Problem was that the mix of handlers was not consistent between
CREATE and ALTER

changed so that it works like:
    - All partitions must use the same engine
      AND it must be the same as the table.
    - if one does NOT specify an engine on the table level
      then one must either NOT specify any engine on any
      partition/subpartition OR for ALL partitions/subpartitions

Note: that after a table have been created, the storage engine
is specified for all parts of the table (table/partition/subpartition)
and so when using alter, one does not need to specify it (unless one
wants to change the storage engine, then one have to specify it on the
table level)


mysql-test/r/partition.result:
  Bug#31931 Partitions: unjustified 'mix of handlers' error message
  test result updated
mysql-test/r/partition_innodb.result:
  Bug#31931 Partitions: unjustified 'mix of handlers' error message
  test result updated
mysql-test/suite/ndb/r/ndb_partition_key.result:
  Bug#31931 Partitions: unjustified 'mix of handlers' error message
  test result updated
mysql-test/suite/ndb/t/ndb_partition_key.test:
  Bug#31931 Partitions: unjustified 'mix of handlers' error message
  test case update
mysql-test/suite/parts/inc/partition_engine.inc:
  Bug#31931 Partitions: unjustified 'mix of handlers' error message
  test case updated
mysql-test/suite/parts/r/ndb_partition_key.result:
  Bug#31931 Partitions: unjustified 'mix of handlers' error message
  test result updated
mysql-test/suite/parts/r/partition_engine_innodb.result:
  Bug#31931 Partitions: unjustified 'mix of handlers' error message
  test result updated
mysql-test/suite/parts/r/partition_engine_myisam.result:
  Bug#31931 Partitions: unjustified 'mix of handlers' error message
  test result updated
mysql-test/suite/parts/t/ndb_partition_key.test:
  Bug#31931 Partitions: unjustified 'mix of handlers' error message
  test case updated
mysql-test/t/partition.test:
  Bug#31931 Partitions: unjustified 'mix of handlers' error message
  test case updated
mysql-test/t/partition_innodb.test:
  Bug#31931 Partitions: unjustified 'mix of handlers' error message
  test case updated
sql/partition_info.cc:
  Bug#31931 Partitions: unjustified 'mix of handlers' error message
  moved the check_engine_condition here from sql_partition.cc
  created a new check_engine_mix from check_native_partitioned in
  sql_partition.cc
sql/partition_info.h:
  Bug#31931 Partitions: unjustified 'mix of handlers' error message
  non static function check_engine_mix (now used in sql_partition.cc)
sql/sql_partition.cc:
  Bug#31931 Partitions: unjustified 'mix of handlers' error message
  moved check_engine_condition to partition_info.cc and moved out some
  common code in check_native_partitioned to check_engine_mix in
  partition_info.cc

Problem: when alter to partitioned table,
it does not see it as change of engine.

Solution: If alter includes partitioning, check if it is possible
to change engines (eg. is the table referenced by a FK)


mysql-test/r/partition_innodb.result:
  Bug#32948: FKs allowed to reference partitioned table
  test result
mysql-test/t/partition_innodb.test:
  Bug#32948: FKs allowed to reference partitioned table
  test case
sql/sql_table.cc:
  Bug#32948: FKs allowed to reference partitioned table
  
  if alter to partitioning, it is the same as changing engine.

mysql-test/r/partition_innodb.result:
  result fixed
mysql-test/t/partition_innodb.test:
  number of subpartitions fixed

mysql-test/r/partition.result:
  test result fixed
mysql-test/r/partition_innodb.result:
  test result fixed
mysql-test/t/partition.test:
  test moved to partition_innodb
mysql-test/t/partition_innodb.test:
  test moved from partition.test

Issuing SELECT COUNT(*) against partitioned InnoDB table may cause
server crash.

Fixed that not all required fields were included into read_set.


mysql-test/r/partition_innodb.result:
  A test case for BUG#30583.
mysql-test/t/partition_innodb.test:
  A test case for BUG#30583.
sql/ha_partition.cc:
  Ensure that all fields of current key are included into read_set,
  as partitioning requires them for sorting
  (see ha_partition::handle_ordered_index_scan).

fixed an assert crash


include/mysql/plugin.h:
  more wl#2936 fixes: no implicit ha_thd() calls
mysql-test/mysql-test-run.pl:
  don't load system-wide plugins
mysql-test/r/partition_innodb.result:
  fix the test
mysql-test/t/partition_innodb.test:
  fix the test
sql/handler.cc:
  more wl#2936 fixes: no implicit ha_thd() calls
sql/handler.h:
  more wl#2936 fixes: no implicit ha_thd() calls
sql/sql_class.cc:
  more wl#2936 fixes: no implicit ha_thd() calls
sql/sql_plugin.cc:
  more wl#2936 fixes: assert crash
storage/innobase/handler/ha_innodb.cc:
  more wl#2936 fixes: no implicit ha_thd() calls

  "Server Variables for Plugins"
  Implement support for plugins to declare server variables.
  Demonstrate functionality by removing InnoDB specific code from sql/*
  New feature for HASH - HASH_UNIQUE flag
  New feature for DYNAMIC_ARRAY - initializer accepts preallocated ptr.
  Completed support for plugin reference counting.


include/hash.h:
  New flag for HASH
    HASH_UNIQUE
include/my_getopt.h:
  New data types for options: ENUM and SET.
  Use typelib to enumerate possible values.
  New flag variable:
    my_getopt_skip_unknown
include/my_sys.h:
  change to DYNAMIC_ARRAY init functions to support pre-allocated buffers
include/mysql.h:
  relocate inclusion of typelib due to longlong requirement
include/mysql/plugin.h:
  wl2936
    New declarations for plugin server variable support.
    New functions for use by plugins
include/mysys_err.h:
  new my_getopt return value: EXIT_ARGUMENT_INVALID
include/typelib.h:
  new typelib function: find_typeset(), returns an int which is a SET of
  the elements in the typelib
mysql-test/r/im_utils.result:
  change to more specific command line settings
  --skip-innodb => --skip-plugin-innodb
  etc.
mysql-test/r/log_tables.result:
  set default storage engine to MEMORY so that test will succeed even
  when some of the other named storage engines are not present
mysql-test/r/ndb_dd_basic.result:
  change in error message
mysql-test/r/partition_innodb.result:
  change in results
mysql-test/r/ps_1general.result:
  bdb doesn't exist, use myisam for a non-transactional engine
mysql-test/r/variables.result:
  information schema doesn't sort row results for server variables.
mysql-test/t/log_tables.test:
  set default storage engine to MEMORY so that test will succeed even
  when some of the other named storage engines are not present
mysql-test/t/ndb_dd_basic.test:
  ALTER LOGFILE GROUP no longer silently fail here
mysql-test/t/partition_innodb.test:
  ALTER TABLE no longer silently fails for unknown storage engine
mysql-test/t/ps_1general.test:
  remove unneccessary parts
  use myisam as it is an always present non-transactional engine
mysql-test/t/variables.test:
  information schema doesn't sort row results for server variables.
mysql-test/t/warnings_engine_disabled-master.opt:
  use the new style command line option
mysys/array.c:
  change to DYNAMIC_ARRAY init functions to support pre-allocated buffers
mysys/hash.c:
  New flag for HASH
    HASH_UNIQUE
  Implement HASH_UNIQUE functionality by performing a hash_search
mysys/my_getopt.c:
  New data types for options: ENUM and SET.
  Use typelib to enumerate possible values.
  New flag variable:
    my_getopt_skip_unknown
mysys/typelib.c:
  new typelib function: find_typeset(), returns an int which is a SET of
  the elements in the typelib
sql/ha_ndbcluster.cc:
  use ha_statistic_increment() method instead of
  statistic_increment() function
  ha_ndbcluster variable has gone away.
sql/ha_partition.cc:
  fix for reference counting
sql/ha_partition.h:
  fix for reference counting
sql/handler.cc:
  fixes for reference counting
sql/handler.h:
  fixes for reference counting
  some new methods to aid storage engine writers
sql/item_func.cc:
  find_sys_var() function now requires thd
sql/item_sum.cc:
  fixes for ref counting
sql/mysql_priv.h:
  remove unneccessary globals.
  new lock: LOCK_system_variables_hash
sql/mysqld.cc:
  Remove InnoBase specific code.
  Support plugin command line processing.
sql/set_var.cc:
  Remove InnoBase specific declarations
  Remove redundant declarations
  changes to permit new variables at run time
  changes for ref counting
sql/set_var.h:
  changes to permit new variables at run time
  changes for ref counting
sql/sql_base.cc:
  changes for ref counting
sql/sql_cache.cc:
  mark code as needing work in the future
sql/sql_class.cc:
  new functions to aid plugin authors
  initialize variables for dynamic plugin variables
sql/sql_class.h:
  remove InnoBase specific declarations
  New declarations for plugin variables.
sql/sql_connect.cc:
  initialization and cleanup of plugin variables
sql/sql_delete.cc:
  change for ref counting
sql/sql_insert.cc:
  change for ref counting
sql/sql_lex.cc:
  changes for ref counting and plugin variables
sql/sql_lex.h:
  add properties for plugin ref counting,
  add to distructor to clean up
sql/sql_partition.cc:
  changes for ref counting
sql/sql_plugin.cc:
  WL2936
    Plugin Variables
    New methods and code to support server variables for plugins.
    New code to complete plugin reference counting
    Debug code adds further indirection so that malloc debugging can be
    used to aid finding plugin ref count leaks
sql/sql_plugin.h:
  WL2936
    Plugin Variables
    New methods and code to support server variables for plugins.
    New code to complete plugin reference counting
    Debug code adds further indirection so that malloc debugging can be
    used to aid finding plugin ref count leaks
sql/sql_repl.cc:
  replication system variables moved here from set_var.cc
sql/sql_repl.h:
  new function to initialise replication server variables
sql/sql_select.cc:
  changes for ref counting
sql/sql_show.cc:
  changes for ref counting
sql/sql_table.cc:
  changes for ref counting
sql/sql_tablespace.cc:
  use supplied functions instead of digging into data structures manually
sql/sql_yacc.yy:
  changes for ref counting
  find_sys_var() now requires thd parameter
  changes on reporting errors to keep user-visible behaviour the same.
sql/structs.h:
  changes for ref counting
sql/table.cc:
  changes for ref counting
sql/table.h:
  changes for ref counting
storage/federated/ha_federated.cc:
  use ha_statistic_increment() method instead of statistic_increment()
  function
storage/heap/ha_heap.cc:
  use ha_statistic_increment() method instead of statistic_increment()
  function
storage/innobase/handler/ha_innodb.cc:
  use ha_statistic_increment() method instead of statistic_increment()
  function
  WL2936
    Move InnoBase specific code out of mysqld.cc and into here
    Declare all required server variables for InnoBase
storage/innobase/include/trx0trx.h:
  store a bit more state so that InnoBase does not have to dig into
  mysqld internal data structures.
storage/myisam/ha_myisam.cc:
  use ha_statistic_increment() method instead of statistic_increment()
  function
storage/myisammrg/ha_myisammrg.cc:
  use ha_statistic_increment() method instead of statistic_increment()
  function

Before the fix: 
  ha_partition objects had ha_partition::m_part_info==NULL and that caused
  crash
After: 
- The new ha_partition::clone() function makes the clones use parent's
  m_part_info value.
- The parent ha_partition object remains responsible for deallocation of
  m_part_info.


mysql-test/r/partition_innodb.result:
  BUG#26117 "index_merge sort-union over partitioned table crashes"
   - Testcase
mysql-test/t/partition_innodb.test:
  BUG#26117 "index_merge sort-union over partitioned table crashes"
   - Testcase

mysql-test/r/partition_innodb.result:
  New test case
mysql-test/t/partition_innodb.test:
  New test case
sql/ha_partition.cc:
  Two more handler interface variables not taken care of

The problem appeared because the same values produced different hash
during INSERT and SELECT for VARCHAR data type.
Fix:
VARCHAR required special treatment to avoid hashing of length bytes
(leftmost one or two bytes) as well as trailing bytes beyond real length,
which could contain garbage. Fix is done by introducing hash() - new method
in the Field class.


mysql-test/r/partition_innodb.result:
  Adding test case
mysql-test/r/partition_pruning.result:
  Fixing test results (results differ due to changes in hash function)
mysql-test/t/partition_innodb.test:
  Adding test case
sql/field.cc:
  Adding generic hash() method, and a special
  method for VARCHAR.
sql/field.h:
  Adding prototypes for new methods
sql/key.cc:
  Mark columns for write before executinf of set_key_image().
  Thanks for Mikael for suggesting this fix.
sql/sql_partition.cc:
  Removing old hash code.
  Using new methid field->hash() instead.

Added new test case


mysql-test/r/partition_innodb.result:
  Added new test case
mysql-test/t/partition_innodb.test:
  Added new test case

Added new test cases

mysql-test/r/partition_innodb.result:
  New BitKeeper file ``mysql-test/r/partition_innodb.result''
mysql-test/t/partition_innodb.test:
  New BitKeeper file ``mysql-test/t/partition_innodb.test''