path: root/doc/src/sgml/release-9.4.sgml

<!-- doc/src/sgml/release-9.4.sgml -->
<!-- See header comment in release.sgml about typical markup -->

 <sect1 id="release-9-4">
  <title>Release 9.4</title>

  <note>
   <title>Release Date</title>
   <simpara>2014-??-??</simpara>
   <simpara>Current as of 2014-08-17</simpara>
  </note>

  <sect2>
   <title>Overview</title>

   <para>
    Major enhancements in <productname>PostgreSQL</> 9.4 include:
   </para>

   <!-- This list duplicates items below, but without authors or details-->

   <itemizedlist>

     <listitem>
      <para>
       Allow <link linkend="rules-materializedviews">materialized views</>
       to be refreshed without blocking reads
      </para>
     </listitem>

     <listitem>
      <para>
       Add support for <link linkend="logicaldecoding">logical decoding</>
       of WAL data, to allow database changes to be streamed out in a
       customizable format
      </para>
     </listitem>

     <listitem>
      <para>
        Allow <link linkend="bgworker">background worker processes</>
        to be dynamically registered, started and terminated
      </para>
     </listitem>

     <listitem>
      <para>
       Add <link linkend="datatype-json"><type>jsonb</></link>, a more
       capable and efficient data type for for storing <acronym>JSON</> data
      </para>
     </listitem>

     <listitem>
      <para>
       Add new <acronym>SQL</> command <xref linkend="SQL-ALTERSYSTEM">
       for updating <filename>postgresql.conf</> configuration file entries
      </para>
     </listitem>

     <listitem>
      <para>
       Reduce lock strength for some <xref linkend="SQL-ALTERTABLE">
       commands
      </para>
     </listitem>

   </itemizedlist>

   <para>
    The above items are explained in more detail in the sections below.
   </para>

  </sect2>

  <sect2>

  <title>Migration to Version 9.4</title>

   <para>
    A dump/restore using <xref linkend="app-pg-dumpall">, or use
    of <xref linkend="pgupgrade">, is required for those wishing to migrate
    data from any previous release.
   </para>

   <para>
    Version 9.4 contains a number of changes that may affect compatibility
    with previous releases.  Observe the following incompatibilities:
   </para>

   <itemizedlist>

    <listitem>
     <para>
      Tighten checks for multidimensional <link
      linkend="arrays">array</link> input (Bruce Momjian)
     </para>

     <para>
      Previously, an input array string that started with a single-element
      array dimension could later contain multidimensional segments,
      e.g. <literal>'{{1}, {2,3}}'::int[]</>.
     </para>
    </listitem>

    <listitem>
     <para>
      Unicode escapes in <link linkend="datatype-json"><type>JSON</type></link>
      text values are no longer rendered with the backslash escaped
      (Andrew Dunstan)
     </para>

     <para>
      Previously, all backslashes in text values being formed into JSON
      were escaped. Now a backslash followed by <literal>u</> and four
      hexadecimal digits is not escaped, as this is a legal sequence in a
      JSON string value, and escaping the backslash led to some perverse
      results.
     </para>
    </listitem>

    <listitem>
     <para>
      When converting values of type <type>timestamp</>
      or <type>timestamptz</>
      to <link linkend="datatype-json"><type>JSON</type></link>, render the
      values in a format compliant with ISO 8601 (Andrew Dunstan)
     </para>

     <para>
      Previously such values were rendered according to the current
      <xref linkend="guc-datestyle"> setting; but many JSON processors
      require timestamps to be in ISO 8601 format.
     </para>
    </listitem>

    <listitem>
     <para>
      Cause consecutive whitespace in <link
      linkend="functions-formatting-table"><function>to_timestamp()</></link>
      and <function>to_date()</> format strings to consume a corresponding
      number of characters in the input string (whitespace or not), then
      conditionally consume adjacent whitespace, if not in <literal>FX</>
      mode (Jeevan Chalke)
     </para>

     <para>
      Previously, consecutive whitespace characters in a non-<literal>FX</>
      format string behaved like a single whitespace character and consumed
      all adjacent whitespace in the input string.  For example, previously
      a format string of three spaces would consume only the first space in
      <literal>' 12'</>, but it will now consume all three characters.
     </para>
    </listitem>

    <listitem>
     <para>
      Fix <link
      linkend="textsearch-functions-table"><function>ts_rank_cd()</></link>
      to ignore stripped lexemes (Alex Hill)
     </para>

     <para>
      Previously, stripped lexemes were treated as if they had a default
      location, producing a rank of dubious usefulness.
     </para>
    </listitem>

    <listitem>
     <para>
      For functions declared to
      take <link linkend="xfunc-sql-variadic-functions"><literal>VARIADIC
      "any"</></link>, an actual parameter marked as <literal>VARIADIC</>
      must be of a determinable array type (Pavel Stehule)
     </para>

     <para>
      Such parameters can no longer be written as an undecorated string
      literal or <literal>NULL</>; a cast to an appropriate array data type
      will now be required.  Note that this does not affect parameters not
      marked <literal>VARIADIC</>.
     </para>
    </listitem>

    <listitem>
     <para>
      <xref linkend="sql-discard"> now also discards sequence-related state
      (Fabr&iacute;zio de Royes Mello, Robert Haas)
     </para>
    </listitem>

    <listitem>
     <para>
      Rename <link linkend="SQL-EXPLAIN"><command>EXPLAIN
      ANALYZE</></link>'s <quote>total runtime</quote> output
      to <quote>execution time</quote> (Tom Lane)
     </para>

     <para>
      Now that planning time is also reported, the previous name was
      confusing.
     </para>
    </listitem>

    <listitem>
     <para>
      <link linkend="SQL-SHOW"><command>SHOW TIME ZONE</></link> now
      outputs simple numeric UTC offsets in <acronym>POSIX</> timezone
      format (Tom Lane)
     </para>

     <para>
      Previously, such timezone settings were displayed as <link
      linkend="datatype-interval-output"><type>interval</></link> values.
      The new output is properly interpreted by <command>SET TIME ZONE</>
      when passed as a simple string, whereas the old output required
      special treatment to be re-parsed correctly.
     </para>
    </listitem>

    <listitem>
     <para>
      Prevent <link
      linkend="ddl-constraints-check-constraints"><literal>CHECK</></link>
      constraints from referencing system columns, except
      <structfield>tableoid</> (Amit Kapila)
     </para>

     <para>
      Previously such check constraints were allowed, but they would often
      cause errors during restores.
     </para>
    </listitem>

    <listitem>
     <para>
      Use the last specified <xref linkend="recovery-target"> if multiple
      values are specified (Heikki Linnakangas)
     </para>
    </listitem>

    <listitem>
     <para>
      On Windows, automatically preserve quotes in command strings supplied
      by the user (Heikki Linnakangas)
     </para>

     <para>
      User commands that did their own quote preservation might need
      adjustment.  This is likely to be an issue for commands used in
      <xref linkend="guc-archive-command">, <xref linkend="restore-command">,
      and <link linkend="sql-copy"><command>COPY TO/FROM PROGRAM</></link>.
     </para>
    </listitem>

    <listitem>
     <para>
      Remove catalog column <link
      linkend="catalog-pg-class"><structfield>pg_class.reltoastidxid</></link>
      (Michael Paquier)
     </para>
    </listitem>

    <listitem>
     <para>
      Remove catalog column <link
      linkend="catalog-pg-rewrite"><structfield>pg_rewrite.ev_attr</></link>
      (Kevin Grittner)
     </para>

     <para>
      Per-column rules have not been supported since
      <application>PostgreSQL</> 7.3.
     </para>
    </listitem>

    <listitem>
     <para>
      Remove native support for <application>Kerberos</> authentication
      (<option>--with-krb5</>, etc)
      (Magnus Hagander)
     </para>

     <para>
      The supported way to use <application>Kerberos</> authentication is
      with <acronym>GSSAPI</>.  The native code has been deprecated since
      <productname>PostgreSQL</> 8.3.
     </para>
    </listitem>

    <listitem>
     <para>
      In <application>PL/Python</>, handle domains over arrays like the
      underlying array type (Rodolfo Campero)
     </para>

     <para>
      Previously such values were treated as strings.
     </para>
    </listitem>

    <listitem>
     <para>
      Make libpq's <link
      linkend="libpq-pqconnectdbparams"><function>PQconnectdbParams()</></link>
      and <link
      linkend="libpq-pqpingparams"><function>PQpingParams()</></link>
      functions process zero-length strings as defaults (Adrian
      Vondendriesch)
     </para>

     <para>
      Previously, these functions treated zero-length string values as
      selecting the default in only some cases.
     </para>
    </listitem>

    <listitem>
     <para>
      Change empty arrays returned by the <xref linkend="intarray"> module
      to be zero-dimensional arrays (Bruce Momjian)
     </para>

     <para>
      Previously, empty arrays were returned as zero-length one-dimensional
      arrays, whose text representation looked the same as zero-dimensional
      arrays (<literal>{}</>), but they acted differently in array
      operations.  <application>intarray</>'s behavior in this area now
      matches the built-in array operators.
     </para>
    </listitem>

    <listitem>
     <para>
      <xref linkend="pgupgrade"> now uses <option>-U</>
      or <option>--username</> to specify the user name (Bruce Momjian)
     </para>

     <para>
      Previously this option was spelled <option>-u</> or <option>--user</>,
      but that was inconsistent with other tools.
     </para>
    </listitem>

   </itemizedlist>

  </sect2>

  <sect2>
   <title>Changes</title>

   <para>
    Below you will find a detailed account of the changes between
    <productname>PostgreSQL</productname> 9.4 and the previous major
    release.
   </para>

   <sect3>
    <title>Server</title>

     <itemizedlist>

      <listitem>
       <para>
        Allow <link linkend="bgworker">background worker processes</link>
        to be dynamically registered, started and terminated (Robert Haas)
       </para>

       <para>
        The <filename>contrib/worker_spi</> module shows an example of use
        of this feature.
       </para>
      </listitem>

      <listitem>
       <para>
        Allow dynamic allocation of shared memory segments (Robert Haas,
        Amit Kapila)
       </para>

       <para>
        This feature is illustrated in <xref linkend="test-shm-mq">.
       </para>
      </listitem>

      <listitem>
       <para>
        During crash recovery or immediate shutdown, send uncatchable
        termination signals (<systemitem>SIGKILL</>) to child processes
        that do not shut down promptly (MauMau, &Aacute;lvaro Herrera)
       </para>

       <para>
        This reduces the likelihood of leaving orphaned child processes
        behind after <xref linkend="app-postmaster"> shutdown, as well
        as ensuring that crash recovery can proceed if some child processes
        have become <quote>stuck</>.
       </para>
      </listitem>

      <listitem>
       <para>
        Improve randomness of the database system identifier (Tom Lane)
       </para>
      </listitem>

      <listitem>
       <para>
        Make <xref linkend="SQL-VACUUM"> properly report dead but
        not-yet-removable rows to the statistics collector (Hari Babu)
       </para>

       <para>
        Previously these were reported as live rows.
       </para>
      </listitem>

     </itemizedlist>

    <sect4>
     <title>Indexes</title>

     <itemizedlist>

      <listitem>
       <para>
        Reduce <link linkend="GIN"><acronym>GIN</></link> index size
        (Alexander Korotkov, Heikki Linnakangas)
       </para>

       <para>
        Indexes upgraded via <xref linkend="pgupgrade"> will work fine
        but will still be in the old, larger <acronym>GIN</> format.
        Use <xref linkend="SQL-REINDEX"> to recreate such an index in the
        new format.
       </para>
      </listitem>

      <listitem>
       <para>
        Improve speed of multi-key <link
        linkend="GIN"><acronym>GIN</></link> lookups (Alexander Korotkov,
        Heikki Linnakangas)
       </para>
      </listitem>

      <listitem>
       <para>
        Add <link linkend="GiST"><acronym>GiST</></link> index support
        for <link linkend="datatype-inet"><type>inet</></link> and
        <link linkend="datatype-cidr"><type>cidr</></link> data types
        (Emre Hasegeli)
       </para>

       <para>
        Such indexes improve <link
        linkend="cidr-inet-operators-table">subnet and supernet</link>
        lookups and ordering comparisons.
       </para>
      </listitem>

      <listitem>
       <para>
        Fix rare race condition in B-tree page deletion (Heikki Linnakangas)
       </para>
      </listitem>

      <listitem>
       <para>
        Make the handling of interrupted B-tree page splits more robust
        (Heikki Linnakangas)
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

    <sect4>
     <title>General Performance</title>

     <itemizedlist>

      <listitem>
       <para>
        Allow multiple backends to insert
        into <link linkend="wal"><acronym>WAL</></link> buffers
        concurrently (Heikki Linnakangas)
       </para>

       <para>
        This improves parallel write performance.
       </para>
      </listitem>

      <listitem>
       <para>
        Conditionally write only the modified portion of updated rows to
        <link linkend="wal"><acronym>WAL</></link> (Amit Kapila)
       </para>
      </listitem>

      <listitem>
       <para>
        Improve performance of aggregate functions used as <link
        linkend="syntax-window-functions">window functions</link>
        (David Rowley, Florian Pflug, Tom Lane)
       </para>
      </listitem>

      <listitem>
       <para>
        Improve speed of aggregates that
        use <link linkend="datatype-numeric"><type>numeric</></link> state
        values (Hadi Moshayedi)
       </para>
      </listitem>

      <listitem>
       <para>
        Attempt to <link linkend="vacuum-for-wraparound">freeze</link>
        tuples when tables are rewritten with <xref
        linkend="SQL-CLUSTER"> or <link
        linkend="SQL-VACUUM"><command>VACUUM FULL</></link> (Robert Haas,
        Andres Freund)
       </para>

       <para>
        This can avoid the need to freeze the tuples in the future.
       </para>
      </listitem>

      <listitem>
       <para>
        Improve speed of <xref linkend="SQL-COPY">
        with <literal>DEFAULT</> <link
        linkend="functions-sequence-table"><function>nextval()</></link>
        columns (Simon Riggs)
       </para>
      </listitem>

      <listitem>
       <para>
        Improve speed of accessing many different <link
        linkend="SQL-CREATESEQUENCE">sequences</link> in the same session
        (David Rowley)
       </para>
      </listitem>

      <listitem>
       <para>
        Raise hard limit on the number of tuples held in memory during sorting
        and B-tree index builds (Noah Misch)
       </para>
      </listitem>

      <listitem>
       <para>
        Reduce memory allocated by <application>PL/pgSQL</>
        <xref linkend="SQL-DO"> blocks (Tom Lane)
       </para>
      </listitem>

      <listitem>
       <para>
        Make the planner more aggressive in extracting restriction clauses
        from mixed <literal>AND</>/<literal>OR</> clauses (Tom Lane)
       </para>
      </listitem>

      <listitem>
       <para>
        Disallow pushing volatile <literal>WHERE</> clauses down
        into <literal>DISTINCT</> subqueries (Tom Lane)
       </para>

       <para>
        Pushing down a <literal>WHERE</> clause can produce a more
        efficient plan overall, but at the cost of evaluating the clause
        more often than is implied by the text of the query; so don't do it
        if the clause contains any volatile functions.
       </para>
      </listitem>

      <listitem>
       <para>
        Auto-resize the catalog caches (Heikki Linnakangas)
       </para>

       <para>
        This reduces memory consumption for sessions accessing only a few
        tables, and improves performance for sessions accessing many tables.
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

    <sect4>
     <title>Monitoring</title>

     <itemizedlist>

      <listitem>
       <para>
        Add <xref linkend="pg-stat-archiver-view"> system view to
        report <link linkend="wal"><acronym>WAL</></link> archiver activity
        (Gabriele Bartolini)
       </para>
      </listitem>

      <listitem>
       <para>
        Add <structfield>n_mod_since_analyze</> columns to
        <xref linkend="pg-stat-all-tables-view"> and related system views
        (Mark Kirkwood)
       </para>

       <para>
        These columns expose the system's estimate of the number of changed
        tuples since the table's last <xref linkend="sql-analyze">.  This
        estimate drives decisions about when to auto-analyze.
       </para>
      </listitem>

      <listitem>
       <para>
        Add <structfield>backend_xid</> and <structfield>backend_xmin</>
        columns to the system view <xref linkend="pg-stat-activity-view">,
        and a <structfield>backend_xmin</> column to
        <xref linkend="pg-stat-replication-view"> (Christian Kruse)
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

    <sect4>
     <title><acronym>SSL</></title>

     <itemizedlist>

      <listitem>
       <para>
        Add support for <acronym>SSL</> <acronym>ECDH</> key exchange
        (Marko Kreen)
       </para>

       <para>
        This allows use of Elliptic Curve keys for server authentication.
        Such keys are faster and have better security than <acronym>RSA</>
        keys. The new configuration parameter
        <xref linkend="guc-ssl-ecdh-curve">
        controls which curve is used for <acronym>ECDH</>.
       </para>
      </listitem>

      <listitem>
       <para>
        Improve the default <xref linkend="guc-ssl-ciphers"> setting
        (Marko Kreen)
       </para>
      </listitem>

      <listitem>
       <para>
        By default, the server not the client now controls the preference
        order of <acronym>SSL</> ciphers
        (Marko Kreen)
       </para>

       <para>
        Previously, the order specified by <xref linkend="guc-ssl-ciphers">
        was usually ignored in favor of client-side defaults, which are not
        configurable in most <productname>PostgreSQL</> clients.  If
        desired, the old behavior can be restored via the new configuration
        parameter <xref linkend="guc-ssl-prefer-server-ciphers">.
       </para>
      </listitem>

      <listitem>
       <para>
        Make <xref linkend="guc-log-connections"> show <acronym>SSL</>
        encryption information (Andreas Kunert)
       </para>
      </listitem>

      <listitem>
       <para>
        Improve <acronym>SSL</> renegotiation handling (&Aacute;lvaro
        Herrera)
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

    <sect4>
     <title>Server Settings</title>

     <itemizedlist>

      <listitem>
       <para>
        Add new <acronym>SQL</> command <xref linkend="SQL-ALTERSYSTEM">
        for updating <filename>postgresql.conf</> configuration file entries
        (Amit Kapila)
       </para>

       <para>
        Previously such settings could only be changed by manually
        editing <filename>postgresql.conf</>.
       </para>
      </listitem>

      <listitem>
       <para>
        Add <xref linkend="guc-autovacuum-work-mem"> configuration parameter
        to control the amount of memory used by autovacuum workers
        (Peter Geoghegan)
       </para>
      </listitem>

      <listitem>
       <para>
        Add <xref linkend="guc-huge-pages"> parameter to allow using huge
        memory pages on Linux (Christian Kruse, Richard Poole, Abhijit
        Menon-Sen)
       </para>

       <para>
        This can improve performance on large-memory systems.
       </para>
      </listitem>

      <listitem>
       <para>
        Add <xref linkend="guc-max-worker-processes"> parameter
        to limit the number of background workers (Robert Haas)
       </para>

       <para>
        This is helpful in configuring a standby server to have the
        required number of worker processes (the same as the primary).
       </para>
      </listitem>

      <listitem>
       <para>
        Add superuser-only <xref linkend="guc-session-preload-libraries">
        parameter to load libraries at session start (Peter Eisentraut)
       </para>

       <para>
        In contrast to <xref linkend="guc-local-preload-libraries">, this
        parameter can load any shared library, not just those in
        the <filename>$libdir/plugins</> directory.
       </para>
      </listitem>

      <listitem>
       <para>
        Add <xref linkend="guc-wal-log-hints"> parameter to enable WAL
        logging of hint-bit changes (Sawada Masahiko)
       </para>

       <para>
        Hint bit changes are not normally logged, except when checksums are
        enabled.  This is useful for external tools
        like <application>pg_rewind</>.
       </para>
      </listitem>

      <listitem>
       <para>
        Increase the default settings of <xref linkend="guc-work-mem">
        and <xref linkend="guc-maintenance-work-mem"> by four times (Bruce
        Momjian)
       </para>

       <para>
        The new defaults are 4MB and 64MB respectively.
       </para>
      </listitem>

      <listitem>
       <para>
        Increase the default setting of <xref
        linkend="guc-effective-cache-size">
        to 4GB (Bruce Momjian, Tom Lane)
       </para>
      </listitem>

      <listitem>
       <para>
        Allow <function>printf</function>-style space padding to be
        specified in <xref linkend="guc-log-line-prefix"> (David Rowley)
       </para>
      </listitem>

      <listitem>
       <para>
        Allow terabyte units (<literal>TB</>) to be used when specifying
        configuration variable values (Simon Riggs)
       </para>
      </listitem>

      <listitem>
       <para>
        Show <acronym>PID</>s of lock holders and waiters and improve
        information about relations in <xref linkend="guc-log-lock-waits">
        log messages (Christian Kruse)
       </para>
      </listitem>

      <listitem>
       <para>
        Reduce server logging level when loading shared libraries (Peter
        Geoghegan)
       </para>

       <para>
        The previous level was <literal>LOG</>, which was too verbose
        for per-session libraries.
       </para>
      </listitem>

      <listitem>
       <para>
        On Windows, make <literal>SQL_ASCII</>-encoded databases and server
        processes (e.g., <xref linkend="app-postmaster">) emit messages in
        the character encoding of the server's Windows user locale
        (Alexander Law, Noah Misch)
       </para>

       <para>
        Previously these messages were output in the Windows
        <acronym>ANSI</> code page.
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

   </sect3>

   <sect3>
    <title>Replication and Recovery</title>

     <itemizedlist>

      <listitem>
       <para>
        Add <link linkend="streaming-replication-slots">replication
        slots</link> to coordinate activity on streaming standbys with the
        node they are streaming from (Andres Freund, Robert Haas)
       </para>

       <para>
        Replication slots allow preservation of resources like
        <acronym>WAL</> files on the primary until they are no longer
        needed by standby servers.
       </para>
      </listitem>

      <listitem>
       <para>
        Add <link linkend="recovery-config"><filename>recovery.conf</></link>
        parameter <xref linkend="recovery-min-apply-delay">
        to delay replication (Robert Haas, Fabr&iacute;zio de Royes Mello,
        Simon Riggs)
       </para>

       <para>
        Delaying replay on standby servers can be useful for recovering
        from user errors.
       </para>
      </listitem>

      <listitem>
       <para>
        Add <xref linkend="recovery-target">
        option <option>immediate</> to stop <link
        linkend="wal"><acronym>WAL</></link> recovery as soon as a
        consistent state is reached (MauMau, Heikki Linnakangas)
       </para>
      </listitem>

      <listitem>
       <para>
        Improve recovery target processing (Heikki Linnakangas)
       </para>

       <para>
        The timestamp reported
        by <link linkend="functions-recovery-info-table"><function>pg_last_xact_replay_timestamp()</></link>
        now reflects already-committed records, not transactions about to
        be committed. Recovering to a restore point now replays the restore
        point, rather than stopping just before the restore point.
       </para>
      </listitem>

      <listitem>
       <para>
        <link
        linkend="functions-admin-backup-table"><function>pg_switch_xlog()</></link>
        now clears any unused trailing space in the old <acronym>WAL</> file
        (Heikki Linnakangas)
       </para>

       <para>
        This improves the compression ratio for <acronym>WAL</> files.
       </para>
      </listitem>

      <listitem>
       <para>
        Report failure return codes from <link
        linkend="archive-recovery-settings">external recovery commands</>
        (Peter Eisentraut)
       </para>
      </listitem>

      <listitem>
       <para>
        Reduce spinlock contention during <acronym>WAL</> replay (Heikki
        Linnakangas)
       </para>
      </listitem>

      <listitem>
       <para>
        Write <acronym>WAL</> records of running transactions more
        frequently (Andres Freund)
       </para>

       <para>
        This allows standby servers to start faster and clean up resources
        more aggressively.
       </para>
      </listitem>

     </itemizedlist>

     <sect4>
      <title><link linkend="logicaldecoding">Logical Decoding</></title>

      <para>
       Logical decoding allows database changes to be streamed in a
       configurable format. The data is read from
       the <link linkend="wal"><acronym>WAL</></link> and transformed into the
       desired target format. To implement this feature, the following changes
       were made:
      </para>

      <itemizedlist>

       <listitem>
        <para>
         Add support for <link linkend="logicaldecoding">logical decoding</>
         of WAL data, to allow database changes to be streamed out in a
         customizable format
         (Andres Freund)
        </para>
       </listitem>

       <listitem>
        <para>
         Add new <xref linkend="guc-wal-level"> setting <option>logical</>
         to enable logical change-set encoding in <acronym>WAL</> (Andres
         Freund)
        </para>
       </listitem>

       <listitem>
        <para>
         Add table-level parameter <link
         linkend="catalog-pg-class"><literal>REPLICA IDENTITY</></link>
         to control logical replication (Andres Freund)
        </para>
       </listitem>

       <listitem>
        <para>
         Add relation option <link
         linkend="SQL-CREATETABLE-storage-parameters"><option>user_catalog_table</></link>
         to identify user-created tables involved in logical change-set
         encoding (Andres Freund)
        </para>
       </listitem>

       <listitem>
        <para>
         Add <xref linkend="app-pgrecvlogical"> application to receive
         logical-decoding data (Andres Freund)
        </para>
       </listitem>

       <listitem>
        <para>
         Add <xref linkend="test-decoding"> module to illustrate logical
         decoding at the <acronym>SQL</> level (Andres Freund)
        </para>
       </listitem>

      </itemizedlist>

     </sect4>

   </sect3>

   <sect3>
    <title>Queries</title>

     <itemizedlist>

      <listitem>
       <para>
        Add <link linkend="queries-tablefunctions"><literal>WITH
        ORDINALITY</></link> syntax to number rows returned from
        set-returning functions in the <literal>FROM</> clause
        (Andrew Gierth, David Fetter)
       </para>

       <para>
        This is particularly useful for functions like
        <function>unnest()</>.
       </para>
      </listitem>

      <listitem>
       <para>
        Add <link linkend="queries-tablefunctions"><literal>ROWS
        FROM()</></link> syntax to allow horizontal concatenation of
        set-returning functions in the <literal>FROM</> clause (Andrew Gierth)
       </para>
      </listitem>

      <listitem>
       <para>
        Allow <xref linkend="SQL-SELECT"> to have
        an empty target list (Tom Lane)
       </para>

       <para>
        This was added so that views that select from a table with zero
        columns can be dumped and restored correctly.
       </para>
      </listitem>

     </itemizedlist>

   </sect3>

   <sect3>
    <title>Utility Commands</title>

     <itemizedlist>

      <listitem>
       <para>
        Add <link linkend="SQL-DISCARD"><command>DISCARD
        SEQUENCES</></link> command to discard cached sequence-related state
        (Fabr&iacute;zio de Royes Mello, Robert Haas)
       </para>

       <para>
        <command>DISCARD ALL</> will now also discard such information.
       </para>
      </listitem>

       <listitem>
        <para>
         Add <literal>FORCE NULL</> option
         to <link linkend="SQL-COPY"><command>COPY FROM</></link>, which
         causes quoted strings matching the specified null string to be
         converted to NULLs in <literal>CSV</> mode (Ian Barwick, Michael
         Paquier)
        </para>

        <para>
         Without this option, only unquoted matching strings will be imported
         as null values.
        </para>
       </listitem>

      <listitem>
       <para>
        Issue warnings for commands used outside of transaction blocks
        when they can have no effect (Bruce Momjian)
       </para>

       <para>
        New warnings are issued for <command>SET
        LOCAL</>, <command>SET CONSTRAINTS</>, <command>SET TRANSACTION</> and
        <command>ABORT</> when used outside a transaction block.
       </para>
      </listitem>

     </itemizedlist>

     <sect4>
      <title><xref linkend="SQL-EXPLAIN"></title>

      <itemizedlist>

       <listitem>
        <para>
         Make <command>EXPLAIN ANALYZE</> show planning time (Andreas
         Karlsson)
        </para>
       </listitem>

       <listitem>
        <para>
         Make <command>EXPLAIN</> show the grouping columns in Agg and
         Group nodes (Tom Lane)
        </para>
       </listitem>

       <listitem>
        <para>
         Make <command>EXPLAIN ANALYZE</> show exact and lossy
         block counts in bitmap heap scans (Etsuro Fujita)
        </para>
       </listitem>

      </itemizedlist>

     </sect4>

     <sect4>
      <title>Views</title>

      <itemizedlist>

      <listitem>
       <para>
        Allow <link linkend="rules-materializedviews">materialized views</>
        to be refreshed without blocking reads
        (Kevin Grittner)
       </para>

       <para>
        This is done with <link
        linkend="SQL-REFRESHMATERIALIZEDVIEW"><command>REFRESH MATERIALIZED
        VIEW CONCURRENTLY</></link>.
       </para>
      </listitem>

      <listitem>
       <para>
        Allow views to be <link
        linkend="SQL-CREATEVIEW-updatable-views">automatically
        updated</link> even if they contain some non-updatable columns
        (Dean Rasheed)
       </para>

       <para>
        Previously the presence of non-updatable output columns such as
        expressions, literals, and function calls prevented automatic
        updates.  Now <command>INSERT</>s, <command>UPDATE</>s and
        <command>DELETE</>s are supported, provided that they do not
        attempt to assign new values to any of the non-updatable columns.
       </para>
      </listitem>

      <listitem>
       <para>
        Allow control over whether <command>INSERT</>s and
        <command>UPDATE</>s can add rows to an auto-updatable view that
        would not appear in the view (Dean Rasheed)
       </para>

       <para>
        This is controlled with the new <xref linkend="SQL-CREATEVIEW">
        clause <literal>WITH CHECK OPTION</>.
       </para>
      </listitem>

      <listitem>
       <para>
        Allow <link linkend="rules-privileges">security barrier views</>
        to be automatically updatable (Dean Rasheed)
       </para>
      </listitem>

      </itemizedlist>

     </sect4>

   </sect3>

   <sect3>
    <title>Object Manipulation</title>

     <itemizedlist>

      <listitem>
       <para>
        Support triggers on <link linkend="SQL-CREATEFOREIGNTABLE">foreign
        tables</> (Ronan Dunklau)
       </para>
      </listitem>

      <listitem>
       <para>
        Allow moving groups of objects from one tablespace to another
        using <xref linkend="SQL-ALTERTABLESPACE"> ... <literal>MOVE</>
        (Stephen Frost)
       </para>
      </listitem>

      <listitem>
       <para>
        Allow changing foreign key constraint deferrability
        via <xref linkend="SQL-ALTERTABLE"> ... <literal>ALTER
        CONSTRAINT</> (Simon Riggs)
       </para>
      </listitem>

      <listitem>
       <para>
        Reduce lock strength for some <xref linkend="SQL-ALTERTABLE">
        commands
        (Simon Riggs, Noah Misch, Robert Haas)
       </para>

       <para>
        Specifically, <literal>VALIDATE CONSTRAINT</>, <literal>CLUSTER
        ON</>, <literal>SET WITHOUT CLUSTER</>, <literal>ALTER COLUMN
        SET STATISTICS</>, <literal>ALTER COLUMN</> <literal>SET</>
        <option>(attribute_option)</>, <literal>ALTER COLUMN RESET</>
        <option>(attribute_option)</> no longer require <literal>ACCESS
        EXCLUSIVE</> locks.
       </para>
      </listitem>

      <listitem>
       <para>
        Allow tablespace options to be set
        in <xref linkend="SQL-CREATETABLESPACE"> (Vik Fearing)
       </para>

       <para>
        Formerly these options could only be set
        via <xref linkend="SQL-ALTERTABLESPACE">.
       </para>
      </listitem>

      <listitem>
       <para>
        Allow <xref linkend="SQL-CREATEAGGREGATE"> to define the estimated
        size of the aggregate's transition state data (Hadi Moshayedi)
       </para>

       <para>
        Proper use of this feature allows the planner to better estimate
        how much memory will be used by aggregates.
       </para>
      </listitem>

      <listitem>
       <para>
        Fix <command>DROP IF EXISTS</> to avoid errors for non-existent
        objects in more cases (Pavel Stehule, Dean Rasheed)
       </para>
      </listitem>

      <listitem>
       <para>
        Improve how system relations are identified (Andres Freund,
        Robert Haas)
       </para>

       <para>
        Previously, relations once moved into the <literal>pg_catalog</>
        schema could no longer be modified or dropped.
       </para>
      </listitem>

     </itemizedlist>

   </sect3>

   <sect3>
    <title>Data Types</title>

    <itemizedlist>

      <listitem>
       <para>
        Fully implement the <link
        linkend="datatype-line"><type>line</></link> data type (Peter
        Eisentraut)
       </para>

       <para>
        The line <emphasis>segment</> data type (<link
        linkend="datatype-lseg"><type>lseg</></link>) has always been
        fully supported.  The previous <type>line</> data type (enabled
        only via a compile-time option) is not binary or dump-compatible.
       </para>
      </listitem>

      <listitem>
       <para>
        Add <link linkend="datatype-pg-lsn"><type>pg_lsn</></link>
        data type to represent a <acronym>WAL</> log sequence number
        (<acronym>LSN</>) (Robert Haas, Michael Paquier)
       </para>
      </listitem>

      <listitem>
       <para>
        Allow single-point <link
        linkend="datatype-polygon"><type>polygon</></link>s to be converted
        to <link linkend="datatype-circle"><type>circle</></link>s
        (Bruce Momjian)
       </para>
      </listitem>

      <listitem>
       <para>
        Allow 5+ digit years for non-<acronym>ISO</> <link
        linkend="datatype-datetime"><type>timestamp</></link> and
        <type>date</> strings, where appropriate (Bruce Momjian)
       </para>
      </listitem>

      <listitem>
       <para>
        Add checks for overflow/underflow of <link
        linkend="datatype-datetime"><type>interval</></link> values
        (Bruce Momjian)
       </para>
      </listitem>

    </itemizedlist>

    <sect4>
     <title><link linkend="datatype-json"><acronym>JSON</></link></title>

     <itemizedlist>

       <listitem>
        <para>
         Add <link linkend="datatype-json"><type>jsonb</></link>, a more
         capable and efficient data type for for storing <acronym>JSON</> data
         (Oleg Bartunov, Teodor Sigaev, Alexander
         Korotkov, Peter Geoghegan, Andrew Dunstan)
        </para>

        <para>
         This new type allows faster access to values in a JSON
         document and faster and more useful indexing of JSON columns.
         Scalar values in <type>jsonb</> documents are stored as appropriate
         scalar SQL types, and the JSON document structure is pre-parsed
         rather than being stored as text as in the original <type>json</>
         data type.
        </para>
       </listitem>

       <listitem>
        <para>
         Add new JSON functions to allow for the construction
         of arbitrarily complex JSON trees (Andrew Dunstan, Laurence Rowe)
        </para>

        <para>
         New functions include <link
         linkend="functions-json-processing-table"><function>json_array_elements_text()</></link>,
         <function>json_build_array()</>, <function>json_object()</>,
         <function>json_object_agg()</>, <function>json_to_record()</>,
         and <function>json_to_recordset()</>.
        </para>
       </listitem>

       <listitem>
        <para>
         Add <link
         linkend="functions-json-processing-table"><function>json_typeof()</></link>
         to return the data type of a <type>json</> value (Andrew Tipton)
        </para>
       </listitem>

     </itemizedlist>

    </sect4>

   </sect3>

   <sect3>
    <title>Functions</title>

     <itemizedlist>

      <listitem>
       <para>
        Add <link
        linkend="functions-datetime-delay"><function>pg_sleep_for(interval)</></link>
        and <function>pg_sleep_until(timestamp)</> to specify
        delays more flexibly (Vik Fearing, Julien Rouhaud)
       </para>

       <para>
        The existing <function>pg_sleep()</> function only supports delays
        specified in seconds.
       </para>
      </listitem>

      <listitem>
       <para>
        Add <link
        linkend="array-functions-table"><function>cardinality()</></link>
        function for arrays (Marko Tiikkaja)
       </para>

       <para>
        This returns the total number of elements in the array, or zero
        for an array with no elements.
       </para>
      </listitem>

      <listitem>
       <para>
        Add <acronym>SQL</> functions to allow <link linkend="lo-funcs">large
        object reads/writes</link> at arbitrary offsets (Pavel Stehule)
       </para>
      </listitem>

      <listitem>
       <para>
        Allow <link
        linkend="array-functions-table"><function>unnest()</></link>
        to take multiple arguments, which are individually unnested then
        horizontally concatenated (Andrew Gierth)
       </para>
      </listitem>

      <listitem>
       <para>
        Add functions to construct <type>time</>s, <type>date</>s,
        <type>timestamp</>s, <type>timestamptz</>s, and <type>interval</>s
        from individual values, rather than strings (Pavel Stehule)
       </para>

       <para>
        These functions' names are prefixed with <literal>make_</>,
        e.g. <link linkend="functions-datetime-table"><function>make_date()</></link>.
       </para>
      </listitem>

      <listitem>
       <para>
        Make <link
        linkend="functions-formatting-table"><function>to_char()</></link>'s
        <literal>TZ</> format specifier return a useful value for simple
        numeric time zone offsets (Tom Lane)
       </para>

       <para>
        Previously, <literal>to_char(CURRENT_TIMESTAMP, 'TZ')</> returned
        an empty string if the <literal>timezone</> was set to a constant
        like <literal>-4</>.
       </para>
      </listitem>

      <listitem>
       <para>
        Add timezone offset format specifier <literal>OF</> to <link
        linkend="functions-formatting-table"><function>to_char()</></link>
        (Bruce Momjian)
       </para>
      </listitem>

      <listitem>
       <para>
        Improve the random seed used for <link
        linkend="functions-math-random-table"><function>random()</></link>
        (Honza Horak)
       </para>
      </listitem>

      <listitem>
       <para>
        Tighten validity checking for Unicode code points in <link
        linkend="functions-string-other"><function>chr(int)</></link>
        (Tom Lane)
       </para>

       <para>
        This function now only accepts values that are valid UTF8 characters
        according to RFC 3629.
       </para>
      </listitem>

     </itemizedlist>

    <sect4>
     <title>System Information Functions</title>

     <itemizedlist>

      <listitem>
       <para>
        Add functions for looking up objects in <structname>pg_class</>,
        <structname>pg_proc</>, <structname>pg_type</>, and
        <structname>pg_operator</> that do not generate errors for
        non-existent objects (Yugo Nagata, Nozomi Anzai,
        Robert Haas)
       </para>

       <para>
        For example, <link
        linkend="functions-info-catalog-table"><function>to_regclass()</></link>
        does a lookup in <structname>pg_class</> similarly to
        the <type>regclass</> input function, but it returns NULL for a
        non-existent object instead of failing.
       </para>
      </listitem>

      <listitem>
       <para>
        Add function <link
        linkend="functions-admin-dblocation"><function>pg_filenode_relation()</></link>
        to allow for more efficient lookup of relation names from filenodes
        (Andres Freund)
       </para>
      </listitem>

      <listitem>
       <para>
        Add <structfield>parameter_default</> column to <link
        linkend="infoschema-parameters"><structname>information_schema.parameters</></link>
        view (Peter Eisentraut)
       </para>
      </listitem>

      <listitem>
       <para>
        Make <link
        linkend="infoschema-schemata"><structname>information_schema.schemata</></link>
        show all accessible schemas (Peter Eisentraut)
       </para>

       <para>
        Previously it only showed schemas owned by the current user.
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

    <sect4>
     <title>Aggregates</title>

     <itemizedlist>

      <listitem>
       <para>
        Add control over which rows are passed
        into aggregate functions using the <link
        linkend="syntax-aggregates"><literal>FILTER</></link> clause
        (David Fetter)
       </para>
      </listitem>

      <listitem>
       <para>
        Support ordered-set (<link
        linkend="syntax-aggregates"><literal>WITHIN GROUP</></link>)
        aggregates (Atri Sharma, Andrew Gierth, Tom Lane)
       </para>
      </listitem>

      <listitem>
       <para>
        Add aggregates <link
        linkend="functions-orderedset-table"><function>percentile_cont()</></link>,
        <function>percentile_disc()</>, <function>mode()</>, <link
        linkend="functions-hypothetical-table"><function>rank()</></link>,
        <function>dense_rank()</>, <function>percent_rank()</>, and
        <function>cume_dist()</>
        (Atri Sharma, Andrew Gierth)
       </para>
      </listitem>

      <listitem>
       <para>
        Support <link
        linkend="xfunc-sql-variadic-functions"><literal>VARIADIC</></link>
        aggregate functions (Tom Lane)
       </para>
      </listitem>

      <listitem>
       <para>
        Allow polymorphic aggregates to have non-polymorphic state data
        types (Tom Lane)
       </para>
       <para>
        This allows proper declaration of aggregates like the built-in
        aggregate <function>array_agg()</> in SQL.
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

   </sect3>

   <sect3>
    <title>Server-Side Languages</title>

    <itemizedlist>

     <listitem>
      <para>
       Add event trigger support to <link linkend="plperl">PL/Perl</>
       and <link linkend="pltcl">PL/Tcl</> (Dimitri Fontaine)
      </para>
     </listitem>

     <listitem>
      <para>
       Convert <link linkend="datatype-numeric"><type>numeric</></link>
       values to <type>decimal</> in <link linkend="plpython">PL/Python</link>
       (Szymon Guz, Ronan Dunklau)
      </para>

      <para>
       Previously such values were converted to Python <type>float</> values,
       risking loss of precision.
      </para>
     </listitem>

    </itemizedlist>

    <sect4>
     <title><link linkend="plpgsql">PL/pgSQL</link> Server-Side Language</title>

     <itemizedlist>

      <listitem>
       <para>
        Add ability to store the PL/PgSQL
        call stack into a variable using <link
        linkend="plpgsql-get-diagnostics-context"><literal>PG_CONTEXT</></link>
        (Pavel Stehule, Stephen Frost)
       </para>
      </listitem>

      <listitem>
       <para>
        Add option <link
        linkend="plpgsql-statements-assignment"><option>print_strict_params</></link>
        to output parameters passed to queries generating <link
        linkend="plpgsql-statements-sql-onerow"><literal>STRICT</></link>
        errors (Marko Tiikkaja)
       </para>
      </listitem>

      <listitem>
       <para>
        Add variables <link
        linkend="plpgsql-extra-checks"><varname>plpgsql.extra_warnings</></link>
        and <varname>plpgsql.extra_errors</> to enable additional PL/pgSQL
        warnings and errors (Marko Tiikkaja, Petr Jelinek)
       </para>

       <para>
        Currently only shadowed variable errors/warnings are available.
       </para>
      </listitem>

    </itemizedlist>

    </sect4>

   </sect3>

   <sect3>
    <title><link linkend="libpq"><application>libpq</></link></title>
     <itemizedlist>

      <listitem>
       <para>
        Add libpq function <link
        linkend="libpq-pqhostaddr"><function>PQhostaddr()</></link>
        to return the server's <acronym>IP</> address (Fujii Masao)
       </para>
      </listitem>

      <listitem>
       <para>
        Make libpq's <link
        linkend="libpq-pqconndefaults"><function>PQconndefaults()</></link>
        function ignore invalid service files (Steve Singer, Bruce Momjian)
       </para>

       <para>
        Previously it returned NULL if an incorrect service file was
        encountered.
       </para>
      </listitem>

      <listitem>
       <para>
        Accept <acronym>TLS</> protocol versions beyond <literal>TLSv1</>
        in libpq (Marko Kreen)
       </para>
      </listitem>

     </itemizedlist>

   </sect3>

   <sect3>
    <title>Client Applications</title>

    <itemizedlist>

      <listitem>
       <para>
        Add <xref linkend="APP-CREATEUSER"> option <option>-g</>
        to specify role membership (Chistopher Browne)
       </para>
      </listitem>

      <listitem>
       <para>
        Add <xref linkend="APP-VACUUMDB">
        option <option>--analyze-in-stages</> to analyze in stages of
        increasing granularity (Peter Eisentraut)
       </para>

       <para>
        This allows minimal statistics to be created quickly.
       </para>
      </listitem>

      <listitem>
       <para>
        Make <xref linkend="APP-PGRESETXLOG"> with option <option>-n</>
        output current and potentially changed values (Rajeev Rastogi)
       </para>
      </listitem>

      <listitem>
       <para>
        Make <xref linkend="app-initdb"> throw error for incorrect locale
        settings, rather than silently falling back to a default choice
        (Tom Lane)
       </para>
      </listitem>

      <listitem>
       <para>
        Make <xref linkend="app-pg-ctl"> return exit code <literal>4</> for
        an inaccessible data directory (Amit Kapila, Bruce Momjian)
       </para>

       <para>
        This behavior more closely matches the Linux Standard Base
        (<acronym>LSB</>) Core Specification.
       </para>
      </listitem>

      <listitem>
       <para>
        On Windows, ensure that a non-absolute <option>-D</> path
        specification is interpreted relative
        to <xref linkend="app-pg-ctl">'s current directory
        (Kumar Rajeev Rastogi)
       </para>

       <para>
        Previously it would be interpreted relative to whichever directory
        the underlying Windows service was started in.
       </para>
      </listitem>

      <listitem>
       <para>
        Allow <function>sizeof()</> in <link linkend="ecpg">ECPG</link>
        C array definitions (Michael Meskes)
       </para>
      </listitem>

      <listitem>
       <para>
        Make <link linkend="ecpg">ECPG</link> properly handle nesting
        of C-style comments in both C and <acronym>SQL</> text
        (Michael Meskes)
       </para>
      </listitem>

    </itemizedlist>

    <sect4>
     <title><xref linkend="APP-PSQL"></title>

     <itemizedlist>

      <listitem>
       <para>
        Add ability to wrap long lines in <application>psql</>'s
        <literal>expanded</> mode by using <command>\pset format wrapped</>
        (Sergey Muraviov)
       </para>
      </listitem>

      <listitem>
       <para>
        Suppress <quote>No rows</quote> output in <application>psql</> <link
        linkend="APP-PSQL-meta-commands"><option>expanded</></link>
        mode when the footer is disabled (Bruce Momjian)
       </para>
      </listitem>

      <listitem>
       <para>
        Allow Control-C to abort <application>psql</> when hung at connection
        startup (Peter Eisentraut)
       </para>
      </listitem>

     </itemizedlist>

     <sect5>
      <title><link linkend="APP-PSQL-meta-commands">Backslash Commands</link></title>

      <itemizedlist>

       <listitem>
        <para>
         Make <application>psql</> <command>\db+</> show tablespace options
         (Magnus Hagander)
        </para>
       </listitem>

       <listitem>
        <para>
         Make <application>psql</> <command>\do+</> display the functions
         which implement the operators (Marko Tiikkaja)
        </para>
       </listitem>

       <listitem>
        <para>
         Make <application>psql</> <command>\d+</> output an
         <literal>OID</> line only if an <literal>oid</literal> column
         exists in a table (Bruce Momjian)
        </para>

        <para>
         Previously, the presence or absence of an <literal>oid</literal>
         column was always reported.
        </para>
       </listitem>

       <listitem>
        <para>
         Make <command>\d</> show disabled system triggers (Bruce
         Momjian)
        </para>

        <para>
         Previously, if you disabled all triggers, only user triggers
         would show as disabled.
        </para>
       </listitem>

       <listitem>
        <para>
         Fix <application>psql</> <command>\copy</> to no longer require
         a space between <literal>stdin</> and a semicolon (Etsuro Fujita)
        </para>
       </listitem>

       <listitem>
        <para>
         Output the row count at the end
         of <application>psql</> <command>\copy</> just
         like <xref linkend="SQL-COPY"> (Kumar Rajeev Rastogi)
        </para>
       </listitem>

       <listitem>
        <para>
         Fix <application>psql</> <command>\conninfo</> to display the
         server's <acronym>IP</> address for clients that connect using
         <literal>hostaddr</> (Fujii Masao)
        </para>

        <para>
         Previously <command>\conninfo</> could not display the server's
         <acronym>IP</> address in such cases.
        </para>
       </listitem>

       <listitem>
        <para>
         Mention the <acronym>SSL</> protocol version in
         <application>psql</>'s <command>\conninfo</> (Marko Kreen)
        </para>
       </listitem>

       <listitem>
        <para>
         Add <application>psql</> tab completion for <command>\pset</>
         (Pavel Stehule)
        </para>
       </listitem>

       <listitem>
        <para>
         Allow <application>psql</> <command>\pset</> with no arguments
         to show all settings (Gilles Darold)
        </para>
       </listitem>

       <listitem>
        <para>
         In <application>psql</>, display the history file name written by
         <command>\s</> without converting it to an absolute path (Tom Lane)
        </para>

        <para>
         The code previously attempted to convert a relative file name to
         an absolute path for display, but frequently got it wrong.
        </para>
       </listitem>

      </itemizedlist>

     </sect5>

    </sect4>

    <sect4>
     <title><xref linkend="APP-PGDUMP"></title>

     <itemizedlist>

      <listitem>
       <para>
        Allow <xref linkend="APP-PGRESTORE"> options
        <option>-I</>, <option>-P</>, <option>-T</> and <option>-n</>
        to be specified multiple times (Heikki Linnakangas)
       </para>

       <para>
        This allows multiple objects to be restored in one operation.
       </para>
      </listitem>

      <listitem>
       <para>
        Optionally add <literal>IF EXISTS</> clauses to the <command>DROP</>
        commands emitted when removing old objects during a restore (Pavel
        Stehule)
       </para>

       <para>
        This change prevents unnecessary errors when removing old objects.
        The new <option>--if-exists</> option
        for <xref linkend="APP-PGDUMP">, <xref linkend="APP-PG-DUMPALL">,
        and <xref linkend="APP-PGRESTORE"> is only available
        when <option>--clean</> is also specified.
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

    <sect4>
     <title><xref linkend="app-pgbasebackup"></title>

     <itemizedlist>

      <listitem>
       <para>
        Add <application>pg_basebackup</> option <option>--xlogdir</>
        to specify the <filename>pg_xlog</> directory location (Haribabu
        Kommi)
       </para>
      </listitem>

      <listitem>
       <para>
        Allow <application>pg_basebackup</> to relocate tablespaces in
        the backup copy (Steeve Lennmark)
       </para>

       <para>
        This is particularly useful for using <application>pg_basebackup</>
        on the same machine as the primary.
       </para>
      </listitem>

      <listitem>
       <para>
        Allow network-stream base backups to be throttled (Antonin Houska)
       </para>

       <para>
        This can be controlled with the <application>pg_basebackup</>
        <option>--max-rate</> parameter.
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

   </sect3>

   <sect3>
    <title>Source Code</title>

     <itemizedlist>

      <listitem>
       <para>
        Improve the way tuples are frozen to preserve forensic information
        (Robert Haas, Andres Freund)
       </para>

       <para>
        This change removes the main objection to freezing tuples as soon
        as possible.  Code that inspects tuple flag bits will need to be
        modified.
       </para>
      </listitem>

      <listitem>
       <para>
        No longer require function prototypes for functions marked with the
        <link linkend="xfunc-c"><function>PG_FUNCTION_INFO_V1</></link>
        macro (Peter Eisentraut)
       </para>

       <para>
        This change eliminates the need to write boilerplate prototypes.
        Note that the <function>PG_FUNCTION_INFO_V1</> macro must appear
        before the corresponding function definition to avoid compiler
        warnings.
       </para>
      </listitem>

      <listitem>
       <para>
        Remove <varname>SnapshotNow</> and
        <function>HeapTupleSatisfiesNow()</> (Robert Haas)
       </para>

       <para>
        All existing uses have been switched to more appropriate snapshot
        types.  Catalog scans now use <acronym>MVCC</> snapshots.
       </para>
      </listitem>

      <listitem>
       <para>
        Add an <acronym>API</> to allow memory allocations over one gigabyte
        (Noah Misch)
       </para>
      </listitem>

      <listitem>
       <para>
        Add <function>psprintf()</> to simplify memory allocation during
        string composition (Peter Eisentraut, Tom Lane)
       </para>
      </listitem>

      <listitem>
       <para>
        Support <function>printf()</> size modifier <literal>z</> to
        print <type>size_t</> values (Andres Freund)
       </para>
      </listitem>

      <listitem>
       <para>
        Change <acronym>API</> of <function>appendStringInfoVA()</>
        to better use <function>vsnprintf()</> (David Rowley, Tom Lane)
       </para>
      </listitem>

      <listitem>
       <para>
        Allow new types of external toast datums to be created (Andres
        Freund)
       </para>
      </listitem>

      <listitem>
       <para>
        Add single-reader, single-writer, lightweight shared message queue
        (Robert Haas)
       </para>
      </listitem>

      <listitem>
       <para>
        Improve spinlock speed on x86_64 <acronym>CPU</>s (Heikki
        Linnakangas)
       </para>
      </listitem>

      <listitem>
       <para>
        Remove spinlock support for unsupported platforms
        <productname>SINIX</>, <productname>Sun3</>, and
        <productname>NS32K</> (Robert Haas)
       </para>
      </listitem>

      <listitem>
       <para>
        Remove <acronym>IRIX</> port (Robert Haas)
       </para>
      </listitem>

      <listitem>
       <para>
        Reduce the number of semaphores required by
        <option>--disable-spinlocks</> builds (Robert Haas)
       </para>
      </listitem>

      <listitem>
       <para>
        Rewrite <application>duplicate_oids</> Unix shell script in
        <application>Perl</> (Andrew Dunstan)
       </para>
      </listitem>

      <listitem>
       <para>
        Add Test Anything Protocol (<acronym>TAP</>) tests for client
        programs (Peter Eisentraut)
       </para>
      </listitem>

      <listitem>
       <para>
        Add make targets <option>check-tests</> and
        <option>installcheck-tests</>, which allow selection of individual
        tests to be run (Andrew Dunstan)
       </para>
      </listitem>

      <listitem>
       <para>
        Remove <option>maintainer-check</> makefile rule (Peter Eisentraut)
       </para>

       <para>
        The default build rules now include all the formerly-optional tests.
       </para>
      </listitem>

      <listitem>
       <para>
        Improve support for <envar>VPATH</> builds of <acronym>PGXS</>
        modules  (C&eacute;dric Villemain, Andrew Dunstan)
       </para>
      </listitem>

      <listitem>
       <para>
        Upgrade to Autoconf 2.69 (Peter Eisentraut)
       </para>
      </listitem>

      <listitem>
       <para>
        Add a <application>configure</> flag that appends custom text to the
        <envar>PG_VERSION</> string (Oskari Saarenmaa)
       </para>

       <para>
        This is useful for packagers building custom binaries.
       </para>
      </listitem>

      <listitem>
       <para>
        Improve DocBook <acronym>XML</> validity (Peter Eisentraut)
       </para>
      </listitem>

      <listitem>
       <para>
        Various minor security and sanity fixes reported by the
        <productname>Coverity</> scanner (Stephen Frost)
       </para>
      </listitem>

      <listitem>
       <para>
        Improve <application>Valgrind</> detection of invalid memory usage
        (Noah Misch)
       </para>
      </listitem>

      <listitem>
       <para>
        Improve sample <application>Emacs</> configuration file
        <filename>emacs.samples</> (Peter Eisentraut)
       </para>

       <para>
        Also add <filename>.dir-locals.el</> to the top of the source tree.
       </para>
      </listitem>

      <listitem>
       <para>
        Allow <application>pgindent</> to accept a command-line list
        of typedefs (Bruce Momjian)
       </para>

       <para>
        <application>pgindent</> is also now smarter about blank lines
        around preprocessor conditionals.
       </para>
      </listitem>

      <listitem>
       <para>
        Avoid most uses of <command>dlltool</command>
        in <productname>Cygwin</> and
        <productname>Mingw</> builds (Marco Atzeri, Hiroshi Inoue)
       </para>
      </listitem>

      <listitem>
       <para>
        Support client-only installs in <acronym>MSVC</> (Windows) builds
        (MauMau)
       </para>
      </listitem>

     </itemizedlist>

   </sect3>

   <sect3>
    <title>Additional Modules</title>

    <itemizedlist>

      <listitem>
       <para>
        Add <xref linkend="pgprewarm"> extension to preload relation data
        into the shared buffer cache at server start (Robert Haas)
       </para>

       <para>
        This allows reaching full operating performance more quickly.
       </para>
      </listitem>

      <listitem>
       <para>
        Add <acronym>UUID</> random number generator
        <function>gen_random_uuid()</> to <xref linkend="pgcrypto">
        (Oskari Saarenmaa)
       </para>

       <para>
        This allows creation of version 4 <acronym>UUID</>s without
        requiring installation of <xref linkend="uuid-ossp">.
       </para>
      </listitem>

      <listitem>
       <para>
        Allow <xref linkend="uuid-ossp"> to work with
        the <systemitem>BSD</> or <systemitem>e2fsprogs</> UUID libraries,
        not only the <systemitem>OSSP</> UUID library (Matteo Beccati)
       </para>

       <para>
        This improves the <application>uuid-ossp</> module's portability
        since it no longer has to have the increasingly-obsolete OSSP
        library.  The module's name is now rather a misnomer, but we won't
        change it.
       </para>
      </listitem>

      <listitem>
       <para>
        Add option to <xref linkend="auto-explain"> to include trigger
        execution time (Horiguchi Kyotaro)
       </para>
      </listitem>

      <listitem>
       <para>
        Fix <xref linkend="pgstattuple"> to not report rows from
        uncommitted transactions as dead (Robert Haas)
       </para>
      </listitem>

      <listitem>
       <para>
        Make <xref linkend="pgstattuple"> functions
        use <type>regclass</type>-type arguments (Satoshi Nagayasu)
       </para>

       <para>
        While <type>text</type>-type arguments are still supported, they
        may be removed in a future major release.
       </para>
      </listitem>

      <listitem>
       <para>
        Improve consistency of <xref linkend="pgrowlocks"> output to honor
        snapshot rules more consistently (Robert Haas)
       </para>
      </listitem>

      <listitem>
       <para>
        Improve <xref linkend="pgtrgm">'s choice of trigrams for indexed
        regular expression searches by discouraging the selection of
        trigrams containing whitespace (Alexander Korotkov)
       </para>
      </listitem>

      <listitem>
       <para>
        Allow <xref linkend="pgxlogdump"> to report a live log stream
        with <option>--follow</> (Heikki Linnakangas)
       </para>
      </listitem>

      <listitem>
       <para>
        Store <xref linkend="cube"> data more compactly (Stas Kelvich)
       </para>

       <para>
        Existing data must be dumped/restored to use the new format.
        The old format can still be read.
       </para>
      </listitem>

      <listitem>
       <para>
        Reduce <xref linkend="vacuumlo"> client-side memory usage by using
        a cursor (Andrew Dunstan)
       </para>
      </listitem>

      <listitem>
       <para>
        Dramatically reduce memory consumption
        in <xref linkend="pgupgrade"> (Bruce Momjian)
       </para>
      </listitem>

      <listitem>
       <para>
        Pass <xref linkend="pgupgrade"> user name (<option>-U</> option) to
        analyze scripts (Bruce Momjian)
       </para>
      </listitem>

    </itemizedlist>

    <sect4>
     <title><xref linkend="pgbench"></title>

     <itemizedlist>

      <listitem>
       <para>
        Remove line length limit for <application>pgbench</> scripts (Sawada
        Masahiko)
       </para>

       <para>
        The previous line limit was <envar>BUFSIZ</>.
       </para>
      </listitem>

      <listitem>
       <para>
        Add long option names to <application>pgbench</> (Fabien Coelho)
       </para>
      </listitem>

      <listitem>
       <para>
        Add <application>pgbench</> option <option>--rate</> to control
        the transaction rate (Fabien Coelho)
       </para>
      </listitem>

      <listitem>
       <para>
        Add <application>pgbench</> option <option>--progress</> to
        print periodic progress reports
        (Fabien Coelho)
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

    <sect4>
     <title><xref linkend="pgstatstatements"></title>

     <itemizedlist>

      <listitem>
       <para>
        Make <application>pg_stat_statements</> use a file, rather than
        shared memory, for query text storage (Peter Geoghegan)
       </para>

       <para>
        This removes the previous limitation on query text length, and
        allows a higher number of unique statements to be tracked by default.
       </para>
      </listitem>

      <listitem>
       <para>
        Allow reporting of <application>pg_stat_statements</>'s internal
        query hash identifier (Daniel Farina, Sameer Thakur, Peter
        Geoghegan)
       </para>
      </listitem>

      <listitem>
       <para>
        Add the ability to retrieve all <application>pg_stat_statements</>
        information except the query text (Peter Geoghegan)
       </para>

       <para>
        This allows monitoring tools to only fetch query text for newly
        created entries, improving performance for repeated querying of the
        statistics.
       </para>
      </listitem>

      <listitem>
       <para>
        Save the statistics file into <filename>$PGDATA/pg_stat</> at server
        shutdown, rather than <filename>$PGDATA/global</> (Fujii Masao)
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

   </sect3>

  </sect2>
 </sect1>