~ubuntu-branches/ubuntu/trusty/postgresql-9.3/trusty-updates : contents of doc/src/sgml/html/release-9-2.html at revision 27

~ubuntu-branches/ubuntu/trusty/postgresql-9.3/trusty-updates : (revision 27)
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<HTML
><HEAD
><TITLE
>Release 9.2</TITLE
><META
NAME="GENERATOR"
CONTENT="Modular DocBook HTML Stylesheet Version 1.79"><LINK
REV="MADE"
HREF="mailto:pgsql-docs@postgresql.org"><LINK
REL="HOME"
TITLE="PostgreSQL 9.3.13 Documentation"
HREF="index.html"><LINK
REL="UP"
TITLE="Release Notes"
HREF="release.html"><LINK
REL="PREVIOUS"
TITLE="Release 9.2.1"
HREF="release-9-2-1.html"><LINK
REL="NEXT"
TITLE="Release 9.1.22"
HREF="release-9-1-22.html"><LINK
REL="STYLESHEET"
TYPE="text/css"
HREF="stylesheet.css"><META
HTTP-EQUIV="Content-Type"
CONTENT="text/html; charset=ISO-8859-1"><META
NAME="creation"
CONTENT="2016-05-09T21:13:26"></HEAD
><BODY
CLASS="SECT1"
><DIV
CLASS="NAVHEADER"
><TABLE
SUMMARY="Header navigation table"
WIDTH="100%"
BORDER="0"
CELLPADDING="0"
CELLSPACING="0"
><TR
><TH
COLSPAN="5"
ALIGN="center"
VALIGN="bottom"
><A
HREF="index.html"
>PostgreSQL 9.3.13 Documentation</A
></TH
></TR
><TR
><TD
WIDTH="10%"
ALIGN="left"
VALIGN="top"
><A
TITLE="Release 9.2.1"
HREF="release-9-2-1.html"
ACCESSKEY="P"
>Prev</A
></TD
><TD
WIDTH="10%"
ALIGN="left"
VALIGN="top"
><A
HREF="release.html"
ACCESSKEY="U"
>Up</A
></TD
><TD
WIDTH="60%"
ALIGN="center"
VALIGN="bottom"
>Appendix E. Release Notes</TD
><TD
WIDTH="20%"
ALIGN="right"
VALIGN="top"
><A
TITLE="Release 9.1.22"
HREF="release-9-1-22.html"
ACCESSKEY="N"
>Next</A
></TD
></TR
></TABLE
><HR
ALIGN="LEFT"
WIDTH="100%"></DIV
><DIV
CLASS="SECT1"
><H1
CLASS="SECT1"
><A
NAME="RELEASE-9-2"
>E.32. Release 9.2</A
></H1
><DIV
CLASS="NOTE"
><BLOCKQUOTE
CLASS="NOTE"
><P
><B
>Release Date: </B
>2012-09-10</P
></BLOCKQUOTE
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="AEN120624"
>E.32.1. Overview</A
></H2
><P
>    This release has been largely focused on performance improvements, though
    new SQL features are not lacking.  Work also continues in the area of
    replication support.  Major enhancements include:
   </P
><P
></P
><UL
><LI
><P
>        Allow queries to retrieve data only from indexes, avoiding heap
        access (<I
CLASS="FIRSTTERM"
>index-only scans</I
>)
       </P
></LI
><LI
><P
>        Allow the planner to generate custom plans for specific parameter
        values even when using prepared statements
       </P
></LI
><LI
><P
>        Improve the planner's ability to use nested loops with inner
        index scans
       </P
></LI
><LI
><P
>        Allow streaming replication slaves to forward data to other slaves
        (<A
HREF="warm-standby.html#CASCADING-REPLICATION"
><I
CLASS="FIRSTTERM"
>cascading
        replication</I
></A
>)
       </P
></LI
><LI
><P
>        Allow <A
HREF="app-pgbasebackup.html"
><SPAN
CLASS="APPLICATION"
>pg_basebackup</SPAN
></A
>
        to make base backups from standby servers
       </P
></LI
><LI
><P
>        Add a <A
HREF="app-pgreceivexlog.html"
><SPAN
CLASS="APPLICATION"
>pg_receivexlog</SPAN
></A
>
        tool to archive WAL file changes as they are written
       </P
></LI
><LI
><P
>        Add the <A
HREF="spgist.html"
>SP-GiST</A
> (Space-Partitioned
        GiST) index access method
       </P
></LI
><LI
><P
>        Add support for <A
HREF="rangetypes.html"
>range data types</A
>
       </P
></LI
><LI
><P
>        Add a <A
HREF="datatype-json.html"
><TT
CLASS="TYPE"
>JSON</TT
></A
>
        data type
       </P
></LI
><LI
><P
>        Add a <A
HREF="sql-createview.html"
><TT
CLASS="LITERAL"
>security_barrier</TT
></A
>
        option for views
       </P
></LI
><LI
><P
>        Allow <SPAN
CLASS="APPLICATION"
>libpq</SPAN
> connection strings to have the format of a
        <A
HREF="libpq-connect.html#LIBPQ-CONNSTRING"
><ACRONYM
CLASS="ACRONYM"
>URI</ACRONYM
></A
>
       </P
></LI
><LI
><P
>        Add a <A
HREF="libpq-single-row-mode.html"
>single-row processing
        mode</A
> to <SPAN
CLASS="APPLICATION"
>libpq</SPAN
> for better handling of large
        result sets
       </P
></LI
></UL
><P
>    The above items are explained in more detail in the sections below.
   </P
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="AEN120671"
>E.32.2. Migration to Version 9.2</A
></H2
><P
>    A dump/restore using <SPAN
CLASS="APPLICATION"
>pg_dump</SPAN
>, or use of
    <SPAN
CLASS="APPLICATION"
>pg_upgrade</SPAN
>, is required for those wishing
    to migrate data from any previous release.
   </P
><P
>    Version 9.2 contains a number of changes that may affect compatibility
    with previous releases.  Observe the following incompatibilities:
   </P
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN120677"
>E.32.2.1. System Catalogs</A
></H3
><P
></P
><UL
><LI
><P
>        Remove the <TT
CLASS="STRUCTFIELD"
>spclocation</TT
> field from <A
HREF="catalog-pg-tablespace.html"
><TT
CLASS="STRUCTNAME"
>pg_tablespace</TT
></A
>
        (Magnus Hagander)
       </P
><P
>        This field was duplicative of the symbolic links that actually define
        tablespace locations, and thus risked errors of omission when moving
        a tablespace.  This change allows tablespace directories to be moved
        while the server is down, by manually adjusting the symbolic links.
        To replace this field, we have added <A
HREF="functions-info.html#FUNCTIONS-INFO-CATALOG-TABLE"
><CODE
CLASS="FUNCTION"
>pg_tablespace_location()</CODE
></A
>
        to allow querying of the symbolic links.
       </P
></LI
><LI
><P
>        Move <TT
CLASS="TYPE"
>tsvector</TT
> most-common-element statistics to new
        <A
HREF="view-pg-stats.html"
><TT
CLASS="STRUCTNAME"
>pg_stats</TT
></A
> columns
        (Alexander Korotkov)
       </P
><P
>        Consult <TT
CLASS="STRUCTFIELD"
>most_common_elems</TT
>
        and <TT
CLASS="STRUCTFIELD"
>most_common_elem_freqs</TT
> for the data formerly
        available in <TT
CLASS="STRUCTFIELD"
>most_common_vals</TT
>
        and <TT
CLASS="STRUCTFIELD"
>most_common_freqs</TT
> for a <TT
CLASS="TYPE"
>tsvector</TT
> column.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN120699"
>E.32.2.2. Functions</A
></H3
><P
></P
><UL
><LI
><P
>        Remove <A
HREF="hstore.html"
>hstore</A
>'s <TT
CLASS="LITERAL"
>=&gt;</TT
>
        operator (Robert Haas)
       </P
><P
>        Users should now use <CODE
CLASS="FUNCTION"
>hstore(text, text)</CODE
>.  Since
        <SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> 9.0, a warning message has been
        emitted when an operator named <TT
CLASS="LITERAL"
>=&gt;</TT
> is created because
        the <ACRONYM
CLASS="ACRONYM"
>SQL</ACRONYM
> standard reserves that token for
        another use.
       </P
></LI
><LI
><P
>        Ensure that <A
HREF="functions-xml.html#FUNCTIONS-XML-PROCESSING"
><CODE
CLASS="FUNCTION"
>xpath()</CODE
></A
>
        escapes special characters in string values (Florian Pflug)
       </P
><P
>        Without this it is possible for the result not to be valid
        <ACRONYM
CLASS="ACRONYM"
>XML</ACRONYM
>.
       </P
></LI
><LI
><P
>        Make <A
HREF="functions-admin.html#FUNCTIONS-ADMIN-DBOBJECT"
><CODE
CLASS="FUNCTION"
>pg_relation_size()</CODE
></A
>
        and friends return NULL if the object does not exist (Phil Sorber)
       </P
><P
>        This prevents queries that call these functions from returning
        errors immediately after a concurrent <TT
CLASS="COMMAND"
>DROP</TT
>.
       </P
></LI
><LI
><P
>        Make <A
HREF="functions-datetime.html#FUNCTIONS-DATETIME-EXTRACT"
><CODE
CLASS="FUNCTION"
>EXTRACT(EPOCH FROM
        <TT
CLASS="REPLACEABLE"
><I
>timestamp without time zone</I
></TT
>)</CODE
></A
>
        measure the epoch from local midnight, not <ACRONYM
CLASS="ACRONYM"
>UTC</ACRONYM
>
        midnight (Tom Lane)
       </P
><P
>        This change reverts an ill-considered change made in release 7.3.
        Measuring from <ACRONYM
CLASS="ACRONYM"
>UTC</ACRONYM
> midnight was inconsistent
        because it made the result dependent on the <A
HREF="runtime-config-client.html#GUC-TIMEZONE"
><TT
CLASS="VARNAME"
>timezone</TT
></A
> setting, which
        computations for <TT
CLASS="TYPE"
>timestamp without time zone</TT
> should not be.
        The previous behavior remains available by casting the input value
        to <TT
CLASS="TYPE"
>timestamp with time zone</TT
>.
       </P
></LI
><LI
><P
>        Properly parse time strings with trailing <TT
CLASS="LITERAL"
>yesterday</TT
>,
        <TT
CLASS="LITERAL"
>today</TT
>, and <TT
CLASS="LITERAL"
>tomorrow</TT
> (Dean Rasheed)
       </P
><P
>        Previously, <TT
CLASS="LITERAL"
>SELECT '04:00:00 yesterday'::timestamp</TT
>
        returned yesterday's date at midnight.
       </P
></LI
><LI
><P
>        Fix <A
HREF="functions-formatting.html"
><CODE
CLASS="FUNCTION"
>to_date()</CODE
></A
> and
        <CODE
CLASS="FUNCTION"
>to_timestamp()</CODE
> to wrap incomplete dates toward 2020
        (Bruce Momjian)
       </P
><P
>        Previously, supplied years and year masks of less than four digits
        wrapped inconsistently.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN120748"
>E.32.2.3. Object Modification</A
></H3
><P
></P
><UL
><LI
><P
>        Prevent  <A
HREF="sql-alterdomain.html"
><TT
CLASS="COMMAND"
>ALTER
        DOMAIN</TT
></A
> from working on non-domain types (Peter
        Eisentraut)
       </P
><P
>        Owner and schema changes were previously possible on non-domain
        types.
       </P
></LI
><LI
><P
>        No longer forcibly lowercase procedural language names in <A
HREF="sql-createfunction.html"
><TT
CLASS="COMMAND"
>CREATE FUNCTION</TT
></A
>
        (Robert Haas)
       </P
><P
>        While unquoted language identifiers are still lowercased, strings
        and quoted identifiers are no longer forcibly down-cased.
        Thus for example <TT
CLASS="LITERAL"
>CREATE FUNCTION ... LANGUAGE 'C'</TT
>
        will no longer work; it must be spelled <TT
CLASS="LITERAL"
>'c'</TT
>, or better
        omit the quotes.
       </P
></LI
><LI
><P
>        Change system-generated names of foreign key enforcement triggers
        (Tom Lane)
       </P
><P
>        This change ensures that the triggers fire in the correct order in
        some corner cases involving self-referential foreign key constraints.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN120766"
>E.32.2.4. Command-Line Tools</A
></H3
><P
></P
><UL
><LI
><P
>        Provide consistent backquote, variable
        expansion, and quoted substring behavior in <A
HREF="app-psql.html"
><SPAN
CLASS="APPLICATION"
>psql</SPAN
></A
> meta-command
        arguments (Tom Lane)
       </P
><P
>        Previously, such references were treated oddly when not separated by
        whitespace from adjacent text.  For example <TT
CLASS="LITERAL"
>'FOO'BAR</TT
> was
        output as <TT
CLASS="LITERAL"
>FOO BAR</TT
> (unexpected insertion of a space) and
        <TT
CLASS="LITERAL"
>FOO'BAR'BAZ</TT
> was output unchanged (not removing the quotes
        as most would expect).
       </P
></LI
><LI
><P
>        No longer treat <A
HREF="app-clusterdb.html"
><SPAN
CLASS="APPLICATION"
>clusterdb</SPAN
></A
>
        table names as double-quoted; no longer treat <A
HREF="app-reindexdb.html"
><SPAN
CLASS="APPLICATION"
>reindexdb</SPAN
></A
> table
        and index names as double-quoted (Bruce Momjian)
       </P
><P
>        Users must now include double-quotes in the command arguments if
        quoting is wanted.
       </P
></LI
><LI
><P
>        <A
HREF="app-createuser.html"
><SPAN
CLASS="APPLICATION"
>createuser</SPAN
></A
>
        no longer prompts for option settings by default (Peter Eisentraut)
       </P
><P
>        Use <TT
CLASS="OPTION"
>--interactive</TT
> to obtain the old behavior.
       </P
></LI
><LI
><P
>        Disable prompting for the user name in <A
HREF="app-dropuser.html"
><SPAN
CLASS="APPLICATION"
>dropuser</SPAN
></A
> unless
        <TT
CLASS="OPTION"
>--interactive</TT
> is specified (Peter Eisentraut)
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN120795"
>E.32.2.5. Server Settings</A
></H3
><P
></P
><UL
><LI
><P
>        Add server parameters for specifying the <A
HREF="runtime-config-connection.html#GUC-SSL-CA-FILE"
>locations of server-side
        <ACRONYM
CLASS="ACRONYM"
>SSL</ACRONYM
> files</A
> (Peter Eisentraut)
       </P
><P
>        This allows changing the names and locations of the files that were
        previously hard-coded as <TT
CLASS="FILENAME"
>server.crt</TT
>,
        <TT
CLASS="FILENAME"
>server.key</TT
>, <TT
CLASS="FILENAME"
>root.crt</TT
>, and
        <TT
CLASS="FILENAME"
>root.crl</TT
> in the data directory.
        <SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>The server will no longer examine <TT
CLASS="FILENAME"
>root.crt</TT
> or
        <TT
CLASS="FILENAME"
>root.crl</TT
> by default</I
></SPAN
>; to load these files, the
        associated parameters must be set to non-default values.
       </P
></LI
><LI
><P
>        Remove the <TT
CLASS="VARNAME"
>silent_mode</TT
> parameter (Heikki Linnakangas)
       </P
><P
>        Similar behavior can be obtained with <TT
CLASS="COMMAND"
>pg_ctl start
        -l postmaster.log</TT
>.
       </P
></LI
><LI
><P
>        Remove the <TT
CLASS="VARNAME"
>wal_sender_delay</TT
> parameter,
        as it is no longer needed (Tom Lane)
       </P
></LI
><LI
><P
>        Remove the <TT
CLASS="VARNAME"
>custom_variable_classes</TT
> parameter (Tom Lane)
       </P
><P
>        The checking provided by this setting was dubious.  Now any
        setting can be prefixed by any class name.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN120822"
>E.32.2.6. Monitoring</A
></H3
><P
></P
><UL
><LI
><P
>        Rename <A
HREF="monitoring-stats.html#MONITORING-STATS-VIEWS-TABLE"
><TT
CLASS="STRUCTNAME"
>pg_stat_activity</TT
></A
><TT
CLASS="STRUCTFIELD"
>.procpid</TT
>
        to <TT
CLASS="STRUCTFIELD"
>pid</TT
>, to match other system tables (Magnus Hagander)
       </P
></LI
><LI
><P
>        Create a separate <TT
CLASS="STRUCTFIELD"
>pg_stat_activity</TT
> column to
        report process state (Scott Mead, Magnus Hagander)
       </P
><P
>        The previous <TT
CLASS="STRUCTFIELD"
>query</TT
> and <TT
CLASS="STRUCTFIELD"
>query_start</TT
>
        values now remain available for an idle session, allowing enhanced
        analysis.
       </P
></LI
><LI
><P
>        Rename <TT
CLASS="STRUCTNAME"
>pg_stat_activity</TT
>.<TT
CLASS="STRUCTFIELD"
>current_query</TT
> to
        <TT
CLASS="STRUCTFIELD"
>query</TT
> because it is not cleared when the query
        completes (Magnus Hagander)
       </P
></LI
><LI
><P
>        Change all <ACRONYM
CLASS="ACRONYM"
>SQL</ACRONYM
>-level statistics timing values
        to be <TT
CLASS="TYPE"
>float8</TT
> columns measured in milliseconds (Tom Lane)
       </P
><P
>        This change eliminates the designed-in assumption that the values
        are accurate to microseconds and no more (since the <TT
CLASS="TYPE"
>float8</TT
>
        values can be fractional).
        The columns affected are
        <TT
CLASS="STRUCTNAME"
>pg_stat_user_functions</TT
>.<TT
CLASS="STRUCTFIELD"
>total_time</TT
>,
        <TT
CLASS="STRUCTNAME"
>pg_stat_user_functions</TT
>.<TT
CLASS="STRUCTFIELD"
>self_time</TT
>,
        <TT
CLASS="STRUCTNAME"
>pg_stat_xact_user_functions</TT
>.<TT
CLASS="STRUCTFIELD"
>total_time</TT
>,
        and
        <TT
CLASS="STRUCTNAME"
>pg_stat_xact_user_functions</TT
>.<TT
CLASS="STRUCTFIELD"
>self_time</TT
>.
        The statistics functions underlying these columns now also return
        <TT
CLASS="TYPE"
>float8</TT
> milliseconds, rather than <TT
CLASS="TYPE"
>bigint</TT
>
        microseconds.
        <TT
CLASS="FILENAME"
>contrib/pg_stat_statements</TT
>'
        <TT
CLASS="STRUCTFIELD"
>total_time</TT
> column is now also measured in
        milliseconds.
       </P
></LI
></UL
></DIV
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="AEN120860"
>E.32.3. Changes</A
></H2
><P
>    Below you will find a detailed account of the changes between
    <SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> 9.2 and the previous major
    release.
   </P
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN120864"
>E.32.3.1. Server</A
></H3
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN120866"
>E.32.3.1.1. Performance</A
></H4
><P
></P
><UL
><LI
><P
>        Allow queries to retrieve data only from indexes, avoiding heap
        access (Robert Haas, Ibrar Ahmed, Heikki Linnakangas, Tom Lane)
       </P
><P
>        This feature is often called <I
CLASS="FIRSTTERM"
>index-only scans</I
>.
        Heap access can be skipped for heap pages containing only tuples that
        are visible to all sessions, as reported by the visibility map; so
        the benefit applies mainly to mostly-static data.  The visibility map
        was made crash-safe as a necessary part of implementing this feature.
       </P
></LI
><LI
><P
>        Add the <A
HREF="spgist.html"
>SP-GiST</A
> (Space-Partitioned
        GiST) index access method (Teodor Sigaev, Oleg Bartunov, Tom
        Lane)
       </P
><P
>        SP-GiST is comparable to GiST in flexibility, but supports
        unbalanced partitioned search structures rather than balanced
        trees.  For suitable problems, SP-GiST can be faster than GiST in both
        index build time and search time.
       </P
></LI
><LI
><P
>        Allow group commit to work effectively under heavy load (Peter
        Geoghegan, Simon Riggs, Heikki Linnakangas)
       </P
><P
>        Previously, batching of commits became ineffective as the write
        workload increased, because of internal lock contention.
       </P
></LI
><LI
><P
>        Allow uncontended locks to be managed using a new
        fast-path lock mechanism (Robert Haas)
       </P
></LI
><LI
><P
>        Reduce overhead of creating virtual transaction ID locks (Robert
        Haas)
       </P
></LI
><LI
><P
>        Reduce the overhead of serializable isolation level locks (Dan
        Ports)
       </P
></LI
><LI
><P
>        Improve PowerPC and Itanium spinlock performance (Manabu Ori,
        Robert Haas, Tom Lane)
       </P
></LI
><LI
><P
>        Reduce overhead for shared invalidation cache messages (Robert
        Haas)
       </P
></LI
><LI
><P
>        Move the frequently accessed members of the <TT
CLASS="STRUCTNAME"
>PGPROC</TT
>
        shared memory array to a separate array (Pavan
        Deolasee, Heikki Linnakangas, Robert Haas)
       </P
></LI
><LI
><P
>        Improve <TT
CLASS="COMMAND"
>COPY</TT
> performance by adding tuples to
        the heap in batches (Heikki Linnakangas)
       </P
></LI
><LI
><P
>        Improve GiST index performance for geometric data types by producing
        better trees with less memory allocation overhead (Alexander Korotkov)
       </P
></LI
><LI
><P
>        Improve GiST index build times (Alexander Korotkov, Heikki
        Linnakangas)
       </P
></LI
><LI
><P
>        Allow hint bits to be set sooner for temporary and unlogged tables
        (Robert Haas)
       </P
></LI
><LI
><P
>        Allow sorting to be performed by inlined,
        non-<ACRONYM
CLASS="ACRONYM"
>SQL</ACRONYM
>-callable comparison functions (Peter
        Geoghegan, Robert Haas, Tom Lane)
       </P
></LI
><LI
><P
>        Make the number of CLOG buffers scale based on <A
HREF="runtime-config-resource.html#GUC-SHARED-BUFFERS"
><TT
CLASS="VARNAME"
>shared_buffers</TT
></A
>
        (Robert Haas, Simon Riggs, Tom Lane)
       </P
></LI
><LI
><P
>        Improve performance of buffer pool scans that occur when tables or
        databases are dropped (Jeff Janes, Simon Riggs)
       </P
></LI
><LI
><P
>        Improve performance of checkpointer's fsync-request queue
        when many tables are being dropped or truncated (Tom Lane)
       </P
></LI
><LI
><P
>        Pass the safe number of file descriptors to child processes on Windows
        (Heikki Linnakangas)
       </P
><P
>        This allows Windows sessions to use more open file descriptors than
        before.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN120916"
>E.32.3.1.2. Process Management</A
></H4
><P
></P
><UL
><LI
><P
>        Create a dedicated background process to perform checkpoints (Simon
        Riggs)
       </P
><P
>        Formerly the background writer did both dirty-page writing and
        checkpointing.  Separating this into two processes allows each goal
        to be accomplished more predictably.
       </P
></LI
><LI
><P
>        Improve asynchronous commit behavior by waking the walwriter sooner
        (Simon Riggs)
       </P
><P
>        Previously, only <A
HREF="runtime-config-wal.html#GUC-WAL-WRITER-DELAY"
><TT
CLASS="VARNAME"
>wal_writer_delay</TT
></A
>
        triggered <ACRONYM
CLASS="ACRONYM"
>WAL</ACRONYM
> flushing to disk; now filling a
        <ACRONYM
CLASS="ACRONYM"
>WAL</ACRONYM
> buffer also triggers <ACRONYM
CLASS="ACRONYM"
>WAL</ACRONYM
>
        writes.
       </P
></LI
><LI
><P
>        Allow the bgwriter, walwriter, checkpointer, statistics collector,
        log collector, and archiver background processes to sleep more
        efficiently during periods of inactivity (Peter Geoghegan, Tom Lane)
       </P
><P
>        This series of changes reduces the frequency of process wake-ups when
        there is nothing to do, dramatically reducing power consumption on
        idle servers.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN120933"
>E.32.3.1.3. Optimizer</A
></H4
><P
></P
><UL
><LI
><P
>        Allow the planner to generate custom plans for specific parameter
        values even when using prepared statements
        (Tom Lane)
       </P
><P
>        In the past, a prepared statement always had a single
        <SPAN
CLASS="QUOTE"
>"generic"</SPAN
> plan that was used for all parameter values, which
        was frequently much inferior to the plans used for non-prepared
        statements containing explicit constant values.  Now, the planner
        attempts to generate custom plans for specific parameter values.
        A generic plan will only be used after custom plans have repeatedly
        proven to provide no benefit.  This change should eliminate the
        performance penalties formerly seen from use of prepared statements
        (including non-dynamic statements in PL/pgSQL).
       </P
></LI
><LI
><P
>        Improve the planner's ability to use nested loops with inner
        index scans (Tom Lane)
       </P
><P
>        The new <SPAN
CLASS="QUOTE"
>"parameterized path"</SPAN
> mechanism allows inner
        index scans to use values from relations that are more than one join
        level up from the scan.  This can greatly improve performance in
        situations where semantic restrictions (such as outer joins) limit
        the allowed join orderings.
       </P
></LI
><LI
><P
>        Improve the planning <ACRONYM
CLASS="ACRONYM"
>API</ACRONYM
> for foreign data wrappers
        (Etsuro Fujita, Shigeru Hanada, Tom Lane)
       </P
><P
>        Wrappers can now provide multiple access <SPAN
CLASS="QUOTE"
>"paths"</SPAN
> for their
        tables, allowing more flexibility in join planning.
       </P
></LI
><LI
><P
>        Recognize self-contradictory restriction clauses for non-table
        relations (Tom Lane)
       </P
><P
>        This check is only performed when <A
HREF="runtime-config-query.html#GUC-CONSTRAINT-EXCLUSION"
><TT
CLASS="VARNAME"
>constraint_exclusion</TT
></A
>
        is <TT
CLASS="LITERAL"
>on</TT
>.
       </P
></LI
><LI
><P
>        Allow <TT
CLASS="LITERAL"
>indexed_col op ANY(ARRAY[...])</TT
> conditions to be
        used in plain index scans and index-only scans (Tom Lane)
       </P
><P
>        Formerly such conditions could only be used in bitmap index scans.
       </P
></LI
><LI
><P
>        Support <CODE
CLASS="FUNCTION"
>MIN</CODE
>/<CODE
CLASS="FUNCTION"
>MAX</CODE
> index optimizations on
        <TT
CLASS="TYPE"
>boolean</TT
> columns (Marti Raudsepp)
       </P
></LI
><LI
><P
>        Account for set-returning functions in <TT
CLASS="COMMAND"
>SELECT</TT
> target
        lists when setting row count estimates (Tom Lane)
       </P
></LI
><LI
><P
>        Fix planner to handle indexes with duplicated columns more reliably
        (Tom Lane)
       </P
></LI
><LI
><P
>        Collect and use element-frequency statistics for arrays (Alexander
        Korotkov, Tom Lane)
       </P
><P
>        This change improves selectivity estimation for the array
        <TT
CLASS="LITERAL"
>&lt;@</TT
>, <TT
CLASS="LITERAL"
>&amp;&amp;</TT
>, and
        <TT
CLASS="LITERAL"
>@&gt;</TT
> operators (array containment and overlaps).
       </P
></LI
><LI
><P
>        Allow statistics to be collected for foreign tables
        (Etsuro Fujita)
       </P
></LI
><LI
><P
>        Improve cost estimates for use of partial indexes (Tom Lane)
       </P
></LI
><LI
><P
>        Improve the planner's ability to use statistics for columns
        referenced in subqueries (Tom Lane)
       </P
></LI
><LI
><P
>        Improve statistical estimates for subqueries using
        <TT
CLASS="LITERAL"
>DISTINCT</TT
> (Tom Lane)
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN120984"
>E.32.3.1.4. Authentication</A
></H4
><P
></P
><UL
><LI
><P
>        Do not treat role names and <TT
CLASS="LITERAL"
>samerole</TT
> specified in <A
HREF="auth-pg-hba-conf.html"
><TT
CLASS="FILENAME"
>pg_hba.conf</TT
></A
>
        as automatically including superusers (Andrew Dunstan)
       </P
><P
>        This makes it easier to use <TT
CLASS="LITERAL"
>reject</TT
> lines with group roles.
       </P
></LI
><LI
><P
>        Adjust <TT
CLASS="FILENAME"
>pg_hba.conf</TT
> processing to handle token
        parsing more consistently (Brendan Jurd, &Aacute;lvaro Herrera)
       </P
></LI
><LI
><P
>        Disallow empty <TT
CLASS="FILENAME"
>pg_hba.conf</TT
> files (Tom Lane)
       </P
><P
>        This was done to more quickly detect misconfiguration.
       </P
></LI
><LI
><P
>        Make superuser privilege imply replication privilege (Noah Misch)
       </P
><P
>        This avoids the need to explicitly assign such privileges.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121004"
>E.32.3.1.5. Monitoring</A
></H4
><P
></P
><UL
><LI
><P
>        Attempt to log the current query string during a backend crash
        (Marti Raudsepp)
       </P
></LI
><LI
><P
>        Make logging of autovacuum I/O activity more verbose (Greg
        Smith, Noah Misch)
       </P
><P
>        This logging is triggered by <A
HREF="runtime-config-autovacuum.html#GUC-LOG-AUTOVACUUM-MIN-DURATION"
><TT
CLASS="VARNAME"
>log_autovacuum_min_duration</TT
></A
>.
       </P
></LI
><LI
><P
>        Make <ACRONYM
CLASS="ACRONYM"
>WAL</ACRONYM
> replay report failures sooner
        (Fujii Masao)
       </P
><P
>        There were some cases where failures were only reported once the
        server went into master mode.
       </P
></LI
><LI
><P
>        Add <A
HREF="functions-admin.html#FUNCTIONS-ADMIN-BACKUP"
><CODE
CLASS="FUNCTION"
>pg_xlog_location_diff()</CODE
></A
>
        to simplify WAL location comparisons (Euler Taveira de Oliveira)
       </P
><P
>        This is useful for computing replication lag.
       </P
></LI
><LI
><P
>        Support configurable event log application names on Windows
        (MauMau, Magnus Hagander)
       </P
><P
>        This allows different instances to use the event log
        with different identifiers, by setting the <A
HREF="runtime-config-logging.html#GUC-EVENT-SOURCE"
><TT
CLASS="VARNAME"
>event_source</TT
></A
>
        server parameter, which is similar to how <A
HREF="runtime-config-logging.html#GUC-SYSLOG-IDENT"
><TT
CLASS="VARNAME"
>syslog_ident</TT
></A
> works.
       </P
></LI
><LI
><P
>        Change <SPAN
CLASS="QUOTE"
>"unexpected EOF"</SPAN
> messages to <TT
CLASS="LITERAL"
>DEBUG1</TT
> level,
        except when there is an open transaction (Magnus Hagander)
       </P
><P
>        This change reduces log chatter caused by applications that close
        database connections ungracefully.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121035"
>E.32.3.1.6. Statistical Views</A
></H4
><P
></P
><UL
><LI
><P
>        Track temporary file sizes and file counts in the <A
HREF="monitoring-stats.html#PG-STAT-DATABASE-VIEW"
><TT
CLASS="STRUCTNAME"
>pg_stat_database</TT
></A
>
        system view (Tomas Vondra)
       </P
></LI
><LI
><P
>        Add a deadlock counter to the <TT
CLASS="STRUCTNAME"
>pg_stat_database</TT
>
        system view (Magnus Hagander)
       </P
></LI
><LI
><P
>        Add a server parameter <A
HREF="runtime-config-statistics.html#GUC-TRACK-IO-TIMING"
><TT
CLASS="VARNAME"
>track_io_timing</TT
></A
>
        to track I/O timings (Ants Aasma, Robert Haas)
       </P
></LI
><LI
><P
>        Report checkpoint timing information in <A
HREF="monitoring-stats.html#PG-STAT-BGWRITER-VIEW"
><TT
CLASS="STRUCTNAME"
>pg_stat_bgwriter</TT
></A
>
        (Greg Smith, Peter Geoghegan)
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121053"
>E.32.3.1.7. Server Settings</A
></H4
><P
></P
><UL
><LI
><P
>        Silently ignore nonexistent schemas specified in <A
HREF="runtime-config-client.html#GUC-SEARCH-PATH"
><TT
CLASS="VARNAME"
>search_path</TT
></A
> (Tom Lane)
       </P
><P
>        This makes it more convenient to use generic path settings, which
        might include some schemas that don't exist in all databases.
       </P
></LI
><LI
><P
>        Allow superusers to set <A
HREF="runtime-config-locks.html#GUC-DEADLOCK-TIMEOUT"
><TT
CLASS="VARNAME"
>deadlock_timeout</TT
></A
>
        per-session, not just per-cluster (Noah Misch)
       </P
><P
>        This allows <TT
CLASS="VARNAME"
>deadlock_timeout</TT
> to be reduced for
        transactions that are likely to be involved in a deadlock, thus
        detecting the failure more quickly.  Alternatively, increasing the
        value can be used to reduce the chances of a session being chosen for
        cancellation due to a deadlock.
       </P
></LI
><LI
><P
>        Add a server parameter <A
HREF="runtime-config-resource.html#GUC-TEMP-FILE-LIMIT"
><TT
CLASS="VARNAME"
>temp_file_limit</TT
></A
>
        to constrain temporary file space usage per session (Mark Kirkwood)
       </P
></LI
><LI
><P
>        Allow a superuser to <TT
CLASS="COMMAND"
>SET</TT
> an extension's
        superuser-only custom variable before loading the associated
        extension (Tom Lane)
       </P
><P
>        The system now remembers whether a <TT
CLASS="COMMAND"
>SET</TT
> was
        performed by a superuser, so that proper privilege checking can be
        done when the extension is loaded.
       </P
></LI
><LI
><P
>        Add <A
HREF="app-postmaster.html"
>postmaster</A
> <TT
CLASS="OPTION"
>-C</TT
>
        option to query configuration parameters (Bruce Momjian)
       </P
><P
>        This allows <SPAN
CLASS="APPLICATION"
>pg_ctl</SPAN
> to better handle cases where
        <TT
CLASS="ENVAR"
>PGDATA</TT
> or <TT
CLASS="OPTION"
>-D</TT
> points to a configuration-only
        directory.
       </P
></LI
><LI
><P
>        Replace an empty locale name with the implied value in
        <TT
CLASS="COMMAND"
>CREATE DATABASE</TT
>
        (Tom Lane)
       </P
><P
>        This prevents cases where
        <TT
CLASS="STRUCTNAME"
>pg_database</TT
>.<TT
CLASS="STRUCTFIELD"
>datcollate</TT
> or
        <TT
CLASS="STRUCTFIELD"
>datctype</TT
> could be interpreted differently after a
        server restart.
       </P
></LI
></UL
><DIV
CLASS="SECT5"
><H5
CLASS="SECT5"
><A
NAME="AEN121091"
>E.32.3.1.7.1. <TT
CLASS="FILENAME"
>postgresql.conf</TT
></A
></H5
><P
></P
><UL
><LI
><P
>        Allow multiple errors in <TT
CLASS="FILENAME"
>postgresql.conf</TT
>
        to be reported, rather than just the first one (Alexey Klyukin,
        Tom Lane)
       </P
></LI
><LI
><P
>        Allow a reload of <TT
CLASS="FILENAME"
>postgresql.conf</TT
> to be
        processed by all sessions, even if there are some settings that
        are invalid for particular sessions (Alexey Klyukin)
       </P
><P
>        Previously, such not-valid-within-session values would cause all
        setting changes to be ignored by that session.
       </P
></LI
><LI
><P
>        Add an <TT
CLASS="LITERAL"
>include_if_exists</TT
> facility for configuration
        files (Greg Smith)
       </P
><P
>        This works the same as <TT
CLASS="LITERAL"
>include</TT
>, except that an error
        is not thrown if the file is missing.
       </P
></LI
><LI
><P
>        Identify the server time zone during <SPAN
CLASS="APPLICATION"
>initdb</SPAN
>, and set
        <TT
CLASS="FILENAME"
>postgresql.conf</TT
> entries
        <A
HREF="runtime-config-client.html#GUC-TIMEZONE"
><TT
CLASS="VARNAME"
>timezone</TT
></A
> and
        <A
HREF="runtime-config-logging.html#GUC-LOG-TIMEZONE"
><TT
CLASS="VARNAME"
>log_timezone</TT
></A
>
        accordingly (Tom Lane)
       </P
><P
>        This avoids expensive time zone probes during server start.
       </P
></LI
><LI
><P
>        Fix <A
HREF="view-pg-settings.html"
><TT
CLASS="STRUCTNAME"
>pg_settings</TT
></A
> to
        report <TT
CLASS="FILENAME"
>postgresql.conf</TT
> line numbers on Windows
        (Tom Lane)
       </P
></LI
></UL
></DIV
></DIV
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN121121"
>E.32.3.2. Replication and Recovery</A
></H3
><P
></P
><UL
><LI
><P
>        Allow streaming replication slaves to forward data to other slaves
        (<A
HREF="warm-standby.html#CASCADING-REPLICATION"
><I
CLASS="FIRSTTERM"
>cascading
        replication</I
></A
>) (Fujii Masao)
       </P
><P
>        Previously, only the master server could supply streaming
        replication log files to standby servers.
       </P
></LI
><LI
><P
>        Add new <A
HREF="runtime-config-wal.html#GUC-SYNCHRONOUS-COMMIT"
><TT
CLASS="VARNAME"
>synchronous_commit</TT
></A
>
        mode <TT
CLASS="LITERAL"
>remote_write</TT
> (Fujii Masao, Simon Riggs)
       </P
><P
>        This mode waits for the standby server to write transaction data to
        its own operating system, but does not wait for the data to be
        flushed to the standby's disk.
       </P
></LI
><LI
><P
>        Add a <A
HREF="app-pgreceivexlog.html"
><SPAN
CLASS="APPLICATION"
>pg_receivexlog</SPAN
></A
>
        tool to archive WAL file changes as they are written, rather
        than waiting for completed WAL files (Magnus Hagander)
       </P
></LI
><LI
><P
>        Allow <A
HREF="app-pgbasebackup.html"
><SPAN
CLASS="APPLICATION"
>pg_basebackup</SPAN
></A
>
        to make base backups from standby servers (Jun Ishizuka, Fujii Masao)
       </P
><P
>        This feature lets the work of making new base backups be off-loaded
        from the primary server.
       </P
></LI
><LI
><P
>        Allow streaming of WAL files while <SPAN
CLASS="APPLICATION"
>pg_basebackup</SPAN
>
        is performing a backup (Magnus Hagander)
       </P
><P
>        This allows passing of WAL files to the standby before they are
        discarded on the primary.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN121148"
>E.32.3.3. Queries</A
></H3
><P
></P
><UL
><LI
><P
>        Cancel the running query if the client gets disconnected
        (Florian Pflug)
       </P
><P
>        If the backend detects loss of client connection during a query, it
        will now cancel the query rather than attempting to finish it.
       </P
></LI
><LI
><P
>        Retain column names at run time for row expressions
        (Andrew Dunstan, Tom Lane)
       </P
><P
>        This change allows better results when a row value is converted to
        <TT
CLASS="TYPE"
>hstore</TT
> or <TT
CLASS="TYPE"
>json</TT
> type: the fields of the resulting
        value will now have the expected names.
       </P
></LI
><LI
><P
>        Improve column labels used for sub-<TT
CLASS="COMMAND"
>SELECT</TT
> results
        (Marti Raudsepp)
       </P
><P
>        Previously, the generic label <TT
CLASS="LITERAL"
>?column?</TT
> was used.
       </P
></LI
><LI
><P
>        Improve heuristics for determining the types of unknown values
        (Tom Lane)
       </P
><P
>        The longstanding rule that an unknown constant might have the
        same type as the value on the other side of the operator using it
        is now applied when considering polymorphic operators, not only
        for simple operator matches.
       </P
></LI
><LI
><P
>        Warn about creating casts to or from domain types (Robert Haas)
       </P
><P
>        Such casts have no effect.
       </P
></LI
><LI
><P
>        When a row fails a <TT
CLASS="LITERAL"
>CHECK</TT
> or <TT
CLASS="LITERAL"
>NOT NULL</TT
>
        constraint, show the row's contents as error detail (Jan
        Kundr&aacute;t)
       </P
><P
>        This should make it easier to identify which row is problematic
        when an insert or update is processing many rows.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN121175"
>E.32.3.4. Object Manipulation</A
></H3
><P
></P
><UL
><LI
><P
>        Provide more reliable operation during concurrent
        <ACRONYM
CLASS="ACRONYM"
>DDL</ACRONYM
> (Robert Haas, Noah Misch)
       </P
><P
>        This change adds locking that should eliminate <SPAN
CLASS="QUOTE"
>"cache lookup
        failed"</SPAN
> errors in many scenarios.  Also, it is no longer possible
        to add relations to a schema that is being concurrently dropped, a
        scenario that formerly led to inconsistent system catalog contents.
       </P
></LI
><LI
><P
>        Add <TT
CLASS="LITERAL"
>CONCURRENTLY</TT
> option to <A
HREF="sql-dropindex.html"
><TT
CLASS="COMMAND"
>DROP INDEX</TT
></A
>
        (Simon Riggs)
       </P
><P
>        This allows index removal without blocking other sessions.
       </P
></LI
><LI
><P
>        Allow foreign data wrappers to have per-column options (Shigeru Hanada)
       </P
></LI
><LI
><P
>        Improve pretty-printing of view definitions (Andrew Dunstan)
       </P
></LI
></UL
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121193"
>E.32.3.4.1. Constraints</A
></H4
><P
></P
><UL
><LI
><P
>        Allow <A
HREF="ddl-constraints.html"
><TT
CLASS="LITERAL"
>CHECK</TT
></A
>
        constraints to be declared <TT
CLASS="LITERAL"
>NOT VALID</TT
> (&Aacute;lvaro
        Herrera)
       </P
><P
>        Adding a <TT
CLASS="LITERAL"
>NOT VALID</TT
> constraint does not cause the table to
        be scanned to verify that existing rows meet the constraint.
        Subsequently, newly added or updated rows are checked.
        Such constraints are ignored by the planner when considering
        <TT
CLASS="VARNAME"
>constraint_exclusion</TT
>, since it is not certain that all
        rows meet the constraint.
       </P
><P
>        The new <TT
CLASS="COMMAND"
>ALTER TABLE VALIDATE</TT
> command allows <TT
CLASS="LITERAL"
>NOT
        VALID</TT
> constraints to be checked for existing rows, after which
        they are converted into ordinary constraints.
       </P
></LI
><LI
><P
>        Allow <TT
CLASS="LITERAL"
>CHECK</TT
> constraints to be declared <TT
CLASS="LITERAL"
>NO
        INHERIT</TT
> (Nikhil Sontakke, Alex Hunsaker, &Aacute;lvaro Herrera)
       </P
><P
>        This makes them enforceable only on the parent table, not on
        child tables.
       </P
></LI
><LI
><P
>        Add the ability to <A
HREF="sql-altertable.html"
>rename</A
>
        constraints (Peter Eisentraut)
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121215"
>E.32.3.4.2. <TT
CLASS="COMMAND"
>ALTER</TT
></A
></H4
><P
></P
><UL
><LI
><P
>        Reduce need to rebuild tables and indexes for certain <A
HREF="sql-altertable.html"
><TT
CLASS="COMMAND"
>ALTER TABLE</TT
></A
>
        ... <TT
CLASS="LITERAL"
>ALTER COLUMN TYPE</TT
> operations (Noah Misch)
       </P
><P
>        Increasing the length limit for a <TT
CLASS="TYPE"
>varchar</TT
> or <TT
CLASS="TYPE"
>varbit</TT
>
        column, or removing the limit altogether, no longer requires a table
        rewrite.  Similarly, increasing the allowable precision of a
        <TT
CLASS="TYPE"
>numeric</TT
> column, or changing a column from constrained
        <TT
CLASS="TYPE"
>numeric</TT
> to unconstrained <TT
CLASS="TYPE"
>numeric</TT
>, no longer
        requires a table rewrite.  Table rewrites are also avoided in similar
        cases involving the <TT
CLASS="TYPE"
>interval</TT
>, <TT
CLASS="TYPE"
>timestamp</TT
>, and
        <TT
CLASS="TYPE"
>timestamptz</TT
> types.
       </P
></LI
><LI
><P
>        Avoid having <A
HREF="sql-altertable.html"
><TT
CLASS="COMMAND"
>ALTER
        TABLE</TT
></A
> revalidate foreign key constraints in some
        cases where it is not necessary (Noah Misch)
       </P
></LI
><LI
><P
>        Add <TT
CLASS="LITERAL"
>IF EXISTS</TT
> options to some <TT
CLASS="COMMAND"
>ALTER</TT
>
        commands (Pavel Stehule)
       </P
><P
>        For example, <TT
CLASS="COMMAND"
>ALTER FOREIGN TABLE IF EXISTS foo RENAME
        TO bar</TT
>.
       </P
></LI
><LI
><P
>        Add <A
HREF="sql-alterforeigndatawrapper.html"
><TT
CLASS="COMMAND"
>ALTER
        FOREIGN DATA WRAPPER</TT
></A
> ... <TT
CLASS="LITERAL"
>RENAME</TT
>
        and <A
HREF="sql-alterserver.html"
><TT
CLASS="COMMAND"
>ALTER
        SERVER</TT
></A
> ... <TT
CLASS="LITERAL"
>RENAME</TT
> (Peter Eisentraut)
       </P
></LI
><LI
><P
>        Add <A
HREF="sql-alterdomain.html"
><TT
CLASS="COMMAND"
>ALTER
        DOMAIN</TT
></A
> ... <TT
CLASS="LITERAL"
>RENAME</TT
> (Peter Eisentraut)
       </P
><P
>        You could already rename domains using <TT
CLASS="COMMAND"
>ALTER
        TYPE</TT
>.
       </P
></LI
><LI
><P
>        Throw an error for <TT
CLASS="COMMAND"
>ALTER DOMAIN</TT
> ... <TT
CLASS="LITERAL"
>DROP
        CONSTRAINT</TT
> on a nonexistent constraint (Peter Eisentraut)
       </P
><P
>        An <TT
CLASS="LITERAL"
>IF EXISTS</TT
> option has been added to provide the
        previous behavior.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121264"
>E.32.3.4.3. <A
HREF="sql-createtable.html"
><TT
CLASS="COMMAND"
>CREATE TABLE</TT
></A
></A
></H4
><P
></P
><UL
><LI
><P
>        Allow <TT
CLASS="COMMAND"
>CREATE TABLE (LIKE ...)</TT
> from foreign
        tables, views, and composite types (Peter Eisentraut)
       </P
><P
>        For example, this allows a table to be created whose schema matches a
        view.
       </P
></LI
><LI
><P
>        Fix <TT
CLASS="COMMAND"
>CREATE TABLE (LIKE ...)</TT
> to avoid index name
        conflicts when copying index comments (Tom Lane)
       </P
></LI
><LI
><P
>        Fix <TT
CLASS="COMMAND"
>CREATE TABLE</TT
> ... <TT
CLASS="LITERAL"
>AS EXECUTE</TT
>
        to handle <TT
CLASS="LITERAL"
>WITH NO DATA</TT
> and column name specifications
        (Tom Lane)
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121281"
>E.32.3.4.4. Object Permissions</A
></H4
><P
></P
><UL
><LI
><P
>        Add a <A
HREF="sql-createview.html"
><TT
CLASS="LITERAL"
>security_barrier</TT
></A
>
        option for views (KaiGai Kohei, Robert Haas)
       </P
><P
>        This option prevents optimizations that might allow view-protected
        data to be exposed to users, for example pushing a clause involving
        an insecure function into the <TT
CLASS="LITERAL"
>WHERE</TT
> clause of the view.
        Such views can be expected to perform more poorly than ordinary
        views.
       </P
></LI
><LI
><P
>        Add a new <A
HREF="sql-createfunction.html"
><TT
CLASS="LITERAL"
>LEAKPROOF</TT
></A
> function
        attribute to mark functions that can safely be pushed down
        into <TT
CLASS="LITERAL"
>security_barrier</TT
> views (KaiGai Kohei)
       </P
></LI
><LI
><P
>        Add support for privileges on data types (Peter Eisentraut)
       </P
><P
>        This adds support for the <ACRONYM
CLASS="ACRONYM"
>SQL</ACRONYM
>-conforming
        <TT
CLASS="LITERAL"
>USAGE</TT
> privilege on types and domains.  The intent is
        to be able to restrict which users can create dependencies on types,
        since such dependencies limit the owner's ability to alter the type.
       </P
></LI
><LI
><P
>        Check for <TT
CLASS="COMMAND"
>INSERT</TT
> privileges in <TT
CLASS="COMMAND"
>SELECT
        INTO</TT
> / <TT
CLASS="COMMAND"
>CREATE TABLE AS</TT
> (KaiGai Kohei)
       </P
><P
>        Because the object is being created by <TT
CLASS="COMMAND"
>SELECT INTO</TT
>
        or <TT
CLASS="COMMAND"
>CREATE TABLE AS</TT
>, the creator would ordinarily
        have insert permissions; but there are corner cases where this is not
        true, such as when <TT
CLASS="LITERAL"
>ALTER DEFAULT PRIVILEGES</TT
> has removed
        such permissions.
       </P
></LI
></UL
></DIV
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN121309"
>E.32.3.5. Utility Operations</A
></H3
><P
></P
><UL
><LI
><P
>        Allow <A
HREF="sql-vacuum.html"
><TT
CLASS="COMMAND"
>VACUUM</TT
></A
> to more
        easily skip pages that cannot be locked (Simon Riggs, Robert Haas)
       </P
><P
>        This change should greatly reduce the incidence of <TT
CLASS="COMMAND"
>VACUUM</TT
>
        getting <SPAN
CLASS="QUOTE"
>"stuck"</SPAN
> waiting for other sessions.
       </P
></LI
><LI
><P
>        Make <A
HREF="sql-explain.html"
><TT
CLASS="COMMAND"
>EXPLAIN</TT
></A
>
        <TT
CLASS="LITERAL"
>(BUFFERS)</TT
> count blocks dirtied and written (Robert Haas)
       </P
></LI
><LI
><P
>        Make <TT
CLASS="COMMAND"
>EXPLAIN ANALYZE</TT
> report the number of rows
        rejected by filter steps (Marko Tiikkaja)
       </P
></LI
><LI
><P
>        Allow <TT
CLASS="COMMAND"
>EXPLAIN ANALYZE</TT
> to avoid timing overhead when
        time values are not wanted (Tomas Vondra)
       </P
><P
>        This is accomplished by setting the new <TT
CLASS="LITERAL"
>TIMING</TT
> option to
        <TT
CLASS="LITERAL"
>FALSE</TT
>.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN121333"
>E.32.3.6. Data Types</A
></H3
><P
></P
><UL
><LI
><P
>        Add support for <A
HREF="rangetypes.html"
>range data types</A
>
        (Jeff Davis, Tom Lane, Alexander Korotkov)
       </P
><P
>        A range data type stores a lower and upper bound belonging to its
        base data type.  It supports operations like contains, overlaps, and
        intersection.
       </P
></LI
><LI
><P
>        Add a <A
HREF="datatype-json.html"
><TT
CLASS="TYPE"
>JSON</TT
></A
>
        data type (Robert Haas)
       </P
><P
>        This type stores <ACRONYM
CLASS="ACRONYM"
>JSON</ACRONYM
> (JavaScript Object Notation)
        data with proper validation.
       </P
></LI
><LI
><P
>        Add <A
HREF="functions-json.html"
><CODE
CLASS="FUNCTION"
>array_to_json()</CODE
></A
>
        and <CODE
CLASS="FUNCTION"
>row_to_json()</CODE
> (Andrew Dunstan)
       </P
></LI
><LI
><P
>        Add a <A
HREF="datatype-numeric.html#DATATYPE-SERIAL"
><TT
CLASS="TYPE"
>SMALLSERIAL</TT
></A
>
        data type (Mike Pultz)
       </P
><P
>        This is like <TT
CLASS="TYPE"
>SERIAL</TT
>, except it stores the sequence in
        a two-byte integer column (<TT
CLASS="TYPE"
>int2</TT
>).
       </P
></LI
><LI
><P
>        Allow <A
HREF="sql-createdomain.html"
>domains</A
> to be
        declared <TT
CLASS="LITERAL"
>NOT VALID</TT
> (&Aacute;lvaro Herrera)
       </P
><P
>        This option can be set at domain creation time, or via <TT
CLASS="COMMAND"
>ALTER
        DOMAIN</TT
> ... <TT
CLASS="LITERAL"
>ADD CONSTRAINT</TT
> ... <TT
CLASS="LITERAL"
>NOT
        VALID</TT
>. <TT
CLASS="COMMAND"
>ALTER DOMAIN</TT
> ... <TT
CLASS="LITERAL"
>VALIDATE
        CONSTRAINT</TT
> fully validates the constraint.
       </P
></LI
><LI
><P
>        Support more locale-specific formatting options for the <A
HREF="datatype-money.html"
><TT
CLASS="TYPE"
>money</TT
></A
> data type (Tom Lane)
       </P
><P
>        Specifically, honor all the POSIX options for ordering of the value,
        sign, and currency symbol in monetary output.  Also, make sure that
        the thousands separator is only inserted to the left of the decimal
        point, as required by POSIX.
       </P
></LI
><LI
><P
>        Add bitwise <SPAN
CLASS="QUOTE"
>"and"</SPAN
>, <SPAN
CLASS="QUOTE"
>"or"</SPAN
>, and <SPAN
CLASS="QUOTE"
>"not"</SPAN
>
        operators for the <TT
CLASS="TYPE"
>macaddr</TT
> data type (Brendan Jurd)
       </P
></LI
><LI
><P
>        Allow <A
HREF="functions-xml.html#FUNCTIONS-XML-PROCESSING"
><CODE
CLASS="FUNCTION"
>xpath()</CODE
></A
> to
        return a single-element <ACRONYM
CLASS="ACRONYM"
>XML</ACRONYM
> array when supplied a
        scalar value (Florian Pflug)
       </P
><P
>        Previously, it returned an empty array.  This change will also
        cause <CODE
CLASS="FUNCTION"
>xpath_exists()</CODE
> to return true, not false,
        for such expressions.
       </P
></LI
><LI
><P
>        Improve <ACRONYM
CLASS="ACRONYM"
>XML</ACRONYM
> error handling to be more robust
        (Florian Pflug)
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN121389"
>E.32.3.7. Functions</A
></H3
><P
></P
><UL
><LI
><P
>        Allow non-superusers to use <A
HREF="functions-admin.html#FUNCTIONS-ADMIN-SIGNAL"
><CODE
CLASS="FUNCTION"
>pg_cancel_backend()</CODE
></A
>
        and <A
HREF="functions-admin.html#FUNCTIONS-ADMIN-SIGNAL"
><CODE
CLASS="FUNCTION"
>pg_terminate_backend()</CODE
></A
>
        on other sessions belonging to the same user
        (Magnus Hagander, Josh Kupershmidt, Dan Farina)
       </P
><P
>        Previously only superusers were allowed to use these functions.
       </P
></LI
><LI
><P
>        Allow importing and exporting of transaction snapshots (Joachim
        Wieland, Tom Lane)
       </P
><P
>        This allows multiple transactions to share identical views of the
        database state.
        Snapshots are exported via <A
HREF="functions-admin.html#FUNCTIONS-SNAPSHOT-SYNCHRONIZATION"
><CODE
CLASS="FUNCTION"
>pg_export_snapshot()</CODE
></A
>
        and imported via <A
HREF="sql-set-transaction.html"
><TT
CLASS="COMMAND"
>SET
        TRANSACTION SNAPSHOT</TT
></A
>.  Only snapshots from
        currently-running transactions can be imported.
       </P
></LI
><LI
><P
>        Support <A
HREF="functions-info.html#FUNCTIONS-INFO-CATALOG-TABLE"
><TT
CLASS="LITERAL"
>COLLATION
        FOR</TT
></A
> on expressions (Peter Eisentraut)
       </P
><P
>        This returns a string representing the collation of the expression.
       </P
></LI
><LI
><P
>        Add <A
HREF="functions-info.html#FUNCTIONS-INFO-SCHEMA-TABLE"
><CODE
CLASS="FUNCTION"
>pg_opfamily_is_visible()</CODE
></A
>
        (Josh Kupershmidt)
       </P
></LI
><LI
><P
>        Add a <TT
CLASS="TYPE"
>numeric</TT
> variant of <A
HREF="functions-admin.html#FUNCTIONS-ADMIN-DBSIZE"
><CODE
CLASS="FUNCTION"
>pg_size_pretty()</CODE
></A
>
        for use with <CODE
CLASS="FUNCTION"
>pg_xlog_location_diff()</CODE
> (Fujii Masao)
       </P
></LI
><LI
><P
>        Add a <A
HREF="functions-info.html#FUNCTIONS-INFO-SESSION-TABLE"
><CODE
CLASS="FUNCTION"
>pg_trigger_depth()</CODE
></A
>
        function (Kevin Grittner)
       </P
><P
>        This reports the current trigger call depth.
       </P
></LI
><LI
><P
>        Allow <A
HREF="functions-aggregate.html#FUNCTIONS-AGGREGATE-TABLE"
><CODE
CLASS="FUNCTION"
>string_agg()</CODE
></A
>
        to process <TT
CLASS="TYPE"
>bytea</TT
> values (Pavel Stehule)
       </P
></LI
><LI
><P
>        Fix regular expressions in which a back-reference occurs within
        a larger quantified subexpression (Tom Lane)
       </P
><P
>        For example, <TT
CLASS="LITERAL"
>^(\w+)( \1)+$</TT
>.  Previous releases did not
        check that the back-reference actually matched the first occurrence.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN121435"
>E.32.3.8. <A
HREF="information-schema.html"
>Information Schema</A
></A
></H3
><P
></P
><UL
><LI
><P
>        Add information schema views
        <TT
CLASS="STRUCTNAME"
>role_udt_grants</TT
>, <TT
CLASS="STRUCTNAME"
>udt_privileges</TT
>,
        and <TT
CLASS="STRUCTNAME"
>user_defined_types</TT
> (Peter Eisentraut)
       </P
></LI
><LI
><P
>        Add composite-type attributes to the
        information schema <TT
CLASS="STRUCTNAME"
>element_types</TT
> view
        (Peter Eisentraut)
       </P
></LI
><LI
><P
>        Implement <TT
CLASS="STRUCTFIELD"
>interval_type</TT
> columns in the information
        schema (Peter Eisentraut)
       </P
><P
>        Formerly these columns read as nulls.
       </P
></LI
><LI
><P
>        Implement collation-related columns in the information schema
        <TT
CLASS="STRUCTNAME"
>attributes</TT
>, <TT
CLASS="STRUCTNAME"
>columns</TT
>,
        <TT
CLASS="STRUCTNAME"
>domains</TT
>, and <TT
CLASS="STRUCTNAME"
>element_types</TT
>
        views (Peter Eisentraut)
       </P
></LI
><LI
><P
>        Implement the <TT
CLASS="STRUCTFIELD"
>with_hierarchy</TT
> column in the
        information schema <TT
CLASS="STRUCTNAME"
>table_privileges</TT
> view (Peter
        Eisentraut)
       </P
></LI
><LI
><P
>        Add display of sequence <TT
CLASS="LITERAL"
>USAGE</TT
> privileges to information
        schema (Peter Eisentraut)
       </P
></LI
><LI
><P
>        Make the information schema show default privileges (Peter
        Eisentraut)
       </P
><P
>        Previously, non-empty default permissions were not represented in the
        views.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN121467"
>E.32.3.9. Server-Side Languages</A
></H3
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121469"
>E.32.3.9.1. <A
HREF="plpgsql.html"
>PL/pgSQL</A
> Server-Side Language</A
></H4
><P
></P
><UL
><LI
><P
>        Allow the PL/pgSQL <TT
CLASS="COMMAND"
>OPEN</TT
> cursor command to supply
        parameters by name (Yeb Havinga)
       </P
></LI
><LI
><P
>        Add a <TT
CLASS="COMMAND"
>GET STACKED DIAGNOSTICS</TT
> PL/pgSQL command
        to retrieve exception info (Pavel Stehule)
       </P
></LI
><LI
><P
>        Speed up PL/pgSQL array assignment by caching type information
        (Pavel Stehule)
       </P
></LI
><LI
><P
>        Improve performance and memory consumption for long chains of
        <TT
CLASS="LITERAL"
>ELSIF</TT
> clauses (Tom Lane)
       </P
></LI
><LI
><P
>        Output the function signature, not just the name, in PL/pgSQL
        error messages (Pavel Stehule)
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121486"
>E.32.3.9.2. <A
HREF="plpython.html"
>PL/Python</A
> Server-Side Language</A
></H4
><P
></P
><UL
><LI
><P
>        Add PL/Python <ACRONYM
CLASS="ACRONYM"
>SPI</ACRONYM
> cursor support (Jan
        Urbanski)
       </P
><P
>        This allows PL/Python to read partial result sets.
       </P
></LI
><LI
><P
>        Add result metadata functions to PL/Python (Peter Eisentraut)
       </P
><P
>        Specifically, this adds result object functions
        <TT
CLASS="LITERAL"
>.colnames</TT
>, <TT
CLASS="LITERAL"
>.coltypes</TT
>, and
        <TT
CLASS="LITERAL"
>.coltypmods</TT
>.
       </P
></LI
><LI
><P
>        Remove support for Python 2.2 (Peter Eisentraut)
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121502"
>E.32.3.9.3. <A
HREF="xfunc-sql.html"
>SQL</A
> Server-Side Language</A
></H4
><P
></P
><UL
><LI
><P
>        Allow <ACRONYM
CLASS="ACRONYM"
>SQL</ACRONYM
>-language functions to reference
        parameters by name (Matthew Draper)
       </P
><P
>        To use this, simply name the function arguments and then reference
        the argument names in the <ACRONYM
CLASS="ACRONYM"
>SQL</ACRONYM
> function body.
       </P
></LI
></UL
></DIV
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN121511"
>E.32.3.10. Client Applications</A
></H3
><P
></P
><UL
><LI
><P
>        Add <A
HREF="app-initdb.html"
><SPAN
CLASS="APPLICATION"
>initdb</SPAN
></A
>
        options <TT
CLASS="OPTION"
>--auth-local</TT
> and <TT
CLASS="OPTION"
>--auth-host</TT
>
        (Peter Eisentraut)
       </P
><P
>        This allows separate control of <TT
CLASS="LITERAL"
>local</TT
> and
        <TT
CLASS="LITERAL"
>host</TT
> <TT
CLASS="FILENAME"
>pg_hba.conf</TT
> authentication
        settings.  <TT
CLASS="OPTION"
>--auth</TT
> still controls both.
       </P
></LI
><LI
><P
>        Add <TT
CLASS="OPTION"
>--replication</TT
>/<TT
CLASS="OPTION"
>--no-replication</TT
> flags to
        <A
HREF="app-createuser.html"
><SPAN
CLASS="APPLICATION"
>createuser</SPAN
></A
>
        to control replication permission (Fujii Masao)
       </P
></LI
><LI
><P
>        Add the <TT
CLASS="OPTION"
>--if-exists</TT
> option to <A
HREF="app-dropdb.html"
><SPAN
CLASS="APPLICATION"
>dropdb</SPAN
></A
> and <A
HREF="app-dropuser.html"
><SPAN
CLASS="APPLICATION"
>dropuser</SPAN
></A
> (Josh
        Kupershmidt)
       </P
></LI
><LI
><P
>        Give command-line tools the ability to specify the name of the
        database to connect to, and fall back to <TT
CLASS="LITERAL"
>template1</TT
>
        if a <TT
CLASS="LITERAL"
>postgres</TT
> database connection fails (Robert Haas)
       </P
></LI
></UL
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121542"
>E.32.3.10.1. <A
HREF="app-psql.html"
><SPAN
CLASS="APPLICATION"
>psql</SPAN
></A
></A
></H4
><P
></P
><UL
><LI
><P
>        Add a display mode to auto-expand output based on the
        display width (Peter Eisentraut)
       </P
><P
>        This adds the <TT
CLASS="LITERAL"
>auto</TT
> option to the <TT
CLASS="COMMAND"
>\x</TT
>
        command, which switches to the expanded mode when the normal
        output would be wider than the screen.
       </P
></LI
><LI
><P
>        Allow inclusion of a script file that is named relative to the
        directory of the file from which it was invoked (Gurjeet Singh)
       </P
><P
>        This is done with a new command <TT
CLASS="COMMAND"
>\ir</TT
>.
       </P
></LI
><LI
><P
>        Add support for non-<ACRONYM
CLASS="ACRONYM"
>ASCII</ACRONYM
> characters in
        <SPAN
CLASS="APPLICATION"
>psql</SPAN
> variable names (Tom Lane)
       </P
></LI
><LI
><P
>        Add support for major-version-specific <TT
CLASS="FILENAME"
>.psqlrc</TT
> files
        (Bruce Momjian)
       </P
><P
>        <SPAN
CLASS="APPLICATION"
>psql</SPAN
> already supported minor-version-specific
        <TT
CLASS="FILENAME"
>.psqlrc</TT
> files.
       </P
></LI
><LI
><P
>        Provide environment variable overrides for <SPAN
CLASS="APPLICATION"
>psql</SPAN
>
        history and startup file locations (Andrew Dunstan)
       </P
><P
>        <TT
CLASS="ENVAR"
>PSQL_HISTORY</TT
> and <TT
CLASS="ENVAR"
>PSQLRC</TT
> now
        determine these file names if set.
       </P
></LI
><LI
><P
>        Add a <TT
CLASS="COMMAND"
>\setenv</TT
> command to modify
        the environment variables passed to child processes (Andrew Dunstan)
       </P
></LI
><LI
><P
>        Name <SPAN
CLASS="APPLICATION"
>psql</SPAN
>'s temporary editor files with a
        <TT
CLASS="FILENAME"
>.sql</TT
> extension (Peter Eisentraut)
       </P
><P
>        This allows extension-sensitive editors to select the right mode.
       </P
></LI
><LI
><P
>        Allow <SPAN
CLASS="APPLICATION"
>psql</SPAN
> to use zero-byte field and record
        separators (Peter Eisentraut)
       </P
><P
>        Various shell tools use zero-byte (NUL) separators,
        e.g. <SPAN
CLASS="APPLICATION"
>find</SPAN
>.
       </P
></LI
><LI
><P
>        Make the <TT
CLASS="COMMAND"
>\timing</TT
> option report times for
        failed queries (Magnus Hagander)
       </P
><P
>        Previously times were reported only for successful queries.
       </P
></LI
><LI
><P
>        Unify and tighten <SPAN
CLASS="APPLICATION"
>psql</SPAN
>'s treatment of <TT
CLASS="COMMAND"
>\copy</TT
>
        and SQL <TT
CLASS="COMMAND"
>COPY</TT
> (Noah Misch)
       </P
><P
>        This fix makes failure behavior more predictable and honors
        <TT
CLASS="COMMAND"
>\set ON_ERROR_ROLLBACK</TT
>.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121596"
>E.32.3.10.2. Informational Commands</A
></H4
><P
></P
><UL
><LI
><P
>        Make <TT
CLASS="COMMAND"
>\d</TT
> on a sequence show the
        table/column name owning it (Magnus Hagander)
       </P
></LI
><LI
><P
>        Show statistics target for columns in <TT
CLASS="COMMAND"
>\d+</TT
> (Magnus
        Hagander)
       </P
></LI
><LI
><P
>        Show role password expiration dates in <TT
CLASS="COMMAND"
>\du</TT
>
        (Fabr&iacute;zio de Royes Mello)
       </P
></LI
><LI
><P
>        Display comments for casts, conversions, domains, and languages
        (Josh Kupershmidt)
       </P
><P
>        These are included in the output of <TT
CLASS="COMMAND"
>\dC+</TT
>,
        <TT
CLASS="COMMAND"
>\dc+</TT
>, <TT
CLASS="COMMAND"
>\dD+</TT
>, and <TT
CLASS="COMMAND"
>\dL</TT
> respectively.
       </P
></LI
><LI
><P
>        Display comments for <ACRONYM
CLASS="ACRONYM"
>SQL</ACRONYM
>/<ACRONYM
CLASS="ACRONYM"
>MED</ACRONYM
>
        objects (Josh Kupershmidt)
       </P
><P
>        These are included in the output of <TT
CLASS="COMMAND"
>\des+</TT
>,
        <TT
CLASS="COMMAND"
>\det+</TT
>, and <TT
CLASS="COMMAND"
>\dew+</TT
> for foreign servers, foreign
        tables, and foreign data wrappers respectively.
       </P
></LI
><LI
><P
>        Change <TT
CLASS="COMMAND"
>\dd</TT
> to display comments only for object types
        without their own backslash command (Josh Kupershmidt)
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121626"
>E.32.3.10.3. Tab Completion</A
></H4
><P
></P
><UL
><LI
><P
>        In <SPAN
CLASS="APPLICATION"
>psql</SPAN
> tab completion, complete <ACRONYM
CLASS="ACRONYM"
>SQL</ACRONYM
>
        keywords in either upper or lower case according to the new <A
HREF="app-psql.html#APP-PSQL-VARIABLES"
><TT
CLASS="LITERAL"
>COMP_KEYWORD_CASE</TT
></A
>
        setting (Peter Eisentraut)
       </P
></LI
><LI
><P
>        Add tab completion support for
        <TT
CLASS="COMMAND"
>EXECUTE</TT
> (Andreas Karlsson)
       </P
></LI
><LI
><P
>        Allow tab completion of role references in
        <TT
CLASS="COMMAND"
>GRANT</TT
>/<TT
CLASS="COMMAND"
>REVOKE</TT
> (Peter
        Eisentraut)
       </P
></LI
><LI
><P
>        Allow tab completion of file names to supply quotes, when necessary
        (Noah Misch)
       </P
></LI
><LI
><P
>        Change tab completion support for
        <TT
CLASS="COMMAND"
>TABLE</TT
> to also include views (Magnus Hagander)
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121647"
>E.32.3.10.4. <A
HREF="app-pgdump.html"
><SPAN
CLASS="APPLICATION"
>pg_dump</SPAN
></A
></A
></H4
><P
></P
><UL
><LI
><P
>        Add an <TT
CLASS="OPTION"
>--exclude-table-data</TT
> option to
        <SPAN
CLASS="APPLICATION"
>pg_dump</SPAN
> (Andrew Dunstan)
       </P
><P
>        This allows dumping of a table's definition but not its data,
        on a per-table basis.
       </P
></LI
><LI
><P
>        Add a <TT
CLASS="OPTION"
>--section</TT
> option to <SPAN
CLASS="APPLICATION"
>pg_dump</SPAN
>
        and <SPAN
CLASS="APPLICATION"
>pg_restore</SPAN
> (Andrew Dunstan)
       </P
><P
>        Valid values are <TT
CLASS="LITERAL"
>pre-data</TT
>, <TT
CLASS="LITERAL"
>data</TT
>,
        and <TT
CLASS="LITERAL"
>post-data</TT
>. The option can be
        given more than once to select two or more sections.
       </P
></LI
><LI
><P
>        Make <A
HREF="app-pg-dumpall.html"
><SPAN
CLASS="APPLICATION"
>pg_dumpall</SPAN
></A
> dump all
        roles first, then all configuration settings on roles (Phil Sorber)
       </P
><P
>        This allows a role's configuration settings to mention other
        roles without generating an error.
       </P
></LI
><LI
><P
>        Allow <SPAN
CLASS="APPLICATION"
>pg_dumpall</SPAN
> to avoid errors if the
        <TT
CLASS="LITERAL"
>postgres</TT
> database is missing in the new cluster
        (Robert Haas)
       </P
></LI
><LI
><P
>        Dump foreign server user mappings in user name order (Peter
        Eisentraut)
       </P
><P
>        This helps produce deterministic dump files.
       </P
></LI
><LI
><P
>        Dump operators in a predictable order (Peter Eisentraut)
       </P
></LI
><LI
><P
>        Tighten rules for when extension configuration tables are dumped
        by <SPAN
CLASS="APPLICATION"
>pg_dump</SPAN
> (Tom Lane)
       </P
></LI
><LI
><P
>        Make <SPAN
CLASS="APPLICATION"
>pg_dump</SPAN
> emit more useful dependency
        information (Tom Lane)
       </P
><P
>        The dependency links included in archive-format dumps were formerly
        of very limited use, because they frequently referenced objects that
        appeared nowhere in the dump.  Now they represent actual dependencies
        (possibly indirect) among the dumped objects.
       </P
></LI
><LI
><P
>        Improve <SPAN
CLASS="APPLICATION"
>pg_dump</SPAN
>'s performance when dumping many
        database objects (Tom Lane)
       </P
></LI
></UL
></DIV
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN121690"
>E.32.3.11. <A
HREF="libpq.html"
><SPAN
CLASS="APPLICATION"
>libpq</SPAN
></A
></A
></H3
><P
></P
><UL
><LI
><P
>        Allow <SPAN
CLASS="APPLICATION"
>libpq</SPAN
> connection strings to have the format of a
        <A
HREF="libpq-connect.html#LIBPQ-CONNSTRING"
><ACRONYM
CLASS="ACRONYM"
>URI</ACRONYM
></A
>
        (Alexander Shulgin)
       </P
><P
>        The syntax begins with <TT
CLASS="LITERAL"
>postgres://</TT
>.  This can allow
        applications to avoid implementing their own parser for URIs
        representing database connections.
       </P
></LI
><LI
><P
>        Add a <A
HREF="libpq-connect.html#LIBPQ-CONNECT-SSLCOMPRESSION"
>connection
        option</A
> to disable <ACRONYM
CLASS="ACRONYM"
>SSL</ACRONYM
> compression
        (Laurenz Albe)
       </P
><P
>        This can be used to remove the overhead of <ACRONYM
CLASS="ACRONYM"
>SSL</ACRONYM
>
        compression on fast networks.
       </P
></LI
><LI
><P
>        Add a <A
HREF="libpq-single-row-mode.html"
>single-row processing
        mode</A
> for better handling of large result sets
        (Kyotaro Horiguchi, Marko Kreen)
       </P
><P
>        Previously, <SPAN
CLASS="APPLICATION"
>libpq</SPAN
> always collected the entire query
        result in memory before passing it back to the application.
       </P
></LI
><LI
><P
>        Add <TT
CLASS="LITERAL"
>const</TT
> qualifiers to the declarations of the functions
        <CODE
CLASS="FUNCTION"
>PQconnectdbParams</CODE
>, <CODE
CLASS="FUNCTION"
>PQconnectStartParams</CODE
>,
        and <CODE
CLASS="FUNCTION"
>PQpingParams</CODE
> (Lionel Elie Mamane)
       </P
></LI
><LI
><P
>        Allow the <TT
CLASS="FILENAME"
>.pgpass</TT
> file to include escaped characters
        in the password field (Robert Haas)
       </P
></LI
><LI
><P
>        Make library functions use <CODE
CLASS="FUNCTION"
>abort()</CODE
> instead of
        <CODE
CLASS="FUNCTION"
>exit()</CODE
> when it is necessary to terminate the process
        (Peter Eisentraut)
       </P
><P
>        This choice does not interfere with the normal exit codes used by the
        program, and generates a signal that can be caught by the caller.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN121727"
>E.32.3.12. Source Code</A
></H3
><P
></P
><UL
><LI
><P
>        Remove dead ports (Peter Eisentraut)
       </P
><P
>        The following platforms are no longer supported: dgux,
        nextstep, sunos4, svr4, ultrix4, univel, bsdi.
       </P
></LI
><LI
><P
>        Add support for building with <A
HREF="install-windows.html"
>MS
        Visual Studio 2010</A
> (Brar Piening)
       </P
></LI
><LI
><P
>        Enable compiling with the MinGW-w64 32-bit compiler (Lars Kanis)
       </P
></LI
><LI
><P
>        Install <TT
CLASS="FILENAME"
>plpgsql.h</TT
> into <TT
CLASS="FILENAME"
>include/server</TT
> during installation
        (Heikki Linnakangas)
       </P
></LI
><LI
><P
>        Improve the latch facility to include detection of postmaster death
        (Peter Geoghegan, Heikki Linnakangas, Tom Lane)
       </P
><P
>        This eliminates one of the main reasons that background processes
        formerly had to wake up to poll for events.
       </P
></LI
><LI
><P
>        Use C flexible array members, where supported (Peter Eisentraut)
       </P
></LI
><LI
><P
>        Improve the concurrent transaction regression tests
        (<SPAN
CLASS="APPLICATION"
>isolationtester</SPAN
>) (Noah Misch)
       </P
></LI
><LI
><P
>        Modify <SPAN
CLASS="APPLICATION"
>thread_test</SPAN
> to create its test files in
        the current directory, rather than <TT
CLASS="FILENAME"
>/tmp</TT
> (Bruce Momjian)
       </P
></LI
><LI
><P
>        Improve flex and bison warning and error reporting (Tom Lane)
       </P
></LI
><LI
><P
>        Add memory barrier support (Robert Haas)
       </P
><P
>        This is currently unused.
       </P
></LI
><LI
><P
>        Modify pgindent to use a typedef file (Bruce Momjian)
       </P
></LI
><LI
><P
>        Add a hook for processing messages due to be sent to the server
        log (Martin Pihlak)
       </P
></LI
><LI
><P
>        Add object access hooks for <TT
CLASS="COMMAND"
>DROP</TT
> commands
        (KaiGai Kohei)
       </P
></LI
><LI
><P
>        Centralize <TT
CLASS="COMMAND"
>DROP</TT
> handling for some object types
        (KaiGai Kohei)
       </P
></LI
><LI
><P
>        Add a <SPAN
CLASS="APPLICATION"
>pg_upgrade</SPAN
> test suite (Peter Eisentraut)
       </P
></LI
><LI
><P
>        Sync regular expression code with <ACRONYM
CLASS="ACRONYM"
>TCL</ACRONYM
> 8.5.11
        and improve internal processing (Tom Lane)
       </P
></LI
><LI
><P
>        Move <ACRONYM
CLASS="ACRONYM"
>CRC</ACRONYM
> tables to libpgport, and provide them
        in a separate include file (Daniel Farina)
       </P
></LI
><LI
><P
>        Add options to <SPAN
CLASS="APPLICATION"
>git_changelog</SPAN
> for use in major
        release note creation (Bruce Momjian)
       </P
></LI
><LI
><P
>        Support Linux's <TT
CLASS="FILENAME"
>/proc/self/oom_score_adj</TT
> API (Tom Lane)
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN121784"
>E.32.3.13. Additional Modules</A
></H3
><P
></P
><UL
><LI
><P
>        Improve efficiency of <A
HREF="dblink.html"
>dblink</A
> by using
        libpq's new single-row processing mode (Kyotaro Horiguchi, Marko
        Kreen)
       </P
><P
>        This improvement does not apply to
        <CODE
CLASS="FUNCTION"
>dblink_send_query()</CODE
>/<CODE
CLASS="FUNCTION"
>dblink_get_result()</CODE
>.
       </P
></LI
><LI
><P
>        Support <TT
CLASS="LITERAL"
>force_not_null</TT
> option in <A
HREF="file-fdw.html"
>file_fdw</A
> (Shigeru Hanada)
       </P
></LI
><LI
><P
>        Implement dry-run mode for <A
HREF="pgarchivecleanup.html"
><SPAN
CLASS="APPLICATION"
>pg_archivecleanup</SPAN
></A
>
        (Gabriele Bartolini)
       </P
><P
>        This only outputs the names of files to be deleted.
       </P
></LI
><LI
><P
>        Add new <A
HREF="pgbench.html"
>pgbench</A
> switches
        <TT
CLASS="OPTION"
>--unlogged-tables</TT
>, <TT
CLASS="OPTION"
>--tablespace</TT
>, and
        <TT
CLASS="OPTION"
>--index-tablespace</TT
> (Robert Haas)
       </P
></LI
><LI
><P
>        Change <A
HREF="pgtestfsync.html"
><SPAN
CLASS="APPLICATION"
>pg_test_fsync</SPAN
></A
> to test
        for a fixed amount of time, rather than a fixed number of cycles
        (Bruce Momjian)
       </P
><P
>        The <TT
CLASS="OPTION"
>-o</TT
>/cycles option was removed, and
        <TT
CLASS="OPTION"
>-s</TT
>/seconds added.
       </P
></LI
><LI
><P
>        Add a <A
HREF="pgtesttiming.html"
><SPAN
CLASS="APPLICATION"
>pg_test_timing</SPAN
></A
>
        utility to measure clock monotonicity and timing overhead (Ants
        Aasma, Greg Smith)
       </P
></LI
><LI
><P
>        Add a <A
HREF="tcn.html"
>tcn</A
> (triggered change notification)
        module to generate <TT
CLASS="COMMAND"
>NOTIFY</TT
> events on table changes
        (Kevin Grittner)
       </P
></LI
></UL
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121823"
>E.32.3.13.1. <A
HREF="pgupgrade.html"
><SPAN
CLASS="APPLICATION"
>pg_upgrade</SPAN
></A
></A
></H4
><P
></P
><UL
><LI
><P
>        Adjust <SPAN
CLASS="APPLICATION"
>pg_upgrade</SPAN
> environment variables (Bruce
        Momjian)
       </P
><P
>        Rename data, bin, and port environment
        variables to begin with <TT
CLASS="LITERAL"
>PG</TT
>, and support
        <TT
CLASS="ENVAR"
>PGPORTOLD</TT
>/<TT
CLASS="ENVAR"
>PGPORTNEW</TT
>, to replace
        <TT
CLASS="ENVAR"
>PGPORT</TT
>.
       </P
></LI
><LI
><P
>        Overhaul <SPAN
CLASS="APPLICATION"
>pg_upgrade</SPAN
> logging and failure reporting
        (Bruce Momjian)
       </P
><P
>        Create four append-only log files, and delete them on success.
        Add <TT
CLASS="OPTION"
>-r</TT
>/<TT
CLASS="OPTION"
>--retain</TT
> option to unconditionally
        retain these files.  Also remove <SPAN
CLASS="APPLICATION"
>pg_upgrade</SPAN
> options
        <TT
CLASS="OPTION"
>-g</TT
>/<TT
CLASS="OPTION"
>-G</TT
>/<TT
CLASS="OPTION"
>-l</TT
> options as unnecessary,
        and tighten log file permissions.
       </P
></LI
><LI
><P
>        Make <SPAN
CLASS="APPLICATION"
>pg_upgrade</SPAN
> create a script to incrementally
        generate more accurate optimizer statistics (Bruce Momjian)
       </P
><P
>        This reduces the time needed to generate minimal cluster statistics
        after an upgrade.
       </P
></LI
><LI
><P
>        Allow <SPAN
CLASS="APPLICATION"
>pg_upgrade</SPAN
> to upgrade an old cluster that
        does not have a <TT
CLASS="LITERAL"
>postgres</TT
> database (Bruce Momjian)
       </P
></LI
><LI
><P
>        Allow <SPAN
CLASS="APPLICATION"
>pg_upgrade</SPAN
> to handle cases where some
        old or new databases are missing, as long as they are empty
        (Bruce Momjian)
       </P
></LI
><LI
><P
>        Allow <SPAN
CLASS="APPLICATION"
>pg_upgrade</SPAN
> to handle configuration-only
        directory installations (Bruce Momjian)
       </P
></LI
><LI
><P
>        In <SPAN
CLASS="APPLICATION"
>pg_upgrade</SPAN
>, add <TT
CLASS="OPTION"
>-o</TT
>/<TT
CLASS="OPTION"
>-O</TT
>
        options to pass parameters to the servers (Bruce Momjian)
       </P
><P
>        This is useful for configuration-only directory installs.
       </P
></LI
><LI
><P
>        Change <SPAN
CLASS="APPLICATION"
>pg_upgrade</SPAN
> to use port 50432 by default
        (Bruce Momjian)
       </P
><P
>        This helps avoid unintended client connections during the upgrade.
       </P
></LI
><LI
><P
>        Reduce cluster locking in <SPAN
CLASS="APPLICATION"
>pg_upgrade</SPAN
> (Bruce
        Momjian)
       </P
><P
>        Specifically, only lock the old cluster if link mode is used,
        and do it right after the schema is restored.
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121874"
>E.32.3.13.2. <A
HREF="pgstatstatements.html"
><SPAN
CLASS="APPLICATION"
>pg_stat_statements</SPAN
></A
></A
></H4
><P
></P
><UL
><LI
><P
>        Allow <SPAN
CLASS="APPLICATION"
>pg_stat_statements</SPAN
> to aggregate similar
        queries via SQL text normalization (Peter Geoghegan, Tom Lane)
       </P
><P
>        Users with applications that use non-parameterized SQL will now
        be able to monitor query performance without detailed log analysis.
       </P
></LI
><LI
><P
>        Add dirtied and written block counts and read/write times to
        <SPAN
CLASS="APPLICATION"
>pg_stat_statements</SPAN
> (Robert Haas, Ants Aasma)
       </P
></LI
><LI
><P
>        Prevent <SPAN
CLASS="APPLICATION"
>pg_stat_statements</SPAN
> from double-counting
        <TT
CLASS="COMMAND"
>PREPARE</TT
> and <TT
CLASS="COMMAND"
>EXECUTE</TT
> commands
        (Tom Lane)
       </P
></LI
></UL
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="AEN121891"
>E.32.3.13.3. <A
HREF="sepgsql.html"
>sepgsql</A
></A
></H4
><P
></P
><UL
><LI
><P
>        Support <TT
CLASS="LITERAL"
>SECURITY LABEL</TT
> on global objects (KaiGai
        Kohei, Robert Haas)
       </P
><P
>        Specifically, add security labels to databases,
        tablespaces, and roles.
       </P
></LI
><LI
><P
>        Allow sepgsql to honor database labels (KaiGai Kohei)
       </P
></LI
><LI
><P
>        Perform sepgsql permission checks during the creation of various
        objects (KaiGai Kohei)
       </P
></LI
><LI
><P
>        Add <CODE
CLASS="FUNCTION"
>sepgsql_setcon()</CODE
> and related functions to control
        the sepgsql security domain (KaiGai Kohei)
       </P
></LI
><LI
><P
>        Add a user space access cache to sepgsql to improve performance
        (KaiGai Kohei)
       </P
></LI
></UL
></DIV
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN121908"
>E.32.3.14. Documentation</A
></H3
><P
></P
><UL
><LI
><P
>        Add a rule to optionally build HTML documentation using the
        stylesheet from the website (Magnus Hagander)
       </P
><P
>        Use <TT
CLASS="COMMAND"
>gmake STYLE=website draft</TT
>.
       </P
></LI
><LI
><P
>        Improve <TT
CLASS="COMMAND"
>EXPLAIN</TT
> documentation (Tom Lane)
       </P
></LI
><LI
><P
>        Document that user/database names are preserved with double-quoting
        by command-line tools like <SPAN
CLASS="APPLICATION"
>vacuumdb</SPAN
>  (Bruce
        Momjian)
       </P
></LI
><LI
><P
>        Document the actual string returned by the client for MD5
        authentication (Cyan Ogilvie)
       </P
></LI
><LI
><P
>        Deprecate use of <TT
CLASS="LITERAL"
>GLOBAL</TT
> and <TT
CLASS="LITERAL"
>LOCAL</TT
> in
        <TT
CLASS="COMMAND"
>CREATE TEMP TABLE</TT
> (Noah Misch)
       </P
><P
>        <SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> has long treated these keyword as no-ops,
        and continues to do so; but in future they might mean what the SQL
        standard says they mean, so applications should avoid using them.
       </P
></LI
></UL
></DIV
></DIV
></DIV
><DIV
CLASS="NAVFOOTER"
><HR
ALIGN="LEFT"
WIDTH="100%"><TABLE
SUMMARY="Footer navigation table"
WIDTH="100%"
BORDER="0"
CELLPADDING="0"
CELLSPACING="0"
><TR
><TD
WIDTH="33%"
ALIGN="left"
VALIGN="top"
><A
HREF="release-9-2-1.html"
ACCESSKEY="P"
>Prev</A
></TD
><TD
WIDTH="34%"
ALIGN="center"
VALIGN="top"
><A
HREF="index.html"
ACCESSKEY="H"
>Home</A
></TD
><TD
WIDTH="33%"
ALIGN="right"
VALIGN="top"
><A
HREF="release-9-1-22.html"
ACCESSKEY="N"
>Next</A
></TD
></TR
><TR
><TD
WIDTH="33%"
ALIGN="left"
VALIGN="top"
>Release 9.2.1</TD
><TD
WIDTH="34%"
ALIGN="center"
VALIGN="top"
><A
HREF="release.html"
ACCESSKEY="U"
>Up</A
></TD
><TD
WIDTH="33%"
ALIGN="right"
VALIGN="top"
>Release 9.1.22</TD
></TR
></TABLE
></DIV
></BODY
></HTML
>