Added a test case for bug #10031.
opt_range.cc:
Fixed bug #10031: range condition was not used with
views. Range analyzer did not take into account that
view columns were always referred through Item_ref.
The source of the problem is in Field_longlong::cmp. If 'this' is
an unsigned number, the method casts both the current value, and
the constant that we compare with to an unsigned number. As a
result if the constant we compare with is a negative number, it
wraps to some unsigned number, and the comparison is incorrect.
When the optimizer chooses the "range" access method, this problem
causes handler::read_range_next to reject the current key when the
upper bound key is a negative number because handler::compare_key
incorrectly considers the positive and negative keys to be equal.
The current patch does not correct the source of the problem in
Field_longlong::cmp because it is not easy to propagate sign
information about the constant at query execution time. Instead
the patch changes the range optimizer so that it never compares
unsiged fields with negative constants. As an added benefit,
queries that do such comparisons will execute faster because
the range optimizer replaces conditions like:
(a) (unsigned_int [< | <=] negative_constant) == FALSE
(b) (unsigned_int [> | >=] negative_constant) == TRUE
with the corresponding constants.
In some cases this may even result in constant time execution.
mysql-test/r/range.result:
- Added test for BUG#11185
- Added missing test from 4.1. This test also tests the fix for BUG#11185.
mysql-test/t/range.test:
- Added test for BUG#11185
- Added missing test from 4.1. This test also tests the fix for BUG#11185.
sql/opt_range.cc:
Added a new optimization to the range optimizer where we detect that
an UNSIGNED field is compared with a negative constant. Depending on
the comparison operator, we know directly that the result of the
comparison is either TRUE or FALSE for all input values, and we need
not check each value.
This optimization is also necessary so that the index range access
method produces correct results when comparing unsigned fields with
negative constants.
The source of the problem is in Field_longlong::cmp. If 'this' is
an unsigned number, the method casts both the current value, and
the constant that we compare with to an unsigned number. As a
result if the constant we compare with is a negative number, it
wraps to some unsigned number, and the comparison is incorrect.
When the optimizer chooses the "range" access method, this problem
causes handler::read_range_next to reject the current key when the
upper bound key is a negative number because handler::compare_key
incorrectly considers the positive and negative keys to be equal.
The current patch does not correct the source of the problem in
Field_longlong::cmp because it is not easy to propagate sign
information about the constant at query execution time. Instead
the patch changes the range optimizer so that it never compares
unsiged fields with negative constants. As an added benefit,
queries that do such comparisons will execute faster because
the range optimizer replaces conditions like:
(a) (unsigned_int [< | <=] negative_constant) == FALSE
(b) (unsigned_int [> | >=] negative_constant) == TRUE
with the corresponding constants.
In some cases this may even result in constant time execution.
The source of the problem is in Field_longlong::cmp. If 'this' is
an unsigned number, the method casts both the current value, and
the constant that we compare with to an unsigned number. As a
result if the constant we compare with is a negative number, it
wraps to some unsigned number, and the comparison is incorrect.
When the optimizer chooses the "range" access method, this problem
causes handler::read_range_next to reject the current key when the
upper bound key is a negative number because handler::compare_key
incorrectly considers the positive and negative keys to be equal.
The current patch does not correct the source of the problem in
Field_longlong::cmp because it is not easy to propagate sign
information about the constant at query execution time. Instead
the patch changes the range optimizer so that it never compares
unsiged fields with negative constants. As an added benefit,
queries that do such comparisons will execute faster because
the range optimizer replaces conditions like:
(a) (unsigned_int [< | <=] negative_constant) == FALSE
(b) (unsigned_int [> | >=] negative_constant) == TRUE
with the corresponding constants.
In some cases this may even result in constant time execution.
mysql-test/r/range.result:
- Changed incorrect result of an old test
- Added new results for BUG#11185
mysql-test/t/range.test:
- Added new tests for BUG#11185
- Deleted an old comment because now the problem is fixed
sql/opt_range.cc:
Added a new optimization to the range optimizer where we detect that
an UNSIGNED field is compared with a negative constant. Depending on
the comparison operator, we know directly that the result of the
comparison is either TRUE or FALSE for all input values, and we need
not check each value.
This optimization is also necessary so that the index range access
method produces correct results when comparing unsigned fields with
negative constants.
The source of the problem is in Field_longlong::cmp. If 'this' is
an unsigned number, the method casts both the current value, and
the constant that we compare with to an unsigned number. As a
result if the constant we compare with is a negative number, it
wraps to some unsigned number, and the comparison is incorrect.
When the optimizer chooses the "range" access method, this problem
causes handler::read_range_next to reject the current key when the
upper bound key is a negative number because handler::compare_key
incorrectly considers the positive and negative keys to be equal.
The current patch does not correct the source of the problem in
Field_longlong::cmp because it is not easy to propagate sign
information about the constant at query execution time. Instead
the patch changes the range optimizer so that it never compares
unsiged fields with negative constants. As an added benefit,
queries that do such comparisons will execute faster because
the range optimizer replaces conditions like:
(a) (unsigned_int [< | <=] negative_constant) == FALSE
(b) (unsigned_int [> | >=] negative_constant) == TRUE
with the corresponding constants.
In some cases this may even result in constant time execution.
Added test cases for optimization request #10561.
opt_range.cc, sql_select.cc:
Fixed bug #10561: an optimization request to allow
range analysis for NOT IN and NOT BETWEEN.
sql/sql_select.cc:
Fixed bug #10561: an optimization request to allow
range analysis for NOT IN and NOT BETWEEN.
sql/opt_range.cc:
Fixed bug #10561: an optimization request to allow
range analysis for NOT IN and NOT BETWEEN.
mysql-test/t/range.test:
Added test cases for optimization request #10561.
mysql-test/r/range.result:
Added test cases for optimization request #10561.
Added test cases for optimization request #10561.
opt_range.cc, sql_select.cc:
Fixed bug #10561: an optimization request to allow
range analysis for NOT IN and NOT BETWEEN.
mysql-test/t/range.test:
Auto merged
scripts/mysql_install_db.sh:
Auto merged
sql/item_func.h:
Auto merged
sql/item_sum.cc:
Auto merged
sql/item_sum.h:
Auto merged
sql/sql_udf.h:
Auto merged
mysql-test/r/mysqldump.result:
Manual merge.
mysql-test/t/mysqldump.test:
Manual merge.
mysql-test/t/range.test:
Added more rows to test to get predictable results
(if it still fails, we need to add even more rows)
scripts/make_binary_distribution.sh:
Apply Jims patch for mysql.server
support-files/mysql.server.sh:
Apply Jims patch to support mysqld_safe
- If number of records in table is 4, the calculated cost for using "index" and "range" become so close so that any rounding errors becomes visible.
- Added one more record to the tables for heap test and expoect "range" to be selected
- Decrease number of records in t1 for range and expect "index" to be choosen.
mysql-test/r/heap.result:
Add one more record to table t1 and expect explain to use "range"
mysql-test/r/heap_btree.result:
Add one more record to table t1 and expect explain to use "range"
mysql-test/r/heap_hash.result:
Add one more record to table t1 and expect explain to use "range"
mysql-test/r/range.result:
Update results
mysql-test/t/heap.test:
Add one more record to table t1 and expect explain to use "range"
mysql-test/t/heap_btree.test:
Add one more record to table t1 and expect explain to use "range"
mysql-test/t/heap_hash.test:
Add one more record to table t1 in order for optimizer to select use of "range" deterministic
mysql-test/t/range.test:
Remove one record from table t1 to avoid that cost for "index" and "range" are so close that rounding error become visible.
- If number of records in table is 4, the calculated cost for using "index" and "range" become so close so that any rounding errors becomes visible.
- Added one more record to the tables for heap test and expoect "range" to be selected
- Decrease number of records in t1 for range and expect "index" to be choosen.
Added push_back(void *, MEM_ROOT *) to make some list-handling code easier that needs to be allocated in a different mem-root
(Before one had to change thd->mem_root ; push_back(); restore mem_root.
mysql-test/t/range.test:
After merge fixRecR
sql/item_func.cc:
After merge fix
sql/item_sum.cc:
After merge fix
sql/opt_range.cc:
After merge fix
sql/parse_file.cc:
After merge fix
sql/sp.cc:
After merge fix
sql/sp_head.cc:
After merge fix
sql/sp_head.h:
After merge fix
sql/sql_base.cc:
After merge fix
sql/sql_class.h:
After merge fix
sql/sql_list.h:
Added push_back(void *, MEM_ROOT *) to make some list-handling code easier that needs to be allocated in a different mem-root
(Before one had to change thd->mem_root ; push_back(); restore mem_root.
sql/sql_prepare.cc:
After merge fix
sql/sql_select.cc:
After merge fix
sql/sql_table.cc:
After merge fix
Fixed problem with OPTIMIZE on INNODB tables
sql/sql_trigger.cc:
After merge fix
sql/sql_union.cc:
After merge fix
sql/sql_view.cc:
After merge fix
sql/sql_yacc.yy:
After merge fix
sql/table.cc:
After merge fix
Added push_back(void *, MEM_ROOT *) to make some list-handling code easier that needs to be allocated in a different mem-root
(Before one had to change thd->mem_root ; push_back(); restore mem_root.
BitKeeper/etc/logging_ok:
auto-union
Docs/Support/texi2html:
Auto merged
include/mysql.h:
Auto merged
include/mysql_com.h:
Auto merged
libmysql/libmysql.c:
Auto merged
myisam/mi_check.c:
Auto merged
myisam/myisamchk.c:
Auto merged
mysql-test/include/ps_modify.inc:
Auto merged
mysql-test/mysql-test-run.sh:
Auto merged
mysql-test/r/ctype_recoding.result:
Auto merged
mysql-test/r/fulltext.result:
Auto merged
mysql-test/r/gis.result:
Auto merged
mysql-test/r/ndb_blob.result:
Auto merged
mysql-test/r/ps_2myisam.result:
Auto merged
mysql-test/r/ps_3innodb.result:
Auto merged
mysql-test/r/ps_4heap.result:
Auto merged
mysql-test/r/ps_5merge.result:
Auto merged
mysql-test/r/ps_6bdb.result:
Auto merged
mysql-test/r/sql_mode.result:
Auto merged
mysql-test/t/join_outer.test:
Auto merged
mysql-test/t/key.test:
Auto merged
mysql-test/t/range.test:
Auto merged
mysql-test/t/show_check.test:
Auto merged
mysql-test/t/sql_mode.test:
Auto merged
sql/field.cc:
Auto merged
sql/ha_ndbcluster.cc:
Auto merged
sql/ha_ndbcluster.h:
Auto merged
sql/item.h:
Auto merged
sql/item_create.cc:
Auto merged
sql/item_func.cc:
Auto merged
sql/item_func.h:
Auto merged
sql/item_strfunc.cc:
Auto merged
sql/item_timefunc.cc:
Auto merged
sql/mysqld.cc:
Auto merged
sql/net_serv.cc:
Auto merged
sql/protocol.cc:
Auto merged
sql/protocol.h:
Auto merged
sql/sql_class.cc:
Auto merged
sql/sql_handler.cc:
Auto merged
sql/sql_parse.cc:
Auto merged
sql/sql_show.cc:
Auto merged
sql/sql_string.cc:
Auto merged
sql/sql_string.h:
Auto merged
sql/sql_table.cc:
Auto merged
client/mysqltest.c:
Merge with 4.1
mysql-test/r/range.result:
Merge with 4.1
Added missing drop table test
sql/ha_innodb.cc:
Merge with 4.1
sql/item.cc:
Merge with 4.1
sql/item_cmpfunc.cc:
Merge with 4.1
sql/opt_range.cc:
Merge with 4.1
sql/sql_prepare.cc:
Merge with 4.1
tests/client_test.c:
Merge with 4.1
Added code to support --silent
configure.in:
Merge with 4.1
ndb/src/common/util/version.c:
Merge with 4.1
Bug #6045: Binary Comparison regression in MySQL 4.1
Binary searches didn't use a case insensitive index, now they do.
mysql-test/t/range.test:
Bug #6045: Binary Comparison regression in MySQL 4.1
Binary searches didn't use a case insensitive index, now they do.
mysql-test/r/range.result:
Bug #6045: Binary Comparison regression in MySQL 4.1
Binary searches didn't use a case insensitive index, now they do.
sql/opt_range.cc:
Bug #6045: Binary Comparison regression in MySQL 4.1
Binary searches didn't use a case insensitive index, now they do.
BitKeeper/etc/ignore:
auto-union
BitKeeper/etc/logging_ok:
auto-union
BitKeeper/triggers/post-commit:
Auto merged
Docs/Support/texi2html:
Auto merged
Makefile.am:
Auto merged
client/Makefile.am:
Auto merged
client/mysql.cc:
Auto merged
client/mysqldump.c:
Auto merged
include/my_base.h:
Auto merged
include/my_global.h:
Auto merged
include/my_pthread.h:
Auto merged
include/my_sys.h:
Auto merged
include/my_time.h:
Auto merged
include/mysql.h:
Auto merged
include/mysql_com.h:
Auto merged
innobase/buf/buf0buf.c:
Auto merged
innobase/include/row0mysql.h:
Auto merged
innobase/row/row0sel.c:
Auto merged
libmysql/libmysql.c:
Auto merged
libmysqld/examples/Makefile.am:
Auto merged
myisam/mi_check.c:
Auto merged
mysql-test/include/ps_modify.inc:
Auto merged
mysql-test/install_test_db.sh:
Auto merged
mysql-test/r/alter_table.result:
Auto merged
mysql-test/r/auto_increment.result:
Auto merged
mysql-test/r/bdb.result:
Auto merged
mysql-test/r/ctype_latin1_de.result:
Auto merged
mysql-test/r/ctype_recoding.result:
Auto merged
mysql-test/r/fulltext.result:
Auto merged
mysql-test/r/func_gconcat.result:
Auto merged
mysql-test/r/func_group.result:
Auto merged
mysql-test/r/func_if.result:
Auto merged
mysql-test/t/derived.test:
Auto merged
mysql-test/t/insert.test:
merge with 4.1
Fixed test case to not use 'if exists' when it shouldn't
mysql-test/t/range.test:
merge with 4.1
Added missing drop table
sql/ha_ndbcluster.cc:
merge with 4.1
Simple optimization: use max() instead of ? :
sql/item_func.cc:
merge with 4.1
(Added back old variable names for easier merges)
sql/opt_range.cc:
merge with 4.1
Removed argument 'parent_alloc' from QUICK_RANGE_SELECT as this was not used
Added assert if using QUICK_GROUP_MIN_MAX_SELECT with parent_alloc as the init() function can't handle this
Changed back get_quick_select_for_ref() to use it's own alloc root becasue this function may be called several times for one query
sql/sql_handler.cc:
merge with 4.1
change variable 'err' to 'error' as same function had a label named 'err'
sql/sql_update.cc:
Use multi-update code from 5.0 instead of 4.1
We will fix the locking code shortly in 5.0 to be faster than in 4.1
into rurik.mysql.com:/home/igor/dev/mysql-5.0-0
mysql-test/r/innodb.result:
Auto merged
mysql-test/r/range.result:
Auto merged
mysql-test/t/range.test:
Auto merged
sql/opt_range.cc:
Auto merged
BitKeeper/etc/ignore:
auto-union
mysql-test/r/range.result:
Auto merged
mysql-test/r/select.result:
Auto merged
mysql-test/t/range.test:
Auto merged
sql/item.cc:
Auto merged
sql/opt_range.cc:
Auto merged
sql/opt_sum.cc:
Auto merged
sql/sql_select.h:
Auto merged
sql/item.h:
Manual merge
sql/sql_select.cc:
Manual merge
test results corrected
mysql-test/r/range.result:
test results corrected
mysql-test/t/range.test:
test results corrected
sql/opt_range.cc:
compatibility fix
BitKeeper/etc/ignore:
auto-union
mysql-test/r/bdb.result:
Auto merged
mysql-test/r/func_group.result:
Auto merged
mysql-test/r/innodb.result:
Auto merged
mysql-test/r/heap_btree.result:
Auto merged
mysql-test/r/select.result:
Auto merged
mysql-test/r/user_var.result:
Auto merged
mysql-test/t/range.test:
Auto merged
sql/item_func.cc:
Auto merged
sql/item_func.h:
Auto merged
sql/item_row.cc:
Auto merged
sql/item_strfunc.h:
Auto merged
sql/opt_sum.cc:
Auto merged
sql/sql_list.h:
Auto merged
mysql-test/r/func_test.result:
Post-automerge resolution of conflicts
mysql-test/r/index_merge.result:
Post-automerge resolution of conflicts
mysql-test/r/join_outer.result:
Post-automerge resolution of conflicts
mysql-test/r/range.result:
Post-automerge resolution of conflicts
mysql-test/r/subselect.result:
Post-automerge resolution of conflicts
sql/item.cc:
Post-automerge resolution of conflicts
sql/item.h:
Post-automerge resolution of conflicts
sql/item_cmpfunc.cc:
Post-automerge resolution of conflicts
sql/item_cmpfunc.h:
Post-automerge resolution of conflicts
sql/opt_range.cc:
Post-automerge resolution of conflicts
sql/sql_select.cc:
Post-automerge resolution of conflicts
sql/sql_select.h:
Post-automerge resolution of conflicts
Some bigger code changes was necessary becasue of the multi-table-update and the new HANDLER code
include/hash.h:
Added back function that's was used in 4.0
mysql-test/r/delete.result:
Update results after merge
mysql-test/r/flush_table.result:
Update results after merge
mysql-test/r/func_str.result:
Update results after merge
mysql-test/r/handler.result:
Update results after merge
Change is big becasue in MySQL 4.1 you are not allowed to qualify the handler alias with a databasename
mysql-test/r/multi_update.result:
More startup cleanups
mysql-test/r/rename.result:
More startup-cleanups
mysql-test/r/select.result:
More startup cleanups
mysql-test/r/show_check.result:
More startup-cleanups
mysql-test/t/ctype_latin1_de.test:
Cleanup
mysql-test/t/derived.test:
Portability fix
mysql-test/t/handler.test:
Update results after merge
Change is big becasue in MySQL 4.1 you are not allowed to qualify the handler alias with a databasename
mysql-test/t/multi_update.test:
More startup cleanups
mysql-test/t/range.test:
More comments
mysql-test/t/rename.test:
More startup cleanups
mysql-test/t/select.test:
More startup cleanups
mysql-test/t/show_check.test:
More startup cleanups
mysql-test/t/type_timestamp.test:
Add back test deleted during merge
sql/item_cmpfunc.cc:
After merge fixes
sql/item_func.cc:
Remove compiler warning
sql/mysql_priv.h:
After merge fixes
sql/mysqld.cc:
After merge fixes
sql/sql_acl.cc:
More debugging
sql/sql_base.cc:
After merge fixes
(This fix was needed bacause of multi-table-update reopens tables)
sql/sql_handler.cc:
After merge fixes
sql/sql_lex.h:
After merge fixes
sql/sql_select.cc:
After merge fixes
sql/sql_show.cc:
After merge fixes
sql/sql_table.cc:
After merge fixes
Simple cleanup of mysql_discard_or_import_tablespace
sql/sql_update.cc:
After merge fixes
Rework mysql_multi_update to take into account derived tables.
sql/sql_yacc.yy:
After merge fixes
