The variables' ON_CHECK functions relied on set_var's "value"
member, which is NULL for SET ... =default. Fixed by using
save_result instead.

Also, for many wsrep variables, pointers to their respective
global variables were used to provide default values. The patch
fixes this by using appropriate macros and string literals.

packages

* Added "galera" as default option group.
* Updated deb/rpm server config files to include mandatory
  and optional settings (as comments) required to configure
  Galera cluster.

Introduced safety checks to ensure mysqld_safe terminates
if mktemp fails to create a valid file.

Galera tests used default base/SST ports which led to
failures due to port conflicts when run in parallel.
Fixed by setting them to ones generated by mtr framework.

system tables

With wsrep patch, binary logging is implicitly enabled. This
fix makes sure that it is turned off for --skip-write-binlog
by swithcing off wsrep_on.

causes crash

Fixed by properly handling the NULL values.

Analysis: Based on crashed the buffer pool instance identifier is
not correct on block to be freed. Add LRU list mutex holding
on functions calling free and add additional safety checks.

threads but never removes them

When wsrep_slave_threads is changed at runtime, the old "change"
value is not taken into account and thus successive SETs can
make wsrep_slave_threads value inconsistent with the actual
number of applier threads present.

* bzr merge -r3980..3984 codership-mysql/5.5
* bzr merge -r3985..3997 codership-mysql/5.5
* Fixed a segfault.
* Modified wsrep.variables test & fixed a warning
  in mtr script.

Updated wsrep.variables test.

if galera library is not specified or found.

- When range optimizer cannot the lookup value into [VAR]CHAR(n) column,
  it should produce:
  = "Impossible range" for equality
  = "no range" for non-equalities.

"have_warnings" was set to an uninialized value when converting
a negative number to datetime.

MDEV-6099 Bad results for DATE_ADD(.., INTERVAL 2000000000000000000.0 SECOND)
MDEV-6097 Inconsistent results for CAST(int,decimal,double AS DATETIME)
MDEV-6100 No warning on CAST(9000000 AS TIME)

We have to run the derived table prepare before the unique table check to mark the derived table (in this case the unique table check can turn that table to materialized one).

use -ggdb3 if supported

  Bug #18167356: EXPLAIN W/ EXISTS(SELECT* UNION SELECT*)
                 WHERE ONE OF SELECT* IS DISTINCT FAILS.

the bugfix itself was not merged - MariaDB doesn't have this bug.

mysql_derived_merge() made correctly working with views.

Added the EXCEPTIONS-CLIENT (that was lost in the merge),
as the ascii version of http://www.mysql.com/about/legal/licensing/foss-exception/