<itemizedlist>

<listitem>

<para>

Partitioning by a hash key

Add support for partitioning by a hash key

</para>

</listitem>

<listitem>

<para>

Support for <literal>PRIMARY KEY</literal>, <literal>FOREIGN

Add support for <literal>PRIMARY KEY</literal>, <literal>FOREIGN

KEY</literal>, indexes, and triggers on partitioned tables

</para>

</listitem>

<listitem>

<para>

<command>UPDATE</command> statements that change a partition key

column now move affected rows to the appropriate partitions

column now cause affected rows to be moved to the appropriate

partitions

</para>

</listitem>

<listitem>

<para>

Improved <command>SELECT</command> performancedue to enhanced

partition elimination strategies during queryprocessing

Improve <command>SELECT</command> performancethrough enhanced

partition elimination strategies during queryplanning and execution

</para>

</listitem>

</itemizedlist>

<listitem>

<para>

Many other useful performance improvements, includingmaking

<command>ALTER TABLE ... ADD COLUMN</command> with a

non-null column default faster

Many other useful performance improvements, includinga significant

speedup to<command>ALTER TABLE ... ADD COLUMN</command> with a

non-null column default, as it no longer rewrites the table data

</para>

</listitem>

<application>pg_dumpall</application>'s output script will now always

create databases with their original locale and encoding, and hence

will fail if the locale or encoding name is unknown to the

destination system. Previously <command>CREATE DATABASE</command>

destination system. Previously, <command>CREATE DATABASE</command>

would be emitted without these specifications if the database locale

and encoding matched the old cluster's defaults.

</para>

</para>

<para>

Previously they skipped one <emphasis>byte</emphasis> for each byte

Previously, they skipped one <emphasis>byte</emphasis> for each byte

of template character, resulting in strange behavior if either string

contained multibyte characters.

</para>

<para>

Per the SQL standard, relative paths start from the document node of

the XML input document, not the root node as these functions

formerly did it.

previously did.

</para>

</listitem>

<para>

Creation of a trigger on a partitioned table automatically creates

triggers on all existingpartitions,andon newly-created ones.

triggers on all existing andfuture partitions.

This also allows deferred unique constraints on partitioned tables.

</para>

</listitem>

-->

<para>

Allow updated rows to automatically move to new partitions based

on the new row contents (Amit Khandekar)

<command>UPDATE</command> statements that change a partition key

column now cause affected rows to be moved to the appropriate

partitions (Amit Khandekar)

</para>

</listitem>

</para>

<para>

Previously partition elimination only happened at planning

Previously, partition elimination only happened at planning

time, meaning many joins and prepared queries could not use

partition elimination.

</para>

-->

<para>

Allow equality joins between partitioned tables with identically

partitioned child tables to join the child tables directly

(Ashutosh Bapat)

In an equality join between partitioned tables, allow matching

partitions to be joined directly (Ashutosh Bapat)

</para>

<para>

-->

<para>

Allowbtree indexes to be built in parallel (Peter Geoghegan,

Allowparallel building of a btree index (Peter Geoghegan,

Rushabh Lathia, Heikki Linnakangas)

</para>

</listitem>

</para>

<para>

Previously the most common values (<acronym>MCV</acronym>s) were

Previously, the most common values (<acronym>MCV</acronym>s) were

identified based on their frequency compared to all column

values. Now, <acronym>MCV</acronym>s are chosen based on their

frequency compared to the non-<acronym>MCV</acronym> values.

</para>

<para>

Previously such cases used the same selectivity estimates

Previously, such cases used the same selectivity estimates

as <literal>&gt;</literal> and <literal>&lt;</literal>, respectively,

unless the comparison constants are <acronym>MCV</acronym>s.

This change is particularly helpful for queries

</para>

<para>

Previously only non-join <command>UPDATE</command>s and

Previously, only non-join <command>UPDATE</command>s and

<command>DELETE</command>s were pushed.

</para>

<literal>pg_execute_server_program</literal>. These roles now also

control who can use server-side <command>COPY</command> and the <link

linkend="file-fdw"><filename>file_fdw</filename></link> extension.

Previously only superusers could use these functions, and that

Previously, only superusers could use these functions, and that

is still the default behavior.

</para>

</listitem>

</para>

<para>

Specifically, these functions were modified:<link

Specifically, these functions were modified: <link

linkend="functions-admin-genfile-table"><function>pg_ls_dir()</function></link>,

<function>pg_read_file()</function>,

<function>pg_read_binary_file()</function>,

<para>

<productname>PostgreSQL</productname> only allows superusers to

access <filename>postgres_fdw</filename> tables without

passwords, e.g. via <literal>peer</literal>. Previously the

session owner had to be a superuser to allow such access;now

passwords, e.g. via <literal>peer</literal>. Previously, the

session owner had to be a superuser to allow such access; now

the view owner is checked instead.

</para>

</listitem>

</para>

<para>

Previously the 16MB default could only be changed at compile time.

Previously, the 16MB default could only be changed at compile time.

</para>

</listitem>

-->

<para>

No longer retain<acronym>WAL</acronym>that spans two checkpoints

Retain<acronym>WAL</acronym>data for only a single checkpoint

(Simon Riggs)

</para>

<para>

Retention of <acronym>WAL</acronym> records is only required for one

checkpoint.

Previously, <acronym>WAL</acronym> was retained for two checkpoints.

</para>

</listitem>

-->

<para>

Allow heap pages' checksumsto bechecked during streaming base

Allowchecksums ofheap pagesto beverified during streaming base

backup (Michael Banck)

</para>

</listitem>

<para>

Allow <function>to_char()</function> and

<function>to_timestamp()</function> to specify the time zone's

hours and minutesfrom <acronym>UTC</acronym>(Nikita Glukhov,

Andrew Dunstan)

offsetfrom <acronym>UTC</acronym>in hours and minutes

(Nikita Glukhov,Andrew Dunstan)

</para>

<para>

</para>

<para>

Previously such circumstances generated errors.

Previously, such circumstances generated errors.

</para>

</listitem>

</para>

<para>

Previously partition information would not be displayed for a

Previously, partition information would not be displayed for a

partitioned table if it had no partitions. Also indicate which

partitions are themselves partitioned.

</para>

</para>

<para>

Previously tab completion queries could fail against older servers.

Previously, tab completion queries could fail against older servers.

</para>

</listitem>

<para>

Allow <application>pgbench</application> to do exponentiation

with <function>pow()</function> and<function>power()</function>

with <function>pow()</function> and <function>power()</function>

(Ra&uacute;l Mar&iacute;n Rodr&iacute;guez)

</para>

</listitem>

-->

<para>

Make the computation of system column

Make the computation of

<structname>pg_class</structname>.<structfield>reltuples</structfield>

more consistent (Tomas Vondra)

by <command>VACUUM</command> consistent with its computation

by <command>ANALYZE</command> (Tomas Vondra)

</para>

</listitem>

</para>

<para>

Previously only superusers could call <filename>adminpack</filename>

Previously, only superusers could call <filename>adminpack</filename>

functions; now role permissions are checked.

</para>

</listitem>