added fix for keeping "records" up to date when execute() fails


mysql-test/r/ndb_insert.result:
  Added testcase for select count() during transaction with failures
mysql-test/t/ndb_insert.test:
  Added testcase for select count() during transaction with failures
sql/ha_ndbcluster.cc:
  added fix for keeping "records" up to date when execute() fails
sql/ha_ndbcluster.h:
  added fix for keeping "records" up to date when execute() fails

into poseidon.(none):/home/tomas/mysql-4.1-ndb-merge

into poseidon.(none):/home/tomas/mysql-4.1-ndb-merge

clearer configure description texts
changed Ndb_local_table_info to use create, destroy metods and hidden constructor/destructor
move definition if Thd_ndb to .h file and changes seize/release to operate on Thd_ndb instead of Ndb objects
moved allocation/deletion of Ndb objects to Thd_ndb


ndb/include/ndbapi/NdbDictionary.hpp:
  new method to set size of local table data
ndb/src/mgmsrv/ConfigInfo.cpp:
  clearer configure description texts
ndb/src/ndbapi/DictCache.cpp:
  changed Ndb_local_table_info to use create, destroy metods and hidden constructor/destructor
ndb/src/ndbapi/DictCache.hpp:
  changed Ndb_local_table_info to use create, destroy metods and hidden constructor/destructor
ndb/src/ndbapi/NdbDictionary.cpp:
  new method to set size of local table data
ndb/src/ndbapi/NdbDictionaryImpl.cpp:
  new method to set size of local table data
ndb/src/ndbapi/NdbDictionaryImpl.hpp:
  new method to set size of local table data
sql/ha_ndbcluster.cc:
  new method to set size of local table data
  moved allocation/deletion of Ndb objects to Thd_ndb
sql/ha_ndbcluster.h:
  move definition if Thd_ndb to .h file and changes seize/release to operate on Thd_ndb instead of Ndb objects

into mysql.com:/M41/mysql-4.1

into poseidon.(none):/home/tomas/mysql-4.1-ndb-merge

into poseidon.(none):/home/tomas/mysql-4.1-ndb-merge


sql/ha_ndbcluster.cc:
  Auto merged

into gw.mysql.r18.ru:/usr/home/ram/work/4.1.b5497

new methods to keep "records" up to date
unset flag HA_NOT_EXACT_COUNT to make handler read "records" field, for count() optim and join optimization
new methods to keep "records" up to datecorrect record field in ndbcluster handler
new method for ndbcluster handler to store/retrieve table and thread specific data
changed local hash to store new table_info object, with placeholders for local data, instead of TableImpl
hanged deleteKey to return ponter to deleted object
moved heavy global cache fetch from inline to separate method


mysql-test/r/ndb_alter_table.result:
  correct record field in ndbcluster handler
mysql-test/r/ndb_blob.result:
  correct record field in ndbcluster handler
ndb/include/ndbapi/NdbDictionary.hpp:
  new method for ndbcluster handler to store/retrieve table and thread specific data
ndb/src/ndbapi/DictCache.cpp:
  changed local hash to store new table_info object, with placeholders for local data, instead of TableImpl
ndb/src/ndbapi/DictCache.hpp:
  changed local hash to store new table_info object, with placeholders for local data, instead of TableImpl
ndb/src/ndbapi/Ndb.cpp:
  replaced method DictionaryImpl::getTable with DictionaryImpl::get_local_table_info
ndb/src/ndbapi/NdbDictionary.cpp:
  new method for ndbcluster handler to store/retrieve table and thread specific data
ndb/src/ndbapi/NdbDictionaryImpl.cpp:
  changed local hash to store new table_info object, with placeholders for local data, instead of TableImpl
  moved heavy global cache fetch from inline to separate method
ndb/src/ndbapi/NdbDictionaryImpl.hpp:
  replaced method DictionaryImpl::getTable with DictionaryImpl::get_local_table_info
ndb/src/ndbapi/NdbLinHash.hpp:
  changed deleteKey to return ponter to deleted object
sql/ha_ndbcluster.cc:
  moved all ndb thread specific data into new placeholder
  new methods to keep "records" up to date
  unset flag HA_NOT_EXACT_COUNT to make handler read "records" field, for count() optim and join optimization
sql/ha_ndbcluster.h:
  new methods to keep "records" up to date
sql/sql_class.h:
  moved all ndb thread specific data into new placeholder

into gw.mysql.r18.ru:/usr/home/ram/work/4.1.b5497


sql/item_strfunc.cc:
  Auto merged

  Bug#5523 Test failure: 'ctype_uca'
  Type cast should have been applied to shift and bit-and operation results, not to min_sort_char/max_sort_char before the operation.


strings/ctype-ucs2.c:
  Bug#5523 Test failure: 'ctype_uca'
  Type cast should have been applied to shift and bit-and operation results, not to min_sort_char/max_sort_char before the operation.

tests/client_test.c:
  Pointer arithmetic on array names is not allowed in all compilers,
  rather use array elements and the address operator (bug#5560).

into mysql.com:/home/jonas/src/mysql-4.1

into deer.(none):/home/hf/work/mysql-4.1.5396

into deer.(none):/home/hf/work/mysql-4.1.5396

into deer.(none):/home/hf/work/mysql-4.1.4815

sql/sql_table.cc:
  Auto merged

into mysql.com:/M41/mysql-4.1

sql/sql_class.cc:
  we need to check exchange->file_name here

into mysql.com:/home/jonas/src/mysql-4.1

missing prototyped problem (bug#5521).


sql/mysqld.cc:
  Some systems (Unixware, SCO) have no prototypes for 'seteuid()' and
  'setegid()', but have them for 'setreuid()' / 'setregid()'.
  So 'sete_id(__)' calls are replaced by 'setre_id(-1,__)' (bug#5521).

into serg.mylan:/usr/home/serg/Abk/mysql-4.1

into mysql.com:/dbdata/psergey/mysql-4.1-bug-tina


sql/handler.h:
  Auto merged

into gw.mysql.r18.ru:/usr/home/ram/work/4.1.b5428

into mysql.com:/home/jonas/src/mysql-4.1

  Bug #5536: wrong server character set in "status" command
  More character set variables.


client/mysql.cc:
  Bug #5536: wrong server character set in "status" command
  More character set variables.

into gw.mysql.r18.ru:/usr/home/ram/work/4.1.b5428


strings/ctype-simple.c:
  Auto merged

into mysql.com:/home/jonas/src/mysql-4.1

Check g_ndb before deleting it
Wait until ready(10) if get mgm connection

 * Renamed handler::estimate_number_of_rows to handler::estimate_rows_upper_bound function, which can also return "unknown"
 * made filesort to use full sort buffer if number of rows to sort is not known.


sql/examples/ha_tina.h:
  Renamed estimate_number_of_rows -> estimate_rows_upper_bound
sql/filesort.cc:
   * Fix for BUG#5117: made filesort() to use full sort buffer if number of rows to sort is not known.
   * Comments and assertion added
sql/ha_berkeley.cc:
  Renamed estimate_number_of_rows -> estimate_rows_upper_bound
sql/ha_berkeley.h:
  Renamed estimate_number_of_rows -> estimate_rows_upper_bound
sql/ha_innodb.cc:
  Renamed estimate_number_of_rows -> estimate_rows_upper_bound
sql/ha_innodb.h:
  Renamed estimate_number_of_rows -> estimate_rows_upper_bound
sql/handler.h:
  Renamed estimate_number_of_rows -> estimate_rows_upper_bound