The previous patch partially fixed things by waiting for the old dump thread
on the master to exit before injecting the DBUG error. This prevents the error
injection going to the wrong thread.

However, there is still the problem that the old dump thread may never exit,
causing the wait to time out. This happens if the dump thread manages to write
all events down the socket before the socket is closed by the slave. The
master dump thread only checks for slave gone when writing a new event, so if
no new events are generated, old dump threads can hang around forever on the
master after the slave disconnects.

Fix by explicitly killing the old dump thread if it is still around.

(Changeset sp1r-igor@olga.mysql.com-20070526173301-38848 moved this loop from one place 
to another, then the merge of sp1r-gshchepa/uchum@gleb.loc-20070527192244-26330  have 
kept both copies).

- move attempt to evaluate join->exec_const_cond() out of the "Extract constant part of each ON expression" loop
  (it got there by mistake when merging).

mysql-test/suite/maria/t/maria3.test:
  Fix when compiling with safemalloc
storage/maria/ha_maria.cc:
  Fix that MAX_FILE_SIZE is not limited on 32 bit machines.

- Add a testcase (the bug has already been fixed)

- Handle the case when the subquery's join is degenerate and so has zero tables.

from an ad hoc set of limitations
to a correct rule

Fixed reference to not initialized memory detected by valgrind

sql/sql_select.cc:
  A bit better fix for tmp-table problem:
  Use only dynamic_record format for group by and distinct.
storage/maria/ma_create.c:
  DYNAMIC_RECORD format doesn't pack VARCHAR fields.
  This change fixes a non-fatal uninitialized memory copy.

The reason for this is that BLOCK_RECORD format is not good when there is a lot of duplicated keys as it first writes the data (to get the row position) and
then writes the key (and thus checks for duplicates).

Fix Field_time_hires::reset()

  microsecond(TIME)
  alter table datetime<->datetime(6)
  max(TIME), mix(TIME) 

mysql-test/t/func_if.test:
  fix the test case of avoid overflow
sql/field.cc:
  don't use make_date() and make_time()
sql/field.h:
  correct eq_def() for temporal fields
sql/item.cc:
  move datetime caching from Item_cache_int
  to Item_cache_temporal
sql/item.h:
  move datetime caching from Item_cache_int
  to Item_cache_temporal
sql/item_func.cc:
  use existing helper methods, don't duplicate
sql/item_sum.cc:
  argument cache must use argument's cmp_type, not result_type.
sql/item_timefunc.cc:
  use existing methods, don't tuplicate.
  remove unused function.
  fix micorseconds() to support TIME argument
sql/mysql_priv.h:
  dead code
sql/time.cc:
  dead code

Fixed problem that global status variables 'bytes_recieved' and 'binlog_bytes_written' where not correctly updated 

Fixed crash when setting query_cache_type to 0.

client/Makefile.am:
  Added zlib include (needed by checksum.c)
sql/set_var.cc:
  Updated call to disable_query_cache()
sql/sql_cache.cc:
  Don't give warning if we start mysqld with --query_cache_type=0 --query_cache-size=0
  Fixed crash when setting query_cache_type to 0 (we shouldn't call query_cache.disable_query_cache() when there is no current_thd)
sql/sql_cache.h:
  Added THD to disable_query_cache()

strings/ctype-simple.c:
  Fixed some compiler warnings

After talking to HeidiSQL people -  libmysql.dll that comes with their distribution in place, dont replace with our own.
It also will result in less HeidiSQL restarts during MariaDB upgrades (since libmysql.dll won't be replaced)

VS did not like to execute 2 commands in custom build step, workaround is 
to use single COMMAND instead of 2.

Upgraded to latest handlersocket code. This fixed LP:766870  "Assertion `next_insert_id == 0' failed with handlersocket"

sql/handler.cc:
  Added DBUG_ code

sql/mysql_priv.h:
  Make opt_help global
sql/mysqld.cc:
  Don't give (double) warnings about lower case file systems if --help is given.
  Don't give warning about non existing data directory if --help is given.
sql/sql_plugin.cc:
  Give a better warning if --help is used and plugin table doesn't exists

The code that added semi-join transformations missed checking
the state of the fixed flag for the items built with the
and_items function before calls of the fix_fields method.
This could lead to an abort failure when the first argument
of and_items() happened to be NULL.
 

compilation error in mysys/my_getsystime.c fixed
some redundant code removed
sec_to_time, time_to_sec, from_unixtime, unix_timestamp, @@timestamp now
  use decimal, not double for numbers with a fractional part.
purge_master_logs_before_date() fixed
many bugs in corner cases fixed

mysys/my_getsystime.c:
  compilation failure fixed
sql/sql_parse.cc:
  don't cut corners. it backfires.

Strict mode now gives error if one tries to update a virtual column.

mysql-test/suite/vcol/r/vcol_column_def_options_innodb.result:
  Updated test results
mysql-test/suite/vcol/r/vcol_column_def_options_myisam.result:
  Updated test results
mysql-test/suite/vcol/r/vcol_keys_innodb.result:
  Updated test results
mysql-test/suite/vcol/r/vcol_keys_myisam.result:
  Updated test results
mysql-test/suite/vcol/r/vcol_misc.result:
  Added new test for 'show columns' and error handling when trying to update a virtual column.
mysql-test/suite/vcol/t/vcol_misc.test:
  Added new test for 'show columns' and error handling when trying to update a virtual column.
sql/sql_base.cc:
  Strict mode now gives error if one tries to update a virtual column.
sql/sql_show.cc:
  Show PERSISTENT instead of VIRTUAL for persistent columns.

Fixed lock sorting and lock check issues with thr_lock that caused warnings when running test suite.
Safety check that could cause core dump when doing create table with virtual column.

mysql-test/mysql-test-run.pl:
  Show also warnings from thr_lock (which starts with just Warning, not Warning:)
mysql-test/r/lock.result:
  Added test that showed not relevant warning when using table locks.
mysql-test/t/lock.test:
  Added test that showed not relevant warning when using table locks.
mysys/thr_lock.c:
  Fixed sorting of locks.
  (Old sort code didn't handle case where TL_WRITE_CONCURRENT_INSERT must be sorted before TL_WRITE)
  Added more information to check_locks warning output.
  Fixed wrong testing of multiple different write locks for same table.
sql/item_cmpfunc.cc:
  Safety check that could cause core dump when doing create table with virtual column.

- make sp.test work both with and without query_cache (attempt 2)
- fix compile warning in make_sort_key(), as directed by SergeiG