Starting with version 5.0.8, changes for MySQL Cluster can be found in the combined Change History.
Previously, conversion of
DATETIME values to numeric form
by adding zero produced a result in
YYYYMMDDHHMMSS format. The result of
DATETIME+0 is now in
Some data definition statements (
TABLE where the table was not a temporary table,
DROP DATABASE, and
CREATE DATABASE) were not being
written to the binary log after a
also caused problems with replication.
As a result of this fix, the following statements now cause an implicit commit:
GROUP BY query uses a grouping column
from the query's
MySQL now issues a warning. This is done because the SQL
standard states that any grouping column must unambiguously
reference a column of the table resulting from the query's
FROM clause, and permitting columns from the
SELECT clause to be used as
grouping columns is a MySQL extension to the standard.
By way of example, consider the following table:
CREATE TABLE users ( userid INT NOT NULL PRIMARY KEY, username VARCHAR(25), usergroupid INT NOT NULL );
MySQL enables you to use the alias in this query:
SELECT usergroupid AS id, COUNT(userid) AS number_of_users FROM users GROUP BY id;
However, the SQL standard requires that the column name be used, as shown here:
SELECT usergroupid AS id, COUNT(userid) AS number_of_users FROM users GROUP BY usergroupid;
Queries such as the first of the two shown above will continue
to be supported in MySQL; however, beginning with MySQL 5.0.8,
using a column alias in this fashion will generate a warning.
Note that in the event of a collision between column names or
aliases used in joins, MySQL attempts to resolve the conflict by
giving preference to columns arising from tables named in the
Using prepared statements within a stored routine
DEALLOCATE PREPARE) could cause
the client connection to be dropped after the routine returned.
In addition, executing a statement which called a function
deallocating the same statement caused the server to crash. This
is prevented by disabling dynamic SQL within stored routines.
This restriction was lifted in 5.0.13 for stored procedures, but not stored functions or triggers.
(Bug #10975, Bug #10605)
References: See also Bug #7115.
Added support for
B'10' syntax for bit
MEMORY tables now support indexes of up to
500 bytes. See The MEMORY (HEAP) Storage Engine.
Expanded on information provided in general log and slow query log for prepared statements. (Bug #8367, Bug #9334)
On Windows systems, a user with any of the following privileges
*.* could crash mysqld
by issuing a
USE LPT1; or
In addition, any of the commands
USE COM1;, or
USE AUX; would report success even though the
database was not in fact changed.
Although this bug was thought to be fixed previously, it was later discovered to be present in the MySQL 5.0.7-beta release for Windows.
TransactionInactiveTimeout = 0 did
not result in an infinite timeout.
MySQL Cluster: mysqld processes did not reconnect to the cluster following a restart of ndb_mgmd. (Bug #11221)
MySQL Cluster: Insert records were incorrectly applied by ndb_restore, thus making restoring from backup inconsistent if the binary log contained inserts. (Bug #11166)
DELETE performed as part of a
transaction caused an erroneous result.
MySQL Cluster: Connections between data nodes and management nodes were not closed following shutdown of ndb_mgmd. (Bug #11132)
The ndb_mgm client's
SHOW command displayed incorrect
output after master data node failure.
MySQL Cluster: When using dynamically allocated ports on Linux, the cluster would hang on initial startup. (Bug #10893)
Not permitting sufficient parallelism in the cluster's
configuration (for example, by setting
NoOfTransactions too small) caused
ndb_restore to fail without providing any
MySQL Cluster: Running ndb_select_count crashed the cluster when running on Red Hat Enterprise 4/64-bit/Opteron. (Bug #10058)
MySQL Cluster: Data nodes failed to restart on 64-bit Solaris. (Bug #9025)
MySQL Cluster: On 64-bit Solaris 9, the cluster timed out and crashed after the first query was made. (Bug #8918)
Replication: An invalid comparison caused warnings for packet length in replication on 64-bit compilers. (Bug #11064)
Multiple range accesses in a subquery cause server crash. (Bug #11487)
CAST() value could not be
included in a
Server crashed when using
GROUP BY on the
result of a
DIV operation on a
INSERT INTO SELECT FROM produced incorrect result when using
An outer join with an
ON condition that
evaluated to false could return an incorrect result.
An outer join with an empty derived table (a result from a subquery) returned no result. (Bug #11284)
CAST( ... AS DECIMAL) didn't work
Corrected a problem with
returning an incorrect result on 64-bit systems.
SHOW INSTANCE OPTIONS command in MySQL
Instance Manager displayed option values incorrectly for options
for which no value had been given.
The default host name for MySQL server was always
Some internal functions did not take into account that, for
multibyte character sets,
columns could exceed 255 bytes and
VARCHAR columns could exceed
65,535 bytes, which could cause the server to crash.
There were locking problems with multiple-statement
DELETE statements performed
within a stored routine, such as incorrectly locking the table
to be read with a read lock rather than a write lock.
crypt() support caused
compilation problems when using OpenSSL/yaSSL on HP-UX and Mac
(Bug #11150, Bug #10675)
NULLIF() function could
produce incorrect results if the first argument was
mysqld_safe would sometimes fail to remove
the pid file for the old
mysql process after
a crash. As a result, the server failed to start due to a false
A mysqld process already exists... error.
Calling a stored procedure that made use of an
... SELECT ... UNION SELECT ... query caused a server
sql_data_access column of
routines table of
INFORMATION_SCHEMA was empty.
statement would crash the server when no
database was selected.
mysqlshow did not interpret
wildcard characters properly if they were given in the table
The host name cache was not working. (Bug #10931)
A three byte buffer overflow in the client functions caused improper exiting of the client when reading a command from the user. (Bug #10841)
The mysql client would output a prompt twice following input of very long strings, because it incorrectly assumed that a call to the _cgets() function would clear the input buffer. (Bug #10840)
@@sql_mode = NULL caused an erroneous
When using a cursor with a prepared statement, the first execution returned the correct result but was not cleaned up properly, causing subsequent executions to return incorrect results. (Bug #10729)
Note that this bug fix induces a slight change in the behavior
of indexes: If an index is defined to be the same length as a
field (or is left to default to that field's length), and the
length of the field is later changed, then the index will adopt
the new length of the field. Previously, the size of the index
did not change for some field types (such as
VARCHAR) when the field type was
InnoDB: Enforce maximum
CHAR_LENGTH() of UTF-8 data in
ON UPDATE CASCADE.
SELECT * FROM returned incorrect results when called from a stored
table had a primary
The granting and revocation of privileges on a stored routine
was performed when running the server with
--skip-grant-tables even after
SET @@GLOBAL.automatic_sp_privileges =
1; was executed.
A stored procedure run while the query cache was enabled could cause the server to crash. (Bug #9715)
Table names were not handled correctly when
lower_case_table_names = 2 if
the table name lettercase differed in the
SHOW CREATE DATABASE INFORMATION_SCHEMA
returned an “unknown database” error.
SELECT DISTINCT ... GROUP BY
rows (it should return a single row).
An issue with index merging could cause suboptimal index merge
plans to be chosen when searching by indexes created on
DATE columns. The same issue
InnoDB storage engine to issue the
using a partial-field key prefix in
mysqlhotcopy script was not parsing the
SHOW SLAVE STATUS
correctly when called with the
A Boolean full-text search where a query contained more query terms than one-third of the query length caused the server to hang or crash. (Bug #7858)
Views could be created with duplicate column names. (Bug #7448)
ORDER BY clause sometimes had no effect on
the ordering of a result when selecting specific columns (as
opposed to using
SELECT *) from a view.
PREPARE to prepare a
statement that invoked a stored routine that executed the
prepared statement caused a Packets out of order
error the second time the routine was invoked. This
is prevented by disabling dynamic SQL within stored routines.
This restriction was lifted in 5.0.13 for stored procedures, but not for stored functions or triggers.
References: See also Bug #10975, Bug #10605.
Selecting from a view defined using
...) caused an error; attempting to execute a
SELECT * FROM INFORMATION_SCHEMA.TABLES query
after defining such a view crashed the server.