LOCK_thread_count locked when we do threads.append().

Objects of the classes Item_func_isnull and Item_func_isnotnull
must have the flag sargable set to TRUE.
Set the value of the flag sargable only in constructors of the 
classes inherited from Item_int_func.

Make IO thread check for end of event group, so that upon
disconnect at the end of an event group it can report the last
read GTID as expected. Also inject a fake Rotate event at
reconnect when skipping part of an initial event group, to
give SQL thread the correct Read_Master_Log_Pos.

Reported by Pavel Ivanov.

Mark a few PCRE CMake variables as advanced,
so the are not presented in cmake-gui by default.

The bug caused a memory overwrite in the function update_ref_and_keys()
It happened due to a wrong value of SELECT_LEX::cond_count. This value
historically was calculated by the fix_fields method. Now the logic of
calling this method became too complicated and, as a result, this value
is calculated not always correctly.
The patch changes the way how and when  the values of SELECT_LEX::cond_count
and of SELECT_LEX::between_count are calculated. The new code does it just at
the beginning of update_ref_and_keys().
 

        Additional patch for the 5.5.

        The Gis_point::init_from_wkt called the String::realloc(),
        and this call is quite slow in the DEBUG mode. Which makes
        loading the huge polygon hang forever.
        Fixed by using the String::realloc(size, inc_size) version instead
        as it's done for other spatial features.

Do not pass PCRE_UCP flag for binary data.
This makes bytes 0x80..FF not to belong to 
generic character classes \d (digit) and \w (word character).

SELECT 0xFF RLIKE '\\w';
 -> 0

Note, this change does not affect non-binary data,
which is still examined with the PCRE_UCP flag by default.

For aggregated fields from views/derived tables the possible adjustment
of thd->lex->in_sum_func->max_arg_level in the function Item_field::fix_fields
must be done before we leave the function.

Fixing compilation failure on Solaris.
The int64_t type was not defined because stdint.h was not included
due to a missing definition in pcre/config-cmake.h.in.

Clean-up: moving initialization of my_string_stack_guard
and pcre_stack_guard into init_libstrings() and init_pcre().

Adding tests with 0x00 characters from

Bug#70470 REGEXP fails to find matches after NUL character

Adding more tests for case sensitivity,
with various collation and (?i) flags combinations.

PCRE unit tests failed on Ubuntu Precise, because "source" 
is not a known command in "dash" (the default shell in Precise).

Changing "source" to ".", which should be understood in all shells.

Apparently in a general case a short-cut for the distinct optimization
is invalid if join buffers are used to join tables after the tables whose
values are to selected.

MDEV-4425 Regexp enhancements
Adding ${CMAKE_BINARY_DIR}/pcre into search path for *.h files.
Needed for find pcre.h (which is generated from pcre.h.in) when
build directory != source directory.

Adding ${CMAKE_BINARY_DIR}/pcre into search path for *.h files.
Needed for find pcre.h (which is generated from pcre.h.in) when
build directory != source directory.

Adding pcre_stack_guard to avoid crashes in pcre_compile()
on a long recursive patterns with parenthesizes:

SELECT a RLIKE '((((...((((x)))...))))';

Removing pcre.h from the tree, it's generated from pcre.h.in 

- Commenting out unused instructions in pcre/CMakeLists.txt
- Don't print PCRE configuration status by default.

This patch also makes libstrings use my_malloc() and my_free()
in embedded server. Previously, embeddes server used malloc()
and free() in libstrings.

It seems to compile/work fine with CMake-2.6.

to avoid system *.h files to be included (e.g. like on labrador).

--replace_regex /.prefix.sql.share.charsets[/\]/MYSQL_CHARSETSDIR/
select @@global.character_sets_dir;

The intention of the '[/\]' part was to replace both slash
'/' and backslash '\\', so it does not depend on the OS.

The pattern '[/\]' was actually wrong, because ']' is escaped
and should be considered as a part of the class, instead of
being a closing bracket for the class. However, due to some bug
in the old REGEX library it worked fine.

After switching to PCRE, mysqltest correctly complains about unbalaced '[]'.
The expected correct pattern should be '[/\\]'.
However, due to some bug in mysqltest, it eats consequetive baskslashes
in a strange way, so there is no a way to have to consequetive 
backslashes after unescaping.

Workaround:

using [[:punct:]] as a pattern that matches both slash and backslash,
which should be fine for this test purposes.

system installed (if any).

libmysqld.a(pcre_exec.c.o): relocation R_X86_64_32S against
`_pcre_ucd_stage1' can not be used when making a shared object;
make[2]: Leaving directory `/mnt/buildbot/build/mariadb-10.0.3'
recompile with -fPIC

Changing ADD_LIBRARY(pcre) to ADD_CONVENIENCE_LIBRARY(pcre)

from pcre/CMakeLists.txt

- do not install anything from pcre library
- do not build the c++ library

"PCRE_STATIC" must be defined before including pcre.h
to avoid linking errors:
- unresolved external symbol __imp_regerror
- unresolved external symbol __imp_pcre_exec

allow user-defined EXECINFO_ROOT prefix, in case library and header are not placed under /usr/local . This change was requested by FreeBSD maintainer.

Other fix of maybe_null problem and revert of revno: 3608 "MDEV-3873 & MDEV-3876 & MDEV-3912 : Wrong result (extra rows) with ALL subquery from a MERGE view."

Bundled jemalloc can only be on  Linux and OSX without problems.
  
On BSDs, build fails because make does not understand GNU extensions (also BSDs do not need  jemalloc, it is already system malloc).
On Solaris, build fails with compile error.