Create the <filename>aqo</filename> extension using the following query:

</para>

<programlisting>

&quot;CREATE EXTENSION aqo;&quot;

CREATE EXTENSION aqo;

</programlisting>

</listitem>

</orderedlist>

<sect3 id="aqo-choosing-query-optimization-modes">

<title>Choosing Query Optimization Modes</title>

<para>

If you often run queries of the same type, for example,you application limits the number

If you often run queries of the same type, for example,your application limits the number

of possible query types, you can use the <literal>intelligent</literal> mode to

improve planning for these queries. In this mode, <filename>aqo</filename>

analyzes each query execution and stores statistics. Statistics on queries of

When run in the <literal>intelligent</literal> mode, <filename>aqo</filename> assigns a unique hash value

to each query type to separate the collected statistics. If you

switch to the <literal>forced</literal> mode, the statistics for all untracked query

types is stored in common query type with hash 0. You can view all

types is stored inacommon query type with hash 0. You can view all

the processed query types and their corresponding hash values in

the <structname>aqo_query_texts</structname> table:

</para>

SET aqo.mode='intelligent';

SELECT * FROM a, b WHERE a.id=b.id;

SET aqo.mode='manual';

SET aqo.mode='controlled';

-- Disable auto_tuning, enable both learn_aqo and use_aqo

-- for this query type:

</programlisting>

<para>

Alternatively, you can specify a particular query type to reset by

providing its hash value in the <literal>fspase_hash</literal> option. For example:

providing its hash value in the <literal>fspace_hash</literal> option. For example:

</para>

<programlisting>

DELETE FROM aqo_data WHERE fspace_hash = (SELECT fspace_hash FROM aqo_queries

<entry>Optimizes all queries together, regardless of the query type.</entry>

</row>

<row>

<entry><literal>manual</literal></entry>

<entry><literal>controlled</literal></entry>

<entry><emphasis role="strong">Default.</emphasis> Uses the default planner for all new queries, but can reuse the collected statistics for already known query types, if any.</entry>

</row>

<row>

intelligent mode. In other modes, new queries are not

appended to <structname>aqo_queries</structname> automatically.

You can change this behavior by setting the

<literal>auto_tuning</literal> variable toTRUE.</entry>

<literal>auto_tuning</literal> variable to<literal>true</literal>.</entry>

</row>

</tbody>

</tgroup>

<!-- doc/src/sgml/atx.sgml -->

<chapter id="atx">

<title>Autonomoustransactions</title>

<title>AutonomousTransactions</title>

<sect1 id="atx-overview">

<title>Overview of autonomous transactions</title>

<title>Overview</title>

<para>

&productname; supports nested transactions: they are rarely explicitly used by programmer and mostly used for error handling and stored

procedures. It is possible to rollback subtransaction without affecting parent transaction. But commit of subtraction is delayed until

procedures. It is possible to rollbackasubtransaction without affectingtheparent transaction. But commit of a subtraction is delayed until

commit of parent transaction.

</para>

<para>

But in some cases application needs to run several independent transactions in one session.

<quote>Autonomous Subtransactions</quote> (in short <acronym>AST</acronym>) denotes the capability of a single session

to run multiple independent transactions, as if multiple different sessions were executing each transaction.

</para>

<para>

Autonomous transactions are needed mostly for implementing audits, when the fact of performing audit should be reported regardless

status of audit itself: whether it was successfully completed or not.

Autonomous transactions are widely used in Oracle PL-SQL, so porting such procedures to &productname; is problematic without autonomous transaction support.

However, in some cases applications need to run several independent transactions inside a single transaction &mdash; <quote>autonomous transactions</quote>. Autonomous transactions are needed mostly for implementing audits, when the fact of performing an audit should be reported regardless of the

status of the audit itself: whether it was successfully completed or not.

</para>

</sect1>

<sect1><title>Behavior</title>

<para>

An<acronym>AST</acronym> can happen only inside another transaction.

Inside an existing transaction (call it T0), the user can decide to starta subtransaction. Then T0 is paused and pushed in an<acronym>AST</acronym> stack, and a new transaction (call it T1) is started.

Anautonomous transaction can happen only inside another transaction.

Inside an existing transaction (call it T0), the user can decide to startan autonomous transaction. Then T0 is paused and pushed in anautonomous transaction stack, and a new transaction (call it T1) is started.

</para>

<para>

At some point in the future the user can commit thesubtransaction; after T1 is committed then T0 is popped from the<acronym>AST</acronym> stack and resumed.

The user can also decide to <command>COMMIT</command> the parent transaction T0, in which case T1 is committed, then T0 is popped from the<acronym>AST</acronym> stack and then committed.

At some point in the future the user can commit theautonomous transaction; after T1 is committed then T0 is popped from theautonomous transaction stack and resumed.

The user can also decide to <command>COMMIT</command> the parent transaction T0, in which case T1 is committed, then T0 is popped from theautonomous transaction stack and then committed.

</para>

<para>

All the transactions happen synchronously; at any time only one transaction can be active, while in the stack there are zero (or more) paused transactions in the stack.

All the possible combinations of <command>COMMIT</command> / <command>ROLLBACK</command> for T0 and T1 can happen;

for instance, it is possible to COMMIT T1 and ROLLBACK T0.

It is possible to nestsubtransactions, up to a global resource limit (e.g. the<acronym>AST</acronym> stack size) which can be set on the server.

It is possible to nestautonomous transactions, up to a global resource limit (e.g. theautonomous transaction stack size) which can be set on the server.

</para>

</sect1>

<sect1><title>Example</title>

<sect1><title>Examples</title>

<para>

The following figure describes anexample where atransactionexecutes a subtransaction. A continuous line denotes an active transaction, while a dotted line denotes a transaction which has been paused and pushed in the<acronym>AST</acronym> stack. Time flows downwards.

This example illustrates how anautonomoustransactionis executed. A continuous line denotes an active transaction, while a dotted line denotes a transaction which has been paused and pushed in theautonomous transaction stack. Time flows downwards.

</para>

<programlisting>

BEGIN; --start ordinarytx T0

BEGIN; --starts ordinarytransaction T0

|

INSERT INTO t VALUES (1);

:\

: BEGIN AUTONOMOUS TRANSACTION; -- start AST tx T1, pushes T0 into stack

: BEGIN AUTONOMOUS TRANSACTION; -- starts autonomous transaction

: | -- T1, pushes T0 into stack

: |

: INSERT INTO t VALUES (2);

: |

: COMMIT AUTONOMOUS TRANSACTION / ROLLBACK AUTONOMOUS TRANSACTION; -- ends tx T1, pops tx T0 from stack

: COMMIT AUTONOMOUS TRANSACTION / ROLLBACK AUTONOMOUS TRANSACTION;

: | -- ends autonomous transaction

: | -- T1, pops transaction T0 from stack

:/

COMMIT / ROLLBACK; -- endstx T0

COMMIT / ROLLBACK;-- endstransaction T0

</programlisting>

<para>

Depending on the two choices between <command>COMMIT</command> and <command>ROLLBACK</command> we can get4 different outputs from:

Depending on the two choices between <command>COMMIT</command> and <command>ROLLBACK</command>, we can getfour different outputs from:

<programlisting>

SELECT sum(x) from t;

</programlisting>

</para>

</sect1>

<sect1><title>Example 2 (more than one subtransaction)</title>

<para>

The parent transaction can have more than onesubtransaction, just by repeatingtheapplication of thepush/pop cycle.

The parent transaction can have more than oneautonomous transaction ifthe push/pop cycle is repeated.

</para>

<programlisting>

BEGIN; --start ordinarytx T0

BEGIN;--starts ordinarytransaction T0

|

INSERT INTO t VALUES (1);

:\

: BEGIN AUTONOMOUS TRANSACTION; -- start AST tx T1, pushes T0 into stack

: BEGIN AUTONOMOUS TRANSACTION; -- starts autonomous transaction

: | -- T1, pushes T0 into stack

: |

: INSERT INTO t VALUES (2);

: |

: COMMIT AUTONOMOUS TRANSACTION / ROLLBACK AUTONOMOUS TRANSACTION; -- ends tx T1, pops tx T0 from stack

: COMMIT AUTONOMOUS TRANSACTION / ROLLBACK AUTONOMOUS TRANSACTION;

: | -- ends autonomous transaction

: | -- T1, pops T0 from stack

:/

|

:\

: BEGIN AUTONOMOUS TRANSACTION; -- start AST tx T2, pushes T0 into stack

: BEGIN AUTONOMOUS TRANSACTION; -- starts autonomous transaction

: | -- T2, pushes T0 into stack

: |

: INSERT INTO t VALUES (4);

: |

: COMMIT AUTONOMOUS TRANSACTION / ROLLBACK AUTONOMOUS TRANSACTION; -- ends tx T2, pops tx T0 from stack

: COMMIT AUTONOMOUS TRANSACTION / ROLLBACK AUTONOMOUS TRANSACTION;

: | -- ends autonomous transaction

: | -- T2, pops T0 from stack

:/

COMMIT / ROLLBACK; -- endstx T0

COMMIT / ROLLBACK;-- endstransaction T0

</programlisting>

</sect1>

<para>

Visibility rules work as in the case of independent transactions executed via <literal>dblink</literal>.

T1 does not see the effects of T0, because the latter has not been committed yet. T0 might see the effects of T1, depending on its own transaction isolation mode.

In case of Read Committed isolation level the parent transaction will see changes made by autonomoussubtransaction.

But in case of Repeatable Read isolation level the parent transaction will not see changes made by autonomoussubtransaction.

In case of Read Committed isolation level the parent transaction will see changes made by autonomoustransactions.

But in case of Repeatable Read isolation level the parent transaction will not see changes made by autonomoustransactions.

</para>

<para>

Now single-session deadlocks become possible, because an<acronym>AST</acronym>can become entangled with one of the paused transactions init's session.

Now single-session deadlocks become possible, because anautonomous transactioncan become entangled with one of the paused transactions inits session.

Autonomous transaction T1 is assumed to depend on parent transaction T0 and if it attempts to obtain any resource locked by T0, then

deadlock is reported.

</para>

</sect1>

<sect1><title>SQLgrammar extension forautonomous transactions</title>

<sect1><title>SQLGrammar Extension forAutonomous Transactions</title>

<para>

&productname; <literal>BEGIN</literal>/<literal>END</literal> transaction statements are extended by optional keyword <literal>AUTONOMOUS</literal>:

<para>

Specifying <literal>AUTONOMOUS</literal> keyword in <literal>END TRANSACTION</literal> clause is optional.

It is possible to have several nesting levels of autonomous transactions, but top level transactioncan not be autonomous.

It is possible to have several nesting levels of autonomous transactions, but top level transactioncannot be autonomous.

</para>

</sect1>

<sect1><title>PL/pgSQLgrammar extension forautonomous transactions</title>

<sect1><title>PL/pgSQLGrammar Extension forAutonomous Transactions</title>

<para>

Block construction in PL/pgSQL is extended by optional <literal>autonomous</literal> keyword.

It is possible to treatallfunction body as autonomous transaction:

It is possible to treatthe wholefunction body as an autonomous transaction:

</para>

<programlisting>

</sect1>

<sect1><title>PL/Pythonextension forautonomous transactions</title>

<sect1><title>PL/PythonExtension forAutonomous Transactions</title>

<para>

In addition to <varname>subtransaction</varname> method, PL/Python module provides new <varname>autonomous</varname> method

which can be used in <varname>WITH</varname> clause to start autonomous transaction:

which can be used in <varname>WITH</varname> clause to startanautonomous transaction:

</para>

<programlisting>

applications are advised to make use of the error code scheme

instead.

</para>

<para>

Maintaining catalogs of message translations requires the on-going

efforts of many volunteers that want to see

<productname>&productname;</> speak their preferred language well.

If messages in your language are currently not available or not fully

translated, your assistance would be appreciated. If you want to

help, refer to <xref linkend="nls"> or write to the developers'

mailing list.

</para>

</sect2>

</sect1>

</para>

<para>

GSSAPI support has to be enabled when <productname>&productname;</> is built;

see <xref linkend="installation"> for more information.

GSSAPI support has to be enabled when <productname>&productname;</> is built.

</para>

<para>

<para>

When building from the source distribution, these components are not built

automatically, unless you build the "world" target

(see <xref linkend="build">).

automatically, unless you build the "world" target.

You can build and install all of them by running:

<screen>

<userinput>make</userinput>

<para>

The installation instructions are also distributed as plain text,

in case they are needed in a situation where better reading tools

are not available. The <filename>INSTALL</filename> file

are not available.<!-- The <filename>INSTALL</filename> file

corresponds to <xref linkend="installation">, with some minor

changes to account for the different context. To recreate the

changes to account for the different context.--> To recreate the

file, change to the directory <filename>doc/src/sgml</filename>

and enter <userinput>make INSTALL</userinput>.

</para>

<!ENTITY release SYSTEM "release.sgml">

<!ENTITY release-pro-9.6 SYSTEM "release-pro-9.6.sgml">

<!ENTITY release-proee-9.6 SYSTEM "release-proee-9.6.sgml">

<!ENTITY release-9.6 SYSTEM "release-9.6.sgml">

<!ENTITY release-9.5 SYSTEM "release-9.5.sgml">

<!ENTITY release-9.4 SYSTEM "release-9.4.sgml">