removed "of is null" if it is possible
  (this cset should be SCRUM related, but not approved as scrum task yet)

TMP_TABLE_PARAM.
This involved moving things around in include files.
All tests, including the ones with Valgrind passed.

   Added the MAX_LENGTH_FOR_SORT_DATA option
filesort.cc:
  Added the MAX_LENGTH_FOR_SORT_DATA option

Very nasty bug.
It was caused by double free()-ing memory of join->select and 
join->quick. 

I was able to pinpoint it only after using Valgrind.

Plus better fix for bug with TMP_TABLE_PARAM.

Plus new constructor for SELECT_LEX.

Fixed problem with 'wrong packet number' in union

Don't create temporary objects with no table name

  binary comparing alway must be used, because sort_string() is executed for textual data.

INSERT into u SELECT * from t
to the binlog
BEFORE unlocking t (which was read-locked at least in MyISAM),
and not the contrary.

  Some default_charset_info were removed
sql_select.cc:
  Some default_charset_info were removed.

Added ROW_RESULT to switchs(for documentation purposes)

Optimize depending sub querys
Remove valgrind warnings

Added ALL as parameter option for all group functions.
Make join handling uniform. This allows us to use ',', JOIN and INNER JOIN the same way.
Sort NULL last if DESC is used (ANSI SQL 99 requirement)

MIN(key_column) could in some cases return NULL on a column with NULL and other values.
MIN(key_column) and MAX(key_column) could in some cases return wrong values when used in OUTER JOIN.

new fatal_error interface to assign is_fatal_error and ne.report_error
commant about Item_row

fixed length of fields created in fubselect
fixed resolving fields of reduced in derived table subselect

merging with switching on static tables optimization (SCRUM)
fixed subselects with uncacheable results
added test for fixed bugs from bugreports

Call pthread_mutex_destroy() on not used mutex.
Changed comments in .h and .c files from // -> /* */
Added detection of mutex on which one didn't call pthread_mutex_destroy()
Fixed bug in create_tmp_field() which causes a memory overrun in queries that uses "ORDER BY constant_expression"
Added optimisation for ORDER BY NULL

after merge fix of 577 task (SCRUM, pre commit to be able to merge with static tables optimization fix)
fixed derived tables with subselect inside

removed switching off static table optimization for subselects (SCRUM , pre-commit  to be able merge with temporary table fix)

fixed memory leacks