path: root/doc/src/sgml/release-9.6.sgml

<!-- doc/src/sgml/release-9.6.sgml -->
<!-- See header comment in release.sgml about typical markup -->

 <sect1 id="release-9-6">
  <title>Release 9.6</title>

  <note>
   <title>Release Date</title>
   <simpara>2016-??-??</simpara>
   <simpara>Current as of 2016-08-07 (commit 19322c0a7)</simpara>
  </note>

  <sect2>
   <title>Overview</title>

   <para>
    Major enhancements in <productname>PostgreSQL</> 9.6 include:
   </para>

   <!-- Items in this list summarize one or more items below -->

   <itemizedlist>

    <listitem>
     <para>
      Parallel sequential scans, joins and aggregates
     </para>
    </listitem>

    <listitem>
     <para>
      Elimination of repetitive scanning of old data by autovacuum
     </para>
    </listitem>

    <listitem>
     <para>
      Synchronous replication now allows multiple standby servers for
      increased reliability
     </para>
    </listitem>

    <listitem>
     <para>
      Full-text search for phrases
     </para>
    </listitem>

    <listitem>
     <para>
      Support for remote joins, sorts, and updates
      in <filename>postgres_fdw</>
     </para>
    </listitem>

    <listitem>
     <para>
      Substantial performance improvements, especially in the area of
      improving scalability on many-CPU servers
     </para>
    </listitem>

   </itemizedlist>

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

  </sect2>

  <sect2>

  <title>Migration to Version 9.6</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 9.6 contains a number of changes that may affect compatibility
    with previous releases.  Observe the following incompatibilities:
   </para>

   <itemizedlist>

    <listitem>
     <para>
      To be filled in, but issues are called out in relevant sections below
     </para>
    </listitem>

   </itemizedlist>

  </sect2>

  <sect2>
   <title>Changes</title>

   <para>
    Below you will find a detailed account of the changes between
    <productname>PostgreSQL</productname> 9.6 and the previous major
    release.
   </para>

   <sect3>
    <title>Server</title>

    <sect4>
     <title>Parallel Queries</title>

     <itemizedlist>

      <listitem>
<!--
2015-09-18 [4a4e6893a] Glue layer to connect the executor to the shm_mq mechani
2015-09-23 [a0d9f6e43] Add readfuncs.c support for plan nodes.
2015-09-28 [d1b7c1ffe] Parallel executor support.
2015-09-30 [3bd909b22] Add a Gather executor node.
2015-10-16 [ee7ca559f] Add a C API for parallel heap scans.
2015-10-16 [bfc78d719] Rewrite interaction of parallel mode with parallel execu
2015-10-16 [a53c06a13] Prohibit parallel query when the isolation level is seri
2015-11-02 [1efc7e538] Fix problems with ParamListInfo serialization mechanism.
2015-11-06 [6e71dd7ce] Modify tqueue infrastructure to support transient record
2015-11-11 [f0661c4e8] Make sequential scans parallel-aware.
2015-11-11 [80558c1f5] Generate parallel sequential scan plans in simple cases.
2015-12-09 [b287df70e] Allow EXPLAIN (ANALYZE, VERBOSE) to display per-worker s
2016-01-20 [45be99f8c] Support parallel joins, and make related improvements.
2016-02-03 [69d34408e] Allow parallel custom and foreign scans.
2016-02-07 [a1c1af2a1] Introduce group locking to prevent parallel processes fr
2016-02-07 [7c944bd90] Introduce a new GUC force_parallel_mode for testing purp
2016-02-25 [57a6a72b6] Enable parallelism for prepared statements and extended
2016-02-26 [7bea19d0a] On second thought, disable parallelism for prepared stat
2016-03-21 [e06a38965] Support parallel aggregation.
2016-04-05 [11c8669c0] Add parallel query support functions for assorted aggreg
2016-04-08 [25fe8b5f1] Add a 'parallel_degree' reloption.
2016-04-27 [59eb55127] Fix EXPLAIN VERBOSE output for parallel aggregate.
2016-06-09 [c9ce4a1c6] Eliminate "parallel degree" terminology.
2016-06-16 [75be66464] Invent min_parallel_relation_size GUC to replace a hard-
-->
       <para>
        Parallel queries (Robert Haas, Amit Kapila, David Rowley, many others)
       </para>

       <para>
        With 9.6, <productname>PostgreSQL</> introduces initial support for
        parallel execution of large queries.  Only strictly read-only queries
        where the driving table is accessed via a sequential scan can be
        parallelized.  Hash joins and nested loops can be performed in
        parallel, as can aggregation (for supported aggregates).  Much
        remains to be done, but this is already a useful set of features.
       </para>

       <para>
        Use of parallel query execution can be controlled through the new
        configuration parameters
        <xref linkend="guc-max-parallel-workers-per-gather">,
        <xref linkend="guc-force-parallel-mode">,
        <xref linkend="guc-parallel-setup-cost">,
        <xref linkend="guc-parallel-tuple-cost">, and
        <xref linkend="guc-min-parallel-relation-size">.
       </para>
      </listitem>

      <listitem>
<!--
2015-09-16 [7aea8e4f2] Determine whether it's safe to attempt a parallel plan f
-->
       <para>
        Provide infrastructure for marking functions as parallel-safe or not
        (Robert Haas, Amit Kapila)
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

    <sect4>
     <title>Indexes</title>

     <itemizedlist>

      <listitem>
<!--
2015-09-02 [30bb26b5e] Allow usage of huge maintenance_work_mem for GIN build.
-->
       <para>
        Allow GIN index builds to make effective use
        of <varname>maintenance_work_mem</> settings larger than 1GB
        (Robert Abraham, Teodor Sigaev)
       </para>
      </listitem>

      <listitem>
<!--
2015-09-07 [e95680832] Add pages deleted from pending list to FSM
2015-09-23 [dc943ad95] Allow autoanalyze to add pages deleted from pending list
-->
       <para>
        Add pages deleted from a GIN index's pending list to the free space
        map immediately, to reduce bloat if the table isn't vacuumed often
        (Jeff Janes, Teodor Sigaev)
       </para>
      </listitem>

      <listitem>
<!--
2016-01-28 [7f46eaf03] Add gin_clean_pending_list function to clean up GIN pend
-->
       <para>
        Add <function>gin_clean_pending_list()</> function to allow manual
        invocation of pending-list cleanup for a GIN index, separately from
        vacuuming or analyzing the parent table (Jeff Janes)
       </para>
      </listitem>

      <listitem>
<!--
2015-09-09 [013ebc0a7] Microvacuum for GIST
2015-09-17 [22f519c92] Fix bug introduced by microvacuum for GiST
-->
       <para>
        Improve handling of dead index tuples in GiST indexes
        (Anastasia Lubennikova)
       </para>

       <para>
        Dead index tuples are now marked as such when an index scan notices
        that the corresponding heap tuple is dead; and when inserting tuples,
        marked-dead tuples will be removed if needed to make space on the page.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-30 [acdf2a8b3] Introduce SP-GiST operator class over box.
-->
       <para>
        Add an SP-GiST operator class for type <type>box</>
        (Alexander Lebedev)
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

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

     <itemizedlist>

      <listitem>
<!--
2016-03-01 [a892234f8] Change the format of the VM fork to add a second bit per
2016-03-08 [77a1d1e79] Department of second thoughts: remove PD_ALL_FROZEN.
2016-03-10 [fd31cd265] Don't vacuum all-frozen pages.
2016-03-11 [7087166a8] pg_upgrade: Convert old visibility map format to new for
2016-06-17 [ede62e56f] Add VACUUM (DISABLE_PAGE_SKIPPING) for emergencies.
2016-07-18 [eca0f1db1] Clear all-frozen visibilitymap status when locking tuple
2016-08-04 [e7caacf73] Fix hard to hit race condition in heapam's tuple locking
-->
       <para>
        Avoid re-vacuuming pages containing only frozen tuples
        (Masahiko Sawada, Robert Haas, Andres Freund)
       </para>

       <para>
        Formerly, an anti-wraparound vacuum had to visit every page of a
        table whether or not there was anything to do there.  Now, pages
        containing only already-frozen tuples are identified in the table's
        visibility map, and can be skipped by vacuum even when it's doing
        transaction wraparound prevention.  This should greatly reduce the
        cost of maintaining large tables containing mostly-unchanging data.
       </para>

       <para>
        If necessary, vacuum can be forced to process all-frozen pages
        using its new <literal>DISABLE_PAGE_SKIPPING</> option.  This
        should never be needed normally, but it might help for example
        in recovering from visibility-map corruption.
       </para>
      </listitem>

      <listitem>
<!--
2015-12-30 [e84290823] Avoid useless truncation attempts during VACUUM.
-->
       <para>
        Avoid useless heap-truncation attempts during <command>VACUUM</>
        (Jeff Janes, Tom Lane)
       </para>

       <para>
        This change avoids taking an exclusive table lock in some cases where
        no truncation is really possible.  The main benefit comes from
        avoiding unnecessary query cancellations on standby servers.
       </para>
      </listitem>

      <listitem>
<!--
2016-01-09 [687f2cd7a] Avoid pin scan for replay of XLOG_BTREE_VACUUM
2016-04-03 [3e4b7d879] Avoid pin scan for replay of XLOG_BTREE_VACUUM in all ca
-->
       <para>
        Reduce interlocking on standby servers during replay of btree index
        vacuuming operations (Simon Riggs)
       </para>

       <para>
        This change avoids substantial replication delays that sometimes
        occurred while replaying such operations.
       </para>
      </listitem>

      <listitem>
<!--
2016-02-11 [d4c3a156c] Remove GROUP BY columns that are functionally dependent
-->
       <para>
        Drop entries from <literal>GROUP BY</> if they are functionally
        dependent on other entries (David Rowley)
       </para>

       <para>
        If a <literal>GROUP BY</> clause includes all columns of a
        non-deferred primary key, as well as other columns of the same
        relation, those other columns are redundant and can be dropped from
        the grouping.  This saves computation in many common cases.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-11 [9118d03a8] When appropriate, postpone SELECT output expressions til
2016-03-25 [d543170f2] Don't split up SRFs when choosing to postpone SELECT out
-->
       <para>
        When appropriate, postpone evaluation of <command>SELECT</> output
        expressions till after <literal>ORDER BY</> sorting
        (Konstantin Knizhnik)
       </para>

       <para>
        This change ensures that volatile or expensive functions in the
        output list are executed in the order suggested by <literal>ORDER
        BY</>, and that they are not evaluated more times than required when
        there's a <literal>LIMIT</>.  Previously, these properties held if
        the ordering was performed by an index scan or pre-merge-join sort,
        but not if it was performed by a top-level sort step.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-10 [428b1d6b2] Allow to trigger kernel writeback after a configurable n
2016-04-13 [fa11a09fe] Fix assorted portability issues with using msync() for d
2016-04-24 [8f91d87d4] Fix documentation & config inconsistencies around 428b1d
2016-04-26 [72a98a639] Don't open formally non-existent segments in _mdfd_getse
2016-05-04 [a71248708] Fix transient mdsync() errors of truncated relations due
2016-02-16 [7975c5e0a] Allow the WAL writer to flush WAL at a reduced rate.
2016-06-10 [4bc0f165c] Change default of backend_flush_after GUC to 0 (disabled
-->
       <para>
        Where feasible, trigger kernel writeback after a configurable number
        of writes, to prevent accumulation of dirty data in kernel disk
        buffers (Fabien Coelho, Andres Freund)
       </para>

       <para>
        <productname>PostgreSQL</> writes data to the kernel's disk cache,
        from where it should be flushed to physical storage in due time.
        Many operating systems are not very smart about managing this, and
        will allow large amounts of dirty data to accumulate then decide to
        flush it all at once, leading to long delays for new I/O requests.
        This change attempts to alleviate this problem by explicitly
        requesting data flushes after a configurable interval.
       </para>

       <para>
        On Linux, <function>sync_file_range()</> is used for this purpose,
        and the feature is on by default because that function has few
        downsides.  The feature is also available on other platforms that
        have <function>msync()</> or <function>posix_fadvise()</>, but those
        interfaces have some undesirable side-effects so the feature is not
        enabled by default on other platforms.
       </para>

       <para>
        The new configuration parameters
        <xref linkend="guc-backend-flush-after">,
        <xref linkend="guc-bgwriter-flush-after">,
        <xref linkend="guc-checkpoint-flush-after">, and
        <xref linkend="guc-wal-writer-flush-after"> control this behavior.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-10 [9cd00c457] Checkpoint sorting and balancing.
-->
       <para>
        Perform checkpoint writes in sorted order
        (Fabien Coelho, Andres Freund)
       </para>

       <para>
        Previously, checkpoints wrote out dirty pages in whatever order they
        happen to appear in within shared buffers, which usually is nearly
        random.  That performs poorly, especially on rotating media.  This
        change causes checkpoint-driven writes to be done in order by file
        and block number, and to be balanced across tablespaces.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-08 [848ef42bb] Add the "snapshot too old" feature
2016-05-06 [2cc41acd8] Fix hash index vs "snapshot too old" problemms
2016-05-06 [7e3da1c47] Mitigate "snapshot too old" performance regression on NU
2016-08-03 [3e2f3c2e4] Prevent "snapshot too old" from trying to return pruned
2016-08-07 [9ee1cf04a] Fix TOAST access failure in RETURNING queries.
-->
       <para>
        Allow old MVCC snapshots to be invalidated after a configurable
        timeout (Kevin Grittner)
       </para>

       <para>
        Normally, deleted tuples cannot be physically removed by vacuuming
        until the last transaction that could <quote>see</> them is gone.
        A transaction that stays open for a long time can thus cause
        considerable table bloat because space cannot be recycled.  This
        feature allows setting a time-based limit, via the new configuration
        parameter <xref linkend="guc-old-snapshot-threshold">, on how long an
        MVCC snapshot is guaranteed valid.  After that, dead tuples are
        candidates for removal.  A transaction using an outdated snapshot
        will get an error, but only if it attempts to read a page that's been
        modified recently enough that it might have contained such data.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-31 [f9aefcb91] Support using index-only scans with partial indexes in m
-->
       <para>
        Allow using an <link linkend="indexes-index-only-scans">index-only
        scan</link> with a partial index when the index's predicate involves
        column(s) not stored in the index (Tomas Vondra, Kyotaro Horiguchi)
       </para>

       <para>
        An index-only scan is now allowed if the query mentions such columns
        only in <literal>WHERE</> clauses that match the index predicate.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-08 [137805f89] Use Foreign Key relationships to infer multi-column join
2016-06-07 [77ba61080] Revert "Use Foreign Key relationships to infer multi-col
2016-06-18 [100340e2d] Restore foreign-key-aware estimation of join relation si
-->
       <para>
        Use foreign key relationships to infer selectivity for
        join predicates (Tomas Vondra, David Rowley)
       </para>

       <para>
        If a table <literal>t</> has a foreign key restriction,
        say <literal>(a,b) REFERENCES r (x,y)</>, then a <literal>WHERE</>
        condition such as <literal>t.a = r.x AND t.b = r.y</> cannot select
        more than one <literal>r</> row per <literal>t</> row.  The planner
        formerly considered the AND'ed conditions to be independent and would
        often drastically misestimate the selectivity as a result.  Now it
        compares the <literal>WHERE</> conditions to applicable foreign key
        constraints and arrives at a better estimate.
       </para>
      </listitem>

      <listitem>
<!--
2015-08-04 [804163bc2] Share transition state between different aggregates when
-->
       <para>
        Improve aggregate-function performance by sharing calculations across
        multiple aggregates if they have the same arguments and transition
        functions (David Rowley)
       </para>
      </listitem>

      <listitem>
<!--
2015-08-26 [8a7d07018] Speed up HeapTupleSatisfiesMVCC() by replacing the XID-i
-->
       <para>
        Speed up visibility tests for recently-created tuples by checking
        our transaction snapshot, not <structname>pg_clog</>, to decide if the
        source transaction should be considered committed (Jeff Janes, Tom
        Lane)
       </para>
      </listitem>

      <listitem>
<!--
2016-02-15 [db76b1efb] Allow SetHintBits() to succeed if the buffer's LSN is ne
-->
       <para>
        Allow tuple hint bits to be set sooner than before (Andres Freund)
       </para>
      </listitem>

      <listitem>
<!--
2016-01-20 [978b2f65a] Speedup 2PC by skipping two phase state files in normal
2016-03-10 [e0694cf9c] Reduce size of two phase file header
-->
       <para>
        Improve performance of short-lived prepared transactions
        (Stas Kelvich, Simon Riggs, Pavan Deolasee)
       </para>

       <para>
        Two-phase commit information is now written only to WAL
        during <command>PREPARE TRANSACTION</>, and read back from there
        during <command>COMMIT PREPARED</>.  A separate state file is created
        only if the pending transaction does not get committed or aborted by
        the time of the next checkpoint.
       </para>
      </listitem>

      <listitem>
<!--
2015-12-08 [25c539233] Improve performance in freeing memory contexts
-->
       <para>
        Improve performance of memory context destruction (Jan Wieck)
       </para>
      </listitem>

      <listitem>
<!--
2016-01-26 [cc988fbb0] Improve ResourceOwners' behavior for large numbers of ow
-->
       <para>
        Improve performance of resource owners with many tracked objects
        (Aleksander Alekseev)
       </para>
      </listitem>

      <listitem>
<!--
2016-02-06 [aa2387e2f] Improve speed of timestamp/time/date output functions.
-->
       <para>
        Improve speed of the output functions for timestamps, times, and dates
        (David Rowley, Andres Freund)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-10 [37c54863c] Rework wait for AccessExclusiveLocks on Hot Standby
-->
       <para>
        Avoid some unnecessary cancellations of hot-standby queries during
        replay of actions that take AccessExclusiveLocks (Jeff Janes)
       </para>
      </listitem>

      <listitem>
<!--
2016-04-01 [be4b4dc75] Omit null rows when applying the Haas-Stokes estimator f
2016-04-01 [3d3bf62f3] Omit null rows when setting the threshold for what's a m
2016-04-04 [391159e03] Partially revert commit 3d3bf62f30200500637b24fdb7b992a9
-->
       <para>
        Improve <command>ANALYZE</>'s estimates for columns with many nulls
        (Tomas Vondra, Alex Shulgin)
       </para>

       <para>
        Previously the code tended to underestimate the number of non-null
        distinct values in a column with many nulls, and it also might make
        poor decisions about what is a most-common value.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-04 [84f9a35e3] Improve estimate of distinct values in estimate_num_grou
-->
       <para>
        Improve planner's estimate of the number of distinct values in a
        query result (Tomas Vondra)
       </para>
      </listitem>

      <listitem>
<!--
2016-04-08 [719c84c1b] Extend relations multiple blocks at a time to improve sc
-->
       <para>
        Extend relations multiple blocks at a time, when there is contention
        for the relation's extension lock (Dilip Kumar)
       </para>

       <para>
        This improves scalability by decreasing contention.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-08 [071180377] Use quicksort, not replacement selection, for external s
2016-03-17 [0011c0091] Improve memory management for external sorts.
-->
       <para>
        Improve sorting performance by using quicksort, not replacement
        selection, within steps of an external sort (Peter Geoghegan)
       </para>

       <para>
        The new approach makes better use of CPU cache for typical cache
        sizes and data volumes.  Where necessary, the behavior can be
        adjusted via the new configuration
        parameter <xref linkend="guc-replacement-sort-tuples">,
        which see for further details.
       </para>
      </listitem>

      <listitem>
<!--
2015-10-09 [0e57b4d8b] Speed up text sorts where the same strings occur multipl
2015-10-20 [5be94a9eb] Be a bit more rigorous about how we cache strcoll and st
-->
       <para>
        Speed up text sorts where the same strings occur multiple times
        (Peter Geoghegan)
       </para>
      </listitem>

      <listitem>
<!--
2015-11-06 [a76ef15d9] Add sort support routine for the UUID data type.
2016-02-03 [b47b4dbf6] Extend sortsupport for text to more opclasses.
2016-02-17 [f1f5ec1ef] Reuse abbreviated keys in ordered [set] aggregates.
-->
       <para>
        Speed up sorting of <type>uuid</>, <type>bytea</>,
        and <type>char(n)</> fields by using <quote>abbreviated</> keys
        (Peter Geoghegan)
       </para>

       <para>
        Support for abbreviated keys has also been added to the non-default
        operator classes <literal>text_pattern_ops</>,
        <literal>varchar_pattern_ops</>, and <literal>bpchar_pattern_ops</>.
        Processing of ordered-set aggregates can also now exploit
        abbreviated keys.
       </para>
      </listitem>

      <listitem>
<!--
2015-12-16 [b648b7034] Speed up CREATE INDEX CONCURRENTLY's TID sort.
-->
       <para>
        Speed up <command>CREATE INDEX CONCURRENTLY</> by treating TIDs
        as 64-bit integers during the sort phase (Peter Geoghegan)
       </para>
      </listitem>

      <listitem>
<!--
2016-04-08 [5364b357f] Increase maximum number of clog buffers.
-->
       <para>
        Increase the number of clog buffers for better scalability
        (Amit Kapila, Andres Freund)
       </para>
      </listitem>

      <listitem>
<!--
2015-08-06 [0e141c0fb] Reduce ProcArrayLock contention by removing backends in
2015-09-03 [4aec49899] Assorted code review for recent ProcArrayLock patch.
-->
       <para>
        Reduce contention for the ProcArrayLock (Amit Kapila, Robert Haas)
       </para>
      </listitem>

      <listitem>
<!--
2015-12-15 [6150a1b08] Move buffer I/O and content LWLocks out of the main tran
-->
       <para>
        Improve performance by moving buffer content locks into the buffer
        descriptors (Andres Freund, Simon Riggs)
       </para>
      </listitem>

      <listitem>
<!--
2016-04-10 [48354581a] Allow Pin/UnpinBuffer to operate in a lockfree manner.
-->
       <para>
        Replace shared-buffer header spinlocks with atomic operations
        to improve scalability (Alexander Korotkov, Andres Freund)
       </para>
      </listitem>

      <listitem>
<!--
2016-04-10 [008608b9d] Avoid the use of a separate spinlock to protect a LWLock
-->
       <para>
        Use atomic operations, rather than a spinlock, to protect an LWLock's
        wait queue (Andres Freund)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-23 [44ca4022f] Partition the freelist for shared dynahash tables.
-->
       <para>
        Partition the freelist for shared hash tables, to reduce contention
        on many-CPU servers (Aleksander Alekseev)
       </para>
      </listitem>

      <listitem>
<!--
2015-07-05 [6c82d8d1f] Further reduce overhead for passing plpgsql variables to
-->
       <para>
        Speed up expression evaluation in <application>PL/pgSQL</> by keeping
        ParamListInfo entries for simple variables valid at all times
        (Tom Lane)
       </para>
      </listitem>

      <listitem>
<!--
2015-07-06 [4f33621f3] Don't set SO_SNDBUF on recent Windows versions that have
-->
       <para>
        Avoid reducing the <literal>SO_SNDBUF</> setting below its default on
        recent Windows versions (Chen Huajun)
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

    <sect4>
     <title>Monitoring</title>

     <itemizedlist>

      <listitem>
<!--
2016-03-10 [53be0b1ad] Provide much better wait information in pg_stat_activity
-->
       <para>
        Improve the <link
        linkend="pg-stat-activity-view"><structname>pg_stat_activity</></link>
        view's information about what a process is waiting for (Amit Kapila,
        Ildus Kurbangaliev)
       </para>

       <para>
        Historically a process has only been shown as waiting if it was
        waiting for a heavyweight lock.  Now waits for lightweight locks
        and buffer pins are also shown in <structname>pg_stat_activity</>.
        Also, the type of lock being waited for is now visible.
        These changes replace the <structfield>waiting</> column
        with <structfield>wait_event_type</> and <structfield>wait_event</>.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-15 [c16dc1aca] Add simple VACUUM progress reporting.
-->
       <para>
        Add <link linkend="pg-stat-progress-vacuum-view"><structname>pg_stat_progress_vacuum</></link>
        system view to provide progress reporting for <command>VACUUM</>
        operations (Amit Langote, Robert Haas, Vinayak Pokale, Rahila Syed)
       </para>
      </listitem>

      <listitem>
<!--
2016-02-17 [a5c43b886] Add new system view, pg_config
-->
       <para>
        Add <link linkend="view-pg-config"><structname>pg_config</></link>
        system view (Joe Conway)
       </para>

       <para>
        This view exposes the same information available from
        the <application>pg_config</> utility, namely assorted compile-time
        configuration information for <productname>PostgreSQL</>.
       </para>
      </listitem>

      <listitem>
<!--
2015-08-10 [3f811c2d6] Add confirmed_flush column to pg_replication_slots.
-->
       <para>
        Add a <structfield>confirmed_flush_lsn</> column to
        the <link linkend="view-pg-replication-slots"><structname>pg_replication_slots</></link>
        system view (Marko Tiikkaja)
       </para>
      </listitem>

      <listitem>
<!--
2016-01-07 [b1a9bad9e] pgstat: add WAL receiver status view & SRF
2016-06-29 [9ed551e0a] Add conninfo to pg_stat_wal_receiver
2016-07-07 [60d50769b] Rename pg_stat_wal_receiver.conn_info to conninfo.
-->
       <para>
        Add <link linkend="pg-stat-wal-receiver-view"><structname>pg_stat_wal_receiver</></link>
        system view to provide information about the state of a hot-standby
        server's WAL receiver process (Michael Paquier)
       </para>
      </listitem>

      <listitem>
<!--
2016-02-22 [52f5d578d] Create a function to reliably identify which sessions bl
-->
       <para>
        Add <function>pg_blocking_pids()</> function to reliably identify
        which sessions block which others (Tom Lane)
       </para>

       <para>
        This function returns an array of the process IDs of any sessions that
        are blocking the session with the given process ID.  Historically
        users have obtained such information using a self-join on
        the <structname>pg_locks</> view; but it's unreasonably tedious to do
        it that way with any modicum of correctness, and the addition of
        parallel queries has made the approach entirely impractical, since
        locks might be held or awaited by child worker processes rather than
        the session's main process.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-05 [dc7d70ea0] Expose control file data via SQL accessible functions.
-->
       <para>
        Add <function>pg_control_system()</>,
        <function>pg_control_checkpoint()</>,
        <function>pg_control_recovery()</>,
        and <function>pg_control_init()</> functions to expose fields
        of <filename>pg_control</> to SQL (Joe Conway, Michael Paquier)
       </para>
      </listitem>

      <listitem>
<!--
2016-01-12 [e63bb4549] Add new user fn pg_current_xlog_flush_location()
-->
       <para>
        Add function <function>pg_current_xlog_flush_location()</> to expose
        the current transaction log flush location (Tomas Vondra)
       </para>
      </listitem>

      <listitem>
<!--
2015-07-17 [a04bb65f7] Add new function pg_notification_queue_usage.
-->
       <para>
        Add function <function>pg_notification_queue_usage()</> to report
        how full the <command>NOTIFY</> queue is (Brendan Jurd)
       </para>
      </listitem>

      <listitem>
<!--
2015-08-25 [7b5ef8f2d] Limit the verbosity of memory context statistics dumps.
-->
       <para>
        Limit the verbosity of memory context statistics dumps (Tom Lane)
       </para>

       <para>
        The memory usage dump printed to the postmaster log during an
        out-of-memory failure now summarizes statistics when there are a large
        number of memory contexts, rather than possibly printing a very large
        report.  There's also a <quote>grand total</> summary line now.
       </para>
      </listitem>

      <listitem>
<!--
2016-02-12 [59a884e98] Change delimiter used for display of NextXID
-->
       <para>
        Change display format for NextXID in <application>pg_controldata</>
        and related places (Joe Conway, Bruce Momjian)
       </para>

       <para>
        Display epoch-and-transaction-ID values in the
        format <replaceable>number</><literal>:</><replaceable>number</>.
        The previous
        format <replaceable>number</><literal>/</><replaceable>number</> was
        confusingly similar to that used for LSNs.
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

    <sect4>
     <title><acronym>Authentication</></title>

     <itemizedlist>

      <listitem>
<!--
2016-04-08 [34c33a1f0] Add BSD authentication method.
-->
       <para>
        Create a <literal>bsd</> authentication method to allow use of
        the BSD Authentication service for <productname>PostgreSQL</>
        client authentication (Marisa Emerson)
       </para>

       <para>
        BSD Authentication is currently only available on OpenBSD.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-08 [2f1d2b7a7] Set PAM_RHOST item for PAM authentication
-->
       <para>
        When using PAM authentication, provide the client IP address or host
        name to PAM modules via the <literal>PAM_RHOST</> item (Grzegorz
        Sampolski)
       </para>
      </listitem>

      <listitem>
<!--
2016-01-07 [5e0b5dcab] Provide more detail in postmaster log for password authe
-->
       <para>
        Provide detail in the postmaster log for more password authentication
        failures (Tom Lane)
       </para>

       <para>
        All ordinarily-reachable password authentication failure cases should
        now provide specific <literal>DETAIL</> fields in the log.
       </para>
      </listitem>

      <listitem>
<!--
2015-09-06 [643beffe8] Support RADIUS passwords up to 128 characters
-->
       <para>
        Support RADIUS passwords up to 128 characters long (Marko Tiikkaja)
       </para>
      </listitem>

      <listitem>
<!--
2016-04-08 [35e2e357c] Add authentication parameters compat_realm and upn_usena
-->
       <para>
        Add new SSPI authentication parameters <varname>compat_realm</>
        and <varname>upn_username</>, to control whether NetBIOS or Kerberos
        realm names and user names are used during SSPI authentication
        (Christian Ullrich)
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

    <sect4>
     <title>Server Configuration</title>

     <itemizedlist>

      <listitem>
<!--
2016-02-02 [7d17e683f] Add support for systemd service notifications
-->
       <para>
        Add <option>--with-systemd</> configure switch to enable calling
        <function>sd_notify()</> at server start and stop
        (Peter Eisentraut)
       </para>

       <para>
        This allows the use of <application>systemd</> service units of
        type <literal>notify</>, which greatly simplifies management
        of <productname>PostgreSQL</> under <application>systemd</>.
       </para>
      </listitem>

      <listitem>
<!--
2015-09-08 [1aba62ec6] Allow per-tablespace effective_io_concurrency
-->
       <para>
        Allow <varname>effective_io_concurrency</> to be set as a
        tablespace parameter, to support cases where different tablespaces
        have different I/O characteristics (Julien Rouhaud)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-16 [c6dda1f48] Add idle_in_transaction_session_timeout.
-->
       <para>
        Allow sessions to be terminated automatically if they sit too long in
        an idle-in-transaction state (Vik Fearing)
       </para>

       <para>
        This behavior is enabled and controlled by the new configuration
        parameter <xref linkend="guc-idle-in-transaction-session-timeout">.
        It can be useful to prevent forgotten transactions from holding onto
        locks or preventing vacuum cleanup for very long periods.
       </para>
      </listitem>

      <listitem>
<!--
2015-09-07 [f828654e1] Add log_line_prefix option 'n' for Unix epoch.
2015-09-07 [b1e1862a1] Coordinate log_line_prefix options 'm' and 'n' to share
-->
       <para>
        Add <varname>log_line_prefix</> option <literal>%n</> to print the
        time as a Unix epoch, with milliseconds (Tomas Vondra, Jeff Davis)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-16 [f4c454e9b] Add syslog_sequence_numbers parameter
2016-03-16 [fc201dfd9] Add syslog_split_messages parameter
-->
       <para>
        Add <xref linkend="guc-syslog-sequence-numbers">
        and <xref linkend="guc-syslog-split-messages">
        configuration parameters to provide more control over message format
        when logging to <systemitem>syslog</> (Peter Eisentraut)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-18 [b555ed810] Merge wal_level "archive" and "hot_standby" into new nam
-->
       <para>
        Merge the <literal>archive</> and <literal>hot_standby</> values of
        the <xref linkend="guc-wal-level"> configuration parameter into a
        single value <literal>replica</> (Peter Eisentraut)
       </para>

       <para>
        Making a distinction between these settings no longer appears to be a
        good idea, and it's in the way of planned future simplification of
        replication setup.  The old names are still accepted but are
        converted internally.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-19 [9a83564c5] Allow SSL server key file to have group read access if o
-->
       <para>
        Allow the server's SSL key file to have group read access if owned by
        root (Christoph Berg)
       </para>

       <para>
        Formerly we insisted on the key file being owned by the user running
        the <productname>PostgreSQL</> server, but that is inconvenient for
        some systems (such as Debian) that wish to manage certificates
        centrally.  So also allow the case where the key file is owned by
        root and has group read access.  It's up to the root admin to ensure
        that such a group doesn't include any untrusted users.
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

    <sect4>
     <title>Reliability</title>

     <itemizedlist>

      <listitem>
<!--
2015-11-12 [ac1d7945f] Make idle backends exit if the postmaster dies.
-->
       <para>
        Force backends to exit if the postmaster dies
        (Rajeev Rastogi, Robert Haas)
       </para>

       <para>
        Under normal circumstances the postmaster should always outlive its
        child processes.  If for some reason it dies, cause backend sessions
        to exit with an error.  Formerly, existing backends would continue to
        run until their client disconnects; but that is unsafe and
        inefficient, and furthermore it prevents a new postmaster from being
        started until the last old backend is gone.  Backends will detect
        postmaster death when waiting for client I/O, so the exit will not be
        instantaneous, but in most circumstances it should happen no later
        than the end of the current query.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-07 [fcff8a575] Detect SSI conflicts before reporting constraint violati
-->
       <para>
        Check for serializability conflicts before reporting
        constraint-violation failures (Thomas Munro)
       </para>

       <para>
        When using serializable transaction isolation, it is desirable that
        any error due to a concurrent transaction should manifest as a
        serialization failure, thereby cueing the application that a retry
        might succeed.  Unfortunately, this doesn't reliably happen for
        duplicate-key failures caused by concurrent insertions.  This change
        ensures that such an error will be reported as a serialization error,
        if the application explicitly checked for the presence of a
        conflicting key (and didn't find it) earlier in the transaction.
       </para>
      </listitem>

      <listitem>
<!--
XXX this is pending backpatch, may need to remove
2016-04-26 [c6ff84b06] Emit invalidations to standby for transactions without x
-->
       <para>
        Ensure that invalidation messages are recorded in WAL even when
        issued by a transaction that has no XID assigned (Andres Freund)
       </para>

       <para>
        This fixes some corner cases in which transactions on standby
        servers failed to notice changes such as new indexes.
       </para>
      </listitem>

      <listitem>
<!--
XXX this is pending backpatch, may need to remove
2016-04-28 [e2c79e14d] Prevent multiple cleanup process for pending list in GIN
-->
       <para>
        Prevent multiple processes from trying to clean a GIN index's pending
        list concurrently (Teodor Sigaev, Jeff Janes)
       </para>

       <para>
        This had been intentionally allowed, but it causes race conditions
        that can result in vacuum missing index entries it needs to delete.
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

   </sect3>

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

     <itemizedlist>

      <listitem>
<!--
2016-04-06 [989be0810] Support multiple synchronous standby servers.
2016-04-27 [4c804fbdf] Clean up parsing of synchronous_standby_names GUC variab
2016-04-30 [17d5db352] Remove warning about num_sync being too large in synchro
-->
       <para>
        Support synchronous replication with multiple synchronous standby
        servers, not just one (Masahiko Sawada, Beena Emerson, Michael
        Paquier, Fujii Masao, Kyotaro Horiguchi)
       </para>

       <para>
        The number of standby servers that must acknowledge a commit before
        it's considered done is now configurable as part of the
        <xref linkend="guc-synchronous-standby-names"> parameter.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-29 [314cbfc5d] Add new replication mode synchronous_commit = 'remote_ap
-->
       <para>
        Add new setting <literal>remote_apply</> for configuration parameter
        <xref linkend="guc-synchronous-commit"> (Thomas Munro)
       </para>

       <para>
        In this mode, the master waits for the transaction to be applied on
        the standby server, not just written to disk.  That means that you
        can count on a transaction started on the standby to see all commits
        previously acknowledged by the master.
       </para>
      </listitem>

      <listitem>
<!--
2015-08-11 [6fcd88511] Allow pg_create_physical_replication_slot() to reserve W
2015-09-06 [c314ead5b] Add ability to reserve WAL upon slot creation via replic
-->
       <para>
        Add a feature to the replication protocol, and a corresponding option
        to the <function>pg_create_physical_replication_slot()</> function,
        to allow reserving WAL immediately when creating a replication slot
        (Gurjeet Singh, Michael Paquier)
       </para>

       <para>
        This allows creation of a replication slot to guarantee that all the
        WAL for a base backup will be available afterwards.
       </para>
      </listitem>

      <listitem>
<!--
2015-07-28 [0dc848b03] pg_basebackup: Add -slot option
-->
       <para>
        Add a <option>--slot</option> option to <application>pg_basebackup</>
        (Peter Eisentraut)
       </para>

       <para>
        This lets <application>pg_basebackup</> use a replication slot defined
        for WAL streaming.  After the base backup completes, selecting the
        same slot for regular streaming replication allows seamless startup of
        the new standby server.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-05 [711768546] Implement backup API functions for non-exclusive backups
2016-07-11 [87d84d67b] Fix start WAL filename for concurrent backups from stand
-->
       <para>
        Extend <function>pg_start_backup()</> and <function>pg_stop_backup()</>
        to support non-exclusive backups (Magnus Hagander)
       </para>
      </listitem>

     </itemizedlist>

   </sect3>

   <sect3>
    <title>Queries</title>

     <itemizedlist>

      <listitem>
<!--
2016-07-26 [d8411a6c8] Allow functions that return sets of tuples to return sim
-->
       <para>
        Allow functions that return sets of tuples to return simple NULLs
        (Andrew Gierth, Tom Lane)
       </para>

       <para>
        In the context of <literal>SELECT FROM function(...)</>, a function
        that returned a set of composite values was previously not allowed
        to return a plain NULL value as part of the set.  Now that is
        allowed and interpreted as a row of NULLs.  This avoids corner-case
        errors with, for example, unnesting an array of composite values.
       </para>
      </listitem>

      <listitem>
<!--
2016-08-03 [a3c7a993d] Make INSERT-from-multiple-VALUES-rows handle targetlist
-->
       <para>
        Fully support array subscripts and field selections in the target
        column list of an <command>INSERT</> with
        multiple <literal>VALUES</> rows (Tom Lane)
       </para>

       <para>
        Previously, such cases failed if the same target column was
        mentioned more than once, e.g., <literal>INSERT INTO tab (x[1],
        x[2]) VALUES ...</>.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-12 [23a27b039] Widen query numbers-of-tuples-processed counters to uint
2016-03-14 [74a379b98] Use repalloc_huge() to enlarge a SPITupleTable's tuple p
-->
       <para>
        Widen tuples-processed counters to 64 bits (Andreas Scherbaum)
       </para>

       <para>
        This change allows command tags for <command>SELECT</> etc. to
        correctly report tuple counts larger than 4 billion.  So will
        PL/pgSQL's <command>GET DIAGNOSTICS ... ROW_COUNT</> command.
       </para>
      </listitem>

      <listitem>
<!--
2015-11-28 [8d32717b6] Avoid doing encoding conversions by double-conversion vi
-->
       <para>
        Avoid doing encoding conversions by double-conversion
        through <literal>MULE_INTERNAL</> encoding (Tom Lane)
       </para>

       <para>
        Previously, many conversions for Cyrillic and Central European
        single-byte encodings were done by converting to a
        related <literal>MULE_INTERNAL</> coding scheme and then to the
        destination encoding.  Aside from being inefficient, this meant that
        when the conversion encountered an untranslatable character, the error
        message would confusingly complain about failure to convert to or
        from <literal>MULE_INTERNAL</>, rather than the user-visible
        encodings.
       </para>
      </listitem>

      <listitem>
<!--
2016-01-28 [fbe5a3fb7] Only try to push down foreign joins if the user mapping
2016-07-15 [45639a052] Avoid invalidating all foreign-join cached plans when us
-->
       <para>
        Consider performing joins of foreign tables remotely only when the
        tables will be accessed under the same role ID (Shigeru Hanada,
        Ashutosh Bapat, Etsuro Fujita)
       </para>

       <para>
        Previously, the foreign join pushdown infrastructure left the question
        of security entirely up to individual foreign data wrappers, but it
        would be easy for an FDW to inadvertently open up subtle security
        holes that way.  So, make it the core code's job to determine which
        role ID will access each table, and don't attempt join pushdown unless
        it's the same for all relevant relations.
       </para>
      </listitem>

     </itemizedlist>

   </sect3>

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

     <itemizedlist>

      <listitem>
<!--
2015-11-27 [92e38182d] COPY (INSERT/UPDATE/DELETE .. RETURNING ..)
-->
       <para>
        Allow <command>COPY</> to copy the output of
        an <literal>INSERT</>/<literal>UPDATE</>/<literal>DELETE</>
        ... <literal>RETURNING</> query (Marko Tiikkaja)
       </para>

       <para>
        Previously, an intermediate CTE had to be written to get this result.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-05 [f2fcad27d] Support ALTER THING .. DEPENDS ON EXTENSION
-->
       <para>
        Introduce <command>ALTER <replaceable>object</> DEPENDS ON
        EXTENSION</command> (Abhijit Menon-Sen)
       </para>

       <para>
        This command allows a database object to be marked as depending on an
        extension, so that it will automatically go away if the extension is
        dropped (without needing <literal>CASCADE</>).  But the object is not
        part of the extension, and thus for example will be dumped separately
        by <application>pg_dump</>.
       </para>
      </listitem>

      <listitem>
<!--
2015-11-19 [bc4996e61] Make ALTER .. SET SCHEMA do nothing, instead of throwing
-->
       <para>
        Make <command>ALTER <replaceable>object</> SET SCHEMA</> do nothing
        when the object is already in the requested schema, rather than
        throwing an error as it historically has for most object types (Marti
        Raudsepp)
       </para>
      </listitem>

      <listitem>
<!--
2015-07-14 [321eed5f0] Add ALTER OPERATOR command, for changing selectivity est
2015-12-31 [0dab5ef39] Fix ALTER OPERATOR to update dependencies properly.
-->
       <para>
        Add options to <command>ALTER OPERATOR</command> to change the
        selectivity functions associated with an existing operator
        (Yury Zhuravlev)
       </para>
      </listitem>

      <listitem>
<!--
2015-07-29 [2cd40adb8] Add IF NOT EXISTS processing to ALTER TABLE ADD COLUMN
-->
       <para>
        Add an <option>IF NOT EXISTS</> option to <command>ALTER TABLE ADD
        COLUMN</> (Fabr&iacute;zio de Royes Mello)
       </para>
      </listitem>

      <listitem>
<!--
2015-08-14 [47167b790] Reduce lock levels for ALTER TABLE SET autovacuum storag
2016-03-10 [fcb4bfddb] Reduce lock level for altering fillfactor
-->
       <para>
        Reduce the lock strength needed by <command>ALTER TABLE</> when
        setting fillfactor and autovacuum-related relation options
        (Fabr&iacute;zio de Royes Mello, Simon Riggs)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-23 [473b93287] Support CREATE ACCESS METHOD
-->
       <para>
        Introduce <command>CREATE ACCESS METHOD</> to allow extensions to
        create index access methods (Alexander Korotkov, Petr Jel&iacute;nek)
       </para>
      </listitem>

      <listitem>
<!--
2015-10-03 [b67aaf21e] Add CASCADE support for CREATE EXTENSION.
-->
       <para>
        Add a <literal>CASCADE</> option to <command>CREATE
        EXTENSION</command>, to automatically create extensions it depends on
        (Petr Jel&iacute;nek)
       </para>
      </listitem>

      <listitem>
<!--
2015-10-22 [d371bebd3] Remove redundant CREATEUSER/NOCREATEUSER options in CREA
-->
       <para>
        Remove the
        long-deprecated <literal>CREATEUSER</>/<literal>NOCREATEUSER</>
        options from <command>CREATE ROLE</> and allied commands (Tom Lane)
       </para>

       <para>
        <literal>CREATEUSER</> actually meant <literal>SUPERUSER</>, for
        ancient backwards-compatibility reasons.  That's been a constant
        source of confusion for people who (reasonably) expect it to
        mean <literal>CREATEROLE</>.  It's been deprecated for ten years
        now, so fix the problem by removing it.
       </para>
      </listitem>

      <listitem>
<!--
2015-10-05 [b943f502b] Have CREATE TABLE LIKE add OID column if any LIKEd table
-->
       <para>
        Make <command>CREATE TABLE ... LIKE</> include an OID column if any
        source table has one (Bruce Momjian)
       </para>
      </listitem>

      <listitem>
<!--
2015-12-16 [f27a6b15e] Mark CHECK constraints declared NOT VALID valid if creat
-->
       <para>
        If a <literal>CHECK</> constraint is declared <literal>NOT VALID</> in
        a table creation command, automatically mark it valid (Amit Langote,
        Amul Sul)
       </para>

       <para>
        This matches the longstanding behavior of <literal>FOREIGN KEY</>
        constraints.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-25 [c94959d41] Fix DROP OPERATOR to reset oprcom/oprnegate links to the
-->
       <para>
        Fix <command>DROP OPERATOR</> to
        clear <structname>pg_operator</>.<structfield>oprcom</>
        and <structname>pg_operator</>.<structfield>oprnegate</> links to the
        dropped operator (Roma Sokolov)
       </para>

       <para>
        Formerly such links were left as-is, which could pose a problem in
        the somewhat unlikely event that the dropped operator's OID was
        reused for another operator.
       </para>
      </listitem>

      <listitem>
<!--
2016-07-11 [4d042999f] Print a given subplan only once in EXPLAIN.
-->
       <para>
        Do not show the same subplan twice in <command>EXPLAIN</> output
        (Tom Lane)
       </para>

       <para>
        In certain cases, typically involving SubPlan nodes in index
        conditions, <command>EXPLAIN</> would print data for the same
        subplan twice.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-16 [c34df8a00] Disallow creation of indexes on system columns (except f
-->
       <para>
        Disallow creation of indexes on system columns, except for OID
        (David Rowley)
       </para>

       <para>
        Such indexes were never considered supported, and would very possibly
        misbehave since the system might change the system-column fields of a
        tuple without updating indexes.  But there was no error check to
        prevent them from being created.
       </para>
      </listitem>

     </itemizedlist>

   </sect3>

   <sect3>
    <title>Permissions Management</title>

    <itemizedlist>

      <listitem>
<!--
2016-04-06 [1574783b4] Use GRANT system to manage access to sensitive functions
-->
       <para>
        Use the privilege system to manage access to sensitive functions
        (Stephen Frost)
       </para>

       <para>
        Formerly, many security-sensitive functions contained hard-wired
        checks that would throw an error if they were called by a
        non-superuser role.  This forced use of superuser roles for some
        relatively pedestrian tasks.  The hard-wired error checks are now gone
        in favor of making <application>initdb</> revoke the default
        public <literal>EXECUTE</> privilege on these functions.  This allows
        installations to choose to grant usage of such functions to trusted
        roles that need not have full superuser privilege.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-08 [293007898] Reserve the "pg_" namespace for roles
2016-05-06 [a89505fd2] Remove various special checks around default roles
2016-05-08 [7df974ee0] Disallow superuser names starting with 'pg_' in initdb
-->
       <para>
        Treat role names beginning with <literal>pg_</> as reserved
        (Stephen Frost)
       </para>

       <para>
        User creation of such role names is now disallowed.  This prevents
        conflicts with built-in roles created by <application>initdb</>.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-08 [7a542700d] Create default roles
-->
       <para>
        Create some <link linkend="default-roles">built-in roles</link> that
        can be used to grant access to what were previously superuser-only
        functions (Stephen Frost)
       </para>
      </listitem>

    </itemizedlist>

   </sect3>

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

    <itemizedlist>

      <listitem>
<!--
2015-12-22 [6efbded6e] Allow omitting one or both boundaries in an array slice
-->
       <para>
        Allow omitting one or both boundaries in an array slice specifier,
        for example <literal>array_col[3:]</>
        (Yury Zhuravlev)
       </para>

       <para>
        Omitted boundaries are taken as the upper or lower limit of the
        corresponding array subscript.  This allows simpler specification of
        many common use-cases.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-16 [a70e13a39] Be more careful about out-of-range dates and timestamps.
-->
       <para>
        Be more careful about out-of-range dates and timestamps
        (Vitaly Burovoy)
       </para>

       <para>
        This change prevents unexpected out-of-range errors
        for <type>timestamp with time zone</> values very close to the
        implementation limits.  Previously, the <quote>same</> value might be
        accepted or not depending on the <varname>timezone</> setting,
        meaning that a dump and reload could fail on a value that had been
        accepted when presented.  Now the limits are enforced according to
        the equivalent UTC time, not local time, so as to be independent
        of <varname>timezone</>.
       </para>

       <para>
        Also, <productname>PostgreSQL</> is now more careful to detect
        overflow in operations that compute new date or timestamp values,
        such as <type>date</> <literal>+</> <type>integer</>.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-30 [50861cd68] Improve portability of I/O behavior for the geometric ty
-->
       <para>
        In the geometric data types, make sure that infinity and NaN
        component values are treated consistently during input and output
        (Tom Lane)
       </para>

       <para>
        Such values will now always print the same way as they would in a
        simple <type>float8</> column, and be accepted the same way on input
        as well.  Previously the behavior was somewhat platform-dependent.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-07 [bb140506d] Phrase full text search.
2016-06-27 [028350f61] Make exact distance match for FTS phrase operator
2016-06-27 [3dbbd0f02] Do not fallback to AND for FTS phrase operator.
2016-06-27 [6734a1cac] Change predecence of phrase operator.
-->
       <para>
        Improve full-text search to support searching for phrases, that is,
        lexemes appearing adjacent to each other in a specific order, or with
        a specified distance between them
        (Teodor Sigaev, Oleg Bartunov, Dmitry Ivanov)
       </para>

       <para>
        A phrase-search query can be specified in <type>tsquery</> input
        using the new operators <literal>&lt;-&gt;</> and
        <literal>&lt;<replaceable>N</>&gt;</literal>.  The former means that
        the lexemes before and after it must appear adjacent to each other in
        that order.  The latter means they must be exactly <replaceable>N</>
        lexemes apart.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-29 [61d66c44f] Fix support of digits in email/hostnames.
-->
       <para>
        Fix text search parser to allow leading digits in <literal>email</>
        and <literal>host</> tokens (Artur Zakirov)
       </para>

       <para>
        In most cases this will result in few changes in the parsing of text.
        But if you have data where such addresses occur frequently, it may be
        worth rebuilding dependent <type>tsvector</> columns and indexes, so
        that addresses of this form will be found properly by text searches.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-04 [d78a7d9c7] Improve support of Hunspell in ispell dictionary.
2016-03-11 [8829af47e] Fix merge affixes for numeric ones
2016-03-17 [f4ceed6ce] Improve support of Hunspell
-->
       <para>
        Upgrade the <literal>ispell</> dictionary to handle modern Hunspell
        files and support more languages (Artur Zakirov)
       </para>
      </listitem>

      <listitem>
<!--
2015-10-30 [12c9a0400] Implement lookbehind constraints in our regular-expressi
-->
       <para>
        Implement lookbehind constraints in regular expressions
        (Tom Lane)
       </para>

       <para>
        A lookbehind constraint is like a lookahead constraint in that it
        consumes no text; but it checks for existence (or nonexistence) of a
        match ending at the current point in the string, rather than one
        starting at the current point.  Similar features exist in many other
        regular-expression engines.
       </para>
      </listitem>

      <listitem>
<!--
2015-09-16 [b44d92b67] Sync regex code with Tcl 8.6.4.
-->
       <para>
        In regular expressions, if an apparent three-digit octal
        escape <literal>\</><replaceable>nnn</> would exceed 377 (255
        decimal), assume it is a two-digit octal escape instead (Tom Lane)
       </para>

       <para>
        This makes the behavior match current Tcl releases.
       </para>
      </listitem>

      <listitem>
<!--
2015-11-07 [c5e86ea93] Add "xid <> xid" and "xid <> int4" operators.
-->
       <para>
        Add <type>xid</> <literal>&lt;&gt;</> <type>xid</>
        and <type>xid</> <literal>&lt;&gt;</> <type>int4</> operators,
        for consistency with the corresponding <literal>=</> operators
        (Michael Paquier)
       </para>
      </listitem>

    </itemizedlist>

   </sect3>

   <sect3>
    <title>Functions</title>

     <itemizedlist>

      <listitem>
<!--
2016-04-06 [0b62fd036] Add jsonb_insert
-->
       <para>
        Add <function>jsonb_insert()</> function to insert a new element into
        a <type>jsonb</> array, or a not-previously-existing key into
        a <type>jsonb</> object (Dmitry Dolgov)
       </para>
      </listitem>

      <listitem>
<!--
2015-11-14 [7d9a4737c] Improve type numeric's calculations for ln(), log(), exp
2016-05-05 [18a02ad2a] Fix corner-case loss of precision in numeric pow() calcu
-->
       <para>
        Improve the accuracy of the <function>ln()</>, <function>log()</>,
        <function>exp()</>, and <function>pow()</> functions for type
        <type>numeric</> (Dean Rasheed)
       </para>
      </listitem>

      <listitem>
<!--
2016-01-05 [abb173392] Add scale(numeric)
-->
       <para>
        Add a <function>scale(numeric)</> function to extract the display
        scale of a <type>numeric</> value (Marko Tiikkaja)
       </para>
      </listitem>

      <listitem>
<!--
2016-01-22 [e1bd684a3] Add trigonometric functions that work in degrees.
-->
       <para>
        Add trigonometric functions that work in degrees (Dean Rasheed)
       </para>

       <para>
        For example, <function>sind()</> measures its argument in
        degrees, whereas <function>sin()</> measures in radians.
        These functions go to some lengths to deliver exact results for values
        where an exact result can be expected, such
        as <literal>sind(30) = 0.5</literal> exactly.
       </para>
      </listitem>

      <listitem>
<!--
2016-01-22 [fd5200c3d] Improve cross-platform consistency of Inf/NaN handling i
-->
       <para>
        Ensure that trigonometric functions handle infinity and NaN inputs per
        the POSIX standard (Dean Rasheed)
       </para>

       <para>
        The POSIX standard says that these functions should return NaN for NaN
        input, and should throw an error for out-of-range inputs including
        infinity; but previously, our actual behavior varied across platforms.
       </para>
      </listitem>

      <listitem>
<!--
2016-01-21 [647d87c56] Make extract() do something more reasonable with infinit
-->
       <para>
        Make <function>extract()</> behave more reasonably with infinite
        inputs (Vitaly Burovoy)
       </para>

       <para>
        Historically the <function>extract()</> function just returned zero
        given an infinite timestamp, regardless of the given unit name.  Make
        it return <literal>infinity</literal> or <literal>-infinity</literal>
        as appropriate when the requested field is one that is monotonically
        increasing (e.g, year, epoch), or NULL when it is not (e.g., day,
        hour).  Also, throw the expected error for bad unit names.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-29 [e511d878f] Allow to_timestamp(float8) to convert float infinity to
-->
       <para>
        Make <function>to_timestamp(float8)</> convert float infinity to
        timestamp infinity (Vitaly Burovoy)
       </para>

       <para>
        Formerly it just failed on an infinite input.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-11 [6943a946c] Tsvector editing functions
2016-05-05 [0b9a23443] Rename tsvector delete() to ts_delete(), and filter() to
-->
       <para>
        Add new functions for <type>tsvector</> data (Stas Kelvich)
       </para>

       <para>
        These are <function>ts_delete()</>, <function>ts_filter()</>,
        <function>unnest()</>, <function>tsvector_to_array()</>,
        <function>array_to_tsvector()</>, and a variant
        of <function>setweight()</> that sets the weight only for specified
        lexeme(s).
       </para>
      </listitem>

      <listitem>
<!--
2015-09-17 [9acb9007d] Fix oversight in tsearch type check
-->
       <para>
        Allow <function>ts_stat_sql()</>
        and <function>tsvector_update_trigger()</> to operate on values that
        are of types binary-compatible with the expected argument type, not
        just that argument type; for example allow <type>citext</>
        where <type>text</> is expected (Teodor Sigaev)
       </para>
      </listitem>

      <listitem>
<!--
2016-02-04 [6819514fc] Add num_nulls() and num_nonnulls() to count NULL argumen
-->
       <para>
        Add variadic functions <function>num_nulls()</>
        and <function>num_nonnulls()</> that count the number of their
        arguments that are null or nonnull (Marko Tiikkaja)
       </para>

       <para>
        An example usage is <literal>CHECK(num_nonnulls(a,b,c) = 1)</> which
        asserts that exactly one of a,b,c isn't NULL.  These functions can
        also be used to count the number of null or nonnull elements in an
        array.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-18 [3187d6de0] Introduce parse_ident()
-->
       <para>
        Add function <function>parse_ident()</> to split a qualified,
        possibly quoted SQL identifier into its parts (Pavel Stehule)
       </para>
      </listitem>

      <listitem>
<!--
2015-10-05 [2d87eedc1] to_char(): Do not count negative sign as a digit for tim
-->
       <para>
        In <function>to_char()</>, do not count a minus sign (when needed) as
        part of the field width for time-related fields (Bruce Momjian)
       </para>

       <para>
        For example, <literal>to_char('-4 years'::interval, 'YY')</> now
        returns <literal>-04</>, rather than <literal>-4</>.
       </para>
      </listitem>

      <listitem>
<!--
2015-10-05 [28b3a3d41] to_number():  allow 'V' to divide by 10^(the number of d
-->
       <para>
        In <function>to_number()</>, interpret <literal>V</> as dividing by
        10 to the power of the number of digits following <literal>V</>
        (Bruce Momjian)
       </para>

       <para>
        This makes it operate in an inverse fashion to <function>to_char()</>.
       </para>
      </listitem>

      <listitem>
<!--
2016-01-05 [ea0d494da] Make the to_reg*() functions accept text not cstring.
-->
       <para>
        Make the <function>to_reg*()</> functions accept type <type>text</>
        not <type>cstring</> (Petr Korobeinikov)
       </para>

       <para>
        This avoids the need to write an explicit cast in most cases where
        the argument isn't a simple literal constant.
       </para>
      </listitem>

      <listitem>
<!--
2016-02-20 [53874c522] Add pg_size_bytes() to parse human-readable size strings
-->
       <para>
        Add <function>pg_size_bytes()</> function to convert human-readable
        size strings to numbers (Pavel Stehule, Vitaly Burovoy, Dean Rasheed)
       </para>

       <para>
        This function converts strings like those produced
        by <function>pg_size_pretty()</> into sizes in bytes.  An example
        usage is <literal>SELECT oid::regclass FROM pg_class WHERE
        pg_total_relation_size(oid) &gt; pg_size_bytes('10 GB')</>.
       </para>
      </listitem>

      <listitem>
<!--
2015-11-06 [8a1fab36a] pg_size_pretty: Format negative values similar to positi
-->
       <para>
        In <function>pg_size_pretty()</>, format negative numbers similarly
        to positive ones (Adrian Vondendriesch)
       </para>

       <para>
        Previously, negative numbers were never abbreviated, just printed
        in bytes.
       </para>
      </listitem>

      <listitem>
<!--
2015-07-02 [10fb48d66] Add an optional missing_ok argument to SQL function curr
-->
       <para>
        Add an optional <replaceable>missing_ok</> argument to
        the <function>current_setting()</> function (David Christensen)
       </para>

       <para>
        This allows avoiding an error for an unrecognized parameter name;
        instead the result is NULL.
       </para>
      </listitem>

      <listitem>
<!--
2016-07-26 [976b24fb4] Change various deparsing functions to return NULL for in
2016-07-29 [3153b1a52] Eliminate a few more user-visible "cache lookup failed"
2016-08-07 [8a8c6b538] Fix crash when pg_get_viewdef_name_ext() is passed a non
-->
       <para>
        Change various catalog-inspection functions to return NULL for
        invalid input (Michael Paquier)
       </para>

       <para>
        <function>pg_get_viewdef()</> now returns NULL if given an invalid
        view OID, and several similar functions likewise return NULL for
        bad input.  Previously, such cases usually led to <quote>cache
        lookup failed</> errors, which are not meant to occur in
        user-facing cases.
       </para>
      </listitem>

      <listitem>
<!--
2016-08-02 [dd5eb805d] Remove unused arguments from pg_replication_origin_xact_
-->
       <para>
        Fix <function>pg_replication_origin_xact_reset()</> to not have any
        arguments (Fujii Masao)
       </para>

       <para>
        The documentation said that it has no arguments, and the C code
        didn't expect any arguments, but the entry in <structname>pg_proc</>
        mistakenly specified two arguments.
       </para>
      </listitem>

     </itemizedlist>

   </sect3>

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

    <itemizedlist>

      <listitem>
<!--
2015-08-21 [fcdfce682] Detect mismatched CONTINUE and EXIT statements at plpgsq
-->
       <para>
        In <link linkend="plpgsql">PL/pgSQL</link>,
        detect mismatched <command>CONTINUE</> and <command>EXIT</> statements
        while compiling PL/pgSQL functions, rather than at execution
        (Jim Nasby)
       </para>
      </listitem>

      <listitem>
<!--
2015-09-05 [0426f349e] Rearrange the handling of error context reports.
This commit is also listed under libpq and psql
-->
       <para>
        Remove PL/pgSQL's <quote>feature</> that suppressed the innermost line
        of <literal>CONTEXT</> for messages emitted by <command>RAISE</>
        commands (Pavel Stehule)
       </para>

       <para>
        This ancient backwards-compatibility hack was agreed to have outlived
        its usefulness.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-08 [5c3c3cd0a] Enhanced custom error in PLPythonu
2016-06-11 [020140d84] PL/Python: Rename new keyword arguments of plpy.error()
2016-07-02 [3a4a33ad4] PL/Python: Report argument parsing errors using exceptio
-->
       <para>
        Extend PL/Python's error-reporting and message-reporting functions to
        allow specifying additional message fields besides the primary error
        message (Pavel Stehule)
       </para>
      </listitem>

      <listitem>
<!--
2016-04-05 [1d2fe56e4] Fix PL/Python for recursion and interleaved set-returnin
-->
       <para>
        Allow PL/Python functions to call themselves recursively via SPI,
        and fix the behavior when multiple set-returning PL/Python functions
        are called within one query (Alexey Grishchenko, Tom Lane)
       </para>
      </listitem>

      <listitem>
<!--
2015-11-05 [8c75ad436] Fix memory leaks in PL/Python.
-->
       <para>
        Fix session-lifespan memory leaks in PL/Python
        (Heikki Linnakangas, Haribabu Kommi, Tom Lane)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-02 [287822068] Convert PL/Tcl to use Tcl's "object" interfaces.
2016-03-02 [e2609323e] Make PL/Tcl require Tcl 8.4 or later.
-->
       <para>
        Modernize <application>PL/Tcl</> to use Tcl's <quote>object</> APIs
        instead of simple strings (Jim Nasby, Karl Lehenbauer)
       </para>

       <para>
        This can improve performance substantially in some cases.
        Note that <application>PL/Tcl</> now requires Tcl 8.4 or later.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-25 [fb8d2a7f5] In PL/Tcl, make database errors return additional info i
2016-03-25 [cd37bb785] Improve PL/Tcl errorCode facility by providing decoded n
-->
       <para>
        In <application>PL/Tcl</>, make database-reported errors return
        additional information in Tcl's <varname>errorCode</> global variable
        (Jim Nasby, Tom Lane)
       </para>

       <para>
        This feature follows the Tcl convention for returning auxiliary data
        about an error.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-02 [c8c7c93de] Fix PL/Tcl's encoding conversion logic.
-->
       <para>
        Fix <application>PL/Tcl</> to perform encoding conversion between
        the database encoding and UTF-8, which is what Tcl expects (Tom Lane)
       </para>

       <para>
        Previously, strings were passed through without conversion, leading
        to misbehavior with non-ASCII characters when the database encoding
        isn't UTF-8.
       </para>
      </listitem>

    </itemizedlist>

   </sect3>

   <sect3>
    <title>Client Interfaces</title>

    <itemizedlist>

      <listitem>
<!--
2015-09-05 [0426f349e] Rearrange the handling of error context reports.
This commit is also listed under psql and PL/pgSQL
-->
       <para>
        Introduce a feature in <application>libpq</> whereby
        the <literal>CONTEXT</> field of messages can be suppressed, either
        always or only for non-error messages (Pavel Stehule)
       </para>

       <para>
        The default behavior of <function>PQerrorMessage()</> is now to
        print <literal>CONTEXT</> only for errors.  The new
        function <function>PQsetErrorContextVisibility()</> can be used to
        adjust this.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-03 [e3161b231] Add libpq support for recreating an error message with d
-->
       <para>
        Add support in <application>libpq</> for regenerating an error
        message with a different verbosity level (Alex Shulgin)
       </para>

       <para>
        This supports <application>psql</>'s new <literal>\errverbose</>
        feature, and may be useful for other clients as well.
       </para>
      </listitem>

      <listitem>
<!--
2015-11-27 [40cb21f70] Improve PQhost() to return useful data for default Unix-
-->
       <para>
        Improve <application>libpq</>'s <function>PQhost()</> function to
        return useful data for default Unix-socket connections (Tom Lane)
       </para>

       <para>
        Previously it would return NULL if no explicit host specification had
        been given; now it returns the default socket directory path.
       </para>
      </listitem>

      <listitem>
<!--
2016-02-16 [fc1ae7d2e] Change ecpg lexer to accept comments with line breaks in
-->
       <para>
        Fix <application>ecpg</>'s lexer to handle line breaks within comments
        starting on preprocessor directive lines (Michael Meskes)
       </para>
      </listitem>

    </itemizedlist>

   </sect3>

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

    <itemizedlist>

      <listitem>
<!--
2015-09-14 [d02426029] Check existency of table/schema for -t/-n option (pg_dum
-->
       <para>
        Add <option>--strict-names</> option to <application>pg_dump</>
        and <application>pg_restore</> (Pavel Stehule)
       </para>

       <para>
        This option causes the program to complain if there is no match for
        a <option>-t</option> or <option>-n</option> option, rather than
        silently doing nothing.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-06 [a9f0e8e5a] In pg_dump, use a bitmap to represent what to include
2016-04-06 [d217b2c36] In pg_dump, split "dump" into "dump" and "dump_contains"
2016-04-06 [23f34fa4b] In pg_dump, include pg_catalog and extension ACLs, if ch
2016-05-06 [5d589993c] pg_dump performance and other fixes
2016-05-06 [e1b120a8c] Only issue LOCK TABLE commands when necessary
-->
       <para>
        In <application>pg_dump</>,
        dump locally-made changes in privilege assignments for system objects
        (Stephen Frost)
       </para>

       <para>
        While it's always been possible for a superuser to change the
        privilege assignments for built-in or extension-created objects,
        such changes were formerly lost in a dump and reload.  Now,
        <application>pg_dump</> recognizes and dumps such changes.
        (This works only when dumping from a 9.6 or later server, however.)
       </para>
      </listitem>

      <listitem>
<!--
2016-04-06 [3b3fcc4ee] pg_dump: Add table qualifications to some tags
-->
       <para>
        In <application>pg_dump</>, include the table name in object tags
        for object types that are only uniquely named per-table (for
        example, triggers) (Peter Eisentraut)
       </para>
      </listitem>

      <listitem>
<!--
2015-07-02 [5671aaca8] Improve pg_restore's -t switch to match all types of rel
-->
       <para>
        Improve <application>pg_restore</>'s <option>-t</option> switch to
        match all types of relations, not only plain tables (Craig Ringer)
       </para>
      </listitem>

    </itemizedlist>

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

      <itemizedlist>

      <listitem>
<!--
2015-12-08 [d5563d7df] psql: Support multiple -c and -f options, and allow mixi
-->
       <para>
        Support multiple <option>-c</option> and <option>-f</option>
        command-line options (Pavel Stehule, Catalin Iacob)
       </para>

       <para>
        To allow this with sane behavior, one backwards incompatibility had to
        be introduced: <option>-c</option> no longer
        implies <option>--no-psqlrc</option>.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-08 [c09b18f21] Support \crosstabview in psql
-->
       <para>
        Add a <command>\crosstabview</> command that prints the
        results of a query in a cross-tabulated display
        (Daniel V&eacute;rit&eacute;)
       </para>

       <para>
        In the crosstab display, data values from one query result column are
        placed in a grid whose column and row headers come from other query
        result columns.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-03 [3cc38ca7d] Add psql \errverbose command to see last server error at
-->
       <para>
        Add an <literal>\errverbose</> command that shows the last server
        error at full verbosity (Alex Shulgin)
       </para>

       <para>
        This is very handy after getting an unexpected error &mdash; you no
        longer need to adjust the <varname>VERBOSITY</> variable and recreate
        the failure in order to see error fields that aren't shown by
        default.
       </para>
      </listitem>

      <listitem>
<!--
2015-07-03 [8eb6407aa] Add psql \ev and \sv commands for editing and showing vi
2016-05-06 [9b66aa006] Fix psql's \ev and \sv commands so that they handle view
-->
       <para>
        Add <literal>\ev</> and <literal>\sv</> commands
        for editing and showing view definitions (Petr Korobeinikov)
       </para>

       <para>
        These are parallel to the existing <literal>\ef</> and <literal>\sf</>
        commands for functions.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-04 [2bbe9112a] Add a \gexec command to psql for evaluation of computed
-->
       <para>
        Add a <command>\gexec</> command that executes a query and re-submits
        the result(s) as new queries (Corey Huinker)
       </para>
      </listitem>

      <listitem>
<!--
2015-10-05 [2145a7660] psql:  allow \pset C in setting the title, matches \C
-->
       <para>
        Allow <literal>\pset C <replaceable>string</></literal> to set the
        table title, for consistency
        with <literal>\C <replaceable>string</></literal> (Bruce Momjian)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-11 [69ab7b9d6] psql: Don't automatically use expanded format when there
-->
       <para>
        In <literal>\pset expanded auto</> mode, do not use expanded format
        for query results with only one column (Andreas Karlsson, Robert Haas)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-21 [dea2b5960] Improve header output from psql's \watch command.
2016-06-15 [9901d8ac2] Use strftime("%c") to format timestamps in psql's \watch
-->
       <para>
        Improve the headers output by the <command>\watch</> command
        (Michael Paquier, Tom Lane)
       </para>

       <para>
        Include the <command>\pset title</> string if one has been set, and
        shorten the prefabricated part of the header to be
        <literal><replaceable>timestamp</> (every <replaceable>N</>s)</literal>.
        Also, the timestamp format now obeys <application>psql</>'s locale
        environment.
       </para>
      </listitem>

      <listitem>
<!--
2015-12-20 [d854118c8] Teach psql's tab completion to consider the entire input
-->
       <para>
        Improve tab-completion logic to consider the
        entire input query, not only the current line (Tom Lane)
       </para>

       <para>
        Previously, breaking a command into multiple lines defeated any tab
        completion rules that would have needed to see words on earlier lines.
       </para>
      </listitem>

      <listitem>
<!--
2015-08-04 [158e3bc8e] Tab completion for CREATE SEQUENCE.
2015-09-08 [3ae16798f] psql: Generic tab completion support for enum and bool G
2015-09-09 [2f8880704] Improve tab-completion for GRANT and REVOKE.
2015-10-01 [bf4817e4f] Fix incorrect tab-completion for GRANT and REVOKE
2015-10-20 [7c0b49cd0] Tab complete CREATE EXTENSION .. VERSION.
2015-12-25 [8014c44e8] Improve SECURITY LABEL tab completion
2016-01-12 [bc56d5898] psql: Fix CREATE INDEX tab completion
2016-01-12 [b1bfb28b5] psql: Improve CREATE INDEX CONCURRENTLY tab completion
2016-01-16 [4189e3d65] psql: Add completion support for DROP INDEX CONCURRENTLY
2016-01-20 [d0f2f53cd] psql: Add tab completion for COPY with query
2016-01-23 [6ae4c8de0] psql: Improve completion of FDW DDL commands
2016-01-26 [879d71393] Various fixes to REFRESH MATERIALIZED VIEW tab completio
2016-02-01 [89611c4df] Various fixes to "ALTER ... SET/RESET" tab completions
-->
       <para>
        Numerous minor improvements in tab-completion behavior (Peter
        Eisentraut, Vik Fearing, Kevin Grittner, Kyotaro Horiguchi, Jeff
        Janes, Andreas Karlsson, Fujii Masao, Thomas Munro, Masahiko Sawada,
        Pavel Stehule)
       </para>
      </listitem>

      <listitem>
<!--
2015-07-07 [275f05c99] Add psql PROMPT variable showing the pid of the connecte
-->
       <para>
        Add a <literal>PROMPT</> option <literal>%p</> to
        insert the process ID of the connected backend (Julien Rouhaud)
       </para>
      </listitem>

      <listitem>
<!--
2015-09-05 [0426f349e] Rearrange the handling of error context reports.
This commit is also listed under libpq and PL/pgSQL
-->
       <para>
        Introduce a feature whereby the <literal>CONTEXT</> field of messages
        can be suppressed, either always or only for non-error messages
        (Pavel Stehule)
       </para>

       <para>
        Printing <literal>CONTEXT</> only for errors is now the default
        behavior.  This can be changed by setting the special variable
        <varname>SHOW_CONTEXT</varname>.
       </para>
      </listitem>

      <listitem>
<!--
2016-07-11 [a670c24c3] Improve output of psql's \df+ command.
-->
       <para>
        Make <command>\df+</> show function access privileges and
        parallel-safety attributes (Michael Paquier)
       </para>
      </listitem>

      </itemizedlist>

    </sect4>

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

     <itemizedlist>

      <listitem>
<!--
2016-03-20 [68ab8e8ba] SQL commands in pgbench scripts are now ended by semicol
-->
       <para>
        SQL commands in <application>pgbench</> scripts are now ended by
        semicolons, not newlines (Kyotaro Horiguchi, Tom Lane)
       </para>

       <para>
        This change allows SQL commands in scripts to span multiple lines.
        Existing custom scripts will need to be modified to add a semicolon
        at the end of each line that does not have one already.  (Doing so
        does not break the script for use with older versions
        of <application>pgbench</>.)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-01 [7e137f846] Extend pgbench's expression syntax to support a few buil
2016-03-28 [86c43f4e2] pgbench: Support double constants and functions.
2016-05-05 [7a622b273] Rename pgbench min/max to least/greatest, and fix handli
2016-05-06 [951529948] Improve handling of numeric-valued variables in pgbench.
-->
       <para>
        Support floating-point arithmetic, as well as
        some <link linkend="pgbench-builtin-functions">built-in functions</>,
        in expressions in backslash commands (Fabien Coelho)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-29 [ad9566470] pgbench: Remove \setrandom.
-->
       <para>
        Replace <command>\setrandom</> with built-in functions (Fabien Coelho)
       </para>

       <para>
        The new built-in functions include <function>random()</>,
        <function>random_exponential()</>, and <function>random_gaussian()</>,
        which perform the same work as <command>\setrandom</>, but are easier
        to use since they can be embedded in larger expressions.  Since these
        additions have made <command>\setrandom</> obsolete, remove it.
       </para>
      </listitem>

      <listitem>
<!--
2016-01-27 [8bea3d221] pgbench: improve multi-script support
2016-03-03 [d561f1cae] pgbench: accept unambiguous builtin prefixes for -b
-->
       <para>
        Allow invocation of multiple copies of the built-in scripts, not only
        custom scripts (Fabien Coelho)
       </para>

       <para>
        This is done with the new <option>-b</> switch, which works similarly
        to <option>-f</> for custom scripts.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-19 [7bafffea6] pgbench: Allow changing weights for scripts
-->
       <para>
        Allow changing the selection probabilities (weights) for scripts
        (Fabien Coelho)
       </para>

       <para>
        When multiple scripts are specified, each <application>pgbench</>
        transaction randomly chooses one to execute.  Formerly this was
        always done with uniform probability, but now different selection
        probabilities can be specified for different scripts.
       </para>
      </listitem>

      <listitem>
<!--
2016-02-01 [1d0c3b3f8] pgbench: allow per-script statistics
-->
       <para>
        Collect statistics for each script in a multi-script run
        (Fabien Coelho)
       </para>

       <para>
        This feature adds an intermediate level of detail to existing global
        and per-command statistics printouts.
       </para>
      </listitem>

      <listitem>
<!--
2015-09-16 [1def9063c] pgbench progress with timestamp
-->
       <para>
        Add a <option>--progress-timestamp</> option to report progress with
        Unix epoch timestamps, instead of time since the run started
        (Fabien Coelho)
       </para>
      </listitem>

      <listitem>
<!--
2015-07-03 [ba3deeefb] Lift the limitation that # of clients must be a multiple
-->
       <para>
        Allow the number of client connections (<option>-c</>) to not be an
        exact multiple of the number of threads (<option>-t</>)
        (Fabien Coelho)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-09 [accf7616f] pgbench: When -T is used, don't wait for transactions be
-->
       <para>
        When the <option>-T</> option is used, stop promptly at the end of
        the specified time (Fabien Coelho)
       </para>

       <para>
        Previously, specifying a low transaction rate could
        cause <application>pgbench</> to wait significantly longer than that.
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

   </sect3>

   <sect3>
    <title>Server Applications</title>

    <itemizedlist>

      <listitem>
<!--
2015-12-17 [66d947b9d] Adjust behavior of single-user -j mode for better initdb
-->
       <para>
        Improve error reporting during <application>initdb</>'s post-bootstrap
        phase (Tom Lane)
       </para>

       <para>
        Previously, an error here led to reporting the entire input file as
        the <quote>failing query</>; now just the current query is reported.
        To get the desired behavior, queries in <application>initdb</>'s
        input files must be separated by blank lines.
       </para>
      </listitem>

      <listitem>
<!--
2015-12-17 [c4a8812cf] Use just one standalone-backend session for initdb's pos
-->
       <para>
        Speed up <application>initdb</> by using just one standalone-backend
        session for all the post-bootstrap steps (Tom Lane)
       </para>
      </listitem>

      <listitem>
<!--
2015-12-01 [e50cda784] Use pg_rewind when target timeline was switched
-->
       <para>
        Improve <application>pg_rewind</> so that it can work when the
        target timeline changes (Alexander Korotkov)
       </para>

       <para>
        This allows, for example, rewinding a promoted standby back to
        some state of the old master's timeline.
       </para>
      </listitem>

    </itemizedlist>

   </sect3>

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

     <itemizedlist>

      <listitem>
<!--
2015-07-02 [726117243] Remove obsolete heap_formtuple/modifytuple/deformtuple f
-->
       <para>
        Remove obsolete <function>heap_formtuple</>/<function>heap_modifytuple</>/<function>heap_deformtuple</>
        functions (Peter Geoghegan)
       </para>
      </listitem>

      <listitem>
<!--
2015-08-05 [de6fd1c89] Rely on inline functions even if that causes warnings in
-->
       <para>
        Unconditionally use <literal>static inline</> functions in header
        files (Andres Freund)
       </para>

       <para>
        This may result in warnings and/or wasted code space with very old
        compilers, but the notational improvement seems worth it.
       </para>
      </listitem>

      <listitem>
<!--
2015-07-29 [13d856e17] Make TAP tests work on Windows.
2015-12-02 [1caef31d9] Refactor Perl test code
2015-12-07 [9821492ee] Cleanup some problems in new Perl test code
2016-03-03 [2c83f435a] Rework PostgresNode's psql method
2016-02-26 [49148645f] Add a test framework for recovery
2016-02-26 [74d58425c] Apply last revision of recovery patch
2016-05-06 [6bd356c33] Add TAP tests for pg_dump
-->
       <para>
        Improve TAP testing infrastructure
        (Michael Paquier, Craig Ringer, &Aacute;lvaro Herrera, Stephen Frost)
       </para>

       <para>
        Notably, it is now possible to test recovery scenarios using this
        infrastructure.
       </para>
      </listitem>

      <listitem>
<!--
2015-09-11 [aa65de042] When trace_lwlocks is used, identify individual lwlocks
-->
       <para>
        Make <varname>trace_lwlocks</> identify individual locks by name
        (Robert Haas)
       </para>
      </listitem>

      <listitem>
<!--
2015-12-19 [d37b816dc] Adopt a more compact, less error-prone notation for tab
2016-01-04 [9b181b036] In psql's tab completion, change most TailMatches patter
2016-01-05 [4f18010af] Convert psql's tab completion for backslash commands to
-->
       <para>
        Improve <application>psql</>'s tab-completion code infrastructure
        (Thomas Munro, Michael Paquier)
       </para>

       <para>
        Tab-completion rules are now considerably easier to write, and more
        compact.
       </para>
      </listitem>

      <listitem>
<!--
2016-01-05 [efa318bcf] Make pg_shseclabel available in early backend startup
-->
       <para>
        Nail the <structname>pg_shseclabel</> system catalog into cache,
        so that it is available for access during connection authentication
        (Adam Brightwell)
       </para>

       <para>
        The core code doesn't use this catalog for authentication, but
        extensions might wish to consult it.
       </para>
      </listitem>

      <listitem>
<!--
2016-01-17 [65c5fcd35] Restructure index access method API to hide most of it a
2016-01-21 [be44ed27b] Improve index AMs' opclass validation procedures.
-->
       <para>
        Restructure index access method API to hide most of it at the C level
        (Alexander Korotkov)
       </para>

       <para>
        This change modernizes the index AM API to look more like the designs
        we've adopted for foreign data wrappers and tablesample handlers.
        This simplifies the C code and should make it more feasible to define
        index access methods in installable extensions.  A consequence is
        that most of the columns of the <structname>pg_am</> system catalog
        have disappeared.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-06 [6c268df12] Add new catalog called pg_init_privs
-->
       <para>
        Add <structname>pg_init_privs</> system catalog to hold original
        privileges of <application>initdb</>-created and extension-created
        objects (Stephen Frost)
       </para>

       <para>
        This infrastructure allows <application>pg_dump</> to dump changes
        that an installation may have made in privileges attached to system
        objects.  Formerly, such changes would be lost in a dump and reload,
        but now they can be preserved.
       </para>
      </listitem>

      <listitem>
<!--
2016-02-04 [c1772ad92] Change the way that LWLocks for extensions are allocated
-->
       <para>
        Change the way that extensions allocate custom LWLocks
        (Amit Kapila, Robert Haas)
       </para>

       <para>
        The <function>RequestAddinLWLocks()</> function is removed, and
        replaced by <function>RequestNamedLWLockTranche()</>.  This allows
        better identification of custom LWLocks, and is less error-prone.
       </para>
      </listitem>

      <listitem>
<!--
2016-02-11 [38f8bdcac] Modify the isolation tester so that multiple sessions ca
-->
       <para>
        Improve the isolation tester to allow multiple sessions to wait
        concurrently, allowing testing of deadlock scenarios (Robert Haas)
       </para>
      </listitem>

      <listitem>
<!--
2016-02-12 [bcac23de7] Introduce extensible node types.
-->
       <para>
        Introduce extensible node types (KaiGai Kohei)
       </para>

       <para>
        This change allows FDWs or custom scan providers to store data in a
        plan tree in a more convenient format than was previously possible.
       </para>
      </listitem>

      <listitem>
<!--
2015-08-11 [68fa28f77] Postpone extParam/allParam calculations until the very e
2016-01-07 [c44d01383] Delay creation of subplan tlist until after create_plan(
2016-02-18 [19a541143] Add an explicit representation of the output targetlist
2016-03-07 [3fc6e2d7f] Make the upper part of the planner work by generating an
2016-07-01 [5ce5e4a12] Set consider_parallel correctly for upper planner rels.
2016-07-01 [9e703987a] Rethink the GetForeignUpperPaths API (again).
-->
       <para>
        Make the planner deal with post-scan/join query steps by generating
        and comparing Paths, replacing a lot of very ad-hoc logic (Tom Lane)
       </para>

       <para>
        This change provides only marginal user-visible improvements today,
        but it enables future work on a lot of upper-planner improvements
        that were impractical to tackle in the old code structure.
       </para>
      </listitem>

      <listitem>
<!--
2016-01-20 [a7de3dc5c] Support multi-stage aggregation.
2016-06-22 [f8ace5477] Fix type-safety problem with parallel aggregate serial/d
2016-06-26 [19e972d55] Rethink node-level representation of partial-aggregation
-->
       <para>
        Support partial aggregation (David Rowley, Simon Riggs)
       </para>

       <para>
        This change allows the computation of an aggregate function to be
        split into separate parts, for example so that parallel worker
        processes can cooperate on computing an aggregate.  In future it
        might allow aggregation across local and remote data to occur
        partially on the remote end.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-09 [b6fb6471f] Add a generic command progress reporting facility.
-->
       <para>
        Add a generic command progress reporting facility (Vinayak Pokale,
        Rahila Syed, Amit Langote, Robert Haas)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-18 [27199058d] Convert psql's flex lexer to be re-entrant, and make it
2016-03-19 [0ea9efbe9] Split psql's lexer into two separate .l files for SQL an
2016-03-20 [b6afae71a] Use %option bison-bridge in psql/pgbench lexers.
2016-03-24 [588d963b0] Create src/fe_utils/, and move stuff into there from pg_
2016-03-24 [d65bea26a] Move psql's print.c and mbprint.c into src/fe_utils.
2016-03-24 [c1156411a] Move psql's psqlscan.l into src/fe_utils.
-->
       <para>
        Separate out <application>psql</>'s flex lexer to make it usable by
        other client programs too (Tom Lane, Kyotaro Horiguchi)
       </para>

       <para>
        This eliminates code duplication for programs that need to be able
        to parse SQL commands well enough to identify command boundaries.
        Doing that in full generality is more painful than one could wish,
        and up to now only <application>psql</> has really gotten it right
        among our supported client programs.
       </para>

       <para>
        A new source-code subdirectory <filename>src/fe_utils/</> has been
        created to hold this and other code that's shared across our client
        programs.  Formerly such sharing was accomplished by symlinking or
        copying source files at build time, which was ugly and required
        duplicate compilation work.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-21 [98a64d0bd] Introduce WaitEventSet API.
-->
       <para>
        Introduce WaitEventSet API to allow efficient waiting for event
        sets that usually don't change from one wait to the next
        (Andres Freund, Amit Kapila)
       </para>
      </listitem>

      <listitem>
<!--
2016-04-01 [65578341a] Add Generic WAL interface
-->
       <para>
        Add a generic interface for writing WAL records
        (Alexander Korotkov, Petr Jel&iacute;nek, Markus Nullmeier)
       </para>

       <para>
        This change allows extensions to write WAL records for changes to
        pages with standard layout.  The problem of needing to replay WAL
        without access to the extension is solved by having generic replay
        code.  This allows extensions to implement, for example, index access
        methods and have WAL support for them.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-06 [3fe3511d0] Generic Messages for Logical Decoding
-->
       <para>
        Support generic WAL messages for logical decoding
        (Petr Jel&iacute;nek, Andres Freund)
       </para>

       <para>
        This feature allows extensions to insert data into the WAL stream
        that can be read by logical-decoding plugins, but is not connected to
        physical data restoration.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-30 [ccd6eb49a] Introduce traversalValue for SP-GiST scan
2016-03-30 [87545f541] Use traversalValue in SP-GiST range opclass.
-->
       <para>
        Allow SP-GiST operator classes to store an arbitrary <quote>traversal
        value</> while descending the index (Alexander Lebedev, Teodor Sigaev)
       </para>

       <para>
        This is somewhat like the <quote>reconstructed value</>, but it could
        be any arbitrary chunk of data, it need not be of the same data type
        as the indexed column.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-04 [66229ac00] Introduce a LOG_SERVER_ONLY ereport level, which is neve
-->
       <para>
        Introduce a <literal>LOG_SERVER_ONLY</> message level
        for <function>ereport()</> (David Steele)
       </para>

       <para>
        This level acts like <literal>LOG</> except that the message is never
        sent to the client.  It's meant for use in auditing and similar
        applications.
       </para>
      </listitem>

      <listitem>
<!--
2016-07-01 [548af97fc] Provide and use a makefile target to build all generated
-->
       <para>
        Provide a makefile target to build all generated headers (Michael
        Paquier, Tom Lane)
       </para>

       <para>
        <literal>submake-generated-headers</> can now be invoked to ensure
        that generated backend header files are up-to-date.  This is useful in
        subdirectories that might be built <quote>standalone</>.
       </para>
      </listitem>

     </itemizedlist>

   </sect3>

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

    <itemizedlist>

      <listitem>
<!--
2016-03-11 [92f03fe76] Allow setting sample ratio for auto_explain
2016-03-13 [7a8d87483] Rename auto_explain.sample_ratio to sample_rate
-->
       <para>
        Add configuration parameter <literal>auto_explain.sample_rate</>
        to allow <filename>contrib/auto_explain</> to capture just a
        configurable fraction of all queries (Craig Ringer, Julien Rouhaud)
       </para>

       <para>
        This allows reduction of overhead for heavy query traffic, while
        still getting useful information on average.
       </para>
      </listitem>

      <listitem>
<!--
2016-04-01 [9ee014fc8] Bloom index contrib module
-->
       <para>
        Add <filename>contrib/bloom</> module that implements an index access
        method based on Bloom filtering (Teodor Sigaev, Alexander Korotkov)
       </para>

       <para>
        This is primarily a proof-of-concept for non-core index access
        methods, but it could be useful in its own right for queries that
        involve searches on many columns.
       </para>
      </listitem>

      <listitem>
<!--
2015-12-18 [33bd250f6] Cube extension kNN support
2015-12-28 [81ee726d8] Code and docs review for cube kNN support.
-->
       <para>
        In <filename>contrib/cube</>, introduce distance operators for cubes,
        and support kNN-style searches in GiST indexes on cube columns
        (Stas Kelvich)
       </para>
      </listitem>

      <listitem>
<!--
2016-02-03 [41d2c081c] Make hstore_to_jsonb_loose match hstore_to_json_loose on
-->
       <para>
        Make <filename>contrib/hstore</>'s <function>hstore_to_jsonb_loose()</>
        and <function>hstore_to_json_loose()</> functions agree on what's a
        number (Tom Lane)
       </para>

       <para>
        Previously, <function>hstore_to_jsonb_loose()</> would convert
        numeric-looking strings to JSON numbers, rather than strings, even if
        they did not exactly match the JSON syntax specification for numbers.
        This was inconsistent with <function>hstore_to_json_loose()</>, so
        tighten the test to match the JSON syntax.
       </para>
      </listitem>

      <listitem>
<!--
2015-07-21 [c6fbe6d6f] Add selectivity estimation functions for intarray operat
-->
       <para>
        Add selectivity estimation functions for <filename>contrib/intarray</>
        operators, to improve plans for queries using those operators (Yury
        Zhuravlev, Alexander Korotkov)
       </para>
      </listitem>

      <listitem>
<!--
2015-11-25 [d6061f83a] Improve pageinspect module
-->
       <para>
        Make <filename>contrib/pageinspect</>'s <function>heap_page_items()</>
        function show the raw data in each tuple, and add new
        functions <function>tuple_data_split()</>
        and <function>heap_page_item_attrs()</> for inspection of individual
        tuple fields (Nikolay Shaplov)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-09 [188f359d3] pgcrypto: support changing S2K iteration count
-->
       <para>
        Add an optional S2K iteration count parameter
        to <filename>contrib/pgcrypto</>'s <function>pgp_sym_encrypt()</>
        function (Jeff Janes)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-16 [f576b17cd] Add word_similarity to pg_trgm contrib module.
-->
       <para>
        Add support for <quote>word similarity</>
        to <filename>contrib/pg_trgm</>
        (Alexander Korotkov, Artur Zakirov)
       </para>

       <para>
        These functions and operators measure the similarity between one
        string and the most similar single word of another string.
       </para>
      </listitem>

      <listitem>
<!--
2016-03-16 [5871b8848] GUC variable pg_trgm.similarity_threshold insead of set_
2016-06-20 [9c852566a] Fix comparison of similarity to threshold in GIST trigra
-->
       <para>
        Add configuration parameter <varname>pg_trgm.similarity_threshold</>
        for <filename>contrib/pg_trgm</>'s similarity threshold
        (Artur Zakirov)
       </para>

       <para>
        This threshold has always been configurable, but formerly it was
        controlled by special-purpose functions <function>set_limit()</>
        and <function>show_limit()</>.  Those are now deprecated.
       </para>
      </listitem>

      <listitem>
<!--
2015-07-20 [97f301464] This supports the triconsistent function for pg_trgm GIN
-->
       <para>
        Improve <filename>contrib/pg_trgm</>'s GIN operator class to speed up
        index searches in which both common and rare keys appear (Jeff Janes)
       </para>
      </listitem>

      <listitem>
<!--
2015-12-25 [25bfa7efd] Improve the gin index scan performance in pg_trgm.
-->
       <para>
        Improve performance of similarity searches in
        <filename>contrib/pg_trgm</> GIN indexes (Christophe Fornaroli)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-08 [ba0a198fb] Add pg_visibility contrib module.
2016-06-17 [71d05a2c7] pg_visibility: Add pg_truncate_visibility_map function.
-->
       <para>
        Add <filename>contrib/pg_visibility</> module to allow examining
        table visibility maps (Robert Haas)
       </para>
      </listitem>

      <listitem>
<!--
2015-09-07 [49124613f] contrib/sslinfo: add ssl_extension_info SRF
-->
       <para>
        Add <function>ssl_extension_info()</> function
        to <filename>contrib/sslinfo</>, to print information about SSL
        extensions present in the X509 certificate used for the current
        connection (Dmitry Voronin)
       </para>
      </listitem>

      <listitem>
<!--
2015-09-04 [1bbd52cb9] Make unaccent handle all diacritics known to Unicode, an
2016-03-16 [9a206d063] Improve script generating unaccent rules
-->
       <para>
        Extend <filename>contrib/unaccent</>'s
        standard <filename>unaccent.rules</> file to handle all diacritics
        known to Unicode, and expand ligatures correctly
        (Thomas Munro, L&eacute;onard Benedetti)
       </para>

       <para>
        The previous version omitted some less-common letters with diacritic
        marks.  It now also expands ligatures into separate letters.
        Installations that use this rules file may wish to
        rebuild <type>tsvector</> columns and indexes that depend on the
        result.
       </para>
      </listitem>

    </itemizedlist>

    <sect4>
     <title><filename>postgres_fdw</></title>

     <itemizedlist>

      <listitem>
<!--
2015-11-03 [d89494166] Allow postgres_fdw to ship extension funcs/operators for
-->
       <para>
        Allow extension-provided operators and functions to be sent for
        remote execution, if the extension is whitelisted in the foreign
        server's options (Paul Ramsey)
       </para>

       <para>
        Users can enable this feature when the extension is known to exist in
        a compatible version in the remote database.  It allows more efficient
        execution of queries involving extension operators.
       </para>
      </listitem>

      <listitem>
<!--
2015-11-03 [f18c944b6] postgres_fdw: Add ORDER BY to some remote SQL queries.
2015-12-22 [ccd8f9792] postgres_fdw: Consider requesting sorted data so we can
-->
       <para>
        Consider performing sorts on the remote server (Ashutosh Bapat)
       </para>
      </listitem>

      <listitem>
<!--
2016-02-09 [e4106b252] postgres_fdw: Push down joins to remote servers.
2016-03-09 [aa09cd242] postgres_fdw: Consider foreign joining and foreign sorti
-->
       <para>
        Consider performing joins on the remote server (Shigeru Hanada,
        Ashutosh Bapat)
       </para>
      </listitem>

      <listitem>
<!--
2016-03-18 [0bf3ae88a] Directly modify foreign tables.
-->
       <para>
        When feasible, perform <command>UPDATE</> or <command>DELETE</>
        entirely on the remote server (Etsuro Fujita)
       </para>

       <para>
        Formerly, this involved sending a <command>SELECT FOR UPDATE</>
        command and then updating or deleting the selected rows one-by-one.
        While that's still necessary if the operation requires any local
        processing, it can now be done remotely if all elements of the query
        are safe to send to the remote server.
       </para>
      </listitem>

      <listitem>
<!--
2016-02-03 [dc203dc3a] postgres_fdw: Allow fetch_size to be set per-table or pe
-->
       <para>
        Allow the fetch size to be set as a server or table option
        (Corey Huinker)
       </para>

       <para>
        Formerly, <filename>postgres_fdw</> always fetched 100 rows at a time
        from remote queries; now that behavior is adjustable.
       </para>
      </listitem>

      <listitem>
<!--
2016-01-28 [96198d94c] Avoid multiple foreign server connections when all use s
-->
       <para>
        Use a single foreign-server connection for local user IDs that all
        map to the same remote user (Ashutosh Bapat)
       </para>
      </listitem>

      <listitem>
<!--
2016-04-21 [f039eaac7] Allow queries submitted by postgres_fdw to be canceled.
2016-05-16 [1b812afb0] Fix multiple problems in postgres_fdw query cancellation
-->
       <para>
        Transmit query cancellation requests to the remote server
        (Michael Paquier)
       </para>

       <para>
        Previously, a local query cancellation request did nothing to cause
        an already-sent remote query to terminate early.
       </para>
      </listitem>

     </itemizedlist>

    </sect4>

   </sect3>

  </sect2>

 </sect1>