4668       to pcre_exec() should be used to pass in a tables address if a value
 4669       other than the default internal tables were used at compile time.
 4670 
 4671 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
 4672     now diagnosed as an error. Previously, most of the time, a negative number
 4673     would have been treated as zero, but if in addition "ovector" was passed as
 4674     NULL, a crash could occur.
 4675 
 4676 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
 4677     new versions from the libtool 1.5 distribution (the last one is a copy of
 4678     a file called libtool.m4). This seems to have fixed the need to patch
 4679     "configure" to support Darwin 1.3 (which I used to do). However, I still
 4680     had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
 4681     workstation).
 4682 
 4683 24. Changed the PCRE licence to be the more standard "BSD" licence.
 4684 
 4685 
 4686 Version 4.5 01-Dec-03
 4687 ---------------------
 4688 
 4689  1. There has been some re-arrangement of the code for the match() function so
 4690     that it can be compiled in a version that does not call itself recursively.
 4691     Instead, it keeps those local variables that need separate instances for
 4692     each "recursion" in a frame on the heap, and gets/frees frames whenever it
 4693     needs to "recurse". Keeping track of where control must go is done by means
 4694     of setjmp/longjmp. The whole thing is implemented by a set of macros that
 4695     hide most of the details from the main code, and operates only if
 4696     NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
 4697     "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
 4698     operating.
 4699 
 4700     To make it easier for callers to provide specially tailored get/free
 4701     functions for this usage, two new functions, pcre_stack_malloc, and
 4702     pcre_stack_free, are used. They are always called in strict stacking order,
 4703     and the size of block requested is always the same.
 4704 
 4705     The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
 4706     PCRE has been compiled to use the stack or the heap for recursion. The
 4707     -C option of pcretest uses this to show which version is compiled.
 4708 
 4709     A new data escape \S, is added to pcretest; it causes the amounts of store
 4710     obtained and freed by both kinds of malloc/free at match time to be added
 4711     to the output.
 4712 
 4713  2. Changed the locale test to use "fr_FR" instead of "fr" because that's
 4714     what's available on my current Linux desktop machine.
 4715 
 4716  3. When matching a UTF-8 string, the test for a valid string at the start has
 4717     been extended. If start_offset is not zero, PCRE now checks that it points
 4718     to a byte that is the start of a UTF-8 character. If not, it returns
 4719     PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
 4720     this is necessary because there may be backward assertions in the pattern.
 4721     When matching the same subject several times, it may save resources to use
 4722     PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
 4723 
 4724  4. The code for checking the validity of UTF-8 strings has been tightened so
 4725     that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
 4726     containing "overlong sequences".
 4727 
 4728  5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
 4729     I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
 4730     should have been "&", but it just so happened that all the cases this let
 4731     through by mistake were picked up later in the function.
 4732 
 4733  6. I had used a variable called "isblank" - this is a C99 function, causing
 4734     some compilers to warn. To avoid this, I renamed it (as "blankclass").
 4735 
 4736  7. Cosmetic: (a) only output another newline at the end of pcretest if it is
 4737     prompting; (b) run "./pcretest /dev/null" at the start of the test script
 4738     so the version is shown; (c) stop "make test" echoing "./RunTest".
 4739 
 4740  8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
 4741 
 4742  9. The prototype for memmove() for systems that don't have it was using
 4743     size_t, but the inclusion of the header that defines size_t was later. I've
 4744     moved the #includes for the C headers earlier to avoid this.
 4745 
 4746 10. Added some adjustments to the code to make it easier to compiler on certain
 4747     special systems:
 4748 
 4749       (a) Some "const" qualifiers were missing.
 4750       (b) Added the macro EXPORT before all exported functions; by default this
 4751           is defined to be empty.
 4752       (c) Changed the dftables auxiliary program (that builds chartables.c) so
 4753           that it reads its output file name as an argument instead of writing
 4754           to the standard output and assuming this can be redirected.
 4755 
 4756 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
 4757     class containing characters with values greater than 255, PCRE compilation
 4758     went into a loop.
 4759 
 4760 12. A recursive reference to a subpattern that was within another subpattern
 4761     that had a minimum quantifier of zero caused PCRE to crash. For example,
 4762     (x(y(?2))z)? provoked this bug with a subject that got as far as the
 4763     recursion. If the recursively-called subpattern itself had a zero repeat,
 4764     that was OK.
 4765 
 4766 13. In pcretest, the buffer for reading a data line was set at 30K, but the
 4767     buffer into which it was copied (for escape processing) was still set at
 4768     1024, so long lines caused crashes.
 4769 
 4770 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
 4771     "internal error: code overflow...". This applied to any character class
 4772     that was followed by a possessive quantifier.
 4773 
 4774 15. Modified the Makefile to add libpcre.la as a prerequisite for
 4775     libpcreposix.la because I was told this is needed for a parallel build to
 4776     work.
 4777 
 4778 16. If a pattern that contained .* following optional items at the start was
 4779     studied, the wrong optimizing data was generated, leading to matching
 4780     errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
 4781     matching string must start with a or b or c. The correct conclusion for
 4782     this pattern is that a match can start with any character.
 4783 
 4784 
 4785 Version 4.4 13-Aug-03
 4786 ---------------------
 4787 
 4788  1. In UTF-8 mode, a character class containing characters with values between
 4789     127 and 255 was not handled correctly if the compiled pattern was studied.
 4790     In fixing this, I have also improved the studying algorithm for such
 4791     classes (slightly).
 4792 
 4793  2. Three internal functions had redundant arguments passed to them. Removal
 4794     might give a very teeny performance improvement.
 4795 
 4796  3. Documentation bug: the value of the capture_top field in a callout is *one
 4797     more than* the number of the hightest numbered captured substring.
 4798 
 4799  4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
 4800     in incorrectly linking with a previously installed version. They now link
 4801     explicitly with libpcre.la.
 4802 
 4803  5. configure.in no longer needs to recognize Cygwin specially.
 4804 
 4805  6. A problem in pcre.in for Windows platforms is fixed.
 4806 
 4807  7. If a pattern was successfully studied, and the -d (or /D) flag was given to
 4808     pcretest, it used to include the size of the study block as part of its
 4809     output. Unfortunately, the structure contains a field that has a different
 4810     size on different hardware architectures. This meant that the tests that
 4811     showed this size failed. As the block is currently always of a fixed size,
 4812     this information isn't actually particularly useful in pcretest output, so
 4813     I have just removed it.
 4814 
 4815  8. Three pre-processor statements accidentally did not start in column 1.
 4816     Sadly, there are *still* compilers around that complain, even though
 4817     standard C has not required this for well over a decade. Sigh.
 4818 
 4819  9. In pcretest, the code for checking callouts passed small integers in the
 4820     callout_data field, which is a void * field. However, some picky compilers
 4821     complained about the casts involved for this on 64-bit systems. Now
 4822     pcretest passes the address of the small integer instead, which should get
 4823     rid of the warnings.
 4824 
 4825 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
 4826     both compile and run time, and gives an error if an invalid UTF-8 sequence
 4827     is found. There is a option for disabling this check in cases where the
 4828     string is known to be correct and/or the maximum performance is wanted.
 4829 
 4830 11. In response to a bug report, I changed one line in Makefile.in from
 4831 
 4832         -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
 4833     to
 4834         -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
 4835 
 4836     to look similar to other lines, but I have no way of telling whether this
 4837     is the right thing to do, as I do not use Windows. No doubt I'll get told
 4838     if it's wrong...
 4839 
 4840 
 4841 Version 4.3 21-May-03
 4842 ---------------------
 4843 
 4844 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
 4845    Makefile.
 4846 
 4847 2. Some refactoring to improve the quality of the code:
 4848 
 4849    (i)   The utf8_table... variables are now declared "const".
 4850 
 4851    (ii)  The code for \cx, which used the "case flipping" table to upper case
 4852          lower case letters, now just substracts 32. This is ASCII-specific,
 4853          but the whole concept of \cx is ASCII-specific, so it seems
 4854          reasonable.
 4855 
 4856    (iii) PCRE was using its character types table to recognize decimal and
 4857          hexadecimal digits in the pattern. This is silly, because it handles
 4858          only 0-9, a-f, and A-F, but the character types table is locale-
 4859          specific, which means strange things might happen. A private
 4860          table is now used for this - though it costs 256 bytes, a table is
 4861          much faster than multiple explicit tests. Of course, the standard
 4862          character types table is still used for matching digits in subject
 4863          strings against \d.
 4864 
 4865    (iv)  Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
 4866          ending in _t are). So I've renamed it as ESC_tee.
 4867 
 4868 3. The first argument for regexec() in the POSIX wrapper should have been
 4869    defined as "const".
 4870 
 4871 4. Changed pcretest to use malloc() for its buffers so that they can be
 4872    Electric Fenced for debugging.
 4873 
 4874 5. There were several places in the code where, in UTF-8 mode, PCRE would try
 4875    to read one or more bytes before the start of the subject string. Often this
 4876    had no effect on PCRE's behaviour, but in some circumstances it could
 4877    provoke a segmentation fault.
 4878 
 4879 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
 4880    to try to read one or more bytes before the start of the subject string.
 4881 
 4882 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
 4883    UTF-8 support could misbehave in various ways if the subject string
 4884    contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
 4885    area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
 4886    back over UTF-8 characters.)
 4887 
 4888 
 4889 Version 4.2 14-Apr-03
 4890 ---------------------
 4891 
 4892 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
 4893 
 4894 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
 4895      [ON_WINDOWS]: new variable, "#" on non-Windows platforms
 4896      [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
 4897      [WIN_PREFIX]: new variable, "cyg" for Cygwin
 4898      * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
 4899        and BUILD_EXEEXT
 4900      Note: automatic setting of the BUILD variables is not yet working
 4901      set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
 4902        compile-time but not at link-time
 4903      [LINK]: use for linking executables only
 4904      make different versions for Windows and non-Windows
 4905      [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
 4906        libraries
 4907      [LINK_FOR_BUILD]: new variable
 4908      [OBJEXT]: use throughout
 4909      [EXEEXT]: use throughout
 4910      <winshared>: new target
 4911      <wininstall>: new target
 4912      <dftables.o>: use native compiler
 4913      <dftables>: use native linker
 4914      <install>: handle Windows platform correctly
 4915      <clean>: ditto
 4916      <check>: ditto
 4917      copy DLL to top builddir before testing
 4918 
 4919    As part of these changes, -no-undefined was removed again. This was reported
 4920    to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
 4921    in any case.
 4922 
 4923 3. Some tidies to get rid of compiler warnings:
 4924 
 4925    . In the match_data structure, match_limit was an unsigned long int, whereas
 4926      match_call_count was an int. I've made them both unsigned long ints.
 4927 
 4928    . In pcretest the fact that a const uschar * doesn't automatically cast to
 4929      a void * provoked a warning.
 4930 
 4931    . Turning on some more compiler warnings threw up some "shadow" variables
 4932      and a few more missing casts.
 4933 
 4934 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
 4935    option, a class that contained a single character with a value between 128
 4936    and 255 (e.g. /[\xFF]/) caused PCRE to crash.
 4937 
 4938 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
 4939    option, a class that contained several characters, but with at least one
 4940    whose value was between 128 and 255 caused PCRE to crash.
 4941 
 4942 
 4943 Version 4.1 12-Mar-03
 4944 ---------------------
 4945 
 4946 1. Compiling with gcc -pedantic found a couple of places where casts were
 4947 needed, and a string in dftables.c that was longer than standard compilers are
 4948 required to support.
 4949 
 4950 2. Compiling with Sun's compiler found a few more places where the code could
 4951 be tidied up in order to avoid warnings.
 4952 
 4953 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
 4954 first of these names is deprecated in the latest Autoconf in favour of the name
 4955 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
 4956 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
 4957 analogy I have changed it to CFLAGS_FOR_BUILD.
 4958 
 4959 4. Added -no-undefined to the linking command in the Makefile, because this is
 4960 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
 4961 linking step for the pcreposix library.
 4962 
 4963 5. PCRE was failing to diagnose the case of two named groups with the same
 4964 name.
 4965 
 4966 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
 4967 literal character that is needed in the subject for a match, and scans along to
 4968 ensure that it is present before embarking on the full matching process. This
 4969 saves time in cases of nested unlimited repeats that are never going to match.
 4970 Problem: the scan can take a lot of time if the subject is very long (e.g.
 4971 megabytes), thus penalizing straightforward matches. It is now done only if the
 4972 amount of subject to be scanned is less than 1000 bytes.
 4973 
 4974 7. A lesser problem with the same optimization is that it was recording the
 4975 first character of an anchored pattern as "needed", thus provoking a search
 4976 right along the subject, even when the first match of the pattern was going to
 4977 fail. The "needed" character is now not set for anchored patterns, unless it
 4978 follows something in the pattern that is of non-fixed length. Thus, it still
 4979 fulfils its original purpose of finding quick non-matches in cases of nested
 4980 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
 4981 
 4982 
 4983 Version 4.0 17-Feb-03
 4984 ---------------------
 4985 
 4986 1. If a comment in an extended regex that started immediately after a meta-item
 4987 extended to the end of string, PCRE compiled incorrect data. This could lead to
 4988 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
 4989 
 4990 2. Moved to autoconf 2.53 and libtool 1.4.2.
 4991 
 4992 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
 4993 the special perltest8 script is no longer needed - all the tests can be run
 4994 from a single perltest script.
 4995 
 4996 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
 4997 by \s. It has now been removed in PCRE. This means it isn't recognized as
 4998 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
 4999 class [:space:] *does* include VT, thereby creating a mess.
 5000 
 5001 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
 5002 space and tab.
 5003 
 5004 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
 5005 its new features into the main test script, reducing the number of scripts.
 5006 
 5007 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
 5008 were backward compatible, and made the (?i) apply to the whole pattern, as if
 5009 /i were given. Now it behaves more logically, and applies the option setting
 5010 only to what follows. PCRE has been changed to follow suit. However, if it
 5011 finds options settings right at the start of the pattern, it extracts them into
 5012 the global options, as before. Thus, they show up in the info data.
 5013 
 5014 8. Added support for the \Q...\E escape sequence. Characters in between are
 5015 treated as literals. This is slightly different from Perl in that $ and @ are
 5016 also handled as literals inside the quotes. In Perl, they will cause variable
 5017 interpolation. Note the following examples:
 5018 
 5019     Pattern            PCRE matches      Perl matches
 5020 
 5021     \Qabc$xyz\E        abc$xyz           abc followed by the contents of $xyz
 5022     \Qabc\$xyz\E       abc\$xyz          abc\$xyz
 5023     \Qabc\E\$\Qxyz\E   abc$xyz           abc$xyz
 5024 
 5025 For compatibility with Perl, \Q...\E sequences are recognized inside character
 5026 classes as well as outside them.
 5027 
 5028 9. Re-organized 3 code statements in pcretest to avoid "overflow in
 5029 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
 5030 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
 5031 signed/unsigned warnings.
 5032 
 5033 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
 5034 option for pcretest, so I've replaced it by a simple function that does just
 5035 that job.
 5036 
 5037 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
 5038 "pcregrep -".
 5039 
 5040 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
 5041 Java package. This provides some syntactic sugar for simple cases of what my
 5042 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
 5043 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
 5044 item, you can use this simplified notation. Note that only makes sense with
 5045 greedy quantifiers. Consequently, the use of the possessive quantifier forces
 5046 greediness, whatever the setting of the PCRE_UNGREEDY option.
 5047 
 5048 13. A change of greediness default within a pattern was not taking effect at
 5049 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
 5050 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
 5051 was abstracted outside.
 5052 
 5053 14. PCRE now supports the \G assertion. It is true when the current matching
 5054 position is at the start point of the match. This differs from \A when the
 5055 starting offset is non-zero. Used with the /g option of pcretest (or similar
 5056 code), it works in the same way as it does for Perl's /g option. If all
 5057 alternatives of a regex begin with \G, the expression is anchored to the start
 5058 match position, and the "anchored" flag is set in the compiled expression.
 5059 
 5060 15. Some bugs concerning the handling of certain option changes within patterns
 5061 have been fixed. These applied to options other than (?ims). For example,
 5062 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
 5063 been the other way round. Some of this was related to change 7 above.
 5064 
 5065 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
 5066 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
 5067 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
 5068 POSIX classes only within a class (e.g. /[[:alpha:]]/).
 5069 
 5070 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
 5071 mode. Unlike ".", it always matches newline, whatever the setting of
 5072 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
 5073 assertions. Perl allows it, but it doesn't (in general) work because it can't
 5074 calculate the length of the lookbehind. At least, that's the case for Perl
 5075 5.8.0 - I've been told they are going to document that it doesn't work in
 5076 future.
 5077 
 5078 18. Added an error diagnosis for escapes that PCRE does not support: these are
 5079 \L, \l, \N, \P, \p, \U, \u, and \X.
 5080 
 5081 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
 5082 reading past the end of the pattern in cases such as /[abcd/.
 5083 
 5084 20. PCRE was getting more memory than necessary for patterns with classes that
 5085 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
 5086 
 5087 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
 5088 compiling PCRE for use with Virtual Pascal.
 5089 
 5090 22. Small fix to the Makefile to make it work properly if the build is done
 5091 outside the source tree.
 5092 
 5093 23. Added a new extension: a condition to go with recursion. If a conditional
 5094 subpattern starts with (?(R) the "true" branch is used if recursion has
 5095 happened, whereas the "false" branch is used only at the top level.
 5096 
 5097 24. When there was a very long string of literal characters (over 255 bytes
 5098 without UTF support, over 250 bytes with UTF support), the computation of how
 5099 much memory was required could be incorrect, leading to segfaults or other
 5100 strange effects.
 5101 
 5102 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
 5103 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
 5104 there was a subsequent back reference to those brackets. This meant that, for
 5105 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
 5106 possible to check for precisely this case. All we can do is abandon the
 5107 optimization if .* occurs inside capturing brackets when there are any back
 5108 references whatsoever. (See below for a better fix that came later.)
 5109 
 5110 26. The handling of the optimization for finding the first character of a
 5111 non-anchored pattern, and for finding a character that is required later in the
 5112 match were failing in some cases. This didn't break the matching; it just
 5113 failed to optimize when it could. The way this is done has been re-implemented.
 5114 
 5115 27. Fixed typo in error message for invalid (?R item (it said "(?p").
 5116 
 5117 28. Added a new feature that provides some of the functionality that Perl
 5118 provides with (?{...}). The facility is termed a "callout". The way it is done
 5119 in PCRE is for the caller to provide an optional function, by setting
 5120 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
 5121 global variable. By default it is unset, which disables all calling out. To get
 5122 the function called, the regex must include (?C) at appropriate points. This
 5123 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
 5124 This provides a means of identifying different callout points. When PCRE
 5125 reaches such a point in the regex, if pcre_callout has been set, the external
 5126 function is called. It is provided with data in a structure called
 5127 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
 5128 matching continues; if it returns a non-zero value, the match at the current
 5129 point fails. However, backtracking will occur if possible. [This was changed
 5130 later and other features added - see item 49 below.]
 5131 
 5132 29. pcretest is upgraded to test the callout functionality. It provides a
 5133 callout function that displays information. By default, it shows the start of
 5134 the match and the current position in the text. There are some new data escapes
 5135 to vary what happens:
 5136 
 5137     \C+         in addition, show current contents of captured substrings
 5138     \C-         do not supply a callout function
 5139     \C!n        return 1 when callout number n is reached
 5140     \C!n!m      return 1 when callout number n is reached for the mth time
 5141 
 5142 30. If pcregrep was called with the -l option and just a single file name, it
 5143 output "<stdin>" if a match was found, instead of the file name.
 5144 
 5145 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
 5146 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
 5147 pcre_exec(). This saves a malloc/free per call. The default value of
 5148 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
 5149 when configuring.
 5150 
 5151 32. The default maximum size of a compiled pattern is 64K. There have been a
 5152 few cases of people hitting this limit. The code now uses macros to handle the
 5153 storing of links as offsets within the compiled pattern. It defaults to 2-byte
 5154 links, but this can be changed to 3 or 4 bytes by --with-link-size when
 5155 configuring. Tests 2 and 5 work only with 2-byte links because they output
 5156 debugging information about compiled patterns.
 5157 
 5158 33. Internal code re-arrangements:
 5159 
 5160 (a) Moved the debugging function for printing out a compiled regex into
 5161     its own source file (printint.c) and used #include to pull it into
 5162     pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
 5163     separate copies.
 5164 
 5165 (b) Defined the list of op-code names for debugging as a macro in
 5166     internal.h so that it is next to the definition of the opcodes.
 5167 
 5168 (c) Defined a table of op-code lengths for simpler skipping along compiled
 5169     code. This is again a macro in internal.h so that it is next to the
 5170     definition of the opcodes.
 5171 
 5172 34. Added support for recursive calls to individual subpatterns, along the
 5173 lines of Robin Houston's patch (but implemented somewhat differently).
 5174 
 5175 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
 5176 allow it to read and process whole directories in Win32. This code was
 5177 contributed by Lionel Fourquaux; it has not been tested by me.
 5178 
 5179 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
 5180 used to name a group. Names consist of alphanumerics and underscores, and must
 5181 be unique. Back references use the syntax (?P=name) and recursive calls use
 5182 (?P>name) which is a PCRE extension to the Python extension. Groups still have
 5183 numbers. The function pcre_fullinfo() can be used after compilation to extract
 5184 a name/number map. There are three relevant calls:
 5185 
 5186   PCRE_INFO_NAMEENTRYSIZE        yields the size of each entry in the map
 5187   PCRE_INFO_NAMECOUNT            yields the number of entries
 5188   PCRE_INFO_NAMETABLE            yields a pointer to the map.
 5189 
 5190 The map is a vector of fixed-size entries. The size of each entry depends on
 5191 the length of the longest name used. The first two bytes of each entry are the
 5192 group number, most significant byte first. There follows the corresponding
 5193 name, zero terminated. The names are in alphabetical order.
 5194 
 5195 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
 5196 case instead of 255. Making it the same both with and without UTF-8 support
 5197 means that the same test output works with both.
 5198 
 5199 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
 5200 calling malloc() with a zero argument.
 5201 
 5202 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
 5203 optimization. I've improved things by keeping a bitmap of backreferences with
 5204 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
 5205 fact referenced, the optimization can be applied. It is unlikely that a
 5206 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
 5207 the match to follow \n) will appear inside brackets with a number greater than
 5208 31, but if it does, any back reference > 31 suppresses the optimization.
 5209 
 5210 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
 5211 of disabling numbered capturing parentheses. Any opening parenthesis that is
 5212 not followed by ? behaves as if it were followed by ?: but named parentheses
 5213 can still be used for capturing (and they will acquire numbers in the usual
 5214 way).
 5215 
 5216 41. Redesigned the return codes from the match() function into yes/no/error so
 5217 that errors can be passed back from deep inside the nested calls. A malloc
 5218 failure while inside a recursive subpattern call now causes the
 5219 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
 5220 
 5221 42. It is now possible to set a limit on the number of times the match()
 5222 function is called in a call to pcre_exec(). This facility makes it possible to
 5223 limit the amount of recursion and backtracking, though not in a directly
 5224 obvious way, because the match() function is used in a number of different
 5225 circumstances. The count starts from zero for each position in the subject
 5226 string (for non-anchored patterns). The default limit is, for compatibility, a
 5227 large number, namely 10 000 000. You can change this in two ways:
 5228 
 5229 (a) When configuring PCRE before making, you can use --with-match-limit=n
 5230     to set a default value for the compiled library.
 5231 
 5232 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
 5233     a different value is set. See 45 below.
 5234 
 5235 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
 5236 
 5237 43. Added a new function pcre_config(int, void *) to enable run-time extraction
 5238 of things that can be changed at compile time. The first argument specifies
 5239 what is wanted and the second points to where the information is to be placed.
 5240 The current list of available information is:
 5241 
 5242   PCRE_CONFIG_UTF8
 5243 
 5244 The output is an integer that is set to one if UTF-8 support is available;
 5245 otherwise it is set to zero.
 5246 
 5247   PCRE_CONFIG_NEWLINE
 5248 
 5249 The output is an integer that it set to the value of the code that is used for
 5250 newline. It is either LF (10) or CR (13).
 5251 
 5252   PCRE_CONFIG_LINK_SIZE
 5253 
 5254 The output is an integer that contains the number of bytes used for internal
 5255 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
 5256 
 5257   PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
 5258 
 5259 The output is an integer that contains the threshold above which the POSIX
 5260 interface uses malloc() for output vectors. See item 31 above.
 5261 
 5262   PCRE_CONFIG_MATCH_LIMIT
 5263 
 5264 The output is an unsigned integer that contains the default limit of the number
 5265 of match() calls in a pcre_exec() execution. See 42 above.
 5266 
 5267 44. pcretest has been upgraded by the addition of the -C option. This causes it
 5268 to extract all the available output from the new pcre_config() function, and to
 5269 output it. The program then exits immediately.
 5270 
 5271 45. A need has arisen to pass over additional data with calls to pcre_exec() in
 5272 order to support additional features. One way would have been to define
 5273 pcre_exec2() (for example) with extra arguments, but this would not have been
 5274 extensible, and would also have required all calls to the original function to
 5275 be mapped to the new one. Instead, I have chosen to extend the mechanism that
 5276 is used for passing in "extra" data from pcre_study().
 5277 
 5278 The pcre_extra structure is now exposed and defined in pcre.h. It currently
 5279 contains the following fields:
 5280 
 5281   flags         a bitmap indicating which of the following fields are set
 5282   study_data    opaque data from pcre_study()
 5283   match_limit   a way of specifying a limit on match() calls for a specific
 5284                   call to pcre_exec()
 5285   callout_data  data for callouts (see 49 below)
 5286 
 5287 The flag bits are also defined in pcre.h, and are
 5288 
 5289   PCRE_EXTRA_STUDY_DATA
 5290   PCRE_EXTRA_MATCH_LIMIT
 5291   PCRE_EXTRA_CALLOUT_DATA
 5292 
 5293 The pcre_study() function now returns one of these new pcre_extra blocks, with
 5294 the actual study data pointed to by the study_data field, and the
 5295 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
 5296 before. That is, this change is entirely upwards-compatible and requires no
 5297 change to existing code.
 5298 
 5299 If you want to pass in additional data to pcre_exec(), you can either place it
 5300 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
 5301 block.
 5302 
 5303 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
 5304 data string contains the escape sequence \M, pcretest calls pcre_exec() several
 5305 times with different match limits, until it finds the minimum value needed for
 5306 pcre_exec() to complete. The value is then output. This can be instructive; for
 5307 most simple matches the number is quite small, but for pathological cases it
 5308 gets very large very quickly.
 5309 
 5310 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
 5311 returns the size of the data block pointed to by the study_data field in a
 5312 pcre_extra block, that is, the value that was passed as the argument to
 5313 pcre_malloc() when PCRE was getting memory in which to place the information
 5314 created by pcre_study(). The fourth argument should point to a size_t variable.
 5315 pcretest has been extended so that this information is shown after a successful
 5316 pcre_study() call when information about the compiled regex is being displayed.
 5317 
 5318 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
 5319 because what follows is always an absolute path. (Later: it turns out that this
 5320 is more than cosmetic for MinGW, because it doesn't like empty path
 5321 components.)
 5322 
 5323 49. Some changes have been made to the callout feature (see 28 above):
 5324 
 5325 (i)  A callout function now has three choices for what it returns:
 5326 
 5327        0  =>  success, carry on matching
 5328      > 0  =>  failure at this point, but backtrack if possible
 5329      < 0  =>  serious error, return this value from pcre_exec()
 5330 
 5331      Negative values should normally be chosen from the set of PCRE_ERROR_xxx
 5332      values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
 5333      "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
 5334      use by callout functions. It will never be used by PCRE itself.
 5335 
 5336 (ii) The pcre_extra structure (see 45 above) has a void * field called
 5337      callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
 5338      pcre_callout_block structure has a field of the same name. The contents of
 5339      the field passed in the pcre_extra structure are passed to the callout
 5340      function in the corresponding field in the callout block. This makes it
 5341      easier to use the same callout-containing regex from multiple threads. For
 5342      testing, the pcretest program has a new data escape
 5343 
 5344        \C*n        pass the number n (may be negative) as callout_data
 5345 
 5346      If the callout function in pcretest receives a non-zero value as
 5347      callout_data, it returns that value.
 5348 
 5349 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
 5350 there were some redundant $(CFLAGS) in commands that are now specified as
 5351 $(LINK), which already includes $(CFLAGS).
 5352 
 5353 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
 5354 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
 5355 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
 5356 one-byte characters throughout. Note that case-insensitive matching applies
 5357 only to characters whose values are less than 256. PCRE doesn't support the
 5358 notion of cases for higher-valued characters.
 5359 
 5360 (i)   A character class whose characters are all within 0-255 is handled as
 5361       a bit map, and the map is inverted for negative classes. Previously, a
 5362       character > 255 always failed to match such a class; however it should
 5363       match if the class was a negative one (e.g. [^ab]). This has been fixed.
 5364 
 5365 (ii)  A negated character class with a single character < 255 is coded as
 5366       "not this character" (OP_NOT). This wasn't working properly when the test
 5367       character was multibyte, either singly or repeated.
 5368 
 5369 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
 5370       mode, for example: \x{100}{2,3}.
 5371 
 5372 (iv)  The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
 5373       singly or repeated) now correctly test multibyte characters. However,
 5374       PCRE doesn't recognize any characters with values greater than 255 as
 5375       digits, spaces, or word characters. Such characters always match \D, \S,
 5376       and \W, and never match \d, \s, or \w.
 5377 
 5378 (v)   Classes may now contain characters and character ranges with values
 5379       greater than 255. For example: [ab\x{100}-\x{400}].
 5380 
 5381 (vi)  pcregrep now has a --utf-8 option (synonym -u) which makes it call
 5382       PCRE in UTF-8 mode.
 5383 
 5384 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
 5385 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
 5386 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
 5387 value.)
 5388 
 5389 53. The single man page has become too large. I have therefore split it up into
 5390 a number of separate man pages. These also give rise to individual HTML pages;
 5391 these are now put in a separate directory, and there is an index.html page that
 5392 lists them all. Some hyperlinking between the pages has been installed.
 5393 
 5394 54. Added convenience functions for handling named capturing parentheses.
 5395 
 5396 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
 5397 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
 5398 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
 5399 are faulted.
 5400 
 5401 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
 5402 calling configure. These values are used when compiling the dftables.c program
 5403 which is run to generate the source of the default character tables. They
 5404 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
 5405 you will need to set these values.
 5406 
 5407 57. Updated the building process for Windows DLL, as provided by Fred Cox.
 5408 
 5409 
 5410 Version 3.9 02-Jan-02
 5411 ---------------------
 5412 
 5413 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
 5414 
 5415 2. If --disable-static was given, the building process failed when trying to
 5416 build pcretest and pcregrep. (For some reason it was using libtool to compile
 5417 them, which is not right, as they aren't part of the library.)
 5418 
 5419 
 5420 Version 3.8 18-Dec-01
 5421 ---------------------
 5422 
 5423 1. The experimental UTF-8 code was completely screwed up. It was packing the
 5424 bytes in the wrong order. How dumb can you get?
 5425 
 5426 
 5427 Version 3.7 29-Oct-01
 5428 ---------------------
 5429 
 5430 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
 5431 This caused pcretest, when used on the test data, to segfault. Unfortunately,
 5432 this didn't happen under Solaris 8, where I normally test things.
 5433 
 5434 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
 5435 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
 5436 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
 5437 this fix an hour or so after the initial 3.7 release.)
 5438 
 5439 
 5440 Version 3.6 23-Oct-01
 5441 ---------------------
 5442 
 5443 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
 5444 offsets passed as NULL with zero offset count.
 5445 
 5446 2. The config.guess and config.sub files had not been updated when I moved to
 5447 the latest autoconf.
 5448 
 5449 
 5450 Version 3.5 15-Aug-01
 5451 ---------------------
 5452 
 5453 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
 5454 had been forgotten.
 5455 
 5456 2. By using declared but undefined structures, we can avoid using "void"
 5457 definitions in pcre.h while keeping the internal definitions of the structures
 5458 private.
 5459 
 5460 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
 5461 user point of view, this means that both static and shared libraries are built
 5462 by default, but this can be individually controlled. More of the work of
 5463 handling this static/shared cases is now inside libtool instead of PCRE's make
 5464 file.
 5465 
 5466 4. The pcretest utility is now installed along with pcregrep because it is
 5467 useful for users (to test regexs) and by doing this, it automatically gets
 5468 relinked by libtool. The documentation has been turned into a man page, so
 5469 there are now .1, .txt, and .html versions in /doc.
 5470 
 5471 5. Upgrades to pcregrep:
 5472    (i)   Added long-form option names like gnu grep.
 5473    (ii)  Added --help to list all options with an explanatory phrase.
 5474    (iii) Added -r, --recursive to recurse into sub-directories.
 5475    (iv)  Added -f, --file to read patterns from a file.
 5476 
 5477 6. pcre_exec() was referring to its "code" argument before testing that
 5478 argument for NULL (and giving an error if it was NULL).
 5479 
 5480 7. Upgraded Makefile.in to allow for compiling in a different directory from
 5481 the source directory.
 5482 
 5483 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
 5484 options bits, the pointer it was passed was to an int instead of to an unsigned
 5485 long int. This mattered only on 64-bit systems.
 5486 
 5487 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
 5488 generated) instead of pcre.in, which it its source. Also made the same change
 5489 in several of the .c files.
 5490 
 5491 10. A new release of gcc defines printf() as a macro, which broke pcretest
 5492 because it had an ifdef in the middle of a string argument for printf(). Fixed
 5493 by using separate calls to printf().
 5494 
 5495 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
 5496 script, to force use of CR or LF instead of \n in the source. On non-Unix
 5497 systems, the value can be set in config.h.
 5498 
 5499 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
 5500 absolute limit. Changed the text of the error message to make this clear, and
 5501 likewise updated the man page.
 5502 
 5503 13. The limit of 99 on the number of capturing subpatterns has been removed.
 5504 The new limit is 65535, which I hope will not be a "real" limit.
 5505 
 5506 
 5507 Version 3.4 22-Aug-00
 5508 ---------------------
 5509 
 5510 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
 5511 
 5512 2. Diagnose condition (?(0) as an error instead of crashing on matching.
 5513 
 5514 
 5515 Version 3.3 01-Aug-00
 5516 ---------------------
 5517 
 5518 1. If an octal character was given, but the value was greater than \377, it
 5519 was not getting masked to the least significant bits, as documented. This could
 5520 lead to crashes in some systems.
 5521 
 5522 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
 5523 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
 5524 
 5525 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
 5526 These just pass their arguments on to (pcre_free)(), but they are provided
 5527 because some uses of PCRE bind it to non-C systems that can call its functions,
 5528 but cannot call free() or pcre_free() directly.
 5529 
 5530 4. Add "make test" as a synonym for "make check". Corrected some comments in
 5531 the Makefile.
 5532 
 5533 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
 5534 Makefile.
 5535 
 5536 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
 5537 command called pgrep for grepping around the active processes.
 5538 
 5539 7. Added the beginnings of support for UTF-8 character strings.
 5540 
 5541 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
 5542 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
 5543 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
 5544 out for the ar command.)
 5545 
 5546 
 5547 Version 3.2 12-May-00
 5548 ---------------------
 5549 
 5550 This is purely a bug fixing release.
 5551 
 5552 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
 5553 of ZA. This was just one example of several cases that could provoke this bug,
 5554 which was introduced by change 9 of version 2.00. The code for breaking
 5555 infinite loops after an iteration that matches an empty string was't working
 5556 correctly.
 5557 
 5558 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
 5559 when matched against abbab (for example). After matching an empty string, it
 5560 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
 5561 caused it to match further down the string than it should.
 5562 
 5563 3. The code contained an inclusion of sys/types.h. It isn't clear why this
 5564 was there because it doesn't seem to be needed, and it causes trouble on some
 5565 systems, as it is not a Standard C header. It has been removed.
 5566 
 5567 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
 5568 were reported on the Macintosh. The changes were from
 5569 
 5570   while ((c = *(++ptr)) != 0 && c != '\n');
 5571 to
 5572   while ((c = *(++ptr)) != 0 && c != '\n') ;
 5573 
 5574 Totally extraordinary, but if that's what it takes...
 5575 
 5576 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
 5577 available. Added HAVE_BCOPY and an autoconf test for it; if neither
 5578 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
 5579 assumes the way PCRE uses memmove() (always moving upwards).
 5580 
 5581 6. PCRE is being used in one environment where strchr() is not available. There
 5582 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
 5583 faster code anyway.
 5584 
 5585 
 5586 Version 3.1 09-Feb-00
 5587 ---------------------
 5588 
 5589 The only change in this release is the fixing of some bugs in Makefile.in for
 5590 the "install" target:
 5591 
 5592 (1) It was failing to install pcreposix.h.
 5593 
 5594 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
 5595 
 5596 
 5597 Version 3.0 01-Feb-00
 5598 ---------------------
 5599 
 5600 1. Add support for the /+ modifier to perltest (to output $` like it does in
 5601 pcretest).
 5602 
 5603 2. Add support for the /g modifier to perltest.
 5604 
 5605 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
 5606 matches null strings.
 5607 
 5608 4. Fix perltest so that it doesn't do unwanted things when fed an empty
 5609 pattern. Perl treats empty patterns specially - it reuses the most recent
 5610 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
 5611 effect.
 5612 
 5613 5. The POSIX interface was broken in that it was just handing over the POSIX
 5614 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
 5615 required a bigger vector, with some working space on the end. This means that
 5616 the POSIX wrapper now has to get and free some memory, and copy the results.
 5617 
 5618 6. Added some simple autoconf support, placing the test data and the
 5619 documentation in separate directories, re-organizing some of the
 5620 information files, and making it build pcre-config (a GNU standard). Also added
 5621 libtool support for building PCRE as a shared library, which is now the
 5622 default.
 5623 
 5624 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
 5625 09 are not valid octal constants. Single digits will be used for minor values
 5626 less than 10.
 5627 
 5628 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
 5629 existing programs that set these in the POSIX interface can use PCRE without
 5630 modification.
 5631 
 5632 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
 5633 return all that pcre_info() returns, plus additional data. The pcre_info()
 5634 function is retained for compatibility, but is considered to be obsolete.
 5635 
 5636 10. Added experimental recursion feature (?R) to handle one common case that
 5637 Perl 5.6 will be able to do with (?p{...}).
 5638 
 5639 11. Added support for POSIX character classes like [:alpha:], which Perl is
 5640 adopting.
 5641 
 5642 
 5643 Version 2.08 31-Aug-99
 5644 ----------------------
 5645 
 5646 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
 5647 trying to match at the startoffset position, but instead was moving forward to
 5648 the next newline as if a previous match had failed.
 5649 
 5650 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
 5651 and could get into a loop if a null string was matched other than at the start
 5652 of the subject.
 5653 
 5654 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
 5655 be distinguished at compile time, and for completeness also added PCRE_DATE.
 5656 
 5657 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
 5658 in GnuWin32 environments.
 5659 
 5660 
 5661 Version 2.07 29-Jul-99
 5662 ----------------------
 5663 
 5664 1. The documentation is now supplied in plain text form and HTML as well as in
 5665 the form of man page sources.
 5666 
 5667 2. C++ compilers don't like assigning (void *) values to other pointer types.
 5668 In particular this affects malloc(). Although there is no problem in Standard
 5669 C, I've put in casts to keep C++ compilers happy.
 5670 
 5671 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
 5672 should be (const char *).
 5673 
 5674 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
 5675 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
 5676 However, I haven't made this a standard facility. The documentation doesn't
 5677 mention it, and the Makefile doesn't support it.
 5678 
 5679 5. The Makefile now contains an "install" target, with editable destinations at
 5680 the top of the file. The pcretest program is not installed.
 5681 
 5682 6. pgrep -V now gives the PCRE version number and date.
 5683 
 5684 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
 5685 causing the entire string to be ignored, instead of just the last character.
 5686 
 5687 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
 5688 non-matching string, it can take a very, very long time, even for strings of
 5689 quite modest length, because of the nested recursion. PCRE now does better in
 5690 some of these cases. It does this by remembering the last required literal
 5691 character in the pattern, and pre-searching the subject to ensure it is present
 5692 before running the real match. In other words, it applies a heuristic to detect
 5693 some types of certain failure quickly, and in the above example, if presented
 5694 with a string that has no trailing " it gives "no match" very quickly.
 5695 
 5696 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
 5697 other alternatives are tried instead.
 5698 
 5699 
 5700 Version 2.06 09-Jun-99
 5701 ----------------------
 5702 
 5703 1. Change pcretest's output for amount of store used to show just the code
 5704 space, because the remainder (the data block) varies in size between 32-bit and
 5705 64-bit systems.
 5706 
 5707 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
 5708 start matching at. This allows lookbehinds to work when searching for multiple
 5709 occurrences in a string.
 5710 
 5711 3. Added additional options to pcretest for testing multiple occurrences:
 5712 
 5713    /+   outputs the rest of the string that follows a match
 5714    /g   loops for multiple occurrences, using the new startoffset argument
 5715    /G   loops for multiple occurrences by passing an incremented pointer
 5716 
 5717 4. PCRE wasn't doing the "first character" optimization for patterns starting
 5718 with \b or \B, though it was doing it for other lookbehind assertions. That is,
 5719 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
 5720 the letter 'x'. On long subject strings, this gives a significant speed-up.
 5721 
 5722 
 5723 Version 2.05 21-Apr-99
 5724 ----------------------
 5725 
 5726 1. Changed the type of magic_number from int to long int so that it works
 5727 properly on 16-bit systems.
 5728 
 5729 2. Fixed a bug which caused patterns starting with .* not to work correctly
 5730 when the subject string contained newline characters. PCRE was assuming
 5731 anchoring for such patterns in all cases, which is not correct because .* will
 5732 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
 5733 DOTALL is set at top level; otherwise it knows that patterns starting with .*
 5734 must be retried after every newline in the subject.
 5735 
 5736 
 5737 Version 2.04 18-Feb-99
 5738 ----------------------
 5739 
 5740 1. For parenthesized subpatterns with repeats whose minimum was zero, the
 5741 computation of the store needed to hold the pattern was incorrect (too large).
 5742 If such patterns were nested a few deep, this could multiply and become a real
 5743 problem.
 5744 
 5745 2. Added /M option to pcretest to show the memory requirement of a specific
 5746 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
 5747 
 5748 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
 5749 compiled in such a way that the backtracking after subsequent failure was
 5750 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
 5751 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
 5752 
 5753 
 5754 Version 2.03 02-Feb-99
 5755 ----------------------
 5756 
 5757 1. Fixed typo and small mistake in man page.
 5758 
 5759 2. Added 4th condition (GPL supersedes if conflict) and created separate
 5760 LICENCE file containing the conditions.
 5761 
 5762 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
 5763 Perl, that is the internal \ allows the delimiter to be included in the
 5764 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
 5765 the final delimiter, add \ to the end of the pattern (to test the error).
 5766 
 5767 4. Added the convenience functions for extracting substrings after a successful
 5768 match. Updated pcretest to make it able to test these functions.
 5769 
 5770 
 5771 Version 2.02 14-Jan-99
 5772 ----------------------
 5773 
 5774 1. Initialized the working variables associated with each extraction so that
 5775 their saving and restoring doesn't refer to uninitialized store.
 5776 
 5777 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
 5778 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
 5779 fix the problem.
 5780 
 5781 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
 5782 calls, and wasn't printing the correct value for compiling calls. Increased the
 5783 default value of LOOPREPEAT, and the number of significant figures in the
 5784 times.
 5785 
 5786 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
 5787 
 5788 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
 5789 a building problem on Windows NT with a FAT file system.
 5790 
 5791 
 5792 Version 2.01 21-Oct-98
 5793 ----------------------
 5794 
 5795 1. Changed the API for pcre_compile() to allow for the provision of a pointer
 5796 to character tables built by pcre_maketables() in the current locale. If NULL
 5797 is passed, the default tables are used.
 5798 
 5799 
 5800 Version 2.00 24-Sep-98
 5801 ----------------------
 5802 
 5803 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
 5804 it any more.
 5805 
 5806 2. Allow quantification of (?>) groups, and make it work correctly.
 5807 
 5808 3. The first character computation wasn't working for (?>) groups.
 5809 
 5810 4. Correct the implementation of \Z (it is permitted to match on the \n at the
 5811 end of the subject) and add 5.005's \z, which really does match only at the
 5812 very end of the subject.
 5813 
 5814 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
 5815 
 5816 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
 5817 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
 5818 localized options. All options to pcre_study() were also removed.
 5819 
 5820 7. Add other new features from 5.005:
 5821 
 5822    $(?<=           positive lookbehind
 5823    $(?<!           negative lookbehind
 5824    (?imsx-imsx)    added the unsetting capability
 5825                    such a setting is global if at outer level; local otherwise
 5826    (?imsx-imsx:)   non-capturing groups with option setting
 5827    (?(cond)re|re)  conditional pattern matching
 5828 
 5829    A backreference to itself in a repeated group matches the previous
 5830    captured string.
 5831 
 5832 8. General tidying up of studying (both automatic and via "study")
 5833 consequential on the addition of new assertions.
 5834 
 5835 9. As in 5.005, unlimited repeated groups that could match an empty substring
 5836 are no longer faulted at compile time. Instead, the loop is forcibly broken at
 5837 runtime if any iteration does actually match an empty substring.
 5838 
 5839 10. Include the RunTest script in the distribution.
 5840 
 5841 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
 5842 discrepancies, some of which were old and were also with respect to 5.004. They
 5843 have now been fixed.
 5844 
 5845 
 5846 Version 1.09 28-Apr-98
 5847 ----------------------
 5848 
 5849 1. A negated single character class followed by a quantifier with a minimum
 5850 value of one (e.g.  [^x]{1,6}  ) was not compiled correctly. This could lead to
 5851 program crashes, or just wrong answers. This did not apply to negated classes
 5852 containing more than one character, or to minima other than one.
 5853 
 5854 
 5855 Version 1.08 27-Mar-98
 5856 ----------------------
 5857 
 5858 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
 5859 
 5860 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
 5861 latter must appear before anything that relies on it in the pattern.
 5862 
 5863 
 5864 Version 1.07 16-Feb-98
 5865 ----------------------
 5866 
 5867 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
 5868 repeat of a potentially empty string).
 5869 
 5870 
 5871 Version 1.06 23-Jan-98
 5872 ----------------------
 5873 
 5874 1. Added Markus Oberhumer's little patches for C++.
 5875 
 5876 2. Literal strings longer than 255 characters were broken.
 5877 
 5878 
 5879 Version 1.05 23-Dec-97
 5880 ----------------------
 5881 
 5882 1. Negated character classes containing more than one character were failing if
 5883 PCRE_CASELESS was set at run time.
 5884 
 5885 
 5886 Version 1.04 19-Dec-97
 5887 ----------------------
 5888 
 5889 1. Corrected the man page, where some "const" qualifiers had been omitted.
 5890 
 5891 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
 5892 input syntax.
 5893 
 5894 3. Fixed memory leak which occurred when a regex with back references was
 5895 matched with an offsets vector that wasn't big enough. The temporary memory
 5896 that is used in this case wasn't being freed if the match failed.
 5897 
 5898 4. Tidied pcretest to ensure it frees memory that it gets.
 5899 
 5900 5. Temporary memory was being obtained in the case where the passed offsets
 5901 vector was exactly big enough.
 5902 
 5903 6. Corrected definition of offsetof() from change 5 below.
 5904 
 5905 7. I had screwed up change 6 below and broken the rules for the use of
 5906 setjmp(). Now fixed.
 5907 
 5908 
 5909 Version 1.03 18-Dec-97
 5910 ----------------------
 5911 
 5912 1. A erroneous regex with a missing opening parenthesis was correctly
 5913 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
 5914 on some systems.
 5915 
 5916 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
 5917 it was reported that one broken compiler failed on the former because "code" is
 5918 also an independent variable.
 5919 
 5920 3. The erroneous regex a[]b caused an array overrun reference.
 5921 
 5922 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
 5923 fail on data ending with that character. (It was going on too far, and checking
 5924 the next character, typically a binary zero.) This was specific to the
 5925 optimized code for single-character negative classes.
 5926 
 5927 5. Added a contributed patch from the TIN world which does the following:
 5928 
 5929   + Add an undef for memmove, in case the the system defines a macro for it.
 5930 
 5931   + Add a definition of offsetof(), in case there isn't one. (I don't know
 5932     the reason behind this - offsetof() is part of the ANSI standard - but
 5933     it does no harm).
 5934 
 5935   + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
 5936     most of the places where whitespace preceded '#'. I have given up and
 5937     allowed the remaining 2 cases to be at the margin.
 5938 
 5939   + Rename some variables in pcre to eliminate shadowing. This seems very
 5940     pedantic, but does no harm, of course.
 5941 
 5942 6. Moved the call to setjmp() into its own function, to get rid of warnings
 5943 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
 5944 
 5945 7. Constructs such as \d{8,} were compiling into the equivalent of
 5946 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
 5947 outcome, but in this particular case used more store than had been allocated,
 5948 which caused the bug to be discovered because it threw up an internal error.
 5949 
 5950 8. The debugging code in both pcre and pcretest for outputting the compiled
 5951 form of a regex was going wrong in the case of back references followed by
 5952 curly-bracketed repeats.
 5953 
 5954 
 5955 Version 1.02 12-Dec-97
 5956 ----------------------
 5957 
 5958 1. Typos in pcre.3 and comments in the source fixed.
 5959 
 5960 2. Applied a contributed patch to get rid of places where it used to remove
 5961 'const' from variables, and fixed some signed/unsigned and uninitialized
 5962 variable warnings.
 5963 
 5964 3. Added the "runtest" target to Makefile.
 5965 
 5966 4. Set default compiler flag to -O2 rather than just -O.
 5967 
 5968 
 5969 Version 1.01 19-Nov-97
 5970 ----------------------
 5971 
 5972 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
 5973 like /([ab]*)*/, that is, for classes with more than one character in them.
 5974 
 5975 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
 5976 as /((?>a*))*/ (a PCRE_EXTRA facility).
 5977 
 5978 
 5979 Version 1.00 18-Nov-97
 5980 ----------------------
 5981 
 5982 1. Added compile-time macros to support systems such as SunOS4 which don't have
 5983 memmove() or strerror() but have other things that can be used instead.
 5984 
 5985 2. Arranged that "make clean" removes the executables.
 5986 
 5987 
 5988 Version 0.99 27-Oct-97
 5989 ----------------------
 5990 
 5991 1. Fixed bug in code for optimizing classes with only one character. It was
 5992 initializing a 32-byte map regardless, which could cause it to run off the end
 5993 of the memory it had got.
 5994 
 5995 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
 5996 
 5997 
 5998 Version 0.98 22-Oct-97
 5999 ----------------------
 6000 
 6001 1. Fixed bug in code for handling temporary memory usage when there are more
 6002 back references than supplied space in the ovector. This could cause segfaults.
 6003 
 6004 
 6005 Version 0.97 21-Oct-97
 6006 ----------------------
 6007 
 6008 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
 6009 
 6010 2. Optimized negated single characters not to use a bit map.
 6011 
 6012 3. Brought error texts together as macro definitions; clarified some of them;
 6013 fixed one that was wrong - it said "range out of order" when it meant "invalid
 6014 escape sequence".
 6015 
 6016 4. Changed some char * arguments to const char *.
 6017 
 6018 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
 6019 
 6020 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
 6021 pcretest.
 6022 
 6023 
 6024 Version 0.96 16-Oct-97
 6025 ----------------------
 6026 
 6027 1. Added a simple "pgrep" utility to the distribution.
 6028 
 6029 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
 6030 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
 6031 where "ddd" means "one or more decimal digits".
 6032 
 6033 3. Fixed serious bug. If a pattern had a back reference, but the call to
 6034 pcre_exec() didn't supply a large enough ovector to record the related
 6035 identifying subpattern, the match always failed. PCRE now remembers the number
 6036 of the largest back reference, and gets some temporary memory in which to save
 6037 the offsets during matching if necessary, in order to ensure that
 6038 backreferences always work.
 6039 
 6040 4. Increased the compatibility with Perl in a number of ways:
 6041 
 6042   (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
 6043       to request this handling. The option can be set at compile or exec time.
 6044 
 6045   (b) $ matches before a terminating newline by default; an option
 6046       PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
 6047       mode). The option can be set at compile or exec time.
 6048 
 6049   (c) The handling of \ followed by a digit other than 0 is now supposed to be
 6050       the same as Perl's. If the decimal number it represents is less than 10
 6051       or there aren't that many previous left capturing parentheses, an octal
 6052       escape is read. Inside a character class, it's always an octal escape,
 6053       even if it is a single digit.
 6054 
 6055   (d) An escaped but undefined alphabetic character is taken as a literal,
 6056       unless PCRE_EXTRA is set. Currently this just reserves the remaining
 6057       escapes.
 6058 
 6059   (e) {0} is now permitted. (The previous item is removed from the compiled
 6060       pattern).
 6061 
 6062 5. Changed all the names of code files so that the basic parts are no longer
 6063 than 10 characters, and abolished the teeny "globals.c" file.
 6064 
 6065 6. Changed the handling of character classes; they are now done with a 32-byte
 6066 bit map always.
 6067 
 6068 7. Added the -d and /D options to pcretest to make it possible to look at the
 6069 internals of compilation without having to recompile pcre.
 6070 
 6071 
 6072 Version 0.95 23-Sep-97
 6073 ----------------------
 6074 
 6075 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
 6076 \x20 at the start of a run of normal characters. These were being treated as
 6077 real characters, instead of the source characters being re-checked.
 6078 
 6079 
 6080 Version 0.94 18-Sep-97
 6081 ----------------------
 6082 
 6083 1. The functions are now thread-safe, with the caveat that the global variables
 6084 containing pointers to malloc() and free() or alternative functions are the
 6085 same for all threads.
 6086 
 6087 2. Get pcre_study() to generate a bitmap of initial characters for non-
 6088 anchored patterns when this is possible, and use it if passed to pcre_exec().
 6089 
 6090 
 6091 Version 0.93 15-Sep-97
 6092 ----------------------
 6093 
 6094 1. /(b)|(:+)/ was computing an incorrect first character.
 6095 
 6096 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
 6097 but not actually doing anything yet.
 6098 
 6099 3. Treat "-" characters in classes that cannot be part of ranges as literals,
 6100 as Perl does (e.g. [-az] or [az-]).
 6101 
 6102 4. Set the anchored flag if a branch starts with .* or .*? because that tests
 6103 all possible positions.
 6104 
 6105 5. Split up into different modules to avoid including unneeded functions in a
 6106 compiled binary. However, compile and exec are still in one module. The "study"
 6107 function is split off.
 6108 
 6109 6. The character tables are now in a separate module whose source is generated
 6110 by an auxiliary program - but can then be edited by hand if required. There are
 6111 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
 6112 toupper() in the code.
 6113 
 6114 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
 6115 make them global. Abolish the function for setting them, as the caller can now
 6116 set them directly.
 6117 
 6118 
 6119 Version 0.92 11-Sep-97
 6120 ----------------------
 6121 
 6122 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
 6123 (e.g. /a{1,3}/) was broken (I mis-optimized it).
 6124 
 6125 2. Caseless matching was not working in character classes if the characters in
 6126 the pattern were in upper case.
 6127 
 6128 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
 6129 
 6130 4. Make PCRE_ANCHORED public and accept as a compile option.
 6131 
 6132 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
 6133 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
 6134 pass them.
 6135 
 6136 6. Give an error if bad option bits passed at compile or run time.
 6137 
 6138 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
 6139 pcretest to cause it to pass that flag.
 6140 
 6141 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
 6142 options, and the first character, if set.
 6143 
 6144 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
 6145 
 6146 
 6147 Version 0.91 10-Sep-97
 6148 ----------------------
 6149 
 6150 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
 6151 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
 6152 
 6153 2. PCRE was looping on encountering an indefinitely repeated back reference to
 6154 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
 6155 Perl does - treats the match as successful.
 6156 
 6157 ****