Added support for multi-master replication topographies, which can now be specified using the following host definition syntax: address=(host=hostname)(port=3306)(type=[master|slave]).

Added the flag CLIENT_PLUGIN_AUTH_LENENC_CLIENT_DATA and support for authentication data up to a length of 2⁶⁴-1 bytes.

Added support for EXPLAIN for UPDATE and DELETE. Extended slow query warning with query execution plan for INSERT, REPLACE, UPDATE, and DELETE.

When attempting to connect to a server using long, multibyte user name and database name, an Array Index Out Of Bounds Exception was thrown. This fix prevents the problem by increasing the space reserved for those strings. (Bug #17251955)

When setting a connection property with a value for blobSendChuckSize, largeRowSizeThreshold, or locatorFetchBufferSize, the suffix “G” was not recognized and caused an error to be thrown. With this fix, Connector/J now supports using “G” or “GB” for gigabyte for the purpose, as well as “KB” for kilobyte and “MB” for megabyte. (Bug #17015673, Bug #69452)

When there was an attempt to create a second branch with a duplicate ID on the same MySQL server instance, the XAException.errorCode returned by Connector/J was not equal to XAException.XAER_DUPID. (Bug #16979862, Bug #69506)

The SYSTEM TABLE type was handled inconsistently in different methods: DatabaseMetaData.getTableTypes() did not return system tables, and DatabaseMetaData.getTables() did not return all system tables. With this fix, the table types SYSTEM TABLE and SYSTEM VIEW are returned by getTableTypes(), and getTables() now returns tables from the internal schemas mysql and performance_schema as instances of SYSTEM TABLE and tables from information_schema as instances of SYSTEM VIEW. (Bug #16879239, Bug #69290)

JDBC authentication failed when there was a null byte in the scramble, because com.mysql.jdbc.MysqlIO.doHandshake only read up to the first null byte when reading in the first eight bytes of the scramble. (Bug #16723048, Bug #62469)

When rewriteBatchedStatements=true and useAffectedRows=true were set, executeBatch() did not return the number of affected rows as expected, but always returned a “1,” even if no rows were affected (for example, when an ON DUPLICATE KEY UPDATE clause was applicable but the data needed no changes). The “1” was actually a hardwired value, a workaround for a known issue on the server side, which does not return the number of affected rows in such cases. This fix makes Connector/J just return the value given by the server anyway, which means the returned number of affected rows is reliable now if Connector/J returns a “0” [no rows affected]; nonzero values returned should not be trusted, though. (Bug #16442951, Bug #68562)

In a replication-aware deployment, when the master was not reachable, new connections could not be made, or the replication driver just hung. As part of the new multiple-master support feature, users can now set the property allowMasterDownConnections=true to allow connections to be established even when no master hosts are available. (Bug #11757979, Bug #50105, Bug #16443992, Bug #63354)

Calling ResultSet.absolute(0) caused Connector/J to throw an error. This fix makes ResultSet.absolute(0) behave like ResultSet.beforeFirst(), as required by the JDBC specifications. (Bug #11749136, Bug #38252)

After setting the connection property yearIsDateType=false, the method ResultSet.getMetaData().getColumnClassName() still returned the Java type of a YEAR column as java.sql.Date although internally it was being treated as java.sql.Short. This fix corrects the metadata for a YEAR column under this situation, so that the correct Java type of java.sql.Short is returned. Please note that the method ResultSet.getMetaData().getColumnType(), however, returns java.sql.Types.DATE irrespective of the yearIsDateType setting. (Bug #11748097, Bug #35115)

The string returned by DatabaseMetaData.getDriverVersion() contains the unexpanded expression “${svn.Revision}” or “${bzr.revision-id}” instead of the actual revision number of Connector/J. (Bug #50538, Bug #11758345)