This patch adds cost estimation for the queries with ORDER BY / GROUP BY
and LIMIT.
If there was a ref/range access to the table whose rows were required
to be ordered in the result set the optimizer always employed this access
though a scan by a different index that was compatible with the required
order could be cheaper to produce the first L rows of the result set.
Now for such queries the optimizer makes a choice between the cheapest
ref/range accesses not compatible with the given order and index scans
compatible with it.
When the SQL_BIG_RESULT flag is specified SELECT should store items from the
select list in the filesort data and use them when sending to a client.
The get_addon_fields function is responsible for creating necessary structures
for that. But this function was allowed to do so only for SELECT and
INSERT .. SELECT queries. This makes the SQL_BIG_RESULT useless for
the CREATE .. SELECT queries.
Now the get_addon_fields allows storing select list items in the filesort
data for the CREATE .. SELECT queries.
If a primary key is defined over column c of enum type then
the EXPLAIN command for a look-up query of the form
SELECT * FROM t WHERE c=0
said that the query was with an impossible where condition though the
query correctly returned non-empty result set when the table indeed
contained rows with error empty strings for column c.
This kind of misbehavior was due to a bug in the function
Field_enum::store(longlong,bool) that erroneously returned 1 if
the the value to be stored was equal to 0.
Note that the method
Field_enum::store(const char *from,uint length,CHARSET_INFO *cs)
correctly returned 0 if a value of the error empty string
was stored.
This bug manifested itself for join queries with GROUP BY and HAVING clauses
whose SELECT lists contained DISTINCT. It occurred when the optimizer could
deduce that the result set would have not more than one row.
The bug could lead to wrong result sets for queries of this type because
HAVING conditions were erroneously ignored in some cases in the function
remove_duplicates.
After dumping triggers mysqldump copied
the value of the OLD_SQL_MODE variable to the SQL_MODE
variable. If the --compact option of the mysqldump was
not set the OLD_SQL_MODE variable had the value
of the uninitialized SQL_MODE variable. So
usually the NO_AUTO_VALUE_ON_ZERO option of the
SQL_MODE variable was discarded.
This fix is for non-"--compact" mode of the mysqldump,
because mysqldump --compact never set SQL_MODE to the
value of NO_AUTO_VALUE_ON_ZERO.
The dump_triggers_for_table function has been modified
to restore previous value of the SQL_MODE variable after
dumping triggers using the SAVE_SQL_MODE temporary
variable.
ORDER BY primary_key on InnoDB table
Queries that use an InnoDB secondary index to retrieve
data don't need to sort in case of ORDER BY primary key
if the secondary index is compared to constant(s).
They can also skip sorting if ORDER BY contains both the
the secondary key parts and the primary key parts (in
that order).
This is because InnoDB returns the rows in order of the
primary key for rows with the same values of the secondary
key columns.
Fixed by preventing temp table sort for the qualifying
queries.
by long running transaction
On Windows opened files can't be deleted. There was a special
upgraded lock mode (TL_WRITE instead of TL_WRITE_ALLOW_READ)
in ALTER TABLE to make sure nobody has the table opened
when deleting the old table in ALTER TABLE. This special mode
was causing ALTER TABLE to hang waiting on a lock inside InnoDB.
This special lock is no longer necessary as the server is
closing the tables it needs to delete in ALTER TABLE.
Fixed by removing the special lock.
Note that this also reverses the fix for bug 17264 that deals with
another consequence of this special lock mode being used.
The Item_date_typecast::val_int function doesn't reset null_value flag.
This makes all values that follows the first null value to be treated as nulls
and led to a wrong result.
Now the Item_date_typecast::val_int function correctly sets the null_value flag
for both null and non-null values.
a temporary table.
The result string of the Item_func_group_concat wasn't initialized in the
copying constructor of the Item_func_group_concat class. This led to a
wrong charset of GROUP_CONCAT result when the select employs a temporary
table.
The copying constructor of the Item_func_group_concat class now correctly
initializes the charset of the result string.
Optimization of queries with DETERMINISTIC functions in the
WHERE clause was not effective: sequential scan was always
used.
Now a SF with the DETERMINISTIC flags is treated as constant
when it's arguments are constants (or a SF doesn't has arguments).
For each view the mysqldump utility creates a temporary table
with the same name and the same columns as the view
in order to satisfy views that depend on this view.
After the creation of all tables, mysqldump drops all
temporary tables and creates actual views.
However, --skip-add-drop-table and --compact flags disable
DROP TABLE statements for those temporary tables. Thus, it was
impossible to create the views because of existence of the
temporary tables with the same names.
