MySQL 4.1.21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であるMySQL 4.1.21がリリースされました。コミュニティエディションは、http://dev.mysql.com/downloads/とミラーサイトのダウンロード・ページから、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。






Functionality added or changed:
    * For spatial data types, the server formerly returned these as
      VARSTRING values with a binary collation. Now the server
      returns spatial values as BLOB values.
    * Added the --set-charset option to mysqlbinlog to allow the
      character set to be specified for processing binary log files.
    * For a table with an AUTO_INCREMENT column, SHOW CREATE TABLE
      now shows the next AUTO_INCREMENT value to be generated.
    * The mysqldumpslow script has been moved from client RPM
      packages to server RPM packages. This corrects a problem where
      mysqldumpslow could not be used with a client-only RPM
      install, because it depends on my_print_defaults which is in
      the server RPM. (Bug#20216:http://bugs.mysql.com/20216)

Bugs fixed:
    * Security fix: If a user has access to MyISAM table t, that
      user can create a MERGE table m that accesses t. However, if
      the user's privileges on t are subsequently revoked, the user
      can continue to access t by doing so through m. If this
      behavior is undesirable, you can start the server with the new
      --skip-merge option to disable the MERGE storage engine.
    * Security fix: Invalid arguments to DATE_FORMAT() caused a
      server crash. (CVE-2006-3469
      Thanks to Jean-David Maillefer for discovering and reporting
      this problem to the Debian project and to Christian Hammers
      from the Debian Team for notifying us of it.
    * Failure to account for a NULL table pointer on big-endian
      machines could cause a server crash during type conversion.
    * Some memory leaks in the libmysqld embedded server were
      corrected. (Bug#16107:http://bugs.mysql.com/16107)
    * When mysqldump disabled keys and locked a MyISAM table, the
      lock operation happened second. If another client performed a
      query on the table in the interim, it could take a long time
      due to indexes not being used. Now the lock operation happens
      first. (Bug#15977:http://bugs.mysql.com/15977)
    * The length of the pattern string prefix for LIKE operations
      was calculated incorrectly for multi-byte character sets. As a
      result, the the scanned range was wider than necessary if the
      prefix contained any multi-byte characters.
    * For very complex SELECT statements could create temporary
      tables that were too big, but for which the temporary files
      did not get removed, causing subsequent queries to fail.
    * Using SELECT and a table join while running a concurrent
      INSERT operation would join incorrect rows.
    * Using SELECT on a corrupt table using the dynamic record
      format could cause a server crash.
    * Checking a spatial table (using CHECK TABLE) with an index and
      only one row would indicate a table corruption.
    * For SELECT ... FOR UPDATE statements that used DISTINCT or
      GROUP BY over all key parts of a unique index (or primary
      key), the optimizer unnecessarily created a temporary table,
      thus losing the linkage to the underlying unique index values.
      This caused a Result set not updatable error. (The temporary
      table is unnecessary because under these circumstances the
      distinct or grouped columns must also be unique.)
    * Concatenating the results of multiple constant subselects
      produced incorrect results.
    * The use of MIN() and MAX() on columns with a partial index
      produced incorrect results in some queries.
    * Use of MIN() or MAX() with GROUP BY on a ucs2 column could
      cause a server crash. (Bug#20076:http://bugs.mysql.com/20076)
    * INSERT INTO ... SELECT ... LIMIT 1 could be slow because the
      LIMIT was ignored when selecting candidate rows.
    * NDB Cluster: When attempting to restart the cluster following
      a data import, the cluster would fail during Phase 4 of the
      restart with Error 2334: Job buffer congestion.
    * NDB Cluster: A node failure during a scan could sometime cause
      the node to crash when restarting too quickly following the
      failure. (Bug#20197:http://bugs.mysql.com/20197)
    * NDB Cluster: It was possible to use port numbers greater than
      65535 for ServerPort in the config.ini file.
    * The omission of leading zeros in dates could lead to erroneous
      results when these were compared with the output of certain
      date and time functions.
    * Certain queries having a WHERE clause that included conditions
      on multi-part keys with more than 2 key parts could produce
      incorrect results and send [Note] Use_count: Wrong count for
      key at... messages to STDERR.
    * An invalid comparison between keys in partial indexes over
      multi-byte character fields could lead to incorrect result
      sets if the selected query execution plan used a range scan by
      a partial index over a UTF8 character field. This also caused
      incorrect results under similar circumstances with many other
      character sets. (Bug#14896:http://bugs.mysql.com/14896)
    * NDB Cluster: The cluster's data nodes would fail while trying
      to load data when NoOfFrangmentLogFiles was equal to 1.
    * NDB Cluster: A problem with error handling when
      ndb_use_exact_count was enabled could lead to incorrect values
      returned from queries using COUNT(). A warning is now returned
      in such cases. (Bug#19202:http://bugs.mysql.com/19202)
    * NDB Cluster: LOAD DATA LOCAL failed to ignore duplicate keys
      in Cluster tables. (Bug#19496:http://bugs.mysql.com/19496)
    * NDB Cluster: Repeated CREATE - INSERT - DROP operations tables
      could in some circumstances cause the MySQL table definition
      cache to become corrupt, so that some mysqld processes could
      access table information but others could not.
    * NDB Cluster: The mgm client command ALL CLUSTERLOG
      STATISTICS=15; had no effect.
    * NDB Cluster: TRUNCATE TABLE failed to reset the AUTO_INCREMENT
      counter. (Bug#18864:http://bugs.mysql.com/18864)
    * NDB Cluster: SELECT ... FOR UPDATE failed to lock the selected
      rows. (Bug#18184:http://bugs.mysql.com/18184)
    * NDB Cluster: The failure of a data node when preparing to
      commit a transaction (that is, while the node's status was
      CS_PREPARE_TO_COMMIT) could cause the failure of other cluster
      data nodes. (Bug#20185:http://bugs.mysql.com/20185)
    * NDB Cluster: Renaming a table in such a way as to move it to
      to a different database failed to move the table's indexes.
    * NDB Cluster: Resources for unique indexes on Cluster table
      columns were incorrectly allocated, so that only one-fourth as
      many unique indexes as indicated by the value of
      UniqueHashIndexes could be created.
    * NDB Cluster (NDBAPI): On big-endian platforms,
      NdbOperation::write_attr() did not update 32-bit fields
      correctly. (Bug#19537:http://bugs.mysql.com/19537)
    * NDB Cluster: Some queries having a WHERE clause of the form
      c1=val1 OR c2 LIKE 'val2' were not evaluated correctly. (Bug #
    * NDB Cluster: Using "stale" mysqld .FRM files could cause a
      newly-restored cluster to fail. This situation could arise
      when restarting a MySQL Cluster using the --intial option
      while leaving connected mysqld processes running.
    * NDB Cluster: Repeated use of the SHOW and ALL STATUS commands
      in the ndb_mgm client could cause the mgmd process to crash.
    * NDB Cluster: An issue with ndb_mgmd prevented more than 27
      mysqld processes from connecting to a single cluster at one
      time. (Bug#17150:http://bugs.mysql.com/17150)
    * NDB Cluster: Data node failures could cause excessive CPU
      usage by ndb_mgmd. (Bug#13987:http://bugs.mysql.com/13987)
    * NDB Cluster: TRUNCATE failed on tables having BLOB or TEXT
      columns with the error Lock wait timeout exceeded.
    * A cast problem caused incorrect results for prepared
      statements that returned float values when MySQL was compiled
      with gcc 4.0. (Bug#19694:http://bugs.mysql.com/19694)
    * Improper character set initialization in the embedded server
      could result in a server crash.
    * Some queries that used ORDER BY and LIMIT performed quickly in
      MySQL 3.23, but slowly in MySQL 4.x/5.x due to an optimizer
      problem. (Bug#4981:http://bugs.mysql.com/4981)
    * Queries using an indexed column as the argument for the MIN()
      and MAX() functions following an ALTER TABLE .. DISABLE KEYS
      statement returned Got error 124 from storage engine until
      ALTER TABLE ... ENABLE KEYS was run on the table.
    * A number of dependency issues in the RPM bench and test
      packages caused installation of these packages to fail.
    * The MD5() and SHA() functions treat their arguments as
      case-sensitive strings. But when they are compared, their
      arguments were compared as case-insensitive strings, which
      leads to two function calls with different arguments (and thus
      different results) compared as being identical. This can lead
      to a wrong decision made in the range optimizer and thus to an
      incorrect result set. (Bug#15351:http://bugs.mysql.com/15351)
    * InnoDB unlocked its data directory before committing a
      transaction, potentially resulting in non-recoverable tables
      if a server crash occurred before the commit.
    * Multiple-table DELETE statements containing a subquery that
      selected from one of the tables being modified caused a server
      crash. (Bug#19225:http://bugs.mysql.com/19225)
    * The ARCHIVE storage engine does not support TRUNCATE TABLE,
      but the server was not returning an appropriate error when
      truncation of an ARCHIVE table was attempted.
    * An update that used a join of a table to itself and modified
      the table on both sides of the join reported the table as
      crashed. (Bug#18036:http://bugs.mysql.com/18036)
    * The fill_help_tables.sql file did not load properly if the
      ANSI_QUOTES SQL mode was enabled.
    * The fill_help_tables.sql file did not contain a SET NAMES
      'utf8' statement to indicate its encoding. This caused
      problems for some settings of the MySQL character set such as
      big5. (Bug#20551:http://bugs.mysql.com/20551)
    * The MySQL server startup script /etc/init.d/mysql (created
      from mysql.server) is now marked to ensure that the system
      services ypbind, nscd, ldap, and NTP are started first (if
      these are configured on the machine).
    * For a reference to a non-existent index in FORCE INDEX, the
      error message referred to a column, not an index.
    * In a multiple-row INSERT statement, LAST_INSERT_ID() should
      return the same value for each row. However, in some cases,
      the value could change if the table being inserted into had
      its own AUTO_INCREMENT column.
    * Invalid escape sequences in option files caused MySQL programs
      that read them to abort.
    * Binary log lacked character set information for table name
      when dropping temporary tables.
    * InnoDB did not increment the handler_read_prev counter.
    * Slave SQL thread cleanup was not handled properly on Mac OS X
      when a statement was killed, resulting in a slave crash.
    * mysqldump did not respect the order of tables named with the
      --tables option. (Bug#18536:http://bugs.mysql.com/18536)
    * The server no longer uses a signal handler for signal 0
      because it could cause a crash on some platforms.
    * LOAD_FILE() returned an error if the file did not exist,
      rather than NULL as it should according to the manual.
    * Use of uninitialized user variables in a subquery in the FROM
      clause results in bad entries in the binary log.
    * IS_USED_LOCK() could return an incorrect connection
      identifier. (Bug#16501:http://bugs.mysql.com/16501)
    * Concurrent reading and writing of privilege structures could
      crash the server. (Bug#16372:http://bugs.mysql.com/16372)
    * A statement containing GROUP BY and HAVING clauses could
      return incorrect results when the HAVING clause contained
      logic that returned FALSE for every row.
    * MONTHNAME(STR_TO_DATE(NULL, '%m')) could cause a server crash.
    * The ref optimizer could choose the ref_or_null access method
      in cases where it was not applicable. This could cause
      inconsistent EXPLAIN or SELECT results for a given statement.
    * ANALYZE TABLE for TEMPORARY tables had no effect.
    * When myisamchk needed to rebuild a table, AUTO_INCREMENT
      information was lost. (Bug#10405:http://bugs.mysql.com/10405)
    * No error message was being issued for storage engines that do
      not support ALTER TABLE. Now an ER_NOT_SUPPORTED_YET error
      occurs. (Bug#7643:http://bugs.mysql.com/7643)
    * The binary log would create an incorrect DROP query when
      creating temporary tables during replication.