MySQL Community Server 5.1.36がリリースされました。


最も普及しているオープンソースデータベース管理システムの新バージョンであるMySQL Community Server 5.1.36がリリースされました。MySQL 5.1.36は、プロダクションシステムでの使用をお勧めします。

MySQL 5.1の新機能の概要については、以下を参照してください。


新たなサーバにMySQL 5.1.36をインストール、または以前のMySQLリリースからMySQL 5.1.36にアップグレードする際の情報については、以下を参照してください。


MySQL 5.1.36は、http://dev.mysql.com/downloads/から、ソースコード及び多くのプラットフォームのためのバイナリで現在利用可能です。




MySQL 5.1に関するオープンな問題の情報については、以下のエラッタリストを参照してください。







   Functionality added or changed:

     * Important Change: Replication: Previously, incident log
       events were represented as comments in the output from
       mysqlbinlog, making them effectively silent when playing
       back the binlog.
       (An incident log event represents an incident that could
       cause the contents of the database to change without that
       event being recorded in the binary log.)
       This meant that, if the SQL were applied to a server, it
       could potentially lead to the master and the slave having
       different data.  To make it possible to handle incident
       log events without breaking applications that expect the
       previous behavior, the nonsense statement RELOAD DATABASE
       is added to the SQL output for that incident log event,
       which causes an error.
       To use this functionality currently requires hand editing
       of the dump file and handling of each case on an
       individual basis by a database administrator before
       applying the output to a server.

     * The time zone tables for Windows available at
       timezones.html have been updated.

     * The mysqltest program now has a move_file from_file
       to_file command for renaming files.  This should be used
       in test cases rather than invoking an external command
       that might be platform specific.

   Bugs fixed:

     * Incompatible Change: The server can load plugins under
       the control of startup options.  For example, many storage
       engines can be built in pluggable form and loaded when
       the server starts.  In the following descriptions,
       plugin_name stands for a plugin name such as innodb.
       Previously, plugin options were handled like other
       boolean options (see Section, "Program Option
       Modifiers").  That is, any of these options enabled the


       And these options disabled the plugin:


       However, use of a boolean option for plugin loading did
       not provide control over what to do if the plugin failed
       to start properly: Should the server exit, or start with
       the plugin disabled? The actual behavior has been that
       the server starts with the plugin disabled, which can be
       problematic.  For example, if InnoDB fails to start,
       existing InnoDB tables become inaccessible, and attempts
       to create new InnoDB tables result in tables that use the
       default storage engine unless the NO_ENGINE_SUBSTITUTION
       SQL mode has been enabled to cause an error to occur
       Now, there is a change in the options used to control
       plugin loading, such that they have a tristate format:

          + --plugin_name=OFF
            Do not enable the plugin.

          + --plugin_name[=ON]
            Enable the plugin.  If plugin initialization fails,
            start the server anyway, but with the plugin
            disabled.  Specifying the option as --plugin_name
            without a value also enables the plugin.

          + --plugin_name=FORCE
            Enable the plugin.  If plugin initialization fails,
            do not start the server.  In other words, force the
            server to run with the plugin or not at all.

       The values OFF, ON, and FORCE are not case sensitive.
       Suppose that CSV and InnoDB have been built as pluggable
       storage engines and that you want the server to load them
       at startup, subject to these conditions: The server is
       allowed to run if CSV initialization fails, but must
       require that InnoDB initialization succeed.  To accomplish
       that, use these lines in an option file:


       This change is incompatible with the previous
       implementation if you used options of the form
       --plugin_name=0 or --plugin_name=1, which should be
       changed to --plugin_name=OFF or --plugin_name=ON,
       --enable-plugin_name is still supported and is the same
       as --plugin_name=ON. --disable-plugin_name and
       --skip-plugin_name are still supported and are the same
       as --plugin_name=OFF.

     * Replication: When using row-based logging, the length of
       an event for which the field metadata exceeded 255 bytes
       in size was incorrectly calculated.  This could lead to
       corruption of the binary log, or cause the server to
       hang. (Bug#42749:http://bugs.mysql.com/42749)
       See alsoBug#44548:http://bugs.mysql.com/44548,

     * Replication: The warning Statement is not safe to log in
       statement format, issued in situations when it cannot be
       determined that a statement or other database event can
       be written reliably to the binary log using the
       statement-based format, has been changed to Statement may
       not be safe to log in statement format.

     * Replication: When stopping and restarting the slave while
       it was replicating temporary tables, the slave server
       could crash or raise an assertion failure.  This was due
       to the fact that, although temporary tables were saved
       between slave thread restarts, the reference to the
       thread being used (table->in_use) was not being properly
       updated when restarting, continuing to reference the old
       thread instead of the new one.  This issue affected
       statement-based replication only.

     * ALTER TABLE on a view crashed the server.

     * UNCOMPRESSED_LENGTH() returned a garbage result when
       passed a string shorter than 5 bytes.  Now
       UNCOMPRESSED_LENGTH() returns NULL and generates a
       warning. (Bug#44796:http://bugs.mysql.com/44796)

     * Several Valgrind warnings were silenced.

     * Selecting RAND(N) function where N is a column of a
       constant table (table with a single row) failed with a
       SIGFPE signal. (Bug#44768:http://bugs.mysql.com/44768)

     * Conversion of a string to a different character set could
       use the same buffer for input and output, leading to
       incorrect results or warnings.

     * mysqld_safe could fail to find the logger program.

     * A Valgrind warning related to transaction processing was
       silenced. (Bug#44664:http://bugs.mysql.com/44664)

     * innochecksum could incorrectly determine the input file
       name from the arguments.

     * Incorrect time was reported at the end of mysqldump
       output. (Bug#44424:http://bugs.mysql.com/44424)

     * Caching of GROUP BY expressions could lead to mismatches
       between compile-time and runtime calculations and cause a
       server crash. (Bug#44399:http://bugs.mysql.com/44399)

     * Lettercase conversion in multibyte cp932 or sjis
       character sequences could produce incorrect results.

     * InnoDB was missing DB_ROLL_PTR information in Table
       Monitor COLUMNS output.

     * Assertion failure could occur for duplicate-key errors in
       INSERT INTO ... SELECT statements.

     * On 64-bit Windows systems, myisamchk did not handle
       key_buffer_size values larger than 4GB.

     * EXPLAIN EXTENDED could crash for UNION queries in which
       the last SELECT was not parenthesized and included an
       ORDER BY clause. (Bug#43612:http://bugs.mysql.com/43612)

     * Multiple-table updates for InnoDB tables could produce
       incorrect results.

     * For DELETE statements with ORDER BY var, where var was a
       global system variable with a NULL value, the server
       could crash. (Bug#42778:http://bugs.mysql.com/42778)

     * Builds linked against OpenSSL had a memory leak in
       association with use of X509 certificates.

     * There was a race condition when changing
       innodb_commit_concurrency at runtime from zero to nonzero
       or from nonzero to zero.  Now this variable cannot be
       changed at runtime from zero to nonzero or vice versa.
       The value can still be changed from one nonzero value to
       another. (Bug#42101:http://bugs.mysql.com/42101)

     * SELECT ... INTO @var could produce values different from
       SELECT ... without the INTO clause.

     * mysql_zap did not work on Mac OS X.

     * For views created with a column list clause, column
       aliases were not substituted when selecting through the
       view using a HAVING clause.

     * A multiple-table DELETE involving a table self-join could
       cause a server crash.

     * Creating an InnoDB table with a comment containing a '#'
       character caused foreign key constraints to be omitted.

     * The mysql option --ignore-spaces was nonfunctional.

     * If a query was such as to produce the error 1054 Unknown
       column '...' in 'field list', using EXPLAIN EXTENDED with
       the query could cause a server crash.

     * If the MYSQL_HISTFILE environment variable was set to
       /dev/null, the mysql client overwrote the /dev/null
       device file as a normal file.

     * mysqld_safe mishandled certain parameters if they
       contained spaces.

     * mysqladmin kill did not work for thread IDs larger than
       32 bits. (Bug#32457:http://bugs.mysql.com/32457)

     * Several client programs failed to interpret
       --skip-password as "send no password."

     * Output from mysql --html did not encode the <, >, or &
       characters. (Bug#27884:http://bugs.mysql.com/27884)

     * mysql_convert_table_format did not prevent converting
       tables to MEMORY or BLACKHOLE tables, which could result
       in data loss. (Bug#27149:http://bugs.mysql.com/27149)