MySQL Community Server 5.0.77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 5.0.77がリリースされました。
MySQL 5.0.77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。




以下のセクションは、以前のリリースであるMySQL Community Server 5.0.67以来のMySQLソースコードにおけるバージョン間の変更を記載しています。バグ修正を含む変更ログは以下で確認できます。


MySQL 5.0.77は、現在の製品リリースに対するバグ修正版であり、MySQL 5.0.67(バイナリ)およびMySQL 5.0.75(ソースのみ)に代わるものです。


 Functionality added or changed:

  * Security Enhancement: To enable stricter control over the
    location from which user-defined functions can be loaded,
    the plugin_dir system variable has been backported from
    MySQL 5.1. If the value is non-empty, user-defined
    function object files can be loaded only from the
    directory named by this variable. If the value is empty,
    the behavior that is used prior to the inclusion of
    plugin_dir applies: The UDF object files must be located
    in a directory that is searched by your system's dynamic
    linker. (Bug#37428:http://bugs.mysql.com/37428)

  * A new status variable, Queries, indicates the number of
    statements executed by the server. This includes
    statements executed within stored programs, unlike the
    Questions variable which includes only statements sent to
    the server by clients.

  * Previously, index hints did not work for FULLTEXT
    searches. Now they work as follows:
    For natural language mode searches, index hints are
    silently ignored. For example, IGNORE INDEX(i) is ignored
    with no warning and the index is still used.
    For boolean mode searches, index hints are honored.

Bugs fixed:

  * Important Change: Security Fix: Additional corrections
    were made for the symlink-related privilege problem
    originally addressed in MySQL 5.0.60. The original fix
    did not correctly handle the data directory path name if
    it contained symlinked directories in its path, and the
    check was made only at table-creation time, not at
    table-opening time later.
    See alsoBug#39277:http://bugs.mysql.com/39277.

  * Security Enhancement: The server consumed excess memory
    while parsing statements with hundreds or thousands of
    nested boolean conditions (such as OR (OR ... (OR ...
    ))). This could lead to a server crash or incorrect
    statement execution, or cause other client statements to
    fail due to lack of memory. The latter result constitutes
    a denial of service.

  * Incompatible Change: There were some problems using
    DllMain() hook functions on Windows that automatically do
    global and per-thread initialization for libmysqld.dll:

       + Per-thread initialization: MySQL internally counts
         the number of active threads, which causes a delay
         in my_end() if not all threads have exited. But
         there are threads that can be started either by
         Windows internally (often in TCP/IP scenarios) or by
         users. Those threads do not necessarily use
         libmysql.dll functionality but still contribute to
         the open-thread count. (One symptom is a five-second
         delay in times for PHP scripts to finish.)

       + Process-initialization: my_init() calls WSAStartup
         that itself loads DLLs and can lead to a deadlock in
         the Windows loader.

    To correct these problems, DLL initialization code now is
    not invoked from libmysql.dll by default. To obtain the
    previous behavior (DLL initialization code will be
    called), set the LIBMYSQL_DLLINIT environment variable to
    any value. This variable exists only to prevent breakage
    of existing Windows-only applications that do not call
    mysql_thread_init() and work okay today. Use of
    LIBMYSQL_DLLINIT is discouraged and is removed in MySQL
    6.0. (Bug#37226:http://bugs.mysql.com/37226,

  * Incompatible Change: SHOW STATUS took a lot of CPU time
    for calculating the value of the
    Innodb_buffer_pool_pages_latched status variable. Now
    this variable is calculated and included in the output of
    SHOW STATUS only if the UNIV_DEBUG symbol is defined at
    MySQL build time.

  * Incompatible Change: In connection with view creation,
    the server created arc directories inside database
    directories and maintained useless copies of .frm files
    there. Creation and renaming procedures of those copies
    as well as creation of arc directories has been
    This change does cause a problem when downgrading to
    older server versions which manifests itself under these

      1. Create a view v_orig in MySQL 5.0.72 or higher.

      2. Rename the view to v_new and then back to v_orig.

      3. Downgrade to an older 5.0.x server and run

      4. Try to rename v_orig to v_new again. This operation

    As a workaround to avoid this problem, use either of
    these approaches:

       + Dump your data using mysqldump before downgrading
         and reload the dump file after downgrading.

       + Instead of renaming a view after the downgrade, drop
         it and recreate it.

    The downgrade problem introduced by the fix for this bug
    has been addressed as

  * Replication: When rotating relay log files, the slave
    deletes relay log files and then edits the relay log
    index file. Formerly, if the slave shut down unexpectedly
    between these two events, the relay log index file could
    then reference relay logs that no longer existed.
    Depending on the circumstances, this could when
    restarting the slave cause either a race condition or the
    failure of replication.

  * In example option files provided in MySQL distributions,
    the thread_stack value was increased from 64K to 128K.

  * SET PASSWORD caused a server crash if the account name
    was given as CURRENT_USER().

    limited to 7680 rows.

  * In debug builds, obsolete debug code could be used to
    crash the server.

  * CHECK TABLE ... FOR UPGRADE did not check for
    incompatible collation changes made in MySQL 5.0.48
   Bug#29499:http://bugs.mysql.com/29499). This also
    affects mysqlcheck and mysql_upgrade, which cause that
    statement to be executed. See Section 2.18.3, "Checking
    Whether Table Indexes Must Be Rebuilt."
    See alsoBug#39585:http://bugs.mysql.com/39585.

  * Some queries that used a "range checked for each record"
    scan could return incorrect results.

  * Certain SELECT queries could fail with a Duplicate entry
    error. (Bug#40953:http://bugs.mysql.com/40953)

  * The FEDERATED handler had a memory leak.

  * IF(..., CAST(longtext_val AS UNSIGNED), signed_val) as an
    argument to an aggregate function could cause an
    assertion failure.

  * Prepared statements allowed invalid dates to be inserted
    when the ALLOW_INVALID_DATES SQL mode was not enabled.

  * mc.exe is no longer needed to compile MySQL on Windows.
    This makes it possible to build MySQL from source using
    Visual Studio Express 2008.

  * Support for the revision field in .frm files has been
    removed. This addresses the downgrading problem
    introduced by the fix for

  * If the operating system is configured to return leap
    seconds from OS time calls or if the MySQL server uses a
    time zone definition that has leap seconds, functions
    such as NOW() could return a value having a time part
    that ends with :59:60 or :59:61. If such values are
    inserted into a table, they would be dumped as is by
    mysqldump but considered invalid when reloaded, leading
    to backup/restore problems.
    Now leap second values are returned with a time part that
    ends with :59:59. This means that a function such as
    NOW() can return the same value for two or three
    consecutive seconds during the leap second. It remains
    true that literal temporal values having a time part that
    ends with :59:60 or :59:61 are considered invalid.
    For additional details about leap-second handling, see
    Section 9.7.2, "Time Zone Leap Second Support."

  * The server could crash during a sort-order optimization
    of a dependent subquery.

  * With the ONLY_FULL_GROUP_BY SQL mode enabled, the check
    for non-aggregated columns in queries with aggregate
    functions, but without a GROUP BY clause was treating all
    the parts of the query as if they were in the select
    list. This is fixed by ignoring the non-aggregated
    columns in the WHERE clause.

  * The server crashed if an integer field in a CSV file did
    not have delimiting quotes.


  * For a TIMESTAMP column in an InnoDB table, testing the
    column with multiple conditions in the WHERE clause
    caused a server crash.

  * The server returned a column type of VARBINARY rather
    than DATE as the result from the COALESCE(), IFNULL(),
    IF(), GREATEST(), or LEAST() functions or CASE expression
    if the result was obtained using filesort in an anonymous
    temporary table during the query execution.

  * References to local variables in stored procedures are
    replaced with NAME_CONST(name, value) when written to the
    binary log. However, an "illegal mix of collation" error
    might occur when executing the log contents if the
    value's collation differed from that of the variable. Now
    information about the variable collation is written as
    well. (Bug#39182:http://bugs.mysql.com/39182)

  * Some recent releases for Solaris 10 were built on Solaris
    10 U5, which included a new version of libnsl.so that
    does not work on U4 or earlier. To correct this, Solaris
    10 builds now are created on machines that do not have
    that upgraded libnsl.so, so that they will work on
    Solaris 10 installations both with and without the
    upgraded libnsl.so.

  * With binary logging enabled CREATE VIEW was subject to
    possible buffer overwrite and a server crash.

  * Queries of the form SELECT ... REGEXP BINARY NULL could
    lead to a hung or crashed server.

  * Statements of the form INSERT ... SELECT .. ON DUPLICATE
    KEY UPDATE col_name = DEFAULT could result in a server
    crash. (Bug#39002:http://bugs.mysql.com/39002)

  * Column names constructed due to wild-card expansion done
    inside a stored procedure could point to freed memory if
    the expansion was performed after the first call to the
    stored procedure.

  * Repeated CREATE TABLE ... SELECT statements, where the
    created table contained an AUTO_INCREMENT column, could
    lead to an assertion failure.

  * If delayed insert failed to upgrade the lock, it did not
    free the temporary memory storage used to keep newly
    constructed BLOB values in memory, resulting in a memory
    leak. (Bug#38693:http://bugs.mysql.com/38693)

  * A server crash resulted from concurrent execution of a
    multiple-table UPDATE that used a NATURAL or USING join
    for the table being updated.

  * On ActiveState Perl, mysql-test-run.pl --start-and-exit
    started but did not exit.

  * Server-side cursors were not initialized properly, which
    could cause a server crash.

  * Stored procedures involving substrings could crash the
    server on certain platforms due to invalid memory reads.

  * A server crash or Valgrind warnings could result when a
    stored procedure selected from a view that referenced a
    function. (Bug#38291:http://bugs.mysql.com/38291)

  * Incorrect handling of aggregate functions when loose
    index scan was used caused a server crash.

  * Queries containing a subquery with DISTINCT and ORDER BY
    could cause a server crash.

  * Queries with a HAVING clause could return a spurious row.

  * Use of spatial data types in prepared statements could
    cause memory leaks or server crashes.

  * The server crashed if an argument to a stored procedure
    was a subquery that returned more than one row.

  * When analyzing the possible index use cases, the server
    was incorrectly reusing an internal structure, leading to
    a server crash. (Bug#37943:http://bugs.mysql.com/37943)

  * A SELECT with a NULL NOT IN condition containing a
    complex subquery from the same table as in the outer
    select caused an assertion failure.

  * For InnoDB tables, ORDER BY ... DESC sometimes returned
    results in ascending order.

  * If a table has a BIT NOT NULL column c1 with a length
    shorter than 8 bits and some additional NOT NULL columns
    c2, ..., and a SELECT query has a WHERE clause of the
    form (c1 = constant) AND c2 ..., the query could return
    an unexpected result set.

  * Nesting of IF() inside of SUM() could cause an extreme
    server slowdown. (Bug#37662:http://bugs.mysql.com/37662)

  * The MONTHNAME() and DAYNAME() functions returned a binary
    string, so that using LOWER() or UPPER() had no effect.
    Now MONTHNAME() and DAYNAME() return a value in
    character_set_connection character set.

  * TIMEDIFF() was erroneously treated as always returning a
    positive result. Also, CAST() of TIME values to DECIMAL
    dropped the sign of negative values.

  * mysqlcheck used SHOW FULL TABLES to get the list of
    tables in a database. For some problems, such as an empty
    .frm file for a table, this would fail and mysqlcheck
    then would neglect to check other tables in the database.

  * The <=> operator could return incorrect results when
    comparing NULL to DATE, TIME, or DATETIME values.

  * Updating a view with a subquery in the CHECK option could
    cause an assertion failure.

  * Statements that displayed the value of system variables
    (for example, SHOW VARIABLES) expect variable values to
    be encoded in character_set_system. However, variables
    set from the command line such as basedir or datadir were
    encoded using character_set_filesystem and not converted
    correctly. (Bug#37339:http://bugs.mysql.com/37339)

  * For a MyISAM table with CHECKSUM = 1 and ROW_FORMAT =
    DYNAMIC table options, a data consistency check (maximum
    record length) could fail and cause the table to be
    marked as corrupted.

  * The max_length result set metadata value was calculated
    incorrectly under some circumstances.

  * CREATE INDEX could crash with InnoDB plugin 1.0.1.

  * Certain boolean-mode FULLTEXT searches that used the
    truncation operator did not return matching records and
    calculated relevance incorrectly.

  * The NO_BACKSLASH_ESCAPES SQL mode was ignored for LOAD
    DATA INFILE and SELECT INTO ... OUTFILE. The setting is
    taken into account now.

  * On a 32-bit server built without big tables support, the
    offset argument in a LIMIT clause might be truncated due
    to a 64-bit to 32-bit cast.

  * If the server failed to expire binary log files at
    startup, it could crash.

  * The code for the ut_usectime() function in InnoDB did not
    handle errors from the gettimeofday() system call. Now it
    retries gettimeofday() several times and updates the
    value of the Innodb_row_lock_time_max status variable
    only if ut_usectime() was successful.

  * Use of CONVERT() with GROUP BY to convert numeric values
    to CHAR could return truncated results.

  * A query which had an ORDER BY DESC clause that is
    satisfied with a reverse range scan could cause a server
    crash for some specific CPU/compiler combinations.

  * Dumping information about locks in use by sending a
    SIGHUP signal to the server or by invoking the mysqladmin
    debug command could lead to a server crash in debug
    builds or to undefined behavior in production builds.

  * The mysql client, when built with Visual Studio 2005, did
    not display Japanese characters.

  * When the fractional part in a multiplication of DECIMAL
    values overflowed, the server truncated the first operand
    rather than the longest. Now the server truncates so as
    to produce more precise multiplications.

  * A read past the end of the string could occur while
    parsing the value of the --innodb-data-file-path option.

  * Host name values in SQL statements were not being checked
    for '@', which is illegal according to RFC952.

  * The UUID() function returned UUIDs with the wrong time;
    this was because the offset for the time part in UUIDs
    was miscalculated.

  * SHOW CREATE TABLE did not display a printable value for
    the default value of BIT columns.

  * mysql_install_db failed on machines that had the host
    name set to localhost.

  * Dynamic plugins failed to load on i5/OS.

  * Freeing of an internal parser stack during parsing of
    complex stored programs caused a server crash.

  * The max_length metadata value was calculated incorrectly
    for the FORMAT() function, which could cause incorrect
    result set metadata to be sent to clients.

  * Index scans performed with the sort_union() access method
    returned wrong results, caused memory to be leaked, and
    caused temporary files to be deleted when the limit set
    by sort_buffer_size was reached.

  * If the server crashed with an InnoDB error due to
    unavailability of undo slots, errors could persist during
    rollback when the server was restarted: There are two
    UNDO slot caches (for INSERT and UPDATE). If all slots
    end up in one of the slot caches, a request for a slot
    from the other slot cache would fail. This can happen if
    the request is for an UPDATE slot and all slots are in
    the INSERT slot cache, or vice versa.

  * For InnoDB tables, ALTER TABLE DROP failed if the name of
    the column to be dropped began with "foreign".

  * perror on Windows did not know about Win32 system error
    codes. (Bug#34825:http://bugs.mysql.com/34825)

  * EXPLAIN EXTENDED evaluation of aggregate functions that
    required a temporary table caused a server crash.

  * Queries of the form SELECT ... WHERE string = ANY(...)
    failed when the server used a single-byte character set
    and the client used a multi-byte character set.
    See alsoBug#20835:http://bugs.mysql.com/20835.

  * Using OPTIMIZE TABLE as the first statement on an InnoDB
    table with an AUTO_INCREMENT column could cause a server
    crash. (Bug#34286:http://bugs.mysql.com/34286)

  * mysql_install_db failed if the server was running with an
    SQL mode of TRADITIONAL. This program now resets the SQL
    mode internally to avoid this problem.

  * The mysql client incorrectly parsed statements containing
    the word "delimiter" in mid-statement.
    This fix is different from the one applied for this bug
    in MySQL 5.0.66. (Bug#33812:http://bugs.mysql.com/33812)
    See alsoBug#38158:http://bugs.mysql.com/38158.

  * For a stored procedure containing a SELECT * ... RIGHT
    JOIN query, execution failed for the second call.

  * Previously, use of index hints with views (which do not
    have indexes) produced the error ERROR 1221 (HY000):
    Incorrect usage of USE/IGNORE INDEX and VIEW. Now this
    produces ERROR 1176 (HY000): Key '...' doesn't exist in
    table '...', the same error as for base tables without an
    appropriate index.

  * Cached queries that used 256 or more tables were not
    properly cached, so that later query invalidation due to
    a TRUNCATE TABLE for one of the tables caused the server
    to hang. (Bug#33362:http://bugs.mysql.com/33362)

  * Some division operations produced a result with incorrect
    precision. (Bug#31616:http://bugs.mysql.com/31616)

  * mysql_upgrade attempted to use the /proc file system even
    on systems that do not have it.

  * mysqldump could fail to dump views containing a large
    number of columns.

  * Queries executed using join buffering of BIT columns
    could produce incorrect results.

    TINYTEXT or MEDIUMTEXT columns to a longer text type if
    necessary when converting the column to a different
    character set. (Bug#31291:http://bugs.mysql.com/31291)

  * On NetWare, mysql_install_db could appear to execute
    normally even if it failed to create the initial
    databases. (Bug#30129:http://bugs.mysql.com/30129)

  * The Serbian translation for the
    ER_INCORRECT_GLOBAL_LOCAL_VAR error was corrected.

  * XA transaction rollbacks could result in corrupted
    transaction states and a server crash.

  * On Windows, Visual Studio does not take into account some
    x86 hardware limitations, which led to incorrect results
    converting large DOUBLE values to unsigned BIGINT values.

  * SSL support was not included in some "generic" RPM
    packages. (Bug#26760:http://bugs.mysql.com/26760)

  * In some cases, the parser interpreted the ; character as
    the end of input and misinterpreted stored program
    definitions. (Bug#26030:http://bugs.mysql.com/26030)

  * The Questions status variable is intended as a count of
    statements sent by clients to the server, but was also
    counting statements executed within stored routines.

  * For access to the INFORMATION_SCHEMA.VIEWS table, the
    server did not check the SHOW VIEW and SELECT privileges,
    leading to inconsistency between output from that table
    and the SHOW CREATE VIEW statement.

  * The FLUSH PRIVILEGES statement did not produce an error
    when it failed. (Bug#21226:http://bugs.mysql.com/21226)

  * A race condition between the mysqld.exe server and the
    Windows service manager could lead to inability to stop
    the server from the service manager.

  * mysqld_safe would sometimes fail to remove the pid file
    for the old mysql process after a crash. As a result, the
    server would fail to start due to a false A mysqld
    process already exists... error.