Functions specific for native InnoDB partitioning. More...

Functions
int	ha_innopart::open (const char name, int mode, uint test_if_locked, const dd::Table table_def) override
	Open an InnoDB table. More...

int	ha_innopart::close () override
	Closes a handle to an InnoDB table. More...

double	ha_innopart::scan_time () override
	Time estimate for full table scan. More...

bool	ha_innopart::was_semi_consistent_read () override
	Was the last returned row semi consistent read. More...

void	ha_innopart::try_semi_consistent_read (bool yes) override
	Try semi consistent read. More...

void	ha_innopart::unlock_row () override
	Removes a lock on a row. More...

int	ha_innopart::index_init (uint index, bool sorted) override
	Initializes a handle to use an index. More...

int	ha_innopart::index_end () override
	End index cursor. More...

int	ha_innopart::rnd_init (bool scan) override
	Initialize a table scan. More...

int	ha_innopart::rnd_end () override
	Ends a table scan. More...

int	ha_innopart::external_lock (THD *thd, int lock_type) override
	Lock/prepare to lock table. More...

THR_LOCK_DATA **	ha_innopart::store_lock (THD thd, THR_LOCK_DATA *to, thr_lock_type lock_type) override
	Function to store lock for all partitions in native partitioned table. More...

int	ha_innopart::write_row (uchar *record) override
	Stores a row in an InnoDB database, to the table specified in this handle. More...

int	ha_innopart::update_row (const uchar old_record, uchar new_record) override
	Updates a row given as a parameter to a new value. More...

int	ha_innopart::delete_row (const uchar *record) override
	Deletes a row given as the parameter. More...

Detailed Description

Functions specific for native InnoDB partitioning.

See also: handler.h

Function Documentation

◆ close()

int ha_innopart::close ( void )

overrideprivatevirtual

Closes a handle to an InnoDB table.

Returns: 0

Reimplemented from ha_innobase.

◆ delete_row()

int ha_innopart::delete_row ( const uchar * record )

inlineoverrideprivatevirtual

Deletes a row given as the parameter.

Returns: error number or 0

Parameters

record in: a row in MySQL format

Reimplemented from ha_innobase.

◆ external_lock()

int ha_innopart::external_lock	(	THD *	thd,
		int	lock_type
	)

overrideprivatevirtual

Lock/prepare to lock table.

As MySQL will execute an external lock for every new table it uses when it starts to process an SQL statement (an exception is when MySQL calls start_stmt for the handle) we can use this function to store the pointer to the THD in the handle. We will also use this function to communicate to InnoDB that a new SQL statement has started and that we must store a savepoint to our transaction handle, so that we are able to roll back the SQL statement in case of an error.

Parameters

[in]	thd	Handle to the user thread.
[in]	lock_type	Lock type.

Returns: 0 or error number.

Reimplemented from ha_innobase.

◆ index_end()

int ha_innopart::index_end ( void )

overrideprivatevirtual

End index cursor.

Returns: 0 or error code.

Reimplemented from ha_innobase.

◆ index_init()

int ha_innopart::index_init	(	uint	keynr,
		bool	sorted
	)

overrideprivatevirtual

Initializes a handle to use an index.

Parameters

[in]	keynr	Key (index) number.
[in]	sorted	True if result MUST be sorted according to index.

Returns: 0 or error number.

Reimplemented from ha_innobase.

◆ open()

int ha_innopart::open	(	const char *	name,
		int	mode,
		uint	test_if_locked,
		const dd::Table *	table_def
	)

overrideprivatevirtual

Open an InnoDB table.

Parameters

[in]	name	table name
[in]	mode	access mode
[in]	test_if_locked	test if the file to be opened is locked
[in]	table_def	dd::Table describing table to be opened

Return values

1	if error
0	if success

Reimplemented from ha_innobase.

◆ rnd_end()

int ha_innopart::rnd_end ( void )

inlineoverrideprivatevirtual

Ends a table scan.

Returns: 0 or error number

Reimplemented from ha_innobase.

◆ rnd_init()

int ha_innopart::rnd_init ( bool scan )

inlineoverrideprivatevirtual

Initialize a table scan.

Parameters

[in] scan whether this is a second call to rnd_init() without rnd_end() in between

Returns: 0 or error number

Reimplemented from ha_innobase.

◆ scan_time()

double ha_innopart::scan_time ( )

overrideprivatevirtual

Time estimate for full table scan.

How many seeks it will take to read through the table. This is to be comparable to the number returned by records_in_range so that we can decide if we should scan the table or use keys.

Returns: estimated time measured in disk seeks.

Reimplemented from ha_innobase.

◆ store_lock()

THR_LOCK_DATA ** ha_innopart::store_lock	(	THD *	thd,
		THR_LOCK_DATA **	to,
		thr_lock_type	lock_type
	)

overrideprivatevirtual

Function to store lock for all partitions in native partitioned table.

Also look at ha_innobase::store_lock for more details.

Parameters

[in]	thd	user thread handle
[in]	to	pointer to the current element in an array of pointers to lock structs
[in]	lock_type	lock type to store in 'lock'; this may also be TL_IGNORE

Return values

to	pointer to the current element in the 'to' array

Reimplemented from ha_innobase.

◆ try_semi_consistent_read()

void ha_innopart::try_semi_consistent_read ( bool yes )

overrideprivatevirtual

Try semi consistent read.

Tell the engine whether it should avoid unnecessary lock waits. If yes, in an UPDATE or DELETE, if the row under the cursor was locked by another transaction, the engine may try an optimistic read of the last committed row value under the cursor.

See also: handler.h and row0mysql.h

Parameters

[in] yes Should semi-consistent read be used.

Reimplemented from ha_innobase.

◆ unlock_row()

void ha_innopart::unlock_row ( void )

overrideprivatevirtual

Removes a lock on a row.

Removes a new lock set on a row, if it was not read optimistically. This can be called after a row has been read in the processing of an UPDATE or a DELETE query.

See also: ha_innobase::unlock_row().

Reimplemented from ha_innobase.

◆ update_row()

int ha_innopart::update_row	(	const uchar *	old_row,
		uchar *	new_row
	)

inlineoverrideprivatevirtual

Updates a row given as a parameter to a new value.

Note that we are given whole rows, not just the fields which are updated: this incurs some overhead for CPU when we check which fields are actually updated. TODO: currently InnoDB does not prevent the 'Halloween problem': in a searched update a single row can get updated several times if its index columns are updated!

Parameters

[in]	old_row	Old row contents in MySQL format
[out]	new_row	Updated row contents in MySQL format

Returns: error number or 0

Reimplemented from ha_innobase.

◆ was_semi_consistent_read()

bool ha_innopart::was_semi_consistent_read ( void )

overrideprivatevirtual

Was the last returned row semi consistent read.

In an UPDATE or DELETE, if the row under the cursor was locked by another transaction, and the engine used an optimistic read of the last committed row value under the cursor, then the engine returns 1 from this function. MySQL must NOT try to update this optimistic value. If the optimistic value does not match the WHERE condition, MySQL can decide to skip over this row. This can be used to avoid unnecessary lock waits.

If this method returns true, it will also signal the storage engine that the next read will be a locking re-read of the row.

See also: handler.h and row0mysql.h

Returns: true if last read was semi consistent else false.

Reimplemented from ha_innobase.

◆ write_row()

int ha_innopart::write_row ( uchar * record )

inlineoverrideprivatevirtual

Stores a row in an InnoDB database, to the table specified in this handle.

Returns: error code

Parameters

record in: a row in MySQL format

Reimplemented from ha_innobase.