2010-09-20 22:08:53 +02:00
|
|
|
<!-- doc/src/sgml/rules.sgml -->
|
2001-07-10 01:50:32 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<chapter id="rules">
|
|
|
|
<title>The Rule System</title>
|
1998-03-01 09:16:16 +01:00
|
|
|
|
2001-05-13 00:51:36 +02:00
|
|
|
<indexterm zone="rules">
|
2003-08-31 19:32:24 +02:00
|
|
|
<primary>rule</primary>
|
2001-05-13 00:51:36 +02:00
|
|
|
</indexterm>
|
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-09-11 23:42:20 +02:00
|
|
|
This chapter discusses the rule system in
|
|
|
|
<productname>PostgreSQL</productname>. Production rule systems
|
|
|
|
are conceptually simple, but there are many subtle points
|
|
|
|
involved in actually using them.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
Some other database systems define active database rules, which
|
2003-09-11 23:42:20 +02:00
|
|
|
are usually stored procedures and triggers. In
|
|
|
|
<productname>PostgreSQL</productname>, these can be implemented
|
|
|
|
using functions and triggers as well.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
The rule system (more precisely speaking, the query rewrite rule
|
|
|
|
system) is totally different from stored procedures and triggers.
|
|
|
|
It modifies queries to take rules into consideration, and then
|
|
|
|
passes the modified query to the query planner for planning and
|
|
|
|
execution. It is very powerful, and can be used for many things
|
|
|
|
such as query language procedures, views, and versions. The
|
|
|
|
theoretical foundations and the power of this rule system are
|
|
|
|
also discussed in <xref linkend="STON90b"> and <xref
|
|
|
|
linkend="ONG90">.
|
1998-12-29 03:24:47 +01:00
|
|
|
</para>
|
2002-01-07 03:29:15 +01:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<sect1 id="querytree">
|
|
|
|
<title>The Query Tree</title>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-08-31 19:32:24 +02:00
|
|
|
<indexterm zone="querytree">
|
|
|
|
<primary>query tree</primary>
|
|
|
|
</indexterm>
|
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
1998-10-25 02:29:01 +02:00
|
|
|
To understand how the rule system works it is necessary to know
|
2000-12-12 06:07:59 +01:00
|
|
|
when it is invoked and what its input and results are.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
The rule system is located between the parser and the planner.
|
|
|
|
It takes the output of the parser, one query tree, and the user-defined
|
|
|
|
rewrite rules, which are also
|
|
|
|
query trees with some extra information, and creates zero or more
|
2001-09-10 23:58:47 +02:00
|
|
|
query trees as result. So its input and output are always things
|
1998-10-25 02:29:01 +02:00
|
|
|
the parser itself could have produced and thus, anything it sees
|
2005-01-23 01:30:59 +01:00
|
|
|
is basically representable as an <acronym>SQL</acronym> statement.
|
|
|
|
</para>
|
1998-03-01 09:16:16 +01:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2001-09-10 23:58:47 +02:00
|
|
|
Now what is a query tree? It is an internal representation of an
|
2005-01-23 01:30:59 +01:00
|
|
|
<acronym>SQL</acronym> statement where the single parts that it is
|
2003-04-11 15:22:35 +02:00
|
|
|
built from are stored separately. These query trees can be shown
|
|
|
|
in the server log if you set the configuration parameters
|
|
|
|
<varname>debug_print_parse</varname>,
|
|
|
|
<varname>debug_print_rewritten</varname>, or
|
|
|
|
<varname>debug_print_plan</varname>. The rule actions are also
|
|
|
|
stored as query trees, in the system catalog
|
|
|
|
<structname>pg_rewrite</structname>. They are not formatted like
|
|
|
|
the log output, but they contain exactly the same information.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
Reading a raw query tree requires some experience. But since
|
2005-01-23 01:30:59 +01:00
|
|
|
<acronym>SQL</acronym> representations of query trees are
|
2003-04-11 15:22:35 +02:00
|
|
|
sufficient to understand the rule system, this chapter will not
|
|
|
|
teach how to read them.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2010-10-10 19:43:33 +02:00
|
|
|
When reading the <acronym>SQL</acronym> representations of the
|
2003-04-11 15:22:35 +02:00
|
|
|
query trees in this chapter it is necessary to be able to identify
|
2001-09-10 23:58:47 +02:00
|
|
|
the parts the statement is broken into when it is in the query tree
|
|
|
|
structure. The parts of a query tree are
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<variablelist>
|
|
|
|
<varlistentry>
|
|
|
|
<term>
|
2001-09-10 23:58:47 +02:00
|
|
|
the command type
|
2005-01-23 01:30:59 +01:00
|
|
|
</term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
1998-10-25 02:29:01 +02:00
|
|
|
This is a simple value telling which command
|
2003-02-19 04:13:25 +01:00
|
|
|
(<command>SELECT</command>, <command>INSERT</command>,
|
|
|
|
<command>UPDATE</command>, <command>DELETE</command>) produced
|
2003-04-11 15:22:35 +02:00
|
|
|
the query tree.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<varlistentry>
|
|
|
|
<term>
|
2001-09-10 23:58:47 +02:00
|
|
|
the range table
|
2005-01-23 01:30:59 +01:00
|
|
|
</term>
|
2001-11-12 20:19:39 +01:00
|
|
|
<indexterm><primary>range table</></>
|
2005-01-23 01:30:59 +01:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2001-09-10 23:58:47 +02:00
|
|
|
The range table is a list of relations that are used in the query.
|
2004-12-30 04:13:56 +01:00
|
|
|
In a <command>SELECT</command> statement these are the relations given after
|
|
|
|
the <literal>FROM</literal> key word.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2001-09-10 23:58:47 +02:00
|
|
|
Every range table entry identifies a table or view and tells
|
2004-12-30 04:13:56 +01:00
|
|
|
by which name it is called in the other parts of the query.
|
|
|
|
In the query tree, the range table entries are referenced by
|
|
|
|
number rather than by name, so here it doesn't matter if there
|
2005-01-23 01:30:59 +01:00
|
|
|
are duplicate names as it would in an <acronym>SQL</acronym>
|
2004-12-30 04:13:56 +01:00
|
|
|
statement. This can happen after the range tables of rules
|
|
|
|
have been merged in. The examples in this chapter will not have
|
|
|
|
this situation.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<varlistentry>
|
|
|
|
<term>
|
2001-09-10 23:58:47 +02:00
|
|
|
the result relation
|
2005-01-23 01:30:59 +01:00
|
|
|
</term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
2001-09-10 23:58:47 +02:00
|
|
|
This is an index into the range table that identifies the
|
2004-12-30 04:13:56 +01:00
|
|
|
relation where the results of the query go.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2010-10-10 19:43:33 +02:00
|
|
|
<command>SELECT</command> queries don't have a result
|
|
|
|
relation. (The special case of <command>SELECT INTO</command> is
|
|
|
|
mostly identical to <command>CREATE TABLE</command> followed by
|
|
|
|
<literal>INSERT ... SELECT</literal>, and is not discussed
|
|
|
|
separately here.)
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
For <command>INSERT</command>, <command>UPDATE</command>, and
|
2004-12-30 04:13:56 +01:00
|
|
|
<command>DELETE</command> commands, the result relation is the table
|
|
|
|
(or view!) where the changes are to take effect.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<varlistentry>
|
|
|
|
<term>
|
2001-09-10 23:58:47 +02:00
|
|
|
the target list
|
2005-01-23 01:30:59 +01:00
|
|
|
</term>
|
2003-08-31 19:32:24 +02:00
|
|
|
<indexterm><primary>target list</></>
|
2005-01-23 01:30:59 +01:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2003-02-19 04:13:25 +01:00
|
|
|
The target list is a list of expressions that define the
|
2003-04-11 15:22:35 +02:00
|
|
|
result of the query. In the case of a
|
|
|
|
<command>SELECT</command>, these expressions are the ones that
|
|
|
|
build the final output of the query. They correspond to the
|
|
|
|
expressions between the key words <command>SELECT</command>
|
|
|
|
and <command>FROM</command>. (<literal>*</literal> is just an
|
|
|
|
abbreviation for all the column names of a relation. It is
|
|
|
|
expanded by the parser into the individual columns, so the
|
|
|
|
rule system never sees it.)
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2010-10-10 19:43:33 +02:00
|
|
|
<command>DELETE</command> commands don't need a normal target list
|
|
|
|
because they don't produce any result. Instead, the rule system
|
|
|
|
adds a special <acronym>CTID</> entry to the empty target list,
|
|
|
|
to allow the executor to find the row to be deleted.
|
|
|
|
(<acronym>CTID</> is added when the result relation is an ordinary
|
|
|
|
table. If it is a view, a whole-row variable is added instead,
|
|
|
|
as described in <xref linkend="rules-views-update">.)
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
For <command>INSERT</command> commands, the target list describes
|
2004-12-30 04:13:56 +01:00
|
|
|
the new rows that should go into the result relation. It consists of the
|
|
|
|
expressions in the <literal>VALUES</> clause or the ones from the
|
|
|
|
<command>SELECT</command> clause in <literal>INSERT
|
|
|
|
... SELECT</literal>. The first step of the rewrite process adds
|
|
|
|
target list entries for any columns that were not assigned to by
|
|
|
|
the original command but have defaults. Any remaining columns (with
|
|
|
|
neither a given value nor a default) will be filled in by the
|
|
|
|
planner with a constant null expression.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
For <command>UPDATE</command> commands, the target list
|
2004-12-30 04:13:56 +01:00
|
|
|
describes the new rows that should replace the old ones. In the
|
|
|
|
rule system, it contains just the expressions from the <literal>SET
|
2010-10-10 19:43:33 +02:00
|
|
|
column = expression</literal> part of the command. The planner will
|
|
|
|
handle missing columns by inserting expressions that copy the values
|
|
|
|
from the old row into the new one. Just as for <command>DELETE</>,
|
|
|
|
the rule system adds a <acronym>CTID</> or whole-row variable so that
|
|
|
|
the executor can identify the old row to be updated.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2001-09-10 23:58:47 +02:00
|
|
|
Every entry in the target list contains an expression that can
|
2004-12-30 04:13:56 +01:00
|
|
|
be a constant value, a variable pointing to a column of one
|
|
|
|
of the relations in the range table, a parameter, or an expression
|
|
|
|
tree made of function calls, constants, variables, operators, etc.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<varlistentry>
|
|
|
|
<term>
|
1998-10-25 02:29:01 +02:00
|
|
|
the qualification
|
2005-01-23 01:30:59 +01:00
|
|
|
</term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
2003-02-19 04:13:25 +01:00
|
|
|
The query's qualification is an expression much like one of
|
2004-12-30 04:13:56 +01:00
|
|
|
those contained in the target list entries. The result value of
|
|
|
|
this expression is a Boolean that tells whether the operation
|
|
|
|
(<command>INSERT</command>, <command>UPDATE</command>,
|
|
|
|
<command>DELETE</command>, or <command>SELECT</command>) for the
|
|
|
|
final result row should be executed or not. It corresponds to the <literal>WHERE</> clause
|
2005-01-23 01:30:59 +01:00
|
|
|
of an <acronym>SQL</acronym> statement.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<varlistentry>
|
|
|
|
<term>
|
2000-12-12 06:07:59 +01:00
|
|
|
the join tree
|
2005-01-23 01:30:59 +01:00
|
|
|
</term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
The query's join tree shows the structure of the <literal>FROM</> clause.
|
2004-12-30 04:13:56 +01:00
|
|
|
For a simple query like <literal>SELECT ... FROM a, b, c</literal>, the join tree is just
|
|
|
|
a list of the <literal>FROM</> items, because we are allowed to join them in
|
|
|
|
any order. But when <literal>JOIN</> expressions, particularly outer joins,
|
|
|
|
are used, we have to join in the order shown by the joins.
|
|
|
|
In that case, the join tree shows the structure of the <literal>JOIN</> expressions. The
|
|
|
|
restrictions associated with particular <literal>JOIN</> clauses (from <literal>ON</> or
|
|
|
|
<literal>USING</> expressions) are stored as qualification expressions attached
|
|
|
|
to those join-tree nodes. It turns out to be convenient to store
|
|
|
|
the top-level <literal>WHERE</> expression as a qualification attached to the
|
|
|
|
top-level join-tree item, too. So really the join tree represents
|
|
|
|
both the <literal>FROM</> and <literal>WHERE</> clauses of a <command>SELECT</command>.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
2000-12-12 06:07:59 +01:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<varlistentry>
|
|
|
|
<term>
|
1998-10-25 02:29:01 +02:00
|
|
|
the others
|
2005-01-23 01:30:59 +01:00
|
|
|
</term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
The other parts of the query tree like the <literal>ORDER BY</>
|
2004-12-30 04:13:56 +01:00
|
|
|
clause aren't of interest here. The rule system
|
|
|
|
substitutes some entries there while applying rules, but that
|
|
|
|
doesn't have much to do with the fundamentals of the rule
|
|
|
|
system.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
</variablelist>
|
1998-12-29 03:24:47 +01:00
|
|
|
</para>
|
2005-01-23 01:30:59 +01:00
|
|
|
</sect1>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<sect1 id="rules-views">
|
|
|
|
<title>Views and the Rule System</title>
|
2003-08-31 19:32:24 +02:00
|
|
|
|
|
|
|
<indexterm zone="rules-views">
|
|
|
|
<primary>rule</primary>
|
|
|
|
<secondary>and views</secondary>
|
|
|
|
</indexterm>
|
|
|
|
|
|
|
|
<indexterm zone="rules-views">
|
|
|
|
<primary>view</>
|
|
|
|
<secondary>implementation through rules</>
|
|
|
|
</indexterm>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
|
|
|
Views in <productname>PostgreSQL</productname> are implemented
|
2003-04-11 15:22:35 +02:00
|
|
|
using the rule system. In fact, there is essentially no difference
|
2007-02-01 01:28:19 +01:00
|
|
|
between:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE VIEW myview AS SELECT * FROM mytab;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2010-10-10 19:43:33 +02:00
|
|
|
|
2007-02-01 01:28:19 +01:00
|
|
|
compared against the two commands:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
|
|
|
CREATE TABLE myview (<replaceable>same column list as mytab</replaceable>);
|
2002-04-20 01:13:54 +02:00
|
|
|
CREATE RULE "_RETURN" AS ON SELECT TO myview DO INSTEAD
|
2002-01-20 23:19:57 +01:00
|
|
|
SELECT * FROM mytab;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2010-10-10 19:43:33 +02:00
|
|
|
|
2003-02-19 04:13:25 +01:00
|
|
|
because this is exactly what the <command>CREATE VIEW</command>
|
|
|
|
command does internally. This has some side effects. One of them
|
|
|
|
is that the information about a view in the
|
2005-01-23 01:30:59 +01:00
|
|
|
<productname>PostgreSQL</productname> system catalogs is exactly
|
2003-04-11 15:22:35 +02:00
|
|
|
the same as it is for a table. So for the parser, there is
|
2003-02-19 04:13:25 +01:00
|
|
|
absolutely no difference between a table and a view. They are the
|
2003-04-11 15:22:35 +02:00
|
|
|
same thing: relations.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<sect2 id="rules-select">
|
|
|
|
<title>How <command>SELECT</command> Rules Work</title>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2003-08-31 19:32:24 +02:00
|
|
|
<indexterm zone="rules-select">
|
|
|
|
<primary>rule</primary>
|
|
|
|
<secondary sortas="SELECT">for SELECT</secondary>
|
|
|
|
</indexterm>
|
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
Rules <literal>ON SELECT</> are applied to all queries as the last step, even
|
2003-02-19 04:13:25 +01:00
|
|
|
if the command given is an <command>INSERT</command>,
|
|
|
|
<command>UPDATE</command> or <command>DELETE</command>. And they
|
2003-04-11 15:22:35 +02:00
|
|
|
have different semantics from rules on the other command types in that they modify the
|
|
|
|
query tree in place instead of creating a new one. So
|
2003-02-19 04:13:25 +01:00
|
|
|
<command>SELECT</command> rules are described first.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
Currently, there can be only one action in an <literal>ON SELECT</> rule, and it must
|
|
|
|
be an unconditional <command>SELECT</> action that is <literal>INSTEAD</>. This restriction was
|
|
|
|
required to make rules safe enough to open them for ordinary users, and
|
2004-12-30 04:13:56 +01:00
|
|
|
it restricts <literal>ON SELECT</> rules to act like views.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
The examples for this chapter are two join views that do some
|
2003-02-19 04:13:25 +01:00
|
|
|
calculations and some more views using them in turn. One of the
|
|
|
|
two first views is customized later by adding rules for
|
2003-04-11 15:22:35 +02:00
|
|
|
<command>INSERT</command>, <command>UPDATE</command>, and
|
2003-02-19 04:13:25 +01:00
|
|
|
<command>DELETE</command> operations so that the final result will
|
|
|
|
be a view that behaves like a real table with some magic
|
2003-04-11 15:22:35 +02:00
|
|
|
functionality. This is not such a simple example to start from and
|
2003-02-19 04:13:25 +01:00
|
|
|
this makes things harder to get into. But it's better to have one
|
|
|
|
example that covers all the points discussed step by step rather
|
|
|
|
than having many different ones that might mix up in mind.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
For the example, we need a little <literal>min</literal> function that
|
2007-02-01 01:28:19 +01:00
|
|
|
returns the lower of 2 integer values. We create that as:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2004-05-17 01:22:08 +02:00
|
|
|
CREATE FUNCTION min(integer, integer) RETURNS integer AS $$
|
2005-01-22 23:56:36 +01:00
|
|
|
SELECT CASE WHEN $1 < $2 THEN $1 ELSE $2 END
|
2004-05-17 01:22:08 +02:00
|
|
|
$$ LANGUAGE SQL STRICT;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
|
|
|
</para>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2000-12-12 06:07:59 +01:00
|
|
|
The real tables we need in the first two rule system descriptions
|
1998-10-25 02:29:01 +02:00
|
|
|
are these:
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE TABLE shoe_data (
|
2003-04-11 15:22:35 +02:00
|
|
|
shoename text, -- primary key
|
|
|
|
sh_avail integer, -- available number of pairs
|
|
|
|
slcolor text, -- preferred shoelace color
|
|
|
|
slminlen real, -- minimum shoelace length
|
|
|
|
slmaxlen real, -- maximum shoelace length
|
|
|
|
slunit text -- length unit
|
2002-01-20 23:19:57 +01:00
|
|
|
);
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE TABLE shoelace_data (
|
2003-04-11 15:22:35 +02:00
|
|
|
sl_name text, -- primary key
|
|
|
|
sl_avail integer, -- available number of pairs
|
|
|
|
sl_color text, -- shoelace color
|
|
|
|
sl_len real, -- shoelace length
|
|
|
|
sl_unit text -- length unit
|
2002-01-20 23:19:57 +01:00
|
|
|
);
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE TABLE unit (
|
2003-04-11 15:22:35 +02:00
|
|
|
un_name text, -- primary key
|
|
|
|
un_fact real -- factor to transform to cm
|
2002-01-20 23:19:57 +01:00
|
|
|
);
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
As you can see, they represent shoe-store data.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2007-02-01 01:28:19 +01:00
|
|
|
The views are created as:
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE VIEW shoe AS
|
|
|
|
SELECT sh.shoename,
|
|
|
|
sh.sh_avail,
|
|
|
|
sh.slcolor,
|
|
|
|
sh.slminlen,
|
|
|
|
sh.slminlen * un.un_fact AS slminlen_cm,
|
|
|
|
sh.slmaxlen,
|
|
|
|
sh.slmaxlen * un.un_fact AS slmaxlen_cm,
|
|
|
|
sh.slunit
|
|
|
|
FROM shoe_data sh, unit un
|
|
|
|
WHERE sh.slunit = un.un_name;
|
|
|
|
|
|
|
|
CREATE VIEW shoelace AS
|
|
|
|
SELECT s.sl_name,
|
|
|
|
s.sl_avail,
|
|
|
|
s.sl_color,
|
|
|
|
s.sl_len,
|
|
|
|
s.sl_unit,
|
|
|
|
s.sl_len * u.un_fact AS sl_len_cm
|
|
|
|
FROM shoelace_data s, unit u
|
|
|
|
WHERE s.sl_unit = u.un_name;
|
|
|
|
|
|
|
|
CREATE VIEW shoe_ready AS
|
|
|
|
SELECT rsh.shoename,
|
|
|
|
rsh.sh_avail,
|
|
|
|
rsl.sl_name,
|
|
|
|
rsl.sl_avail,
|
|
|
|
min(rsh.sh_avail, rsl.sl_avail) AS total_avail
|
|
|
|
FROM shoe rsh, shoelace rsl
|
|
|
|
WHERE rsl.sl_color = rsh.slcolor
|
2005-01-22 23:56:36 +01:00
|
|
|
AND rsl.sl_len_cm >= rsh.slminlen_cm
|
|
|
|
AND rsl.sl_len_cm <= rsh.slmaxlen_cm;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-02-19 04:13:25 +01:00
|
|
|
The <command>CREATE VIEW</command> command for the
|
|
|
|
<literal>shoelace</literal> view (which is the simplest one we
|
2003-04-11 15:22:35 +02:00
|
|
|
have) will create a relation <literal>shoelace</> and an entry in
|
2003-02-19 04:13:25 +01:00
|
|
|
<structname>pg_rewrite</structname> that tells that there is a
|
2003-04-11 15:22:35 +02:00
|
|
|
rewrite rule that must be applied whenever the relation <literal>shoelace</>
|
2003-02-19 04:13:25 +01:00
|
|
|
is referenced in a query's range table. The rule has no rule
|
2003-04-11 15:22:35 +02:00
|
|
|
qualification (discussed later, with the non-<command>SELECT</> rules, since
|
|
|
|
<command>SELECT</> rules currently cannot have them) and it is <literal>INSTEAD</>. Note
|
|
|
|
that rule qualifications are not the same as query qualifications.
|
|
|
|
The action of our rule has a query qualification.
|
|
|
|
The action of the rule is one query tree that is a copy of the
|
2003-02-19 04:13:25 +01:00
|
|
|
<command>SELECT</command> statement in the view creation command.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
2010-10-10 19:43:33 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<note>
|
|
|
|
<para>
|
1998-10-25 02:29:01 +02:00
|
|
|
The two extra range
|
2009-11-06 00:24:27 +01:00
|
|
|
table entries for <literal>NEW</> and <literal>OLD</> that you can see in
|
2003-02-19 04:13:25 +01:00
|
|
|
the <structname>pg_rewrite</structname> entry aren't of interest
|
2003-04-11 15:22:35 +02:00
|
|
|
for <command>SELECT</command> rules.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</note>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-02-19 04:13:25 +01:00
|
|
|
Now we populate <literal>unit</literal>, <literal>shoe_data</literal>
|
2003-04-11 15:22:35 +02:00
|
|
|
and <literal>shoelace_data</literal> and run a simple query on a view:
|
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
INSERT INTO unit VALUES ('cm', 1.0);
|
|
|
|
INSERT INTO unit VALUES ('m', 100.0);
|
|
|
|
INSERT INTO unit VALUES ('inch', 2.54);
|
|
|
|
|
|
|
|
INSERT INTO shoe_data VALUES ('sh1', 2, 'black', 70.0, 90.0, 'cm');
|
|
|
|
INSERT INTO shoe_data VALUES ('sh2', 0, 'black', 30.0, 40.0, 'inch');
|
|
|
|
INSERT INTO shoe_data VALUES ('sh3', 4, 'brown', 50.0, 65.0, 'cm');
|
|
|
|
INSERT INTO shoe_data VALUES ('sh4', 3, 'brown', 40.0, 50.0, 'inch');
|
|
|
|
|
|
|
|
INSERT INTO shoelace_data VALUES ('sl1', 5, 'black', 80.0, 'cm');
|
|
|
|
INSERT INTO shoelace_data VALUES ('sl2', 6, 'black', 100.0, 'cm');
|
|
|
|
INSERT INTO shoelace_data VALUES ('sl3', 0, 'black', 35.0 , 'inch');
|
|
|
|
INSERT INTO shoelace_data VALUES ('sl4', 8, 'black', 40.0 , 'inch');
|
|
|
|
INSERT INTO shoelace_data VALUES ('sl5', 4, 'brown', 1.0 , 'm');
|
|
|
|
INSERT INTO shoelace_data VALUES ('sl6', 0, 'brown', 0.9 , 'm');
|
|
|
|
INSERT INTO shoelace_data VALUES ('sl7', 7, 'brown', 60 , 'cm');
|
|
|
|
INSERT INTO shoelace_data VALUES ('sl8', 1, 'brown', 40 , 'inch');
|
|
|
|
|
|
|
|
SELECT * FROM shoelace;
|
|
|
|
|
|
|
|
sl_name | sl_avail | sl_color | sl_len | sl_unit | sl_len_cm
|
|
|
|
-----------+----------+----------+--------+---------+-----------
|
|
|
|
sl1 | 5 | black | 80 | cm | 80
|
|
|
|
sl2 | 6 | black | 100 | cm | 100
|
|
|
|
sl7 | 7 | brown | 60 | cm | 60
|
|
|
|
sl3 | 0 | black | 35 | inch | 88.9
|
|
|
|
sl4 | 8 | black | 40 | inch | 101.6
|
|
|
|
sl8 | 1 | brown | 40 | inch | 101.6
|
|
|
|
sl5 | 4 | brown | 1 | m | 100
|
|
|
|
sl6 | 0 | brown | 0.9 | m | 90
|
2002-01-20 23:19:57 +01:00
|
|
|
(8 rows)
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<para>
|
|
|
|
This is the simplest <command>SELECT</command> you can do on our
|
2003-02-19 04:13:25 +01:00
|
|
|
views, so we take this opportunity to explain the basics of view
|
|
|
|
rules. The <literal>SELECT * FROM shoelace</literal> was
|
2007-02-01 01:28:19 +01:00
|
|
|
interpreted by the parser and produced the query tree:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
SELECT shoelace.sl_name, shoelace.sl_avail,
|
|
|
|
shoelace.sl_color, shoelace.sl_len,
|
|
|
|
shoelace.sl_unit, shoelace.sl_len_cm
|
|
|
|
FROM shoelace shoelace;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
|
|
|
and this is given to the rule system. The rule system walks through the
|
2003-04-11 15:22:35 +02:00
|
|
|
range table and checks if there are rules
|
2001-09-10 23:58:47 +02:00
|
|
|
for any relation. When processing the range table entry for
|
2003-02-19 04:13:25 +01:00
|
|
|
<literal>shoelace</literal> (the only one up to now) it finds the
|
2007-02-01 01:28:19 +01:00
|
|
|
<literal>_RETURN</literal> rule with the query tree:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
SELECT s.sl_name, s.sl_avail,
|
2002-01-20 23:19:57 +01:00
|
|
|
s.sl_color, s.sl_len, s.sl_unit,
|
2003-04-11 15:22:35 +02:00
|
|
|
s.sl_len * u.un_fact AS sl_len_cm
|
2009-11-06 00:24:27 +01:00
|
|
|
FROM shoelace old, shoelace new,
|
2002-01-20 23:19:57 +01:00
|
|
|
shoelace_data s, unit u
|
2003-04-11 15:22:35 +02:00
|
|
|
WHERE s.sl_unit = u.un_name;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
To expand the view, the rewriter simply creates a subquery range-table
|
|
|
|
entry containing the rule's action query tree, and substitutes this
|
2010-10-10 19:43:33 +02:00
|
|
|
range table entry for the original one that referenced the view. The
|
2007-02-01 01:28:19 +01:00
|
|
|
resulting rewritten query tree is almost the same as if you had typed:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
SELECT shoelace.sl_name, shoelace.sl_avail,
|
|
|
|
shoelace.sl_color, shoelace.sl_len,
|
|
|
|
shoelace.sl_unit, shoelace.sl_len_cm
|
|
|
|
FROM (SELECT s.sl_name,
|
|
|
|
s.sl_avail,
|
|
|
|
s.sl_color,
|
|
|
|
s.sl_len,
|
|
|
|
s.sl_unit,
|
|
|
|
s.sl_len * u.un_fact AS sl_len_cm
|
|
|
|
FROM shoelace_data s, unit u
|
|
|
|
WHERE s.sl_unit = u.un_name) shoelace;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2000-12-12 06:07:59 +01:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
There is one difference however: the subquery's range table has two
|
2009-11-06 00:24:27 +01:00
|
|
|
extra entries <literal>shoelace old</> and <literal>shoelace new</>. These entries don't
|
2000-12-12 06:07:59 +01:00
|
|
|
participate directly in the query, since they aren't referenced by
|
2003-04-11 15:22:35 +02:00
|
|
|
the subquery's join tree or target list. The rewriter uses them
|
|
|
|
to store the access privilege check information that was originally present
|
2002-01-20 23:19:57 +01:00
|
|
|
in the range-table entry that referenced the view. In this way, the
|
2003-04-11 15:22:35 +02:00
|
|
|
executor will still check that the user has proper privileges to access
|
2000-12-12 06:07:59 +01:00
|
|
|
the view, even though there's no direct use of the view in the rewritten
|
|
|
|
query.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
2000-12-12 06:07:59 +01:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2000-12-12 06:07:59 +01:00
|
|
|
That was the first rule applied. The rule system will continue checking
|
2002-01-20 23:19:57 +01:00
|
|
|
the remaining range-table entries in the top query (in this example there
|
|
|
|
are no more), and it will recursively check the range-table entries in
|
2003-04-11 15:22:35 +02:00
|
|
|
the added subquery to see if any of them reference views. (But it
|
2009-11-06 00:24:27 +01:00
|
|
|
won't expand <literal>old</> or <literal>new</> — otherwise we'd have infinite recursion!)
|
2003-04-11 15:22:35 +02:00
|
|
|
In this example, there are no rewrite rules for <literal>shoelace_data</> or <literal>unit</>,
|
2000-12-12 06:07:59 +01:00
|
|
|
so rewriting is complete and the above is the final result given to
|
|
|
|
the planner.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2005-10-22 16:44:35 +02:00
|
|
|
Now we want to write a query that finds out for which shoes currently in the store
|
2003-04-11 15:22:35 +02:00
|
|
|
we have the matching shoelaces (color and length) and where the
|
1998-10-25 02:29:01 +02:00
|
|
|
total number of exactly matching pairs is greater or equal to two.
|
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2005-01-22 23:56:36 +01:00
|
|
|
SELECT * FROM shoe_ready WHERE total_avail >= 2;
|
2003-04-11 15:22:35 +02:00
|
|
|
|
|
|
|
shoename | sh_avail | sl_name | sl_avail | total_avail
|
|
|
|
----------+----------+---------+----------+-------------
|
|
|
|
sh1 | 2 | sl1 | 5 | 2
|
|
|
|
sh3 | 4 | sl7 | 7 | 4
|
2002-01-20 23:19:57 +01:00
|
|
|
(2 rows)
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2007-02-01 01:28:19 +01:00
|
|
|
The output of the parser this time is the query tree:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
SELECT shoe_ready.shoename, shoe_ready.sh_avail,
|
|
|
|
shoe_ready.sl_name, shoe_ready.sl_avail,
|
|
|
|
shoe_ready.total_avail
|
|
|
|
FROM shoe_ready shoe_ready
|
2005-01-22 23:56:36 +01:00
|
|
|
WHERE shoe_ready.total_avail >= 2;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2010-10-10 19:43:33 +02:00
|
|
|
The first rule applied will be the one for the
|
2003-02-19 04:13:25 +01:00
|
|
|
<literal>shoe_ready</literal> view and it results in the
|
2007-02-01 01:28:19 +01:00
|
|
|
query tree:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
SELECT shoe_ready.shoename, shoe_ready.sh_avail,
|
|
|
|
shoe_ready.sl_name, shoe_ready.sl_avail,
|
|
|
|
shoe_ready.total_avail
|
|
|
|
FROM (SELECT rsh.shoename,
|
|
|
|
rsh.sh_avail,
|
|
|
|
rsl.sl_name,
|
|
|
|
rsl.sl_avail,
|
|
|
|
min(rsh.sh_avail, rsl.sl_avail) AS total_avail
|
|
|
|
FROM shoe rsh, shoelace rsl
|
|
|
|
WHERE rsl.sl_color = rsh.slcolor
|
2005-01-22 23:56:36 +01:00
|
|
|
AND rsl.sl_len_cm >= rsh.slminlen_cm
|
|
|
|
AND rsl.sl_len_cm <= rsh.slmaxlen_cm) shoe_ready
|
|
|
|
WHERE shoe_ready.total_avail >= 2;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2000-12-12 06:07:59 +01:00
|
|
|
|
2003-02-19 04:13:25 +01:00
|
|
|
Similarly, the rules for <literal>shoe</literal> and
|
|
|
|
<literal>shoelace</literal> are substituted into the range table of
|
2003-04-11 15:22:35 +02:00
|
|
|
the subquery, leading to a three-level final query tree:
|
2000-12-12 06:07:59 +01:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
SELECT shoe_ready.shoename, shoe_ready.sh_avail,
|
|
|
|
shoe_ready.sl_name, shoe_ready.sl_avail,
|
|
|
|
shoe_ready.total_avail
|
|
|
|
FROM (SELECT rsh.shoename,
|
|
|
|
rsh.sh_avail,
|
|
|
|
rsl.sl_name,
|
|
|
|
rsl.sl_avail,
|
|
|
|
min(rsh.sh_avail, rsl.sl_avail) AS total_avail
|
|
|
|
FROM (SELECT sh.shoename,
|
|
|
|
sh.sh_avail,
|
|
|
|
sh.slcolor,
|
|
|
|
sh.slminlen,
|
|
|
|
sh.slminlen * un.un_fact AS slminlen_cm,
|
|
|
|
sh.slmaxlen,
|
|
|
|
sh.slmaxlen * un.un_fact AS slmaxlen_cm,
|
|
|
|
sh.slunit
|
|
|
|
FROM shoe_data sh, unit un
|
|
|
|
WHERE sh.slunit = un.un_name) rsh,
|
|
|
|
(SELECT s.sl_name,
|
|
|
|
s.sl_avail,
|
|
|
|
s.sl_color,
|
|
|
|
s.sl_len,
|
|
|
|
s.sl_unit,
|
|
|
|
s.sl_len * u.un_fact AS sl_len_cm
|
|
|
|
FROM shoelace_data s, unit u
|
|
|
|
WHERE s.sl_unit = u.un_name) rsl
|
|
|
|
WHERE rsl.sl_color = rsh.slcolor
|
2005-01-22 23:56:36 +01:00
|
|
|
AND rsl.sl_len_cm >= rsh.slminlen_cm
|
|
|
|
AND rsl.sl_len_cm <= rsh.slmaxlen_cm) shoe_ready
|
|
|
|
WHERE shoe_ready.total_avail > 2;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
</para>
|
2000-12-12 06:07:59 +01:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<para>
|
|
|
|
It turns out that the planner will collapse this tree into a
|
|
|
|
two-level query tree: the bottommost <command>SELECT</command>
|
|
|
|
commands will be <quote>pulled up</quote> into the middle
|
|
|
|
<command>SELECT</command> since there's no need to process them
|
|
|
|
separately. But the middle <command>SELECT</command> will remain
|
|
|
|
separate from the top, because it contains aggregate functions.
|
|
|
|
If we pulled those up it would change the behavior of the topmost
|
|
|
|
<command>SELECT</command>, which we don't want. However,
|
|
|
|
collapsing the query tree is an optimization that the rewrite
|
|
|
|
system doesn't have to concern itself with.
|
|
|
|
</para>
|
2005-01-23 01:30:59 +01:00
|
|
|
</sect2>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<sect2>
|
|
|
|
<title>View Rules in Non-<command>SELECT</command> Statements</title>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
Two details of the query tree aren't touched in the description of
|
2001-09-10 23:58:47 +02:00
|
|
|
view rules above. These are the command type and the result relation.
|
2010-10-10 19:43:33 +02:00
|
|
|
In fact, the command type is not needed by view rules, but the result
|
|
|
|
relation may affect the way in which the query rewriter works, because
|
|
|
|
special care needs to be taken if the result relation is a view.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
There are only a few differences between a query tree for a
|
|
|
|
<command>SELECT</command> and one for any other
|
|
|
|
command. Obviously, they have a different command type and for a
|
|
|
|
command other than a <command>SELECT</command>, the result
|
|
|
|
relation points to the range-table entry where the result should
|
|
|
|
go. Everything else is absolutely the same. So having two tables
|
|
|
|
<literal>t1</> and <literal>t2</> with columns <literal>a</> and
|
2007-02-01 01:28:19 +01:00
|
|
|
<literal>b</>, the query trees for the two statements:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
SELECT t2.b FROM t1, t2 WHERE t1.a = t2.a;
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2006-12-27 17:07:36 +01:00
|
|
|
UPDATE t1 SET b = t2.b FROM t2 WHERE t1.a = t2.a;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
are nearly identical. In particular:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<itemizedlist>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
2004-12-30 04:13:56 +01:00
|
|
|
The range tables contain entries for the tables <literal>t1</> and <literal>t2</>.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</listitem>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2004-12-30 04:13:56 +01:00
|
|
|
The target lists contain one variable that points to column
|
|
|
|
<literal>b</> of the range table entry for table <literal>t2</>.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</listitem>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2004-12-30 04:13:56 +01:00
|
|
|
The qualification expressions compare the columns <literal>a</> of both
|
|
|
|
range-table entries for equality.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</listitem>
|
2000-12-12 06:07:59 +01:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2004-12-30 04:13:56 +01:00
|
|
|
The join trees show a simple join between <literal>t1</> and <literal>t2</>.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</itemizedlist>
|
2003-04-11 15:22:35 +02:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<para>
|
|
|
|
The consequence is, that both query trees result in similar
|
|
|
|
execution plans: They are both joins over the two tables. For the
|
|
|
|
<command>UPDATE</command> the missing columns from <literal>t1</> are added to
|
|
|
|
the target list by the planner and the final query tree will read
|
2007-02-01 01:28:19 +01:00
|
|
|
as:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2006-12-27 17:07:36 +01:00
|
|
|
UPDATE t1 SET a = t1.a, b = t2.b FROM t2 WHERE t1.a = t2.a;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
|
|
|
and thus the executor run over the join will produce exactly the
|
2010-10-10 19:43:33 +02:00
|
|
|
same result set as:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
SELECT t1.a, t2.b FROM t1, t2 WHERE t1.a = t2.a;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2010-10-10 19:43:33 +02:00
|
|
|
|
|
|
|
But there is a little problem in
|
|
|
|
<command>UPDATE</command>: the part of the executor plan that does
|
|
|
|
the join does not care what the results from the join are
|
|
|
|
meant for. It just produces a result set of rows. The fact that
|
|
|
|
one is a <command>SELECT</command> command and the other is an
|
|
|
|
<command>UPDATE</command> is handled higher up in the executor, where
|
|
|
|
it knows that this is an <command>UPDATE</command>, and it knows that
|
|
|
|
this result should go into table <literal>t1</>. But which of the rows
|
|
|
|
that are there has to be replaced by the new row?
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2001-09-10 23:58:47 +02:00
|
|
|
To resolve this problem, another entry is added to the target list
|
2003-02-19 04:13:25 +01:00
|
|
|
in <command>UPDATE</command> (and also in
|
|
|
|
<command>DELETE</command>) statements: the current tuple ID
|
2006-12-27 17:07:36 +01:00
|
|
|
(<acronym>CTID</>).<indexterm><primary>CTID</></>
|
|
|
|
This is a system column containing the
|
2003-02-19 04:13:25 +01:00
|
|
|
file block number and position in the block for the row. Knowing
|
|
|
|
the table, the <acronym>CTID</> can be used to retrieve the
|
2006-12-27 17:07:36 +01:00
|
|
|
original row of <literal>t1</> to be updated. After adding the
|
2007-02-01 01:28:19 +01:00
|
|
|
<acronym>CTID</> to the target list, the query actually looks like:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
SELECT t1.a, t2.b, t1.ctid FROM t1, t2 WHERE t1.a = t2.a;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2010-10-10 19:43:33 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
Now another detail of <productname>PostgreSQL</productname> enters
|
2003-04-11 15:22:35 +02:00
|
|
|
the stage. Old table rows aren't overwritten, and this
|
|
|
|
is why <command>ROLLBACK</command> is fast. In an <command>UPDATE</command>,
|
|
|
|
the new result row is inserted into the table (after stripping the
|
2003-11-01 02:56:29 +01:00
|
|
|
<acronym>CTID</>) and in the row header of the old row, which the
|
2003-04-11 15:22:35 +02:00
|
|
|
<acronym>CTID</> pointed to, the <literal>cmax</> and
|
2003-02-19 04:13:25 +01:00
|
|
|
<literal>xmax</> entries are set to the current command counter
|
2003-04-11 15:22:35 +02:00
|
|
|
and current transaction ID. Thus the old row is hidden, and after
|
2010-10-10 19:43:33 +02:00
|
|
|
the transaction commits the vacuum cleaner can eventually remove
|
|
|
|
the dead row.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2000-12-12 06:07:59 +01:00
|
|
|
Knowing all that, we can simply apply view rules in absolutely
|
1998-10-25 02:29:01 +02:00
|
|
|
the same way to any command. There is no difference.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</sect2>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<sect2>
|
|
|
|
<title>The Power of Views in <productname>PostgreSQL</productname></title>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-02-19 04:13:25 +01:00
|
|
|
The above demonstrates how the rule system incorporates view
|
2003-04-11 15:22:35 +02:00
|
|
|
definitions into the original query tree. In the second example, a
|
2003-02-19 04:13:25 +01:00
|
|
|
simple <command>SELECT</command> from one view created a final
|
2003-04-11 15:22:35 +02:00
|
|
|
query tree that is a join of 4 tables (<literal>unit</> was used twice with
|
2003-02-19 04:13:25 +01:00
|
|
|
different names).
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
1998-10-25 02:29:01 +02:00
|
|
|
The benefit of implementing views with the rule system is,
|
2000-12-12 06:07:59 +01:00
|
|
|
that the planner has all
|
1998-10-25 02:29:01 +02:00
|
|
|
the information about which tables have to be scanned plus the
|
|
|
|
relationships between these tables plus the restrictive
|
|
|
|
qualifications from the views plus the qualifications from
|
|
|
|
the original query
|
2003-04-11 15:22:35 +02:00
|
|
|
in one single query tree. And this is still the situation
|
1998-10-25 02:29:01 +02:00
|
|
|
when the original query is already a join over views.
|
2003-04-11 15:22:35 +02:00
|
|
|
The planner has to decide which is
|
|
|
|
the best path to execute the query, and the more information
|
2000-12-12 06:07:59 +01:00
|
|
|
the planner has, the better this decision can be. And
|
2005-01-23 01:30:59 +01:00
|
|
|
the rule system as implemented in <productname>PostgreSQL</productname>
|
1998-10-25 02:29:01 +02:00
|
|
|
ensures, that this is all information available about the query
|
2003-04-11 15:22:35 +02:00
|
|
|
up to that point.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</sect2>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<sect2 id="rules-views-update">
|
|
|
|
<title>Updating a View</title>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-02-19 04:13:25 +01:00
|
|
|
What happens if a view is named as the target relation for an
|
|
|
|
<command>INSERT</command>, <command>UPDATE</command>, or
|
2010-10-10 19:43:33 +02:00
|
|
|
<command>DELETE</command>? Simply doing the substitutions
|
|
|
|
described above would give a query tree in which the result
|
|
|
|
relation points at a subquery range-table entry, which will not
|
|
|
|
work. Instead, the rewriter assumes that the operation will be
|
|
|
|
handled by an <literal>INSTEAD OF</> trigger on the view.
|
|
|
|
(If there is no such trigger, the executor will throw an error
|
|
|
|
when execution starts.) Rewriting works slightly differently
|
|
|
|
in this case. For <command>INSERT</command>, the rewriter does
|
|
|
|
nothing at all with the view, leaving it as the result relation
|
|
|
|
for the query. For <command>UPDATE</command> and
|
|
|
|
<command>DELETE</command>, it's still necessary to expand the
|
|
|
|
view query to produce the <quote>old</> rows that the command will
|
|
|
|
attempt to update or delete. So the view is expanded as normal,
|
|
|
|
but another unexpanded range-table entry is added to the query
|
|
|
|
to represent the view in its capacity as the result relation.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2010-10-10 19:43:33 +02:00
|
|
|
The problem that now arises is how to identify the rows to be
|
|
|
|
updated in the view. Recall that when the result relation
|
|
|
|
is a table, a special <acronym>CTID</> entry is added to the target
|
|
|
|
list to identify the physical locations of the rows to be updated.
|
|
|
|
This does not work if the result relation is a view, because a view
|
|
|
|
does not have any <acronym>CTID</>, since its rows do not have
|
|
|
|
actual physical locations. Instead, for an <command>UPDATE</command>
|
|
|
|
or <command>DELETE</command> operation, a special <literal>wholerow</>
|
|
|
|
entry is added to the target list, which expands to include all
|
|
|
|
columns from the view. The executor uses this value to supply the
|
|
|
|
<quote>old</> row to the <literal>INSTEAD OF</> trigger. It is
|
|
|
|
up to the trigger to work out what to update based on the old and
|
|
|
|
new row values.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
If there are no <literal>INSTEAD OF</> triggers to update the view,
|
|
|
|
the executor will throw an error, because it cannot automatically
|
|
|
|
update a view by itself. To change this, we can define rules that
|
|
|
|
modify the behavior of <command>INSERT</command>,
|
|
|
|
<command>UPDATE</command>, and <command>DELETE</command> commands on
|
|
|
|
a view. These rules will rewrite the command, typically into a command
|
|
|
|
that updates one or more tables, rather than views. That is the topic
|
|
|
|
of the next section.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
Note that rules are evaluated first, rewriting the original query
|
|
|
|
before it is planned and executed. Therefore, if a view has
|
|
|
|
<literal>INSTEAD OF</> triggers as well as rules on <command>INSERT</>,
|
|
|
|
<command>UPDATE</>, or <command>DELETE</>, then the rules will be
|
|
|
|
evaluated first, and depending on the result, the triggers may not be
|
|
|
|
used at all.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</sect2>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
</sect1>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<sect1 id="rules-update">
|
|
|
|
<title>Rules on <command>INSERT</>, <command>UPDATE</>, and <command>DELETE</></title>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2003-08-31 19:32:24 +02:00
|
|
|
<indexterm zone="rules-update">
|
|
|
|
<primary>rule</primary>
|
|
|
|
<secondary sortas="INSERT">for INSERT</secondary>
|
|
|
|
</indexterm>
|
|
|
|
|
|
|
|
<indexterm zone="rules-update">
|
|
|
|
<primary>rule</primary>
|
|
|
|
<secondary sortas="UPDATE">for UPDATE</secondary>
|
|
|
|
</indexterm>
|
|
|
|
|
|
|
|
<indexterm zone="rules-update">
|
|
|
|
<primary>rule</primary>
|
|
|
|
<secondary sortas="DELETE">for DELETE</secondary>
|
|
|
|
</indexterm>
|
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
Rules that are defined on <command>INSERT</>, <command>UPDATE</>,
|
|
|
|
and <command>DELETE</> are significantly different from the view rules
|
|
|
|
described in the previous section. First, their <command>CREATE
|
|
|
|
RULE</command> command allows more:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<itemizedlist>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
2004-12-30 04:13:56 +01:00
|
|
|
They are allowed to have no action.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</listitem>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2004-12-30 04:13:56 +01:00
|
|
|
They can have multiple actions.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</listitem>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2006-09-02 19:06:52 +02:00
|
|
|
They can be <literal>INSTEAD</> or <literal>ALSO</> (the default).
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</listitem>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2004-12-30 04:13:56 +01:00
|
|
|
The pseudorelations <literal>NEW</> and <literal>OLD</> become useful.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</listitem>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2004-12-30 04:13:56 +01:00
|
|
|
They can have rule qualifications.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</itemizedlist>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
Second, they don't modify the query tree in place. Instead they
|
|
|
|
create zero or more new query trees and can throw away the
|
1998-10-25 02:29:01 +02:00
|
|
|
original one.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
2003-04-11 15:22:35 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<sect2>
|
|
|
|
<title>How Update Rules Work</title>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2007-02-01 01:28:19 +01:00
|
|
|
Keep the syntax:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2005-05-20 03:37:08 +02:00
|
|
|
CREATE [ OR REPLACE ] RULE <replaceable class="parameter">name</replaceable> AS ON <replaceable class="parameter">event</replaceable>
|
|
|
|
TO <replaceable class="parameter">table</replaceable> [ WHERE <replaceable class="parameter">condition</replaceable> ]
|
|
|
|
DO [ ALSO | INSTEAD ] { NOTHING | <replaceable class="parameter">command</replaceable> | ( <replaceable class="parameter">command</replaceable> ; <replaceable class="parameter">command</replaceable> ... ) }
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
|
|
|
in mind.
|
2001-09-13 17:55:24 +02:00
|
|
|
In the following, <firstterm>update rules</> means rules that are defined
|
2003-04-11 15:22:35 +02:00
|
|
|
on <command>INSERT</>, <command>UPDATE</>, or <command>DELETE</>.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
1998-10-25 02:29:01 +02:00
|
|
|
Update rules get applied by the rule system when the result
|
2003-04-11 15:22:35 +02:00
|
|
|
relation and the command type of a query tree are equal to the
|
2003-02-19 04:13:25 +01:00
|
|
|
object and event given in the <command>CREATE RULE</command> command.
|
2003-04-11 15:22:35 +02:00
|
|
|
For update rules, the rule system creates a list of query trees.
|
|
|
|
Initially the query-tree list is empty.
|
|
|
|
There can be zero (<literal>NOTHING</> key word), one, or multiple actions.
|
|
|
|
To simplify, we will look at a rule with one action. This rule
|
2006-09-02 19:06:52 +02:00
|
|
|
can have a qualification or not and it can be <literal>INSTEAD</> or
|
|
|
|
<literal>ALSO</> (the default).
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
1998-10-25 02:29:01 +02:00
|
|
|
What is a rule qualification? It is a restriction that tells
|
|
|
|
when the actions of the rule should be done and when not. This
|
2003-04-11 15:22:35 +02:00
|
|
|
qualification can only reference the pseudorelations <literal>NEW</> and/or <literal>OLD</>,
|
|
|
|
which basically represent the relation that was given as object (but with a
|
1998-10-25 02:29:01 +02:00
|
|
|
special meaning).
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<para>
|
2006-09-02 19:06:52 +02:00
|
|
|
So we have three cases that produce the following query trees for
|
1998-10-25 02:29:01 +02:00
|
|
|
a one-action rule.
|
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<variablelist>
|
|
|
|
<varlistentry>
|
2006-09-02 19:06:52 +02:00
|
|
|
<term>No qualification, with either <literal>ALSO</> or
|
|
|
|
<literal>INSTEAD</></term>
|
2003-04-11 15:22:35 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
the query tree from the rule action with the original query
|
|
|
|
tree's qualification added
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
2004-03-09 06:05:41 +01:00
|
|
|
<term>Qualification given and <literal>ALSO</></term>
|
2003-04-11 15:22:35 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
the query tree from the rule action with the rule
|
|
|
|
qualification and the original query tree's qualification
|
|
|
|
added
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
|
|
|
<term>Qualification given and <literal>INSTEAD</></term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
the query tree from the rule action with the rule
|
|
|
|
qualification and the original query tree's qualification; and
|
|
|
|
the original query tree with the negated rule qualification
|
|
|
|
added
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
</variablelist>
|
|
|
|
|
2004-03-09 06:05:41 +01:00
|
|
|
Finally, if the rule is <literal>ALSO</>, the unchanged original query tree is
|
2003-04-11 15:22:35 +02:00
|
|
|
added to the list. Since only qualified <literal>INSTEAD</> rules already add the
|
|
|
|
original query tree, we end up with either one or two output query trees
|
1998-10-25 02:29:01 +02:00
|
|
|
for a rule with one action.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
For <literal>ON INSERT</> rules, the original query (if not suppressed by <literal>INSTEAD</>)
|
2001-07-10 01:50:32 +02:00
|
|
|
is done before any actions added by rules. This allows the actions to
|
2003-04-11 15:22:35 +02:00
|
|
|
see the inserted row(s). But for <literal>ON UPDATE</> and <literal>ON
|
|
|
|
DELETE</> rules, the original query is done after the actions added by rules.
|
2001-07-10 01:50:32 +02:00
|
|
|
This ensures that the actions can see the to-be-updated or to-be-deleted
|
|
|
|
rows; otherwise, the actions might do nothing because they find no rows
|
|
|
|
matching their qualifications.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
2001-07-10 01:50:32 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
The query trees generated from rule actions are thrown into the
|
|
|
|
rewrite system again, and maybe more rules get applied resulting
|
|
|
|
in more or less query trees.
|
2005-01-30 00:45:36 +01:00
|
|
|
So a rule's actions must have either a different
|
|
|
|
command type or a different result relation than the rule itself is
|
|
|
|
on, otherwise this recursive process will end up in an infinite loop.
|
|
|
|
(Recursive expansion of a rule will be detected and reported as an
|
|
|
|
error.)
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
The query trees found in the actions of the
|
2003-02-19 04:13:25 +01:00
|
|
|
<structname>pg_rewrite</structname> system catalog are only
|
|
|
|
templates. Since they can reference the range-table entries for
|
2003-04-11 15:22:35 +02:00
|
|
|
<literal>NEW</> and <literal>OLD</>, some substitutions have to be made before they can be
|
|
|
|
used. For any reference to <literal>NEW</>, the target list of the original
|
2003-02-19 04:13:25 +01:00
|
|
|
query is searched for a corresponding entry. If found, that
|
2003-04-11 15:22:35 +02:00
|
|
|
entry's expression replaces the reference. Otherwise, <literal>NEW</> means the
|
|
|
|
same as <literal>OLD</> (for an <command>UPDATE</command>) or is replaced by
|
|
|
|
a null value (for an <command>INSERT</command>). Any reference to <literal>OLD</> is
|
|
|
|
replaced by a reference to the range-table entry that is the
|
2003-02-19 04:13:25 +01:00
|
|
|
result relation.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
After the system is done applying update rules, it applies view rules to the
|
|
|
|
produced query tree(s). Views cannot insert new update actions so
|
2000-12-12 06:07:59 +01:00
|
|
|
there is no need to apply update rules to the output of view rewriting.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
2000-12-12 06:07:59 +01:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<sect3>
|
|
|
|
<title>A First Rule Step by Step</title>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
Say we want to trace changes to the <literal>sl_avail</> column in the
|
|
|
|
<literal>shoelace_data</literal> relation. So we set up a log table
|
2003-02-19 04:13:25 +01:00
|
|
|
and a rule that conditionally writes a log entry when an
|
|
|
|
<command>UPDATE</command> is performed on
|
|
|
|
<literal>shoelace_data</literal>.
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE TABLE shoelace_log (
|
2003-04-11 15:22:35 +02:00
|
|
|
sl_name text, -- shoelace changed
|
2002-01-20 23:19:57 +01:00
|
|
|
sl_avail integer, -- new available value
|
|
|
|
log_who text, -- who did it
|
|
|
|
log_when timestamp -- when
|
|
|
|
);
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE RULE log_shoelace AS ON UPDATE TO shoelace_data
|
2003-04-11 15:22:35 +02:00
|
|
|
WHERE NEW.sl_avail <> OLD.sl_avail
|
2002-01-20 23:19:57 +01:00
|
|
|
DO INSERT INTO shoelace_log VALUES (
|
|
|
|
NEW.sl_name,
|
|
|
|
NEW.sl_avail,
|
|
|
|
current_user,
|
|
|
|
current_timestamp
|
|
|
|
);
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
Now someone does:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
UPDATE shoelace_data SET sl_avail = 6 WHERE sl_name = 'sl7';
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
and we look at the log table:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
SELECT * FROM shoelace_log;
|
|
|
|
|
|
|
|
sl_name | sl_avail | log_who | log_when
|
|
|
|
---------+----------+---------+----------------------------------
|
|
|
|
sl7 | 6 | Al | Tue Oct 20 16:14:45 1998 MET DST
|
2002-01-20 23:19:57 +01:00
|
|
|
(1 row)
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<para>
|
1998-10-25 02:29:01 +02:00
|
|
|
That's what we expected. What happened in the background is the following.
|
2007-02-01 01:28:19 +01:00
|
|
|
The parser created the query tree:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
UPDATE shoelace_data SET sl_avail = 6
|
2002-01-20 23:19:57 +01:00
|
|
|
FROM shoelace_data shoelace_data
|
2003-04-11 15:22:35 +02:00
|
|
|
WHERE shoelace_data.sl_name = 'sl7';
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
There is a rule <literal>log_shoelace</literal> that is <literal>ON UPDATE</> with the rule
|
2007-02-01 01:28:19 +01:00
|
|
|
qualification expression:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
NEW.sl_avail <> OLD.sl_avail
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2007-02-01 01:28:19 +01:00
|
|
|
and the action:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
INSERT INTO shoelace_log VALUES (
|
2009-11-06 00:24:27 +01:00
|
|
|
new.sl_name, new.sl_avail,
|
2003-04-11 15:22:35 +02:00
|
|
|
current_user, current_timestamp )
|
2009-11-06 00:24:27 +01:00
|
|
|
FROM shoelace_data new, shoelace_data old;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2000-12-12 06:07:59 +01:00
|
|
|
|
Wording cleanup for error messages. Also change can't -> cannot.
Standard English uses "may", "can", and "might" in different ways:
may - permission, "You may borrow my rake."
can - ability, "I can lift that log."
might - possibility, "It might rain today."
Unfortunately, in conversational English, their use is often mixed, as
in, "You may use this variable to do X", when in fact, "can" is a better
choice. Similarly, "It may crash" is better stated, "It might crash".
2007-02-01 20:10:30 +01:00
|
|
|
(This looks a little strange since you cannot normally write
|
2003-04-11 15:22:35 +02:00
|
|
|
<literal>INSERT ... VALUES ... FROM</>. The <literal>FROM</>
|
|
|
|
clause here is just to indicate that there are range-table entries
|
2009-11-06 00:24:27 +01:00
|
|
|
in the query tree for <literal>new</> and <literal>old</>.
|
2003-04-11 15:22:35 +02:00
|
|
|
These are needed so that they can be referenced by variables in
|
|
|
|
the <command>INSERT</command> command's query tree.)
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2004-03-09 06:05:41 +01:00
|
|
|
The rule is a qualified <literal>ALSO</> rule, so the rule system
|
2003-04-11 15:22:35 +02:00
|
|
|
has to return two query trees: the modified rule action and the original
|
|
|
|
query tree. In step 1, the range table of the original query is
|
|
|
|
incorporated into the rule's action query tree. This results in:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
INSERT INTO shoelace_log VALUES (
|
2009-11-06 00:24:27 +01:00
|
|
|
new.sl_name, new.sl_avail,
|
2003-04-11 15:22:35 +02:00
|
|
|
current_user, current_timestamp )
|
2009-11-06 00:24:27 +01:00
|
|
|
FROM shoelace_data new, shoelace_data old,
|
2002-01-20 23:19:57 +01:00
|
|
|
<emphasis>shoelace_data shoelace_data</emphasis>;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
In step 2, the rule qualification is added to it, so the result set
|
|
|
|
is restricted to rows where <literal>sl_avail</> changes:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
INSERT INTO shoelace_log VALUES (
|
2009-11-06 00:24:27 +01:00
|
|
|
new.sl_name, new.sl_avail,
|
2003-04-11 15:22:35 +02:00
|
|
|
current_user, current_timestamp )
|
2009-11-06 00:24:27 +01:00
|
|
|
FROM shoelace_data new, shoelace_data old,
|
2002-01-20 23:19:57 +01:00
|
|
|
shoelace_data shoelace_data
|
2009-11-06 00:24:27 +01:00
|
|
|
<emphasis>WHERE new.sl_avail <> old.sl_avail</emphasis>;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
(This looks even stranger, since <literal>INSERT ... VALUES</> doesn't have
|
|
|
|
a <literal>WHERE</> clause either, but the planner and executor will have no
|
2000-12-12 06:07:59 +01:00
|
|
|
difficulty with it. They need to support this same functionality
|
2003-04-11 15:22:35 +02:00
|
|
|
anyway for <literal>INSERT ... SELECT</>.)
|
2002-10-15 00:14:35 +02:00
|
|
|
</para>
|
2000-12-12 06:07:59 +01:00
|
|
|
|
2002-10-15 00:14:35 +02:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
In step 3, the original query tree's qualification is added,
|
|
|
|
restricting the result set further to only the rows that would have been touched
|
|
|
|
by the original query:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
INSERT INTO shoelace_log VALUES (
|
2009-11-06 00:24:27 +01:00
|
|
|
new.sl_name, new.sl_avail,
|
2003-04-11 15:22:35 +02:00
|
|
|
current_user, current_timestamp )
|
2009-11-06 00:24:27 +01:00
|
|
|
FROM shoelace_data new, shoelace_data old,
|
2002-01-20 23:19:57 +01:00
|
|
|
shoelace_data shoelace_data
|
2009-11-06 00:24:27 +01:00
|
|
|
WHERE new.sl_avail <> old.sl_avail
|
2003-04-11 15:22:35 +02:00
|
|
|
<emphasis>AND shoelace_data.sl_name = 'sl7'</emphasis>;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<para>
|
|
|
|
Step 4 replaces references to <literal>NEW</> by the target list entries from the
|
|
|
|
original query tree or by the matching variable references
|
|
|
|
from the result relation:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
INSERT INTO shoelace_log VALUES (
|
2002-01-20 23:19:57 +01:00
|
|
|
<emphasis>shoelace_data.sl_name</emphasis>, <emphasis>6</emphasis>,
|
2003-04-11 15:22:35 +02:00
|
|
|
current_user, current_timestamp )
|
2009-11-06 00:24:27 +01:00
|
|
|
FROM shoelace_data new, shoelace_data old,
|
2002-01-20 23:19:57 +01:00
|
|
|
shoelace_data shoelace_data
|
2009-11-06 00:24:27 +01:00
|
|
|
WHERE <emphasis>6</emphasis> <> old.sl_avail
|
2003-04-11 15:22:35 +02:00
|
|
|
AND shoelace_data.sl_name = 'sl7';
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
Step 5 changes <literal>OLD</> references into result relation references:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
INSERT INTO shoelace_log VALUES (
|
2002-01-20 23:19:57 +01:00
|
|
|
shoelace_data.sl_name, 6,
|
2003-04-11 15:22:35 +02:00
|
|
|
current_user, current_timestamp )
|
2009-11-06 00:24:27 +01:00
|
|
|
FROM shoelace_data new, shoelace_data old,
|
2002-01-20 23:19:57 +01:00
|
|
|
shoelace_data shoelace_data
|
2003-04-11 15:22:35 +02:00
|
|
|
WHERE 6 <> <emphasis>shoelace_data.sl_avail</emphasis>
|
|
|
|
AND shoelace_data.sl_name = 'sl7';
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<para>
|
2004-03-09 06:05:41 +01:00
|
|
|
That's it. Since the rule is <literal>ALSO</>, we also output the
|
2003-04-11 15:22:35 +02:00
|
|
|
original query tree. In short, the output from the rule system
|
|
|
|
is a list of two query trees that correspond to these statements:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
INSERT INTO shoelace_log VALUES (
|
2002-01-20 23:19:57 +01:00
|
|
|
shoelace_data.sl_name, 6,
|
2003-04-11 15:22:35 +02:00
|
|
|
current_user, current_timestamp )
|
2002-01-20 23:19:57 +01:00
|
|
|
FROM shoelace_data
|
2003-04-11 15:22:35 +02:00
|
|
|
WHERE 6 <> shoelace_data.sl_avail
|
2002-01-20 23:19:57 +01:00
|
|
|
AND shoelace_data.sl_name = 'sl7';
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2002-01-20 23:19:57 +01:00
|
|
|
UPDATE shoelace_data SET sl_avail = 6
|
|
|
|
WHERE sl_name = 'sl7';
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
These are executed in this order, and that is exactly what
|
|
|
|
the rule was meant to do.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
The substitutions and the added qualifications
|
2007-02-01 01:28:19 +01:00
|
|
|
ensure that, if the original query would be, say:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
UPDATE shoelace_data SET sl_color = 'green'
|
|
|
|
WHERE sl_name = 'sl7';
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
no log entry would get written. In that case, the original query
|
|
|
|
tree does not contain a target list entry for
|
|
|
|
<literal>sl_avail</>, so <literal>NEW.sl_avail</> will get
|
|
|
|
replaced by <literal>shoelace_data.sl_avail</>. Thus, the extra
|
2007-02-01 01:28:19 +01:00
|
|
|
command generated by the rule is:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
INSERT INTO shoelace_log VALUES (
|
2002-01-20 23:19:57 +01:00
|
|
|
shoelace_data.sl_name, <emphasis>shoelace_data.sl_avail</emphasis>,
|
2003-04-11 15:22:35 +02:00
|
|
|
current_user, current_timestamp )
|
2002-01-20 23:19:57 +01:00
|
|
|
FROM shoelace_data
|
2003-04-11 15:22:35 +02:00
|
|
|
WHERE <emphasis>shoelace_data.sl_avail</emphasis> <> shoelace_data.sl_avail
|
2002-01-20 23:19:57 +01:00
|
|
|
AND shoelace_data.sl_name = 'sl7';
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
and that qualification will never be true.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
It will also work if the original query modifies multiple rows. So
|
2007-02-01 01:28:19 +01:00
|
|
|
if someone issued the command:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
UPDATE shoelace_data SET sl_avail = 0
|
|
|
|
WHERE sl_color = 'black';
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
four rows in fact get updated (<literal>sl1</>, <literal>sl2</>, <literal>sl3</>, and <literal>sl4</>).
|
|
|
|
But <literal>sl3</> already has <literal>sl_avail = 0</>. In this case, the original
|
|
|
|
query trees qualification is different and that results
|
2007-02-01 01:28:19 +01:00
|
|
|
in the extra query tree:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
INSERT INTO shoelace_log
|
|
|
|
SELECT shoelace_data.sl_name, 0,
|
2002-01-20 23:19:57 +01:00
|
|
|
current_user, current_timestamp
|
|
|
|
FROM shoelace_data
|
2003-04-11 15:22:35 +02:00
|
|
|
WHERE 0 <> shoelace_data.sl_avail
|
2002-01-20 23:19:57 +01:00
|
|
|
AND <emphasis>shoelace_data.sl_color = 'black'</emphasis>;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
being generated by the rule. This query tree will surely insert
|
|
|
|
three new log entries. And that's absolutely correct.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
Here we can see why it is important that the original query tree
|
|
|
|
is executed last. If the <command>UPDATE</command> had been
|
|
|
|
executed first, all the rows would have already been set to zero, so the
|
|
|
|
logging <command>INSERT</command> would not find any row where
|
|
|
|
<literal>0 <> shoelace_data.sl_avail</literal>.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</sect3>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
</sect2>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<sect2 id="rules-update-views">
|
|
|
|
<title>Cooperation with Views</title>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-08-31 19:32:24 +02:00
|
|
|
<indexterm zone="rules-update-views"><primary>view</><secondary>updating</></>
|
2003-04-11 15:22:35 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
1998-10-25 02:29:01 +02:00
|
|
|
A simple way to protect view relations from the mentioned
|
2003-04-11 15:22:35 +02:00
|
|
|
possibility that someone can try to run <command>INSERT</command>,
|
|
|
|
<command>UPDATE</command>, or <command>DELETE</command> on them is
|
2007-02-01 01:28:19 +01:00
|
|
|
to let those query trees get thrown away. So we could create the rules:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE RULE shoe_ins_protect AS ON INSERT TO shoe
|
|
|
|
DO INSTEAD NOTHING;
|
|
|
|
CREATE RULE shoe_upd_protect AS ON UPDATE TO shoe
|
|
|
|
DO INSTEAD NOTHING;
|
|
|
|
CREATE RULE shoe_del_protect AS ON DELETE TO shoe
|
|
|
|
DO INSTEAD NOTHING;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
If someone now tries to do any of these operations on the view
|
2003-02-19 04:13:25 +01:00
|
|
|
relation <literal>shoe</literal>, the rule system will
|
2003-04-11 15:22:35 +02:00
|
|
|
apply these rules. Since the rules have
|
|
|
|
no actions and are <literal>INSTEAD</>, the resulting list of
|
|
|
|
query trees will be empty and the whole query will become
|
1998-10-25 02:29:01 +02:00
|
|
|
nothing because there is nothing left to be optimized or
|
|
|
|
executed after the rule system is done with it.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
1998-10-25 02:29:01 +02:00
|
|
|
A more sophisticated way to use the rule system is to
|
2003-04-11 15:22:35 +02:00
|
|
|
create rules that rewrite the query tree into one that
|
1998-10-25 02:29:01 +02:00
|
|
|
does the right operation on the real tables. To do that
|
2003-02-19 04:13:25 +01:00
|
|
|
on the <literal>shoelace</literal> view, we create
|
1998-10-25 02:29:01 +02:00
|
|
|
the following rules:
|
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE RULE shoelace_ins AS ON INSERT TO shoelace
|
|
|
|
DO INSTEAD
|
|
|
|
INSERT INTO shoelace_data VALUES (
|
|
|
|
NEW.sl_name,
|
|
|
|
NEW.sl_avail,
|
|
|
|
NEW.sl_color,
|
|
|
|
NEW.sl_len,
|
2003-04-11 15:22:35 +02:00
|
|
|
NEW.sl_unit
|
|
|
|
);
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE RULE shoelace_upd AS ON UPDATE TO shoelace
|
|
|
|
DO INSTEAD
|
2003-04-11 15:22:35 +02:00
|
|
|
UPDATE shoelace_data
|
|
|
|
SET sl_name = NEW.sl_name,
|
2002-01-20 23:19:57 +01:00
|
|
|
sl_avail = NEW.sl_avail,
|
|
|
|
sl_color = NEW.sl_color,
|
|
|
|
sl_len = NEW.sl_len,
|
|
|
|
sl_unit = NEW.sl_unit
|
|
|
|
WHERE sl_name = OLD.sl_name;
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE RULE shoelace_del AS ON DELETE TO shoelace
|
|
|
|
DO INSTEAD
|
|
|
|
DELETE FROM shoelace_data
|
|
|
|
WHERE sl_name = OLD.sl_name;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2006-09-02 19:06:52 +02:00
|
|
|
<para>
|
|
|
|
If you want to support <literal>RETURNING</> queries on the view,
|
|
|
|
you need to make the rules include <literal>RETURNING</> clauses that
|
|
|
|
compute the view rows. This is usually pretty trivial for views on a
|
|
|
|
single table, but it's a bit tedious for join views such as
|
2007-02-01 01:28:19 +01:00
|
|
|
<literal>shoelace</literal>. An example for the insert case is:
|
2006-09-02 19:06:52 +02:00
|
|
|
|
|
|
|
<programlisting>
|
|
|
|
CREATE RULE shoelace_ins AS ON INSERT TO shoelace
|
|
|
|
DO INSTEAD
|
|
|
|
INSERT INTO shoelace_data VALUES (
|
|
|
|
NEW.sl_name,
|
|
|
|
NEW.sl_avail,
|
|
|
|
NEW.sl_color,
|
|
|
|
NEW.sl_len,
|
|
|
|
NEW.sl_unit
|
|
|
|
)
|
|
|
|
RETURNING
|
|
|
|
shoelace_data.*,
|
|
|
|
(SELECT shoelace_data.sl_len * u.un_fact
|
|
|
|
FROM unit u WHERE shoelace_data.sl_unit = u.un_name);
|
|
|
|
</programlisting>
|
|
|
|
|
|
|
|
Note that this one rule supports both <command>INSERT</> and
|
|
|
|
<command>INSERT RETURNING</> queries on the view — the
|
|
|
|
<literal>RETURNING</> clause is simply ignored for <command>INSERT</>.
|
|
|
|
</para>
|
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<para>
|
|
|
|
Now assume that once in a while, a pack of shoelaces arrives at
|
|
|
|
the shop and a big parts list along with it. But you don't want
|
|
|
|
to manually update the <literal>shoelace</literal> view every
|
|
|
|
time. Instead we setup two little tables: one where you can
|
|
|
|
insert the items from the part list, and one with a special
|
|
|
|
trick. The creation commands for these are:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE TABLE shoelace_arrive (
|
2003-04-11 15:22:35 +02:00
|
|
|
arr_name text,
|
2002-01-20 23:19:57 +01:00
|
|
|
arr_quant integer
|
|
|
|
);
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE TABLE shoelace_ok (
|
2003-04-11 15:22:35 +02:00
|
|
|
ok_name text,
|
2002-01-20 23:19:57 +01:00
|
|
|
ok_quant integer
|
|
|
|
);
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE RULE shoelace_ok_ins AS ON INSERT TO shoelace_ok
|
|
|
|
DO INSTEAD
|
2003-04-11 15:22:35 +02:00
|
|
|
UPDATE shoelace
|
|
|
|
SET sl_avail = sl_avail + NEW.ok_quant
|
2002-01-20 23:19:57 +01:00
|
|
|
WHERE sl_name = NEW.ok_name;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
Now you can fill the table <literal>shoelace_arrive</literal> with
|
|
|
|
the data from the parts list:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
SELECT * FROM shoelace_arrive;
|
|
|
|
|
|
|
|
arr_name | arr_quant
|
|
|
|
----------+-----------
|
|
|
|
sl3 | 10
|
|
|
|
sl6 | 20
|
|
|
|
sl8 | 20
|
2002-01-20 23:19:57 +01:00
|
|
|
(3 rows)
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
Take a quick look at the current data:
|
2010-10-10 19:43:33 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
SELECT * FROM shoelace;
|
|
|
|
|
|
|
|
sl_name | sl_avail | sl_color | sl_len | sl_unit | sl_len_cm
|
|
|
|
----------+----------+----------+--------+---------+-----------
|
|
|
|
sl1 | 5 | black | 80 | cm | 80
|
|
|
|
sl2 | 6 | black | 100 | cm | 100
|
|
|
|
sl7 | 6 | brown | 60 | cm | 60
|
|
|
|
sl3 | 0 | black | 35 | inch | 88.9
|
|
|
|
sl4 | 8 | black | 40 | inch | 101.6
|
|
|
|
sl8 | 1 | brown | 40 | inch | 101.6
|
|
|
|
sl5 | 4 | brown | 1 | m | 100
|
|
|
|
sl6 | 0 | brown | 0.9 | m | 90
|
2002-01-20 23:19:57 +01:00
|
|
|
(8 rows)
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
Now move the arrived shoelaces in:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
INSERT INTO shoelace_ok SELECT * FROM shoelace_arrive;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
and check the results:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
SELECT * FROM shoelace ORDER BY sl_name;
|
|
|
|
|
|
|
|
sl_name | sl_avail | sl_color | sl_len | sl_unit | sl_len_cm
|
|
|
|
----------+----------+----------+--------+---------+-----------
|
|
|
|
sl1 | 5 | black | 80 | cm | 80
|
|
|
|
sl2 | 6 | black | 100 | cm | 100
|
|
|
|
sl7 | 6 | brown | 60 | cm | 60
|
|
|
|
sl4 | 8 | black | 40 | inch | 101.6
|
|
|
|
sl3 | 10 | black | 35 | inch | 88.9
|
|
|
|
sl8 | 21 | brown | 40 | inch | 101.6
|
|
|
|
sl5 | 4 | brown | 1 | m | 100
|
|
|
|
sl6 | 20 | brown | 0.9 | m | 90
|
2002-01-20 23:19:57 +01:00
|
|
|
(8 rows)
|
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
SELECT * FROM shoelace_log;
|
|
|
|
|
|
|
|
sl_name | sl_avail | log_who| log_when
|
|
|
|
---------+----------+--------+----------------------------------
|
|
|
|
sl7 | 6 | Al | Tue Oct 20 19:14:45 1998 MET DST
|
|
|
|
sl3 | 10 | Al | Tue Oct 20 19:25:16 1998 MET DST
|
|
|
|
sl6 | 20 | Al | Tue Oct 20 19:25:16 1998 MET DST
|
|
|
|
sl8 | 21 | Al | Tue Oct 20 19:25:16 1998 MET DST
|
2002-01-20 23:19:57 +01:00
|
|
|
(4 rows)
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<para>
|
|
|
|
It's a long way from the one <literal>INSERT ... SELECT</literal>
|
|
|
|
to these results. And the description of the query-tree
|
|
|
|
transformation will be the last in this chapter. First, there is
|
2007-02-01 01:28:19 +01:00
|
|
|
the parser's output:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
INSERT INTO shoelace_ok
|
|
|
|
SELECT shoelace_arrive.arr_name, shoelace_arrive.arr_quant
|
2002-01-20 23:19:57 +01:00
|
|
|
FROM shoelace_arrive shoelace_arrive, shoelace_ok shoelace_ok;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
Now the first rule <literal>shoelace_ok_ins</literal> is applied and turns this
|
2007-02-01 01:28:19 +01:00
|
|
|
into:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
UPDATE shoelace
|
|
|
|
SET sl_avail = shoelace.sl_avail + shoelace_arrive.arr_quant
|
2002-01-20 23:19:57 +01:00
|
|
|
FROM shoelace_arrive shoelace_arrive, shoelace_ok shoelace_ok,
|
2009-11-06 00:24:27 +01:00
|
|
|
shoelace_ok old, shoelace_ok new,
|
2002-01-20 23:19:57 +01:00
|
|
|
shoelace shoelace
|
2003-04-11 15:22:35 +02:00
|
|
|
WHERE shoelace.sl_name = shoelace_arrive.arr_name;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-02-19 04:13:25 +01:00
|
|
|
and throws away the original <command>INSERT</command> on
|
|
|
|
<literal>shoelace_ok</literal>. This rewritten query is passed to
|
2003-04-11 15:22:35 +02:00
|
|
|
the rule system again, and the second applied rule
|
2007-02-01 01:28:19 +01:00
|
|
|
<literal>shoelace_upd</literal> produces:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
UPDATE shoelace_data
|
|
|
|
SET sl_name = shoelace.sl_name,
|
|
|
|
sl_avail = shoelace.sl_avail + shoelace_arrive.arr_quant,
|
2002-01-20 23:19:57 +01:00
|
|
|
sl_color = shoelace.sl_color,
|
|
|
|
sl_len = shoelace.sl_len,
|
|
|
|
sl_unit = shoelace.sl_unit
|
|
|
|
FROM shoelace_arrive shoelace_arrive, shoelace_ok shoelace_ok,
|
2009-11-06 00:24:27 +01:00
|
|
|
shoelace_ok old, shoelace_ok new,
|
|
|
|
shoelace shoelace, shoelace old,
|
|
|
|
shoelace new, shoelace_data shoelace_data
|
2003-04-11 15:22:35 +02:00
|
|
|
WHERE shoelace.sl_name = shoelace_arrive.arr_name
|
|
|
|
AND shoelace_data.sl_name = shoelace.sl_name;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
Again it's an <literal>INSTEAD</> rule and the previous query tree is trashed.
|
2003-02-19 04:13:25 +01:00
|
|
|
Note that this query still uses the view <literal>shoelace</literal>.
|
2003-04-11 15:22:35 +02:00
|
|
|
But the rule system isn't finished with this step, so it continues
|
2007-02-01 01:28:19 +01:00
|
|
|
and applies the <literal>_RETURN</literal> rule on it, and we get:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
UPDATE shoelace_data
|
|
|
|
SET sl_name = s.sl_name,
|
|
|
|
sl_avail = s.sl_avail + shoelace_arrive.arr_quant,
|
2002-01-20 23:19:57 +01:00
|
|
|
sl_color = s.sl_color,
|
|
|
|
sl_len = s.sl_len,
|
|
|
|
sl_unit = s.sl_unit
|
|
|
|
FROM shoelace_arrive shoelace_arrive, shoelace_ok shoelace_ok,
|
2009-11-06 00:24:27 +01:00
|
|
|
shoelace_ok old, shoelace_ok new,
|
|
|
|
shoelace shoelace, shoelace old,
|
|
|
|
shoelace new, shoelace_data shoelace_data,
|
|
|
|
shoelace old, shoelace new,
|
2002-01-20 23:19:57 +01:00
|
|
|
shoelace_data s, unit u
|
2003-04-11 15:22:35 +02:00
|
|
|
WHERE s.sl_name = shoelace_arrive.arr_name
|
|
|
|
AND shoelace_data.sl_name = s.sl_name;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
Finally, the rule <literal>log_shoelace</literal> gets applied,
|
2007-02-01 01:28:19 +01:00
|
|
|
producing the extra query tree:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
INSERT INTO shoelace_log
|
|
|
|
SELECT s.sl_name,
|
|
|
|
s.sl_avail + shoelace_arrive.arr_quant,
|
2002-01-20 23:19:57 +01:00
|
|
|
current_user,
|
|
|
|
current_timestamp
|
|
|
|
FROM shoelace_arrive shoelace_arrive, shoelace_ok shoelace_ok,
|
2009-11-06 00:24:27 +01:00
|
|
|
shoelace_ok old, shoelace_ok new,
|
|
|
|
shoelace shoelace, shoelace old,
|
|
|
|
shoelace new, shoelace_data shoelace_data,
|
|
|
|
shoelace old, shoelace new,
|
2002-01-20 23:19:57 +01:00
|
|
|
shoelace_data s, unit u,
|
2009-11-06 00:24:27 +01:00
|
|
|
shoelace_data old, shoelace_data new
|
2002-01-20 23:19:57 +01:00
|
|
|
shoelace_log shoelace_log
|
2003-04-11 15:22:35 +02:00
|
|
|
WHERE s.sl_name = shoelace_arrive.arr_name
|
|
|
|
AND shoelace_data.sl_name = s.sl_name
|
|
|
|
AND (s.sl_avail + shoelace_arrive.arr_quant) <> s.sl_avail;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2010-10-10 19:43:33 +02:00
|
|
|
|
|
|
|
After that the rule system runs out of rules and returns the
|
2003-04-11 15:22:35 +02:00
|
|
|
generated query trees.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
So we end up with two final query trees that are equivalent to the
|
2007-02-01 01:28:19 +01:00
|
|
|
<acronym>SQL</acronym> statements:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
INSERT INTO shoelace_log
|
|
|
|
SELECT s.sl_name,
|
2002-01-20 23:19:57 +01:00
|
|
|
s.sl_avail + shoelace_arrive.arr_quant,
|
|
|
|
current_user,
|
|
|
|
current_timestamp
|
|
|
|
FROM shoelace_arrive shoelace_arrive, shoelace_data shoelace_data,
|
|
|
|
shoelace_data s
|
|
|
|
WHERE s.sl_name = shoelace_arrive.arr_name
|
|
|
|
AND shoelace_data.sl_name = s.sl_name
|
2003-04-11 15:22:35 +02:00
|
|
|
AND s.sl_avail + shoelace_arrive.arr_quant <> s.sl_avail;
|
2010-10-10 19:43:33 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
UPDATE shoelace_data
|
|
|
|
SET sl_avail = shoelace_data.sl_avail + shoelace_arrive.arr_quant
|
2002-01-20 23:19:57 +01:00
|
|
|
FROM shoelace_arrive shoelace_arrive,
|
|
|
|
shoelace_data shoelace_data,
|
|
|
|
shoelace_data s
|
|
|
|
WHERE s.sl_name = shoelace_arrive.sl_name
|
|
|
|
AND shoelace_data.sl_name = s.sl_name;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
|
|
|
The result is that data coming from one relation inserted into another,
|
|
|
|
changed into updates on a third, changed into updating
|
|
|
|
a fourth plus logging that final update in a fifth
|
|
|
|
gets reduced into two queries.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-02-19 04:13:25 +01:00
|
|
|
There is a little detail that's a bit ugly. Looking at the two
|
2003-04-11 15:22:35 +02:00
|
|
|
queries, it turns out that the <literal>shoelace_data</literal>
|
2003-02-19 04:13:25 +01:00
|
|
|
relation appears twice in the range table where it could
|
|
|
|
definitely be reduced to one. The planner does not handle it and
|
|
|
|
so the execution plan for the rule systems output of the
|
|
|
|
<command>INSERT</command> will be
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<literallayout class="monospaced">
|
1998-10-25 02:29:01 +02:00
|
|
|
Nested Loop
|
2005-01-22 23:06:27 +01:00
|
|
|
-> Merge Join
|
|
|
|
-> Seq Scan
|
|
|
|
-> Sort
|
|
|
|
-> Seq Scan on s
|
|
|
|
-> Seq Scan
|
|
|
|
-> Sort
|
|
|
|
-> Seq Scan on shoelace_arrive
|
|
|
|
-> Seq Scan on shoelace_data
|
2003-04-11 15:22:35 +02:00
|
|
|
</literallayout>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2001-09-10 23:58:47 +02:00
|
|
|
while omitting the extra range table entry would result in a
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<literallayout class="monospaced">
|
1998-10-25 02:29:01 +02:00
|
|
|
Merge Join
|
2005-01-22 23:06:27 +01:00
|
|
|
-> Seq Scan
|
|
|
|
-> Sort
|
|
|
|
-> Seq Scan on s
|
|
|
|
-> Seq Scan
|
|
|
|
-> Sort
|
|
|
|
-> Seq Scan on shoelace_arrive
|
2003-04-11 15:22:35 +02:00
|
|
|
</literallayout>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
which produces exactly the same entries in the log table. Thus,
|
|
|
|
the rule system caused one extra scan on the table
|
|
|
|
<literal>shoelace_data</literal> that is absolutely not
|
|
|
|
necessary. And the same redundant scan is done once more in the
|
2003-02-19 04:13:25 +01:00
|
|
|
<command>UPDATE</command>. But it was a really hard job to make
|
|
|
|
that all possible at all.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
Now we make a final demonstration of the
|
2005-01-23 01:30:59 +01:00
|
|
|
<productname>PostgreSQL</productname> rule system and its power.
|
2003-04-11 15:22:35 +02:00
|
|
|
Say you add some shoelaces with extraordinary colors to your
|
|
|
|
database:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
INSERT INTO shoelace VALUES ('sl9', 0, 'pink', 35.0, 'inch', 0.0);
|
|
|
|
INSERT INTO shoelace VALUES ('sl10', 1000, 'magenta', 40.0, 'inch', 0.0);
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
We would like to make a view to check which
|
|
|
|
<literal>shoelace</literal> entries do not fit any shoe in color.
|
2007-02-01 01:28:19 +01:00
|
|
|
The view for this is:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
CREATE VIEW shoelace_mismatch AS
|
2002-01-20 23:19:57 +01:00
|
|
|
SELECT * FROM shoelace WHERE NOT EXISTS
|
|
|
|
(SELECT shoename FROM shoe WHERE slcolor = sl_color);
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2007-02-01 01:28:19 +01:00
|
|
|
Its output is:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
SELECT * FROM shoelace_mismatch;
|
|
|
|
|
|
|
|
sl_name | sl_avail | sl_color | sl_len | sl_unit | sl_len_cm
|
|
|
|
---------+----------+----------+--------+---------+-----------
|
|
|
|
sl9 | 0 | pink | 35 | inch | 88.9
|
|
|
|
sl10 | 1000 | magenta | 40 | inch | 101.6
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<para>
|
|
|
|
Now we want to set it up so that mismatching shoelaces that are
|
|
|
|
not in stock are deleted from the database.
|
2005-01-23 01:30:59 +01:00
|
|
|
To make it a little harder for <productname>PostgreSQL</productname>,
|
2007-02-01 01:28:19 +01:00
|
|
|
we don't delete it directly. Instead we create one more view:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
CREATE VIEW shoelace_can_delete AS
|
|
|
|
SELECT * FROM shoelace_mismatch WHERE sl_avail = 0;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
|
|
|
and do it this way:
|
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
DELETE FROM shoelace WHERE EXISTS
|
2003-04-11 15:22:35 +02:00
|
|
|
(SELECT * FROM shoelace_can_delete
|
2002-01-20 23:19:57 +01:00
|
|
|
WHERE sl_name = shoelace.sl_name);
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2010-05-13 21:16:14 +02:00
|
|
|
<foreignphrase>Voilà</foreignphrase>:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
SELECT * FROM shoelace;
|
|
|
|
|
|
|
|
sl_name | sl_avail | sl_color | sl_len | sl_unit | sl_len_cm
|
|
|
|
---------+----------+----------+--------+---------+-----------
|
|
|
|
sl1 | 5 | black | 80 | cm | 80
|
|
|
|
sl2 | 6 | black | 100 | cm | 100
|
|
|
|
sl7 | 6 | brown | 60 | cm | 60
|
|
|
|
sl4 | 8 | black | 40 | inch | 101.6
|
|
|
|
sl3 | 10 | black | 35 | inch | 88.9
|
|
|
|
sl8 | 21 | brown | 40 | inch | 101.6
|
|
|
|
sl10 | 1000 | magenta | 40 | inch | 101.6
|
|
|
|
sl5 | 4 | brown | 1 | m | 100
|
|
|
|
sl6 | 20 | brown | 0.9 | m | 90
|
2002-01-20 23:19:57 +01:00
|
|
|
(9 rows)
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<para>
|
|
|
|
A <command>DELETE</command> on a view, with a subquery qualification that
|
1998-10-25 02:29:01 +02:00
|
|
|
in total uses 4 nesting/joined views, where one of them
|
2003-04-11 15:22:35 +02:00
|
|
|
itself has a subquery qualification containing a view
|
1998-10-25 02:29:01 +02:00
|
|
|
and where calculated view columns are used,
|
2010-10-10 19:43:33 +02:00
|
|
|
gets rewritten into
|
2003-04-11 15:22:35 +02:00
|
|
|
one single query tree that deletes the requested data
|
1998-10-25 02:29:01 +02:00
|
|
|
from a real table.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
There are probably only a few situations out in the real world
|
|
|
|
where such a construct is necessary. But it makes you feel
|
|
|
|
comfortable that it works.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</sect2>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
</sect1>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<sect1 id="rules-privileges">
|
|
|
|
<title>Rules and Privileges</title>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-08-31 19:32:24 +02:00
|
|
|
<indexterm zone="rules-privileges">
|
|
|
|
<primary>privilege</primary>
|
|
|
|
<secondary sortas="Regeln">with rules</secondary>
|
|
|
|
</indexterm>
|
|
|
|
|
|
|
|
<indexterm zone="rules-privileges">
|
|
|
|
<primary>privilege</primary>
|
|
|
|
<secondary sortas="Sichten">with views</secondary>
|
|
|
|
</indexterm>
|
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
|
|
|
Due to rewriting of queries by the <productname>PostgreSQL</productname>
|
1998-10-25 02:29:01 +02:00
|
|
|
rule system, other tables/views than those used in the original
|
2003-04-11 15:22:35 +02:00
|
|
|
query get accessed. When update rules are used, this can include write access
|
1998-10-25 02:29:01 +02:00
|
|
|
to tables.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
1998-10-25 02:29:01 +02:00
|
|
|
Rewrite rules don't have a separate owner. The owner of
|
|
|
|
a relation (table or view) is automatically the owner of the
|
|
|
|
rewrite rules that are defined for it.
|
2005-01-23 01:30:59 +01:00
|
|
|
The <productname>PostgreSQL</productname> rule system changes the
|
2001-09-10 23:58:47 +02:00
|
|
|
behavior of the default access control system. Relations that
|
2000-12-12 06:07:59 +01:00
|
|
|
are used due to rules get checked against the
|
2003-04-11 15:22:35 +02:00
|
|
|
privileges of the rule owner, not the user invoking the rule.
|
|
|
|
This means that a user only needs the required privileges
|
|
|
|
for the tables/views that he names explicitly in his queries.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
1998-10-25 02:29:01 +02:00
|
|
|
For example: A user has a list of phone numbers where some of
|
|
|
|
them are private, the others are of interest for the secretary of the office.
|
|
|
|
He can construct the following:
|
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
CREATE TABLE phone_data (person text, phone text, private boolean);
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE VIEW phone_number AS
|
2010-10-08 15:15:17 +02:00
|
|
|
SELECT person, CASE WHEN NOT private THEN phone END AS phone
|
|
|
|
FROM phone_data;
|
2002-01-20 23:19:57 +01:00
|
|
|
GRANT SELECT ON phone_number TO secretary;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2010-10-08 15:15:17 +02:00
|
|
|
|
1998-10-25 02:29:01 +02:00
|
|
|
Nobody except him (and the database superusers) can access the
|
2003-04-11 15:22:35 +02:00
|
|
|
<literal>phone_data</> table. But because of the <command>GRANT</>,
|
|
|
|
the secretary can run a <command>SELECT</command> on the
|
|
|
|
<literal>phone_number</> view. The rule system will rewrite the
|
|
|
|
<command>SELECT</command> from <literal>phone_number</> into a
|
2010-10-08 15:15:17 +02:00
|
|
|
<command>SELECT</command> from <literal>phone_data</>.
|
|
|
|
Since the user is the owner of
|
2003-04-11 15:22:35 +02:00
|
|
|
<literal>phone_number</> and therefore the owner of the rule, the
|
|
|
|
read access to <literal>phone_data</> is now checked against his
|
|
|
|
privileges and the query is permitted. The check for accessing
|
|
|
|
<literal>phone_number</> is also performed, but this is done
|
|
|
|
against the invoking user, so nobody but the user and the
|
|
|
|
secretary can use it.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
The privileges are checked rule by rule. So the secretary is for now the
|
1998-10-25 02:29:01 +02:00
|
|
|
only one who can see the public phone numbers. But the secretary can setup
|
2003-04-11 15:22:35 +02:00
|
|
|
another view and grant access to that to the public. Then, anyone
|
|
|
|
can see the <literal>phone_number</> data through the secretary's view.
|
1998-10-25 02:29:01 +02:00
|
|
|
What the secretary cannot do is to create a view that directly
|
2003-04-11 15:22:35 +02:00
|
|
|
accesses <literal>phone_data</>. (Actually he can, but it will not work since
|
|
|
|
every access will be denied during the permission checks.)
|
1998-10-25 02:29:01 +02:00
|
|
|
And as soon as the user will notice, that the secretary opened
|
2003-04-11 15:22:35 +02:00
|
|
|
his <literal>phone_number</> view, he can revoke his access. Immediately, any
|
|
|
|
access to the secretary's view would fail.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
One might think that this rule-by-rule checking is a security
|
|
|
|
hole, but in fact it isn't. But if it did not work this way, the secretary
|
|
|
|
could set up a table with the same columns as <literal>phone_number</> and
|
1998-10-25 02:29:01 +02:00
|
|
|
copy the data to there once per day. Then it's his own data and
|
2003-02-19 04:13:25 +01:00
|
|
|
he can grant access to everyone he wants. A
|
2003-04-11 15:22:35 +02:00
|
|
|
<command>GRANT</command> command means, <quote>I trust you</quote>.
|
1998-10-25 02:29:01 +02:00
|
|
|
If someone you trust does the thing above, it's time to
|
2003-04-11 15:22:35 +02:00
|
|
|
think it over and then use <command>REVOKE</command>.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2010-10-08 15:15:17 +02:00
|
|
|
Note that while views can be used to hide the contents of certain
|
|
|
|
columns using the technique shown above, they cannot be used to reliably
|
|
|
|
conceal the data in unseen rows. For example, the following view is
|
|
|
|
insecure:
|
|
|
|
<programlisting>
|
|
|
|
CREATE VIEW phone_number AS
|
|
|
|
SELECT person, phone FROM phone_data WHERE phone NOT LIKE '412%';
|
|
|
|
</programlisting>
|
|
|
|
This view might seem secure, since the rule system will rewrite any
|
2010-10-10 19:43:33 +02:00
|
|
|
<command>SELECT</command> from <literal>phone_number</> into a
|
2010-10-08 15:15:17 +02:00
|
|
|
<command>SELECT</command> from <literal>phone_data</> and add the
|
|
|
|
qualification that only entries where <literal>phone</> does not begin
|
|
|
|
with 412 are wanted. But if the user can create his or her own functions,
|
|
|
|
it is not difficult to convince the planner to execute the user-defined
|
|
|
|
function prior to the <function>NOT LIKE</function> expression.
|
2010-10-10 19:43:33 +02:00
|
|
|
For example:
|
2010-10-08 15:15:17 +02:00
|
|
|
<programlisting>
|
|
|
|
CREATE FUNCTION tricky(text, text) RETURNS bool AS $$
|
|
|
|
BEGIN
|
|
|
|
RAISE NOTICE '% => %', $1, $2;
|
|
|
|
RETURN true;
|
|
|
|
END
|
|
|
|
$$ LANGUAGE plpgsql COST 0.0000000000000000000001;
|
2010-10-10 19:43:33 +02:00
|
|
|
|
2010-10-08 15:15:17 +02:00
|
|
|
SELECT * FROM phone_number WHERE tricky(person, phone);
|
|
|
|
</programlisting>
|
|
|
|
Every person and phone number in the <literal>phone_data</> table will be
|
|
|
|
printed as a <literal>NOTICE</literal>, because the planner will choose to
|
|
|
|
execute the inexpensive <function>tricky</function> function before the
|
|
|
|
more expensive <function>NOT LIKE</function>. Even if the user is
|
|
|
|
prevented from defining new functions, built-in functions can be used in
|
2010-10-10 19:43:33 +02:00
|
|
|
similar attacks. (For example, most casting functions include their
|
|
|
|
input values in the error messages they produce.)
|
2010-10-08 15:15:17 +02:00
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
Similar considerations apply to update rules. In the examples of
|
2003-04-11 15:22:35 +02:00
|
|
|
the previous section, the owner of the tables in the example
|
|
|
|
database could grant the privileges <literal>SELECT</>,
|
|
|
|
<literal>INSERT</>, <literal>UPDATE</>, and <literal>DELETE</> on
|
|
|
|
the <literal>shoelace</> view to someone else, but only
|
|
|
|
<literal>SELECT</> on <literal>shoelace_log</>. The rule action to
|
|
|
|
write log entries will still be executed successfully, and that
|
|
|
|
other user could see the log entries. But he cannot create fake
|
2010-10-08 15:15:17 +02:00
|
|
|
entries, nor could he manipulate or remove existing ones. In this
|
|
|
|
case, there is no possibility of subverting the rules by convincing
|
|
|
|
the planner to alter the order of operations, because the only rule
|
|
|
|
which references <literal>shoelace_log</> is an unqualified
|
|
|
|
<literal>INSERT</>. This might not be true in more complex scenarios.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</sect1>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<sect1 id="rules-status">
|
|
|
|
<title>Rules and Command Status</title>
|
2002-10-15 00:14:35 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
|
|
|
The <productname>PostgreSQL</productname> server returns a command
|
2002-10-15 00:14:35 +02:00
|
|
|
status string, such as <literal>INSERT 149592 1</>, for each
|
2003-04-11 15:22:35 +02:00
|
|
|
command it receives. This is simple enough when there are no rules
|
2002-10-15 00:14:35 +02:00
|
|
|
involved, but what happens when the query is rewritten by rules?
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
2002-10-15 00:14:35 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
Rules affect the command status as follows:
|
2002-10-15 00:14:35 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<itemizedlist>
|
2002-10-15 00:14:35 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
If there is no unconditional <literal>INSTEAD</> rule for the query, then
|
2002-10-15 00:14:35 +02:00
|
|
|
the originally given query will be executed, and its command
|
|
|
|
status will be returned as usual. (But note that if there were
|
2003-04-11 15:22:35 +02:00
|
|
|
any conditional <literal>INSTEAD</> rules, the negation of their qualifications
|
Update documentation on may/can/might:
Standard English uses "may", "can", and "might" in different ways:
may - permission, "You may borrow my rake."
can - ability, "I can lift that log."
might - possibility, "It might rain today."
Unfortunately, in conversational English, their use is often mixed, as
in, "You may use this variable to do X", when in fact, "can" is a better
choice. Similarly, "It may crash" is better stated, "It might crash".
Also update two error messages mentioned in the documenation to match.
2007-01-31 21:56:20 +01:00
|
|
|
will have been added to the original query. This might reduce the
|
2002-10-15 00:14:35 +02:00
|
|
|
number of rows it processes, and if so the reported status will
|
|
|
|
be affected.)
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
|
|
|
|
<listitem>
|
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
If there is any unconditional <literal>INSTEAD</> rule for the query, then
|
2002-10-15 00:14:35 +02:00
|
|
|
the original query will not be executed at all. In this case,
|
2003-02-19 04:13:25 +01:00
|
|
|
the server will return the command status for the last query
|
2003-04-11 15:22:35 +02:00
|
|
|
that was inserted by an <literal>INSTEAD</> rule (conditional or
|
|
|
|
unconditional) and is of the same command type
|
2003-02-19 04:13:25 +01:00
|
|
|
(<command>INSERT</command>, <command>UPDATE</command>, or
|
|
|
|
<command>DELETE</command>) as the original query. If no query
|
|
|
|
meeting those requirements is added by any rule, then the
|
|
|
|
returned command status shows the original query type and
|
2003-04-11 15:22:35 +02:00
|
|
|
zeroes for the row-count and OID fields.
|
2002-10-15 00:14:35 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
2003-04-11 15:22:35 +02:00
|
|
|
</itemizedlist>
|
|
|
|
|
2010-02-24 16:54:31 +01:00
|
|
|
(This system was established in <productname>PostgreSQL</> 7.3.
|
|
|
|
In versions before that, the command status might show different
|
|
|
|
results when rules exist.)
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
2002-10-15 00:14:35 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
The programmer can ensure that any desired <literal>INSTEAD</> rule is the one
|
2002-10-15 00:14:35 +02:00
|
|
|
that sets the command status in the second case, by giving it the
|
|
|
|
alphabetically last rule name among the active rules, so that it
|
2003-04-11 15:22:35 +02:00
|
|
|
gets applied last.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</sect1>
|
2002-10-15 00:14:35 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<sect1 id="rules-triggers">
|
2011-01-29 19:00:18 +01:00
|
|
|
<title>Rules Versus Triggers</title>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-08-31 19:32:24 +02:00
|
|
|
<indexterm zone="rules-triggers">
|
|
|
|
<primary>rule</primary>
|
|
|
|
<secondary sortas="Trigger">compared with triggers</secondary>
|
|
|
|
</indexterm>
|
|
|
|
|
|
|
|
<indexterm zone="rules-triggers">
|
|
|
|
<primary>trigger</primary>
|
|
|
|
<secondary sortas="Regeln">compared with rules</secondary>
|
|
|
|
</indexterm>
|
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
1998-10-25 02:29:01 +02:00
|
|
|
Many things that can be done using triggers can also be
|
2005-01-23 01:30:59 +01:00
|
|
|
implemented using the <productname>PostgreSQL</productname>
|
2003-04-11 15:22:35 +02:00
|
|
|
rule system. One of the things that cannot be implemented by
|
|
|
|
rules are some kinds of constraints, especially foreign keys. It is possible
|
|
|
|
to place a qualified rule that rewrites a command to <literal>NOTHING</>
|
1998-10-25 02:29:01 +02:00
|
|
|
if the value of a column does not appear in another table.
|
|
|
|
But then the data is silently thrown away and that's
|
|
|
|
not a good idea. If checks for valid values are required,
|
|
|
|
and in the case of an invalid value an error message should
|
2003-04-11 15:22:35 +02:00
|
|
|
be generated, it must be done by a trigger.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2010-10-10 19:43:33 +02:00
|
|
|
In this chapter, we focused on using rules to update views. All of
|
|
|
|
the update rule examples in this chapter can also be implemented
|
|
|
|
using <literal>INSTEAD OF</> triggers on the views. Writing such
|
|
|
|
triggers is often easier than writing rules, particularly if complex
|
|
|
|
logic is required to perform the update.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2004-12-30 04:13:56 +01:00
|
|
|
For the things that can be implemented by both, which is best
|
|
|
|
depends on the usage of the database.
|
2010-10-10 19:43:33 +02:00
|
|
|
A trigger is fired once for each affected row. A rule modifies
|
2004-12-30 04:13:56 +01:00
|
|
|
the query or generates an additional query. So if many
|
1998-10-25 02:29:01 +02:00
|
|
|
rows are affected in one statement, a rule issuing one extra
|
2004-12-30 04:13:56 +01:00
|
|
|
command is likely to be faster than a trigger that is
|
2010-10-10 19:43:33 +02:00
|
|
|
called for every single row and must re-determine what to do
|
2004-12-30 04:13:56 +01:00
|
|
|
many times. However, the trigger approach is conceptually far
|
|
|
|
simpler than the rule approach, and is easier for novices to get right.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
Here we show an example of how the choice of rules versus triggers
|
|
|
|
plays out in one situation. There are two tables:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE TABLE computer (
|
|
|
|
hostname text, -- indexed
|
|
|
|
manufacturer text -- indexed
|
|
|
|
);
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE TABLE software (
|
|
|
|
software text, -- indexed
|
|
|
|
hostname text -- indexed
|
|
|
|
);
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
Both tables have many thousands of rows and the indexes on
|
|
|
|
<structfield>hostname</> are unique. The rule or trigger should
|
|
|
|
implement a constraint that deletes rows from <literal>software</>
|
|
|
|
that reference a deleted computer. The trigger would use this command:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
DELETE FROM software WHERE hostname = $1;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
Since the trigger is called for each individual row deleted from
|
|
|
|
<literal>computer</>, it can prepare and save the plan for this
|
|
|
|
command and pass the <structfield>hostname</> value in the
|
2007-02-01 01:28:19 +01:00
|
|
|
parameter. The rule would be written as:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
CREATE RULE computer_del AS ON DELETE TO computer
|
|
|
|
DO DELETE FROM software WHERE hostname = OLD.hostname;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<para>
|
2007-02-01 01:28:19 +01:00
|
|
|
Now we look at different types of deletes. In the case of a:
|
2010-10-10 19:43:33 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
DELETE FROM computer WHERE hostname = 'mypc.local.net';
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
the table <literal>computer</> is scanned by index (fast), and the
|
|
|
|
command issued by the trigger would also use an index scan (also fast).
|
2007-02-01 01:28:19 +01:00
|
|
|
The extra command from the rule would be:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
DELETE FROM software WHERE computer.hostname = 'mypc.local.net'
|
|
|
|
AND software.hostname = computer.hostname;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2001-05-17 23:50:18 +02:00
|
|
|
Since there are appropriate indexes setup, the planner
|
1998-10-25 02:29:01 +02:00
|
|
|
will create a plan of
|
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<literallayout class="monospaced">
|
2002-01-20 23:19:57 +01:00
|
|
|
Nestloop
|
2005-01-22 23:06:27 +01:00
|
|
|
-> Index Scan using comp_hostidx on computer
|
|
|
|
-> Index Scan using soft_hostidx on software
|
2003-04-11 15:22:35 +02:00
|
|
|
</literallayout>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
|
|
|
So there would be not that much difference in speed between
|
2003-04-11 15:22:35 +02:00
|
|
|
the trigger and the rule implementation.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
With the next delete we want to get rid of all the 2000 computers
|
|
|
|
where the <structfield>hostname</> starts with
|
|
|
|
<literal>old</>. There are two possible commands to do that. One
|
2007-02-01 01:28:19 +01:00
|
|
|
is:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2005-01-22 23:56:36 +01:00
|
|
|
DELETE FROM computer WHERE hostname >= 'old'
|
|
|
|
AND hostname < 'ole'
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2007-02-01 01:28:19 +01:00
|
|
|
The command added by the rule will be:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2005-01-22 23:56:36 +01:00
|
|
|
DELETE FROM software WHERE computer.hostname >= 'old' AND computer.hostname < 'ole'
|
2003-04-11 15:22:35 +02:00
|
|
|
AND software.hostname = computer.hostname;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
|
|
|
|
with the plan
|
|
|
|
|
|
|
|
<literallayout class="monospaced">
|
2002-01-20 23:19:57 +01:00
|
|
|
Hash Join
|
2005-01-22 23:06:27 +01:00
|
|
|
-> Seq Scan on software
|
|
|
|
-> Hash
|
|
|
|
-> Index Scan using comp_hostidx on computer
|
2003-04-11 15:22:35 +02:00
|
|
|
</literallayout>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2007-02-01 01:28:19 +01:00
|
|
|
The other possible command is:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2002-01-20 23:19:57 +01:00
|
|
|
DELETE FROM computer WHERE hostname ~ '^old';
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
which results in the following executing plan for the command
|
|
|
|
added by the rule:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<literallayout class="monospaced">
|
2002-01-20 23:19:57 +01:00
|
|
|
Nestloop
|
2005-01-22 23:06:27 +01:00
|
|
|
-> Index Scan using comp_hostidx on computer
|
|
|
|
-> Index Scan using soft_hostidx on software
|
2003-04-11 15:22:35 +02:00
|
|
|
</literallayout>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2000-12-12 06:07:59 +01:00
|
|
|
This shows, that the planner does not realize that the
|
2003-04-11 15:22:35 +02:00
|
|
|
qualification for <structfield>hostname</> in
|
|
|
|
<literal>computer</> could also be used for an index scan on
|
|
|
|
<literal>software</> when there are multiple qualification
|
|
|
|
expressions combined with <literal>AND</>, which is what it does
|
|
|
|
in the regular-expression version of the command. The trigger will
|
|
|
|
get invoked once for each of the 2000 old computers that have to be
|
|
|
|
deleted, and that will result in one index scan over
|
|
|
|
<literal>computer</> and 2000 index scans over
|
|
|
|
<literal>software</>. The rule implementation will do it with two
|
|
|
|
commands that use indexes. And it depends on the overall size of
|
|
|
|
the table <literal>software</> whether the rule will still be faster in the
|
|
|
|
sequential scan situation. 2000 command executions from the trigger over the SPI
|
|
|
|
manager take some time, even if all the index blocks will soon be in the cache.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2007-02-01 01:28:19 +01:00
|
|
|
The last command we look at is:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2005-10-15 22:12:33 +02:00
|
|
|
DELETE FROM computer WHERE manufacturer = 'bim';
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
|
|
|
Again this could result in many rows to be deleted from
|
2003-04-11 15:22:35 +02:00
|
|
|
<literal>computer</>. So the trigger will again run many commands
|
2007-02-01 01:28:19 +01:00
|
|
|
through the executor. The command generated by the rule will be:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2005-10-15 22:12:33 +02:00
|
|
|
DELETE FROM software WHERE computer.manufacturer = 'bim'
|
2003-04-11 15:22:35 +02:00
|
|
|
AND software.hostname = computer.hostname;
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
The plan for that command will again be the nested loop over two
|
|
|
|
index scans, only using a different index on <literal>computer</>:
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<programlisting>
|
2003-04-11 15:22:35 +02:00
|
|
|
Nestloop
|
2005-01-22 23:06:27 +01:00
|
|
|
-> Index Scan using comp_manufidx on computer
|
|
|
|
-> Index Scan using soft_hostidx on software
|
2005-01-23 01:30:59 +01:00
|
|
|
</programlisting>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
In any of these cases, the extra commands from the rule system
|
|
|
|
will be more or less independent from the number of affected rows
|
|
|
|
in a command.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2003-04-11 15:22:35 +02:00
|
|
|
<![IGNORE[
|
|
|
|
<!-- What's happening with this? If it doesn't come back, remove this section. -->
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-02-19 04:13:25 +01:00
|
|
|
Another situation is cases on <command>UPDATE</command> where it depends on the
|
1998-10-25 02:29:01 +02:00
|
|
|
change of an attribute if an action should be performed or
|
2006-04-23 05:39:52 +02:00
|
|
|
not. The only way to
|
1998-10-25 02:29:01 +02:00
|
|
|
create a rule as in the shoelace_log example is to do it with
|
|
|
|
a rule qualification. That results in an extra query that is
|
2000-05-02 22:02:03 +02:00
|
|
|
performed always, even if the attribute of interest cannot
|
2001-09-10 23:58:47 +02:00
|
|
|
change at all because it does not appear in the target list
|
1998-10-25 02:29:01 +02:00
|
|
|
of the initial query. When this is enabled again, it will be
|
|
|
|
one more advantage of rules over triggers. Optimization of
|
|
|
|
a trigger must fail by definition in this case, because the
|
2000-12-12 06:07:59 +01:00
|
|
|
fact that its actions will only be done when a specific attribute
|
|
|
|
is updated is hidden in its functionality. The definition of
|
1998-10-25 02:29:01 +02:00
|
|
|
a trigger only allows to specify it on row level, so whenever a
|
2000-12-12 06:07:59 +01:00
|
|
|
row is touched, the trigger must be called to make its
|
1998-10-25 02:29:01 +02:00
|
|
|
decision. The rule system will know it by looking up the
|
2001-09-10 23:58:47 +02:00
|
|
|
target list and will suppress the additional query completely
|
1998-10-25 02:29:01 +02:00
|
|
|
if the attribute isn't touched. So the rule, qualified or not,
|
2000-05-02 22:02:03 +02:00
|
|
|
will only do its scans if there ever could be something to do.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
2003-04-11 15:22:35 +02:00
|
|
|
]]>
|
1998-10-25 02:29:01 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
<para>
|
2003-04-11 15:22:35 +02:00
|
|
|
The summary is, rules will only be significantly slower than
|
|
|
|
triggers if their actions result in large and badly qualified
|
|
|
|
joins, a situation where the planner fails.
|
2005-01-23 01:30:59 +01:00
|
|
|
</para>
|
|
|
|
</sect1>
|
1998-04-28 16:52:46 +02:00
|
|
|
|
2005-01-23 01:30:59 +01:00
|
|
|
</chapter>
|