Linux SRV-16 6.18.44-paas #1 SMP PREEMPT_DYNAMIC Thu Aug 13 10:08:12 UTC 2026 x86_64
/proc/thread-self/root/opt/postgresql-11/doc/html/
//proc/thread-self/root/opt/postgresql-11/doc/html/release-8-3.html
 <?xml version="1.0" encoding="UTF-8" standalone="no"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>E.205. Release 8.3</title><link rel="stylesheet" type="text/css" href="stylesheet.css" /><link rev="made" href="pgsql-docs@postgresql.org" /><meta name="generator" content="DocBook XSL Stylesheets V1.79.1" /><link rel="prev" href="release-8-3-1.html" title="E.204. Release 8.3.1" /><link rel="next" href="release-8-2-23.html" title="E.206. Release 8.2.23" /></head><body><div xmlns="http://www.w3.org/TR/xhtml1/transitional" class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="5" align="center">E.205. Release 8.3</th></tr><tr><td width="10%" align="left"><a accesskey="p" href="release-8-3-1.html" title="E.204. Release 8.3.1">Prev</a> </td><td width="10%" align="left"><a accesskey="u" href="release.html" title="Appendix E. Release Notes">Up</a></td><th width="60%" align="center">Appendix E. Release Notes</th><td width="10%" align="right"><a accesskey="h" href="index.html" title="PostgreSQL 11.1 Documentation">Home</a></td><td width="10%" align="right"> <a accesskey="n" href="release-8-2-23.html" title="E.206. Release 8.2.23">Next</a></td></tr></table><hr></hr></div><div class="sect1" id="RELEASE-8-3"><div class="titlepage"><div><div><h2 class="title" style="clear: both">E.205. Release 8.3</h2></div></div></div><div class="toc"><dl class="toc"><dt><span class="sect2"><a href="release-8-3.html#id-1.11.6.209.3">E.205.1. Overview</a></span></dt><dt><span class="sect2"><a href="release-8-3.html#id-1.11.6.209.4">E.205.2. Migration to Version 8.3</a></span></dt><dt><span class="sect2"><a href="release-8-3.html#id-1.11.6.209.5">E.205.3. Changes</a></span></dt></dl></div><p><strong>Release date: </strong>2008-02-04</p><div class="sect2" id="id-1.11.6.209.3"><div class="titlepage"><div><div><h3 class="title">E.205.1. Overview</h3></div></div></div><p>
    With significant new functionality and performance enhancements,
    this release represents a major leap forward for
    <span class="productname">PostgreSQL</span>. This was made possible by a growing
    community that has dramatically accelerated the pace of
    development. This release adds the following major features:
   </p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
      Full text search is integrated into the core database system
     </p></li><li class="listitem"><p>
      Support for the SQL/XML standard, including new operators and an
      <code class="type">XML</code> data type
     </p></li><li class="listitem"><p>
      Enumerated data types (<code class="type">ENUM</code>)
     </p></li><li class="listitem"><p>
      Arrays of composite types
     </p></li><li class="listitem"><p>
      Universally Unique Identifier (<code class="type">UUID</code>) data type
     </p></li><li class="listitem"><p>
      Add control over whether <code class="literal">NULL</code>s sort first or last
     </p></li><li class="listitem"><p>
      Updatable cursors
     </p></li><li class="listitem"><p>
      Server configuration parameters can now be set on a per-function
      basis
     </p></li><li class="listitem"><p>
      User-defined types can now have type modifiers
     </p></li><li class="listitem"><p>
      Automatically re-plan cached queries when table
      definitions change or statistics are updated
     </p></li><li class="listitem"><p>
      Numerous improvements in logging and statistics collection
     </p></li><li class="listitem"><p>
      Support Security Service Provider Interface (<acronym class="acronym">SSPI</acronym>) for
      authentication on Windows
     </p></li><li class="listitem"><p>
      Support multiple concurrent autovacuum processes, and other
      autovacuum improvements
     </p></li><li class="listitem"><p>
      Allow the whole <span class="productname">PostgreSQL</span> distribution to be compiled
      with <span class="productname">Microsoft Visual C++</span>
     </p></li></ul></div><p>
    Major performance improvements are listed below.  Most of
    these enhancements are automatic and do not require user changes or
    tuning:
   </p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
      Asynchronous commit delays writes to WAL during transaction commit
     </p></li><li class="listitem"><p>
      Checkpoint writes can be spread over a longer time period to smooth
      the I/O spike during each checkpoint
     </p></li><li class="listitem"><p>
      Heap-Only Tuples (<acronym class="acronym">HOT</acronym>) accelerate space reuse for
      most <code class="command">UPDATE</code>s and <code class="command">DELETE</code>s
     </p></li><li class="listitem"><p>
      Just-in-time background writer strategy improves disk write
      efficiency
     </p></li><li class="listitem"><p>
      Using non-persistent transaction IDs for read-only transactions
      reduces overhead and <code class="command">VACUUM</code> requirements
     </p></li><li class="listitem"><p>
      Per-field and per-row storage overhead has been reduced
     </p></li><li class="listitem"><p>
      Large sequential scans no longer force out frequently used
      cached pages
     </p></li><li class="listitem"><p>
      Concurrent large sequential scans can now share disk reads
     </p></li><li class="listitem"><p>
      <code class="literal">ORDER BY ... LIMIT</code> can be done without sorting
     </p></li></ul></div><p>
    The above items are explained in more detail in the sections below.
   </p></div><div class="sect2" id="id-1.11.6.209.4"><div class="titlepage"><div><div><h3 class="title">E.205.2. Migration to Version 8.3</h3></div></div></div><p>
    A dump/restore using <span class="application">pg_dump</span> is
    required for those wishing to migrate data from any previous
    release.
   </p><p>
    Observe the following incompatibilities:
   </p><div class="sect3" id="id-1.11.6.209.4.4"><div class="titlepage"><div><div><h4 class="title">E.205.2.1. General</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Non-character data types are no longer automatically cast to
       <code class="type">TEXT</code> (Peter, Tom)
      </p><p>
       Previously, if a non-character value was supplied to an operator or
       function that requires <code class="type">text</code> input, it was automatically
       cast to <code class="type">text</code>, for most (though not all) built-in data types.
       This no longer happens: an explicit cast to <code class="type">text</code> is now
       required for all non-character-string types.  For example, these
       expressions formerly worked:

</p><pre class="programlisting">
substr(current_date, 1, 4)
23 LIKE '2%'
</pre><p>

       but will now draw <span class="quote">“<span class="quote">function does not exist</span>”</span> and <span class="quote">“<span class="quote">operator
       does not exist</span>”</span> errors respectively.  Use an explicit cast instead:

</p><pre class="programlisting">
substr(current_date::text, 1, 4)
23::text LIKE '2%'
</pre><p>

       (Of course, you can use the more verbose <code class="literal">CAST()</code> syntax too.)
       The reason for the change is that these automatic casts too often caused
       surprising behavior.  An example is that in previous releases, this
       expression was accepted but did not do what was expected:

</p><pre class="programlisting">
current_date &lt; 2017-11-17
</pre><p>

       This is actually comparing a date to an integer, which should be
       (and now is) rejected — but in the presence of automatic
       casts both sides were cast to <code class="type">text</code> and a textual comparison
       was done, because the <code class="literal">text &lt; text</code> operator was able
       to match the expression when no other <code class="literal">&lt;</code> operator could.
      </p><p>
       Types <code class="type">char(<em class="replaceable"><code>n</code></em>)</code> and
       <code class="type">varchar(<em class="replaceable"><code>n</code></em>)</code> still cast to <code class="type">text</code>
       automatically.  Also, automatic casting to <code class="type">text</code> still works for
       inputs to the concatenation (<code class="literal">||</code>) operator, so long as least
       one input is a character-string type.
      </p></li><li class="listitem"><p>
        Full text search features from <code class="filename">contrib/tsearch2</code> have
        been moved into the core server, with some minor syntax changes
      </p><p>
       <code class="filename">contrib/tsearch2</code> now contains a compatibility
       interface.
      </p></li><li class="listitem"><p>
       <code class="literal">ARRAY(SELECT ...)</code>, where the <code class="command">SELECT</code>
       returns no rows, now returns an empty array, rather than NULL
       (Tom)
      </p></li><li class="listitem"><p>
       The array type name for a base data type is no longer always the base
       type's name with an underscore prefix
      </p><p>
       The old naming convention is still honored when possible, but
       application code should no longer depend on it. Instead
       use the new <code class="literal">pg_type.typarray</code> column to
       identify the array data type associated with a given type.
      </p></li><li class="listitem"><p>
       <code class="literal">ORDER BY ... USING</code> <em class="replaceable"><code>operator</code></em> must now
       use a less-than or greater-than <em class="replaceable"><code>operator</code></em> that is
       defined in a btree operator class
      </p><p>
       This restriction was added to prevent inconsistent results.
      </p></li><li class="listitem"><p>
       <code class="command">SET LOCAL</code> changes now persist until
       the end of the outermost transaction, unless rolled back (Tom)
      </p><p>
       Previously <code class="command">SET LOCAL</code>'s effects were lost
       after subtransaction commit (<code class="command">RELEASE SAVEPOINT</code>
       or exit from a PL/pgSQL exception block).
      </p></li><li class="listitem"><p>
       Commands rejected in transaction blocks are now also rejected in
       multiple-statement query strings (Tom)
      </p><p>
       For example, <code class="literal">"BEGIN; DROP DATABASE; COMMIT"</code> will now be
       rejected even if submitted as a single query message.
      </p></li><li class="listitem"><p>
       <code class="command">ROLLBACK</code> outside a transaction block now
       issues <code class="literal">NOTICE</code> instead of <code class="literal">WARNING</code> (Bruce)
      </p></li><li class="listitem"><p>
       Prevent <code class="command">NOTIFY</code>/<code class="command">LISTEN</code>/<code class="command">UNLISTEN</code>
       from accepting schema-qualified names (Bruce)
      </p><p>
       Formerly, these commands accepted <code class="literal">schema.relation</code> but
       ignored the schema part, which was confusing.
      </p></li><li class="listitem"><p>
       <code class="command">ALTER SEQUENCE</code> no longer affects the sequence's
       <code class="function">currval()</code> state (Tom)
      </p></li><li class="listitem"><p>
       Foreign keys now must match indexable conditions for
       cross-data-type references (Tom)
      </p><p>
       This improves semantic consistency and helps avoid
       performance problems.
      </p></li><li class="listitem"><p>
       Restrict object size functions to users who have reasonable
       permissions to view such information (Tom)
      </p><p>
       For example, <code class="function">pg_database_size()</code> now requires
       <code class="literal">CONNECT</code> permission, which is granted to everyone by
       default. <code class="function">pg_tablespace_size()</code> requires
       <code class="literal">CREATE</code> permission in the tablespace, or is allowed if
       the tablespace is the default tablespace for the database.
      </p></li><li class="listitem"><p>
       Remove the undocumented <code class="literal">!!=</code> (not in) operator (Tom)
      </p><p>
       <code class="literal">NOT IN (SELECT ...)</code> is the proper way to
       perform this operation.
      </p></li><li class="listitem"><p>
       Internal hashing functions are now more uniformly-distributed (Tom)
      </p><p>
       If application code was computing and storing hash values using
       internal <span class="productname">PostgreSQL</span> hashing functions, the hash
       values must be regenerated.
      </p></li><li class="listitem"><p>
       C-code conventions for handling variable-length data values
       have changed (Greg Stark, Tom)
      </p><p>
       The new <code class="function">SET_VARSIZE()</code> macro <span class="emphasis"><em>must</em></span> be used
       to set the length of generated <code class="type">varlena</code> values. Also, it
       might be necessary to expand (<span class="quote">“<span class="quote">de-TOAST</span>”</span>) input values
       in more cases.
      </p></li><li class="listitem"><p>
       Continuous archiving no longer reports each successful archive
       operation to the server logs unless <code class="literal">DEBUG</code> level is used
       (Simon)
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.4.5"><div class="titlepage"><div><div><h4 class="title">E.205.2.2. Configuration Parameters</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Numerous changes in administrative server parameters
      </p><p>
       <code class="varname">bgwriter_lru_percent</code>,
       <code class="varname">bgwriter_all_percent</code>,
       <code class="varname">bgwriter_all_maxpages</code>,
       <code class="varname">stats_start_collector</code>, and
       <code class="varname">stats_reset_on_server_start</code> are removed.
       <code class="varname">redirect_stderr</code> is renamed to
       <code class="varname">logging_collector</code>.
       <code class="varname">stats_command_string</code> is renamed to
       <code class="varname">track_activities</code>.
       <code class="varname">stats_block_level</code> and <code class="varname">stats_row_level</code>
       are merged into <code class="varname">track_counts</code>.
       A new boolean configuration parameter, <code class="varname">archive_mode</code>,
       controls archiving. Autovacuum's default settings have changed.
      </p></li><li class="listitem"><p>
       Remove <code class="varname">stats_start_collector</code> parameter (Tom)
      </p><p>
       We now always start the collector process, unless <acronym class="acronym">UDP</acronym>
       socket creation fails.
      </p></li><li class="listitem"><p>
       Remove <code class="varname">stats_reset_on_server_start</code> parameter (Tom)
      </p><p>
       This was removed because <code class="function">pg_stat_reset()</code>
       can be used for this purpose.
      </p></li><li class="listitem"><p>
       Commenting out a parameter in <code class="filename">postgresql.conf</code> now
       causes it to revert to its default value (Joachim Wieland)
      </p><p>
       Previously, commenting out an entry left the parameter's value unchanged
       until the next server restart.
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.4.6"><div class="titlepage"><div><div><h4 class="title">E.205.2.3. Character Encodings</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Add more checks for invalidly-encoded data (Andrew)
      </p><p>
       This change plugs some holes that existed in literal backslash
       escape string processing and <code class="command">COPY</code> escape
       processing. Now the de-escaped string is rechecked to see if the
       result created an invalid multi-byte character.
      </p></li><li class="listitem"><p>
       Disallow database encodings that are inconsistent with the server's
       locale setting (Tom)
      </p><p>
       On most platforms, <code class="literal">C</code> locale is the only locale that
       will work with any database encoding.  Other locale settings imply
       a specific encoding and will misbehave if the database encoding
       is something different.  (Typical symptoms include bogus textual
       sort order and wrong results from <code class="function">upper()</code> or
       <code class="function">lower()</code>.)  The server now rejects attempts to create
       databases that have an incompatible encoding.
      </p></li><li class="listitem"><p>
       Ensure that <code class="function">chr()</code> cannot create
       invalidly-encoded values (Andrew)
      </p><p>
       In UTF8-encoded databases the argument of <code class="function">chr()</code> is
       now treated as a Unicode code point. In other multi-byte encodings
       <code class="function">chr()</code>'s argument must designate a 7-bit ASCII
       character.  Zero is no longer accepted.
       <code class="function">ascii()</code> has been adjusted to match.
      </p></li><li class="listitem"><p>
       Adjust <code class="function">convert()</code> behavior to ensure encoding
       validity (Andrew)
      </p><p>
       The two argument form of <code class="function">convert()</code> has been
       removed. The three argument form now takes a <code class="type">bytea</code>
       first argument and returns a <code class="type">bytea</code>. To cover the
       loss of functionality, three new functions have been added:
      </p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: circle; "><li class="listitem"><p>
         <code class="function">convert_from(bytea, name)</code> returns
         <code class="type">text</code> — converts the first argument from the named
         encoding to the database encoding
        </p></li><li class="listitem"><p>
         <code class="function">convert_to(text, name)</code> returns
         <code class="type">bytea</code> — converts the first argument from the
         database encoding to the named encoding
        </p></li><li class="listitem"><p>
         <code class="function">length(bytea, name)</code> returns
         <code class="type">integer</code> — gives the length of the first
         argument in characters in the named encoding
        </p></li></ul></div></li><li class="listitem"><p>
       Remove <code class="literal">convert(argument USING conversion_name)</code>
       (Andrew)
      </p><p>
       Its behavior did not match the SQL standard.
      </p></li><li class="listitem"><p>
       Make JOHAB encoding client-only (Tatsuo)
      </p><p>
       JOHAB is not safe as a server-side encoding.
      </p></li></ul></div></div></div><div class="sect2" id="id-1.11.6.209.5"><div class="titlepage"><div><div><h3 class="title">E.205.3. Changes</h3></div></div></div><p>
    Below you will find a detailed account of the
    changes between <span class="productname">PostgreSQL</span> 8.3 and
    the previous major release.
   </p><div class="sect3" id="id-1.11.6.209.5.3"><div class="titlepage"><div><div><h4 class="title">E.205.3.1. Performance</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Asynchronous commit delays writes to WAL during transaction commit
       (Simon)
      </p><p>
       This feature dramatically increases performance for short data-modifying
       transactions.  The disadvantage is that because disk writes are delayed,
       if the database or operating system crashes before data is written to
       the disk, committed data will be lost.  This feature is useful for
       applications that can accept some data loss.  Unlike turning off
       <code class="varname">fsync</code>, using asynchronous commit does not put
       database consistency at risk; the worst case is that after a crash the
       last few reportedly-committed transactions might not be committed after
       all.
       This feature is enabled by turning off <code class="varname">synchronous_commit</code>
       (which can be done per-session or per-transaction, if some transactions
       are critical and others are not).
       <code class="varname">wal_writer_delay</code> can be adjusted to control the maximum
       delay before transactions actually reach disk.
      </p></li><li class="listitem"><p>
       Checkpoint writes can be spread over a longer time period to smooth
       the I/O spike during each checkpoint (Itagaki Takahiro and Heikki
       Linnakangas)
      </p><p>
       Previously all modified buffers were forced to disk as quickly as
       possible during a
       checkpoint, causing an I/O spike that decreased server performance.
       This new approach spreads out disk writes during checkpoints,
       reducing peak I/O usage. (User-requested and shutdown checkpoints
       are still written as quickly as possible.)
      </p></li><li class="listitem"><p>
       Heap-Only Tuples (<acronym class="acronym">HOT</acronym>) accelerate space reuse for most
       <code class="command">UPDATE</code>s and <code class="command">DELETE</code>s (Pavan Deolasee, with
       ideas from many others)
      </p><p>
       <code class="command">UPDATE</code>s and <code class="command">DELETE</code>s leave dead tuples
       behind, as do failed <code class="command">INSERT</code>s.  Previously only
       <code class="command">VACUUM</code> could reclaim space taken by dead tuples. With
       <acronym class="acronym">HOT</acronym> dead tuple space can be automatically reclaimed at
       the time of <code class="command">INSERT</code> or <code class="command">UPDATE</code> if no changes
       are made to indexed columns.  This allows for more consistent
       performance.  Also, <acronym class="acronym">HOT</acronym> avoids adding duplicate index
       entries.
      </p></li><li class="listitem"><p>
       Just-in-time background writer strategy improves disk write
       efficiency (Greg Smith, Itagaki Takahiro)
      </p><p>
       This greatly reduces the need for manual tuning of the background
       writer.
      </p></li><li class="listitem"><p>
       Per-field and per-row storage overhead have been reduced
       (Greg Stark, Heikki Linnakangas)
      </p><p>
       Variable-length data types with data values less than 128 bytes long
       will see a storage decrease of 3 to 6 bytes. For example, two adjacent
       <code class="type">char(1)</code> fields now use 4 bytes instead of 16. Row headers
       are also 4 bytes shorter than before.
      </p></li><li class="listitem"><p>
       Using non-persistent transaction IDs for read-only transactions
       reduces overhead and <code class="command">VACUUM</code> requirements (Florian Pflug)
      </p><p>
       Non-persistent transaction IDs do not increment the global
       transaction counter. Therefore, they reduce the load on
       <code class="structname">pg_clog</code> and increase the time between forced
       vacuums to prevent transaction ID wraparound.
       Other performance
       improvements were also made that should improve concurrency.
      </p></li><li class="listitem"><p>
       Avoid incrementing the command counter after a read-only command (Tom)
      </p><p>
       There was formerly a hard limit of 2<sup>32</sup>
       (4 billion) commands per transaction.  Now only commands that
       actually changed the database count, so while this limit still
       exists, it should be significantly less annoying.
      </p></li><li class="listitem"><p>
       Create a dedicated <acronym class="acronym">WAL</acronym> writer process to off-load
       work from backends (Simon)
      </p></li><li class="listitem"><p>
       Skip unnecessary WAL writes for <code class="command">CLUSTER</code> and
       <code class="command">COPY</code> (Simon)
      </p><p>
       Unless WAL archiving is enabled, the system now avoids WAL writes
       for <code class="command">CLUSTER</code> and just <code class="function">fsync()</code>s the
       table at the end of the command.  It also does the same for
       <code class="command">COPY</code> if the table was created in the same
       transaction.
      </p></li><li class="listitem"><p>
       Large sequential scans no longer force out frequently used
       cached pages (Simon, Heikki, Tom)
      </p></li><li class="listitem"><p>
       Concurrent large sequential scans can now share disk reads (Jeff Davis)
      </p><p>
       This is accomplished by starting the new sequential scan in the
       middle of the table (where another sequential scan is already
       in-progress) and wrapping around to the beginning to finish.  This
       can affect the order of returned rows in a query that does not
       specify <code class="literal">ORDER BY</code>.  The <code class="varname">synchronize_seqscans</code>
       configuration parameter can be used to disable this if necessary.
      </p></li><li class="listitem"><p>
       <code class="literal">ORDER BY ... LIMIT</code> can be done without sorting
       (Greg Stark)
      </p><p>
       This is done by sequentially scanning the table and tracking just
       the <span class="quote">“<span class="quote">top N</span>”</span> candidate rows, rather than performing a
       full sort of the entire table.  This is useful when there is no
       matching index and the <code class="literal">LIMIT</code> is not large.
      </p></li><li class="listitem"><p>
       Put a rate limit on messages sent to the statistics
       collector by backends
       (Tom)
      </p><p>
       This reduces overhead for short transactions, but might sometimes
       increase the delay before statistics are tallied.
      </p></li><li class="listitem"><p>
       Improve hash join performance for cases with many NULLs (Tom)
      </p></li><li class="listitem"><p>
       Speed up operator lookup for cases with non-exact datatype matches (Tom)
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.4"><div class="titlepage"><div><div><h4 class="title">E.205.3.2. Server</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Autovacuum is now enabled by default (Alvaro)
      </p><p>
       Several changes were made to eliminate disadvantages of having
       autovacuum enabled, thereby justifying the change in default.
       Several other autovacuum parameter defaults were also modified.
      </p></li><li class="listitem"><p>
       Support multiple concurrent autovacuum processes (Alvaro, Itagaki
       Takahiro)
      </p><p>
       This allows multiple vacuums to run concurrently.  This prevents
       vacuuming of a large table from delaying vacuuming of smaller tables.
      </p></li><li class="listitem"><p>
       Automatically re-plan cached queries when table
       definitions change or statistics are updated (Tom)
      </p><p>
       Previously PL/pgSQL functions that referenced temporary tables
       would fail if the temporary table was dropped and recreated
       between function invocations, unless <code class="literal">EXECUTE</code> was
       used.  This improvement fixes that problem and many related issues.
      </p></li><li class="listitem"><p>
       Add a <code class="varname">temp_tablespaces</code> parameter to control
       the tablespaces for temporary tables and files (Jaime Casanova,
       Albert Cervera, Bernd Helmle)
      </p><p>
       This parameter defines a list of tablespaces to be used.  This
       enables spreading the I/O load across multiple tablespaces. A random
       tablespace is chosen each time a temporary object is created.
       Temporary files are no longer stored in per-database
       <code class="filename">pgsql_tmp/</code> directories but in per-tablespace
       directories.
      </p></li><li class="listitem"><p>
       Place temporary tables' TOAST tables in special schemas named
       <code class="literal">pg_toast_temp_<em class="replaceable"><code>nnn</code></em></code> (Tom)
      </p><p>
       This allows low-level code to recognize these tables as temporary,
       which enables various optimizations such as not WAL-logging changes
       and using local rather than shared buffers for access. This also
       fixes a bug wherein backends unexpectedly held open file references
       to temporary TOAST tables.
      </p></li><li class="listitem"><p>
       Fix problem that a constant flow of new connection requests could
       indefinitely delay the postmaster from completing a shutdown or
       a crash restart (Tom)
      </p></li><li class="listitem"><p>
       Guard against a very-low-probability data loss scenario by preventing
       re-use of a deleted table's relfilenode until after the next
       checkpoint (Heikki)
      </p></li><li class="listitem"><p>
       Fix <code class="command">CREATE CONSTRAINT TRIGGER</code>
       to convert old-style foreign key trigger definitions into regular
       foreign key constraints (Tom)
      </p><p>
       This will ease porting of foreign key constraints carried forward from
       pre-7.3 databases, if they were never converted using
       <code class="filename">contrib/adddepend</code>.
      </p></li><li class="listitem"><p>
       Fix <code class="literal">DEFAULT NULL</code> to override inherited defaults (Tom)
      </p><p>
       <code class="literal">DEFAULT NULL</code> was formerly considered a noise phrase, but it
       should (and now does) override non-null defaults that would otherwise
       be inherited from a parent table or domain.
      </p></li><li class="listitem"><p>
       Add new encodings EUC_JIS_2004 and SHIFT_JIS_2004 (Tatsuo)
      </p><p>
       These new encodings can be converted to and from UTF-8.
      </p></li><li class="listitem"><p>
       Change server startup log message from <span class="quote">“<span class="quote">database system is
       ready</span>”</span> to <span class="quote">“<span class="quote">database system is ready to accept
       connections</span>”</span>, and adjust its timing
      </p><p>
       The message now appears only when the postmaster is really ready
       to accept connections.
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.5"><div class="titlepage"><div><div><h4 class="title">E.205.3.3. Monitoring</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Add <code class="varname">log_autovacuum_min_duration</code> parameter to
       support configurable logging of autovacuum activity (Simon, Alvaro)
      </p></li><li class="listitem"><p>
       Add <code class="varname">log_lock_waits</code> parameter to log lock waiting
       (Simon)
      </p></li><li class="listitem"><p>
       Add <code class="varname">log_temp_files</code> parameter to log temporary
       file usage (Bill Moran)
      </p></li><li class="listitem"><p>
       Add <code class="varname">log_checkpoints</code> parameter to improve logging
       of checkpoints (Greg Smith, Heikki)
      </p></li><li class="listitem"><p>
       <code class="varname">log_line_prefix</code> now supports
       <code class="literal">%s</code> and <code class="literal">%c</code> escapes in all
       processes (Andrew)
      </p><p>
       Previously these escapes worked only for user sessions, not for
       background database processes.
      </p></li><li class="listitem"><p>
       Add <code class="varname">log_restartpoints</code> to control logging of
       point-in-time recovery restart points (Simon)
      </p></li><li class="listitem"><p>
       Last transaction end time is now logged at end of recovery and at
       each logged restart point (Simon)
      </p></li><li class="listitem"><p>
       Autovacuum now reports its activity start time in
       <code class="literal">pg_stat_activity</code> (Tom)
      </p></li><li class="listitem"><p>
       Allow server log output in comma-separated value (CSV) format (Arul
       Shaji, Greg Smith, Andrew Dunstan)
      </p><p>
       CSV-format log files can easily be loaded into a database table for
       subsequent analysis.
      </p></li><li class="listitem"><p>
       Use PostgreSQL-supplied timezone support for formatting timestamps
       displayed in the server log (Tom)
      </p><p>
       This avoids Windows-specific problems with localized time zone
       names that are in the wrong encoding. There is a new
       <code class="varname">log_timezone</code> parameter that controls the timezone
       used in log messages, independently of the client-visible
       <code class="varname">timezone</code> parameter.
      </p></li><li class="listitem"><p>
       New system view <code class="literal">pg_stat_bgwriter</code> displays
       statistics about background writer activity (Magnus)
      </p></li><li class="listitem"><p>
       Add new columns for database-wide tuple statistics to
       <code class="literal">pg_stat_database</code> (Magnus)
      </p></li><li class="listitem"><p>
       Add an <code class="literal">xact_start</code> (transaction start time) column to
       <code class="literal">pg_stat_activity</code> (Neil)
      </p><p>
       This makes it easier to identify long-running transactions.
      </p></li><li class="listitem"><p>
       Add <code class="literal">n_live_tuples</code> and <code class="literal">n_dead_tuples</code> columns
       to <code class="literal">pg_stat_all_tables</code> and related views (Glen
       Parker)
      </p></li><li class="listitem"><p>
       Merge <code class="varname">stats_block_level</code> and <code class="varname">stats_row_level</code>
       parameters into a single parameter <code class="varname">track_counts</code>, which
       controls all messages sent to the statistics collector process
       (Tom)
      </p></li><li class="listitem"><p>
       Rename <code class="varname">stats_command_string</code> parameter to
       <code class="varname">track_activities</code> (Tom)
      </p></li><li class="listitem"><p>
       Fix statistical counting of live and dead tuples to recognize that
       committed and aborted transactions have different effects (Tom)
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.6"><div class="titlepage"><div><div><h4 class="title">E.205.3.4. Authentication</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Support Security Service Provider Interface (<acronym class="acronym">SSPI</acronym>) for
       authentication on Windows (Magnus)
      </p></li><li class="listitem"><p>
       Support GSSAPI authentication (Henry Hotz, Magnus)
      </p><p>
       This should be preferred to native Kerberos authentication because
       GSSAPI is an industry standard.
      </p></li><li class="listitem"><p>
       Support a global SSL configuration file (Victor Wagner)
      </p></li><li class="listitem"><p>
       Add <code class="varname">ssl_ciphers</code> parameter to control accepted SSL ciphers
       (Victor Wagner)
      </p></li><li class="listitem"><p>
       Add a Kerberos realm parameter, <code class="varname">krb_realm</code> (Magnus)
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.7"><div class="titlepage"><div><div><h4 class="title">E.205.3.5. Write-Ahead Log (<acronym class="acronym">WAL</acronym>) and Continuous Archiving</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Change the timestamps recorded in transaction WAL records from
       time_t to TimestampTz representation (Tom)
      </p><p>
       This provides sub-second resolution in WAL, which can be useful for
       point-in-time recovery.
      </p></li><li class="listitem"><p>
       Reduce WAL disk space needed by warm standby servers (Simon)
      </p><p>
       This change allows a warm standby server to pass the name of the earliest
       still-needed WAL file to the recovery script, allowing automatic removal
       of no-longer-needed WAL files.  This is done using <code class="literal">%r</code> in
       the <code class="varname">restore_command</code> parameter of
       <code class="filename">recovery.conf</code>.
      </p></li><li class="listitem"><p>
       New boolean configuration parameter, <code class="varname">archive_mode</code>,
       controls archiving (Simon)
      </p><p>
       Previously setting <code class="varname">archive_command</code> to an empty string
       turned off archiving. Now <code class="varname">archive_mode</code> turns archiving
       on and off, independently of <code class="varname">archive_command</code>. This is
       useful for stopping archiving temporarily.
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.8"><div class="titlepage"><div><div><h4 class="title">E.205.3.6. Queries</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Full text search is integrated into the core database
       system (Teodor, Oleg)
      </p><p>
       Text search has been improved, moved into the core code, and is now
       installed by default.  <code class="filename">contrib/tsearch2</code> now contains
       a compatibility interface.
      </p></li><li class="listitem"><p>
       Add control over whether <code class="literal">NULL</code>s sort first or last (Teodor, Tom)
      </p><p>
       The syntax is <code class="literal">ORDER BY ... NULLS FIRST/LAST</code>.
      </p></li><li class="listitem"><p>
       Allow per-column ascending/descending (<code class="literal">ASC</code>/<code class="literal">DESC</code>)
       ordering options for indexes (Teodor, Tom)
      </p><p>
       Previously a query using <code class="literal">ORDER BY</code> with mixed
       <code class="literal">ASC</code>/<code class="literal">DESC</code> specifiers could not fully use
       an index. Now an index can be fully used in such cases if the
       index was created with matching
       <code class="literal">ASC</code>/<code class="literal">DESC</code> specifications.
       <code class="literal">NULL</code> sort order within an index can be controlled, too.
      </p></li><li class="listitem"><p>
       Allow <code class="literal">col IS NULL</code> to use an index (Teodor)
      </p></li><li class="listitem"><p>
       Updatable cursors (Arul Shaji, Tom)
      </p><p>
       This eliminates the need to reference a primary key to
       <code class="command">UPDATE</code> or <code class="command">DELETE</code> rows returned by a cursor.
       The syntax is <code class="literal">UPDATE/DELETE WHERE CURRENT OF</code>.
      </p></li><li class="listitem"><p>
       Allow <code class="literal">FOR UPDATE</code> in cursors (Arul Shaji, Tom)
      </p></li><li class="listitem"><p>
       Create a general mechanism that supports casts to and from the
       standard string types (<code class="type">TEXT</code>, <code class="type">VARCHAR</code>,
       <code class="type">CHAR</code>) for <span class="emphasis"><em>every</em></span> datatype, by
       invoking the datatype's I/O functions (Tom)
      </p><p>
       Previously, such casts were available only for types that had
       specialized function(s) for the purpose.
       These new casts are assignment-only in the to-string direction,
       explicit-only in the other direction, and therefore should create no
       surprising behavior.
      </p></li><li class="listitem"><p>
       Allow <code class="literal">UNION</code> and related constructs to return a domain
       type, when all inputs are of that domain type (Tom)
      </p><p>
       Formerly, the output would be considered to be of the domain's base
       type.
      </p></li><li class="listitem"><p>
       Allow limited hashing when using two different data types (Tom)
      </p><p>
       This allows hash joins, hash indexes, hashed subplans, and hash
       aggregation to be used in situations involving cross-data-type
       comparisons, if the data types have compatible hash functions.
       Currently, cross-data-type hashing support exists for
       <code class="type">smallint</code>/<code class="type">integer</code>/<code class="type">bigint</code>,
       and for <code class="type">float4</code>/<code class="type">float8</code>.
      </p></li><li class="listitem"><p>
       Improve optimizer logic for detecting when variables are equal
       in a <code class="literal">WHERE</code> clause (Tom)
      </p><p>
       This allows mergejoins to work with descending sort orders, and
       improves recognition of redundant sort columns.
      </p></li><li class="listitem"><p>
       Improve performance when planning large inheritance trees in
       cases where most tables are excluded by constraints (Tom)
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.9"><div class="titlepage"><div><div><h4 class="title">E.205.3.7. Object Manipulation</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Arrays of composite types (David Fetter, Andrew, Tom)
      </p><p>
       In addition to arrays of explicitly-declared composite types,
       arrays of the rowtypes of regular tables and views are now
       supported, except for rowtypes of system catalogs, sequences, and TOAST
       tables.
      </p></li><li class="listitem"><p>
       Server configuration parameters can now be set on a per-function
       basis (Tom)
      </p><p>
       For example, functions can now set their own
       <code class="varname">search_path</code> to prevent unexpected behavior if a
       different <code class="varname">search_path</code> exists at run-time.  Security
       definer functions should set <code class="varname">search_path</code> to
       avoid security loopholes.
      </p></li><li class="listitem"><p>
       <code class="command">CREATE/ALTER FUNCTION</code> now supports
       <code class="literal">COST</code> and <code class="literal">ROWS</code> options (Tom)
      </p><p>
       <code class="literal">COST</code> allows specification of the cost of a
       function call.  <code class="literal">ROWS</code> allows specification of
       the average number or rows returned by a set-returning function.
       These values are used by the optimizer in choosing the best plan.
      </p></li><li class="listitem"><p>
       Implement <code class="command">CREATE TABLE LIKE ...  INCLUDING
       INDEXES</code> (Trevor Hardcastle, Nikhil Sontakke, Neil)
      </p></li><li class="listitem"><p>
       Allow <code class="command">CREATE INDEX CONCURRENTLY</code> to ignore
       transactions in other databases (Simon)
      </p></li><li class="listitem"><p>
       Add <code class="command">ALTER VIEW ... RENAME TO</code> and <code class="command">ALTER
       SEQUENCE ... RENAME TO</code> (David Fetter, Neil)
      </p><p>
       Previously this could only be done via <code class="command">ALTER TABLE ...
       RENAME TO</code>.
      </p></li><li class="listitem"><p>
       Make <code class="command">CREATE/DROP/RENAME DATABASE</code> wait briefly for
       conflicting backends to exit before failing (Tom)
      </p><p>
       This increases the likelihood that these commands will succeed.
      </p></li><li class="listitem"><p>
       Allow triggers and rules to be deactivated in groups using a
       configuration parameter, for replication purposes (Jan)
      </p><p>
       This allows replication systems to disable triggers and rewrite
       rules as a group without modifying the system catalogs directly.
       The behavior is controlled by <code class="command">ALTER TABLE</code> and a new
       parameter <code class="varname">session_replication_role</code>.
      </p></li><li class="listitem"><p>
       User-defined types can now have type modifiers (Teodor, Tom)
      </p><p>
       This allows a user-defined type to take a modifier, like
       <code class="type">ssnum(7)</code>.  Previously only built-in
       data types could have modifiers.
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.10"><div class="titlepage"><div><div><h4 class="title">E.205.3.8. Utility Commands</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Non-superuser database owners now are able to add trusted procedural
       languages to their databases by default (Jeremy Drake)
      </p><p>
       While this is reasonably safe, some administrators might wish to
       revoke the privilege. It is controlled by
       <code class="structname">pg_pltemplate</code>.<code class="structfield">tmpldbacreate</code>.
      </p></li><li class="listitem"><p>
       Allow a session's current parameter setting to be used as the
       default for future sessions (Tom)
      </p><p>
       This is done with <code class="literal">SET ... FROM CURRENT</code> in
       <code class="command">CREATE/ALTER FUNCTION</code>, <code class="command">ALTER
       DATABASE</code>, or <code class="command">ALTER ROLE</code>.
      </p></li><li class="listitem"><p>
       Implement new commands <code class="command">DISCARD ALL</code>,
       <code class="command">DISCARD PLANS</code>, <code class="command">DISCARD
       TEMPORARY</code>, <code class="command">CLOSE ALL</code>, and
       <code class="command">DEALLOCATE ALL</code> (Marko Kreen, Neil)
      </p><p>
       These commands simplify resetting a database session to its initial
       state, and are particularly useful for connection-pooling software.
      </p></li><li class="listitem"><p>
       Make <code class="command">CLUSTER</code> MVCC-safe (Heikki Linnakangas)
      </p><p>
       Formerly, <code class="command">CLUSTER</code> would discard all tuples
       that were committed dead, even if there were still transactions
       that should be able to see them under MVCC visibility rules.
      </p></li><li class="listitem"><p>
       Add new <code class="command">CLUSTER</code> syntax: <code class="literal">CLUSTER
       <em class="replaceable"><code>table</code></em> USING <em class="replaceable"><code>index</code></em></code>
       (Holger Schurig)
     </p><p>
       The old <code class="command">CLUSTER</code> syntax is still supported, but
       the new form is considered more logical.
      </p></li><li class="listitem"><p>
       Fix <code class="command">EXPLAIN</code> so it can show complex plans
       more accurately (Tom)
      </p><p>
       References to subplan outputs are now always shown correctly,
       instead of using <code class="literal">?column<em class="replaceable"><code>N</code></em>?</code>
       for complicated cases.
      </p></li><li class="listitem"><p>
       Limit the amount of information reported when a user is dropped
       (Alvaro)
      </p><p>
       Previously, dropping (or attempting to drop) a user who owned many
       objects could result in large <code class="literal">NOTICE</code> or
       <code class="literal">ERROR</code> messages listing all these objects; this
       caused problems for some client applications.  The length of the
       message is now limited, although a full list is still sent to the
       server log.
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.11"><div class="titlepage"><div><div><h4 class="title">E.205.3.9. Data Types</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Support for the SQL/XML standard, including new operators and an
       <code class="type">XML</code> data type (Nikolay Samokhvalov, Pavel Stehule, Peter)
      </p></li><li class="listitem"><p>
       Enumerated data types (<code class="type">ENUM</code>) (Tom Dunstan)
      </p><p>
       This feature provides convenient support for fields that have a
       small, fixed set of allowed values.  An example of creating an
       <code class="literal">ENUM</code> type is
       <code class="literal">CREATE TYPE mood AS ENUM ('sad', 'ok', 'happy')</code>.
      </p></li><li class="listitem"><p>
       Universally Unique Identifier (<code class="type">UUID</code>) data type (Gevik
       Babakhani, Neil)
      </p><p>
       This closely matches <acronym class="acronym">RFC</acronym> 4122.
      </p></li><li class="listitem"><p>
       Widen the <code class="type">MONEY</code> data type to 64 bits (D'Arcy Cain)
      </p><p>
       This greatly increases the range of supported <code class="type">MONEY</code>
       values.
      </p></li><li class="listitem"><p>
       Fix <code class="type">float4</code>/<code class="type">float8</code> to handle
       <code class="literal">Infinity</code> and <code class="literal">NAN</code> (Not A Number)
       consistently (Bruce)
      </p><p>
       The code formerly was not consistent about distinguishing
       <code class="literal">Infinity</code> from overflow conditions.
      </p></li><li class="listitem"><p>
       Allow leading and trailing whitespace during input of
       <code class="type">boolean</code> values (Neil)
      </p></li><li class="listitem"><p>
       Prevent <code class="command">COPY</code> from using digits and lowercase letters as
       delimiters (Tom)
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.12"><div class="titlepage"><div><div><h4 class="title">E.205.3.10. Functions</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Add new regular expression functions
       <code class="function">regexp_matches()</code>,
       <code class="function">regexp_split_to_array()</code>, and
       <code class="function">regexp_split_to_table()</code> (Jeremy Drake, Neil)
      </p><p>
       These functions provide extraction of regular expression
       subexpressions and allow splitting a string using a POSIX regular
       expression.
      </p></li><li class="listitem"><p>
       Add <code class="function">lo_truncate()</code> for large object truncation
       (Kris Jurka)
      </p></li><li class="listitem"><p>
       Implement <code class="function">width_bucket()</code> for the <code class="type">float8</code>
       data type (Neil)
      </p></li><li class="listitem"><p>
       Add <code class="function">pg_stat_clear_snapshot()</code> to discard
       statistics snapshots collected during the current transaction
       (Tom)
      </p><p>
       The first request for statistics in a transaction takes a statistics
       snapshot that does not change during the transaction.  This function
       allows the snapshot to be discarded and a new snapshot loaded during
       the next statistics query. This is particularly useful for PL/pgSQL
       functions, which are confined to a single transaction.
      </p></li><li class="listitem"><p>
       Add <code class="literal">isodow</code> option to <code class="function">EXTRACT()</code> and
       <code class="function">date_part()</code> (Bruce)
      </p><p>
       This returns the day of the week, with Sunday as seven.
       (<code class="literal">dow</code> returns Sunday as zero.)
      </p></li><li class="listitem"><p>
       Add <code class="literal">ID</code> (ISO day of week) and <code class="literal">IDDD</code> (ISO
       day of year) format codes for <code class="function">to_char()</code>,
       <code class="function">to_date()</code>, and <code class="function">to_timestamp()</code> (Brendan
       Jurd)
      </p></li><li class="listitem"><p>
       Make <code class="function">to_timestamp()</code> and <code class="function">to_date()</code>
       assume <code class="literal">TM</code> (trim) option for potentially
       variable-width fields (Bruce)
      </p><p>
       This matches <span class="productname">Oracle</span>'s behavior.
      </p></li><li class="listitem"><p>
       Fix off-by-one conversion error in
       <code class="function">to_date()</code>/<code class="function">to_timestamp()</code>
       <code class="literal">D</code> (non-ISO day of week) fields (Bruce)
      </p></li><li class="listitem"><p>
       Make <code class="function">setseed()</code> return void, rather than a
       useless integer value (Neil)
      </p></li><li class="listitem"><p>
       Add a hash function for <code class="type">NUMERIC</code> (Neil)
      </p><p>
       This allows hash indexes and hash-based plans to be used with
       <code class="type">NUMERIC</code> columns.
      </p></li><li class="listitem"><p>
       Improve efficiency of
       <code class="literal">LIKE</code>/<code class="literal">ILIKE</code>, especially for
       multi-byte character sets like UTF-8 (Andrew, Itagaki Takahiro)
      </p></li><li class="listitem"><p>
       Make <code class="function">currtid()</code> functions require
       <code class="literal">SELECT</code> privileges on the target table (Tom)
      </p></li><li class="listitem"><p>
       Add several <code class="function">txid_*()</code> functions to query
       active transaction IDs (Jan)
      </p><p>
       This is useful for various replication solutions.
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.13"><div class="titlepage"><div><div><h4 class="title">E.205.3.11. PL/pgSQL Server-Side Language</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Add scrollable cursor support, including directional control in
       <code class="command">FETCH</code> (Pavel Stehule)
      </p></li><li class="listitem"><p>
       Allow <code class="literal">IN</code> as an alternative to
       <code class="literal">FROM</code> in PL/pgSQL's <code class="command">FETCH</code>
       statement, for consistency with the backend's
       <code class="command">FETCH</code> command (Pavel Stehule)
      </p></li><li class="listitem"><p>
       Add <code class="command">MOVE</code> to PL/pgSQL (Magnus, Pavel Stehule,
       Neil)
      </p></li><li class="listitem"><p>
       Implement <code class="command">RETURN QUERY</code> (Pavel Stehule, Neil)
      </p><p>
       This adds convenient syntax for PL/pgSQL set-returning functions
       that want to return the result of a query.  <code class="command">RETURN QUERY</code>
       is easier and more efficient than a loop
       around <code class="command">RETURN NEXT</code>.
      </p></li><li class="listitem"><p>
       Allow function parameter names to be qualified with the
       function's name (Tom)
      </p><p>
       For example, <code class="literal">myfunc.myvar</code>. This is particularly
       useful for specifying variables in a query where the variable
       name might match a column name.
      </p></li><li class="listitem"><p>
       Make qualification of variables with block labels work properly (Tom)
      </p><p>
       Formerly, outer-level block labels could unexpectedly interfere with
       recognition of inner-level record or row references.
      </p></li><li class="listitem"><p>
       Tighten requirements for <code class="literal">FOR</code> loop
       <code class="literal">STEP</code> values (Tom)
      </p><p>
       Prevent non-positive <code class="literal">STEP</code> values, and handle
       loop overflows.
      </p></li><li class="listitem"><p>
       Improve accuracy when reporting syntax error locations (Tom)
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.14"><div class="titlepage"><div><div><h4 class="title">E.205.3.12. Other Server-Side Languages</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Allow type-name arguments to PL/Perl
       <code class="function">spi_prepare()</code> to be data type aliases in
       addition to names found in <code class="literal">pg_type</code> (Andrew)
      </p></li><li class="listitem"><p>
       Allow type-name arguments to PL/Python
       <code class="function">plpy.prepare()</code> to be data type aliases in
       addition to names found in <code class="literal">pg_type</code> (Andrew)
      </p></li><li class="listitem"><p>
       Allow type-name arguments to PL/Tcl <code class="function">spi_prepare</code> to
       be data type aliases in addition to names found in
       <code class="literal">pg_type</code> (Andrew)
      </p></li><li class="listitem"><p>
       Enable PL/PythonU to compile on Python 2.5 (Marko Kreen)
      </p></li><li class="listitem"><p>
       Support a true PL/Python boolean type in compatible Python versions
       (Python 2.3 and later) (Marko Kreen)
      </p></li><li class="listitem"><p>
       Fix PL/Tcl problems with thread-enabled <code class="filename">libtcl</code> spawning
       multiple threads within the backend (Steve Marshall, Paul Bayer,
       Doug Knight)
      </p><p>
       This caused all sorts of unpleasantness.
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.15"><div class="titlepage"><div><div><h4 class="title">E.205.3.13. <a class="link" href="app-psql.html" title="psql"><span class="application">psql</span></a></h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       List disabled triggers separately in <code class="literal">\d</code> output
       (Brendan Jurd)
      </p></li><li class="listitem"><p>
       In <code class="literal">\d</code> patterns, always match <code class="literal">$</code>
       literally (Tom)
      </p></li><li class="listitem"><p>
       Show aggregate return types in <code class="literal">\da</code> output
       (Greg Sabino Mullane)
      </p></li><li class="listitem"><p>
       Add the function's volatility status to the output of
       <code class="literal">\df+</code> (Neil)
      </p></li><li class="listitem"><p>
       Add <code class="literal">\prompt</code> capability (Chad Wagner)
      </p></li><li class="listitem"><p>
       Allow <code class="literal">\pset</code>, <code class="literal">\t</code>, and
       <code class="literal">\x</code> to specify <code class="literal">on</code> or <code class="literal">off</code>,
       rather than just toggling (Chad Wagner)
      </p></li><li class="listitem"><p>
       Add <code class="literal">\sleep</code> capability (Jan)
      </p></li><li class="listitem"><p>
       Enable <code class="literal">\timing</code> output for <code class="literal">\copy</code> (Andrew)
      </p></li><li class="listitem"><p>
       Improve <code class="literal">\timing</code> resolution on Windows
       (Itagaki Takahiro)
      </p></li><li class="listitem"><p>
       Flush <code class="literal">\o</code> output after each backslash command (Tom)
      </p></li><li class="listitem"><p>
       Correctly detect and report errors while reading a <code class="literal">-f</code>
       input file (Peter)
      </p></li><li class="listitem"><p>
       Remove <code class="literal">-u</code> option (this option has long been deprecated)
       (Tom)
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.16"><div class="titlepage"><div><div><h4 class="title">E.205.3.14. <a class="link" href="app-pgdump.html" title="pg_dump"><span class="application">pg_dump</span></a></h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Add <code class="literal">--tablespaces-only</code> and <code class="literal">--roles-only</code>
       options to <span class="application">pg_dumpall</span> (Dave Page)
      </p></li><li class="listitem"><p>
       Add an output file option to
       <span class="application">pg_dumpall</span> (Dave Page)
      </p><p>
       This is primarily useful on Windows, where output redirection of
       child <span class="application">pg_dump</span> processes does not work.
      </p></li><li class="listitem"><p>
       Allow <span class="application">pg_dumpall</span> to accept an initial-connection
       database name rather than the default
       <code class="literal">template1</code> (Dave Page)
      </p></li><li class="listitem"><p>
       In <code class="literal">-n</code> and <code class="literal">-t</code> switches, always match
       <code class="literal">$</code> literally (Tom)
      </p></li><li class="listitem"><p>
       Improve performance when a database has thousands of objects (Tom)
      </p></li><li class="listitem"><p>
       Remove <code class="literal">-u</code> option (this option has long been deprecated)
       (Tom)
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.17"><div class="titlepage"><div><div><h4 class="title">E.205.3.15. Other Client Applications</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       In <span class="application">initdb</span>, allow the location of the
       <code class="filename">pg_xlog</code> directory to be specified
       (Euler Taveira de Oliveira)
      </p></li><li class="listitem"><p>
       Enable server core dump generation in <span class="application">pg_regress</span>
       on supported operating systems (Andrew)
      </p></li><li class="listitem"><p>
       Add a <code class="literal">-t</code> (timeout) parameter to <span class="application">pg_ctl</span>
       (Bruce)
      </p><p>
       This controls how long <span class="application">pg_ctl</span> will wait when waiting
       for server startup or shutdown.  Formerly the timeout was hard-wired
       as 60 seconds.
      </p></li><li class="listitem"><p>
       Add a <span class="application">pg_ctl</span> option to control generation
       of server core dumps (Andrew)
      </p></li><li class="listitem"><p>
       Allow Control-C to cancel <span class="application">clusterdb</span>,
       <span class="application">reindexdb</span>, and <span class="application">vacuumdb</span> (Itagaki
       Takahiro, Magnus)
      </p></li><li class="listitem"><p>
       Suppress command tag output for <span class="application">createdb</span>,
       <span class="application">createuser</span>, <span class="application">dropdb</span>, and
       <span class="application">dropuser</span> (Peter)
      </p><p>
       The <code class="literal">--quiet</code> option is ignored and will be removed in 8.4.
       Progress messages when acting on all databases now go to stdout
       instead of stderr because they are not actually errors.
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.18"><div class="titlepage"><div><div><h4 class="title">E.205.3.16. <a class="link" href="libpq.html" title="Chapter 34. libpq - C Library"><span class="application">libpq</span></a></h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Interpret the <code class="literal">dbName</code> parameter of
       <code class="function">PQsetdbLogin()</code> as a <code class="literal">conninfo</code> string if
       it contains an equals sign (Andrew)
      </p><p>
       This allows use of <code class="literal">conninfo</code> strings in client
       programs that still use <code class="literal">PQsetdbLogin()</code>.
      </p></li><li class="listitem"><p>
       Support a global <acronym class="acronym">SSL</acronym> configuration file (Victor
       Wagner)
      </p></li><li class="listitem"><p>
       Add environment variable <code class="varname">PGSSLKEY</code> to control
       <acronym class="acronym">SSL</acronym> hardware keys (Victor Wagner)
      </p></li><li class="listitem"><p>
       Add <code class="function">lo_truncate()</code> for large object
       truncation (Kris Jurka)
      </p></li><li class="listitem"><p>
       Add <code class="function">PQconnectionNeedsPassword()</code> that returns
       true if the server required a password but none was supplied
       (Joe Conway, Tom)
      </p><p>
       If this returns true after a failed connection attempt, a client
       application should prompt the user for a password.  In the past
       applications have had to check for a specific error message string to
       decide whether a password is needed; that approach is now
       deprecated.
      </p></li><li class="listitem"><p>
       Add <code class="function">PQconnectionUsedPassword()</code> that returns
       true if the supplied password was actually used
       (Joe Conway, Tom)
      </p><p>
       This is useful in some security contexts where it is important
       to know whether a user-supplied password is actually valid.
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.19"><div class="titlepage"><div><div><h4 class="title">E.205.3.17. <a class="link" href="ecpg.html" title="Chapter 36. ECPG - Embedded SQL in C"><span class="application">ecpg</span></a></h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Use V3 frontend/backend protocol (Michael)
      </p><p>
       This adds support for server-side prepared statements.
      </p></li><li class="listitem"><p>
       Use native threads, instead of pthreads, on Windows (Magnus)
      </p></li><li class="listitem"><p>
       Improve thread-safety of ecpglib (Itagaki Takahiro)
      </p></li><li class="listitem"><p>
       Make the ecpg libraries export only necessary API symbols (Michael)
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.20"><div class="titlepage"><div><div><h4 class="title">E.205.3.18. <span class="application">Windows</span> Port</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Allow the whole <span class="productname">PostgreSQL</span> distribution to be compiled
       with <span class="productname">Microsoft Visual C++</span> (Magnus and others)
      </p><p>
       This allows Windows-based developers to use familiar development
       and debugging tools.
       Windows executables made with Visual C++ might also have better
       stability and performance than those made with other tool sets.
       The client-only Visual C++ build scripts have been removed.
      </p></li><li class="listitem"><p>
       Drastically reduce postmaster's memory usage when it has many child
       processes (Magnus)
      </p></li><li class="listitem"><p>
       Allow regression tests to be started by an administrative
       user (Magnus)
      </p></li><li class="listitem"><p>
       Add native shared memory implementation (Magnus)
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.21"><div class="titlepage"><div><div><h4 class="title">E.205.3.19. Server Programming Interface (<acronym class="acronym">SPI</acronym>)</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Add cursor-related functionality in SPI (Pavel Stehule)
      </p><p>
       Allow access to the cursor-related planning options, and add
       <code class="command">FETCH</code>/<code class="command">MOVE</code> routines.
      </p></li><li class="listitem"><p>
       Allow execution of cursor commands through
       <code class="function">SPI_execute</code> (Tom)
      </p><p>
       The macro <code class="literal">SPI_ERROR_CURSOR</code> still exists but will
       never be returned.
      </p></li><li class="listitem"><p>
       SPI plan pointers are now declared as <code class="literal">SPIPlanPtr</code> instead of
       <code class="literal">void *</code> (Tom)
      </p><p>
       This does not break application code, but switching is
       recommended to help catch simple programming mistakes.
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.22"><div class="titlepage"><div><div><h4 class="title">E.205.3.20. Build Options</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Add <span class="application">configure</span> option <code class="literal">--enable-profiling</code>
       to enable code profiling (works only with <span class="application">gcc</span>)
       (Korry Douglas and Nikhil Sontakke)
      </p></li><li class="listitem"><p>
       Add <span class="application">configure</span> option <code class="literal">--with-system-tzdata</code>
       to use the operating system's time zone database (Peter)
      </p></li><li class="listitem"><p>
       Fix <acronym class="acronym">PGXS</acronym> so extensions can be built against PostgreSQL
       installations whose <span class="application">pg_config</span> program does not
       appear first in the <code class="varname">PATH</code> (Tom)
      </p></li><li class="listitem"><p>
       Support <code class="command">gmake draft</code> when building the
       <acronym class="acronym">SGML</acronym> documentation (Bruce)
      </p><p>
       Unless <code class="literal">draft</code> is used, the documentation build will
       now be repeated if necessary to ensure the index is up-to-date.
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.23"><div class="titlepage"><div><div><h4 class="title">E.205.3.21. Source Code</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Rename macro <code class="literal">DLLIMPORT</code> to <code class="literal">PGDLLIMPORT</code> to
       avoid conflicting with third party includes (like Tcl) that
       define <code class="literal">DLLIMPORT</code> (Magnus)
      </p></li><li class="listitem"><p>
       Create <span class="quote">“<span class="quote">operator families</span>”</span> to improve planning of
       queries involving cross-data-type comparisons (Tom)
      </p></li><li class="listitem"><p>
       Update GIN <code class="function">extractQuery()</code> API to allow signalling
       that nothing can satisfy the query (Teodor)
      </p></li><li class="listitem"><p>
       Move <code class="literal">NAMEDATALEN</code> definition from
       <code class="filename">postgres_ext.h</code> to <code class="filename">pg_config_manual.h</code>
       (Peter)
      </p></li><li class="listitem"><p>
       Provide <code class="function">strlcpy()</code> and
       <code class="function">strlcat()</code> on all platforms, and replace
       error-prone uses of <code class="function">strncpy()</code>,
       <code class="function">strncat()</code>, etc (Peter)
      </p></li><li class="listitem"><p>
       Create hooks to let an external plugin monitor (or even replace) the
       planner and create plans for hypothetical situations (Gurjeet
       Singh, Tom)
      </p></li><li class="listitem"><p>
       Create a function variable <code class="literal">join_search_hook</code> to let plugins
       override the join search order portion of the planner (Julius
       Stroffek)
      </p></li><li class="listitem"><p>
       Add <code class="function">tas()</code> support for Renesas' M32R processor
       (Kazuhiro Inaoka)
      </p></li><li class="listitem"><p>
       <code class="function">quote_identifier()</code> and
       <span class="application">pg_dump</span> no longer quote keywords that are
       unreserved according to the grammar (Tom)
      </p></li><li class="listitem"><p>
       Change the on-disk representation of the <code class="type">NUMERIC</code>
       data type so that the <code class="structfield">sign_dscale</code> word comes
       before the weight (Tom)
      </p></li><li class="listitem"><p>
       Use <acronym class="acronym">SYSV</acronym> semaphores rather than POSIX on Darwin
       &gt;= 6.0, i.e., macOS 10.2 and up (Chris Marcellino)
      </p></li><li class="listitem"><p>
       Add <a class="link" href="acronyms.html" title="Appendix K. Acronyms">acronym</a> and <a class="link" href="creating-cluster.html#CREATING-CLUSTER-NFS" title="18.2.2. Use of Network File Systems">NFS</a> documentation
       sections (Bruce)
      </p></li><li class="listitem"><p>
       "Postgres" is now documented as an accepted alias for
       "PostgreSQL" (Peter)
      </p></li><li class="listitem"><p>
       Add documentation about preventing database server spoofing when
       the server is down (Bruce)
      </p></li></ul></div></div><div class="sect3" id="id-1.11.6.209.5.24"><div class="titlepage"><div><div><h4 class="title">E.205.3.22. Contrib</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
       Move <code class="filename">contrib</code> <code class="filename">README</code> content into the
       main <span class="productname">PostgreSQL</span> documentation (Albert Cervera i
       Areny)
      </p></li><li class="listitem"><p>
       Add <code class="filename">contrib/pageinspect</code> module for low-level
       page inspection (Simon, Heikki)
      </p></li><li class="listitem"><p>
       Add <code class="filename">contrib/pg_standby</code> module for controlling
       warm standby operation (Simon)
      </p></li><li class="listitem"><p>
       Add <code class="filename">contrib/uuid-ossp</code> module for generating
       <code class="type">UUID</code> values using the OSSP UUID library (Peter)
      </p><p>
       Use <span class="application">configure</span>
       <code class="literal">--with-ossp-uuid</code> to activate. This takes
       advantage of the new <code class="type">UUID</code> builtin type.
      </p></li><li class="listitem"><p>
       Add <code class="filename">contrib/dict_int</code>,
       <code class="filename">contrib/dict_xsyn</code>, and
       <code class="filename">contrib/test_parser</code> modules to provide
       sample add-on text search dictionary templates and parsers
       (Sergey Karpov)
      </p></li><li class="listitem"><p>
       Allow <span class="application">contrib/pgbench</span> to set the fillfactor (Pavan
       Deolasee)
      </p></li><li class="listitem"><p>
       Add timestamps to <span class="application">contrib/pgbench</span> <code class="literal">-l</code>
       (Greg Smith)
      </p></li><li class="listitem"><p>
       Add usage count statistics to
       <code class="filename">contrib/pgbuffercache</code> (Greg Smith)
      </p></li><li class="listitem"><p>
       Add GIN support for <code class="filename">contrib/hstore</code> (Teodor)
      </p></li><li class="listitem"><p>
       Add GIN support for <code class="filename">contrib/pg_trgm</code> (Guillaume Smet, Teodor)
      </p></li><li class="listitem"><p>
       Update OS/X startup scripts in
       <code class="filename">contrib/start-scripts</code> (Mark Cotner, David
       Fetter)
      </p></li><li class="listitem"><p>
       Restrict <code class="function">pgrowlocks()</code> and
       <code class="function">dblink_get_pkey()</code> to users who have
       <code class="literal">SELECT</code> privilege on the target table (Tom)
      </p></li><li class="listitem"><p>
       Restrict <code class="filename">contrib/pgstattuple</code> functions to
       superusers (Tom)
      </p></li><li class="listitem"><p>
       <code class="filename">contrib/xml2</code> is deprecated and planned for
       removal in 8.4 (Peter)
      </p><p>
       The new XML support in core PostgreSQL supersedes this module.
      </p></li></ul></div></div></div></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="release-8-3-1.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="release.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="release-8-2-23.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">E.204. Release 8.3.1 </td><td width="20%" align="center"><a accesskey="h" href="index.html">Home</a></td><td width="40%" align="right" valign="top"> E.206. Release 8.2.23</td></tr></table></div></body></html>