path: root/doc/src/sgml/query.sgml

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

 <chapter id="tutorial-sql">
  <title>The <acronym>SQL</acronym> Language</title>

  <sect1 id="tutorial-sql-intro">
   <title>Introduction</title>

   <para>
    This chapter provides an overview of how to use
    <acronym>SQL</acronym> to perform simple operations.  This
    tutorial is only intended to give you an introduction and is in no
    way a complete tutorial on <acronym>SQL</acronym>.  Numerous books
    have been written on <acronym>SQL</acronym>, including <xref
    linkend="melt93"/> and <xref linkend="date97"/>.
    You should be aware that some <productname>PostgreSQL</productname>
    language features are extensions to the standard.
   </para>

   <para>
    In the examples that follow, we assume that you have created a
    database named <literal>mydb</literal>, as described in the previous
    chapter, and have been able to start <application>psql</application>.
   </para>

   <para>
    Examples in this manual can also be found in the
    <productname>PostgreSQL</productname> source distribution
    in the directory <filename>src/tutorial/</filename>.  (Binary
    distributions of <productname>PostgreSQL</productname> might not
    provide those files.)  To use those
    files, first change to that directory and run <application>make</application>:

<screen>
<prompt>$</prompt> <userinput>cd <replaceable>...</replaceable>/src/tutorial</userinput>
<prompt>$</prompt> <userinput>make</userinput>
</screen>

    This creates the scripts and compiles the C files containing user-defined
    functions and types.  Then, to start the tutorial, do the following:

<screen>
<prompt>$</prompt> <userinput>psql -s mydb</userinput>
<computeroutput>
...
</computeroutput>
<prompt>mydb=&gt;</prompt> <userinput>\i basics.sql</userinput>
</screen>

    The <literal>\i</literal> command reads in commands from the
    specified file. <command>psql</command>'s <literal>-s</literal> option puts you in
    single step mode which pauses before sending each statement to the
    server.  The commands used in this section are in the file
    <filename>basics.sql</filename>.
   </para>
  </sect1>


  <sect1 id="tutorial-concepts">
   <title>Concepts</title>

   <para>
    <indexterm><primary>relational database</primary></indexterm>
    <indexterm><primary>hierarchical database</primary></indexterm>
    <indexterm><primary>object-oriented database</primary></indexterm>
    <indexterm><primary>relation</primary></indexterm>
    <indexterm><primary>table</primary></indexterm>

    <productname>PostgreSQL</productname> is a <firstterm>relational
    database management system</firstterm> (<acronym>RDBMS</acronym>).
    That means it is a system for managing data stored in
    <firstterm>relations</firstterm>.  Relation is essentially a
    mathematical term for <firstterm>table</firstterm>.  The notion of
    storing data in tables is so commonplace today that it might
    seem inherently obvious, but there are a number of other ways of
    organizing databases.  Files and directories on Unix-like
    operating systems form an example of a hierarchical database.  A
    more modern development is the object-oriented database.
   </para>

   <para>
    <indexterm><primary>row</primary></indexterm>
    <indexterm><primary>column</primary></indexterm>

    Each table is a named collection of <firstterm>rows</firstterm>.
    Each row of a given table has the same set of named
    <firstterm>columns</firstterm>,
    and each column is of a specific data type.  Whereas columns have
    a fixed order in each row, it is important to remember that SQL
    does not guarantee the order of the rows within the table in any
    way (although they can be explicitly sorted for display).
   </para>

   <para>
    <indexterm><primary>database cluster</primary></indexterm>
    <indexterm><primary>cluster</primary><secondary>of databases</secondary><see>database cluster</see></indexterm>

    Tables are grouped into databases, and a collection of databases
    managed by a single <productname>PostgreSQL</productname> server
    instance constitutes a database <firstterm>cluster</firstterm>.
   </para>
  </sect1>


  <sect1 id="tutorial-table">
   <title>Creating a New Table</title>

   <indexterm zone="tutorial-table">
    <primary>CREATE TABLE</primary>
   </indexterm>

   <para>
    You  can  create  a  new  table by specifying the table
    name, along with all column names and their types:

<programlisting>
CREATE TABLE weather (
    city            varchar(80),
    temp_lo         int,           -- low temperature
    temp_hi         int,           -- high temperature
    prcp            real,          -- precipitation
    date            date
);
</programlisting>

    You can enter this into <command>psql</command> with the line
    breaks.  <command>psql</command> will recognize that the command
    is not terminated until the semicolon.
   </para>

   <para>
    White space (i.e., spaces, tabs, and newlines) can be used freely
    in SQL commands.  That means you can type the command aligned
    differently than above, or even all on one line.  Two dashes
    (<quote><literal>--</literal></quote>) introduce comments.
    Whatever follows them is ignored up to the end of the line.  SQL
    is case-insensitive about key words and identifiers, except
    when identifiers are double-quoted to preserve the case (not done
    above).
   </para>

   <para>
    <type>varchar(80)</type> specifies a data type that can store
    arbitrary character strings up to 80 characters in length.
    <type>int</type> is the normal integer type.  <type>real</type> is
    a type for storing single precision floating-point numbers.
    <type>date</type> should be self-explanatory.  (Yes, the column of
    type <type>date</type> is also named <structfield>date</structfield>.
    This might be convenient or confusing &mdash; you choose.)
   </para>

   <para>
    <productname>PostgreSQL</productname> supports the standard
    <acronym>SQL</acronym> types <type>int</type>,
    <type>smallint</type>, <type>real</type>, <type>double
    precision</type>, <type>char(<replaceable>N</replaceable>)</type>,
    <type>varchar(<replaceable>N</replaceable>)</type>, <type>date</type>,
    <type>time</type>, <type>timestamp</type>, and
    <type>interval</type>, as well as other types of general utility
    and a rich set of geometric types.
    <productname>PostgreSQL</productname> can be customized with an
    arbitrary number of user-defined data types.  Consequently, type
    names are not key words in the syntax, except where required to
    support special cases in the <acronym>SQL</acronym> standard.
   </para>

   <para>
    The second example will store cities and their associated
    geographical location:
<programlisting>
CREATE TABLE cities (
    name            varchar(80),
    location        point
);
</programlisting>
    The <type>point</type> type is an example of a
    <productname>PostgreSQL</productname>-specific data type.
   </para>

   <para>
    <indexterm>
     <primary>DROP TABLE</primary>
    </indexterm>

    Finally, it should be mentioned that if you don't need a table any
    longer or want to recreate it differently you can remove it using
    the following command:
<synopsis>
DROP TABLE <replaceable>tablename</replaceable>;
</synopsis>
   </para>
  </sect1>


  <sect1 id="tutorial-populate">
   <title>Populating a Table With Rows</title>

   <indexterm zone="tutorial-populate">
    <primary>INSERT</primary>
   </indexterm>

   <para>
    The <command>INSERT</command> statement is used to populate a table  with
    rows:

<programlisting>
INSERT INTO weather VALUES ('San Francisco', 46, 50, 0.25, '1994-11-27');
</programlisting>

    Note that all data types use rather obvious input formats.
    Constants that are not simple numeric values usually must be
    surrounded by single quotes (<literal>'</literal>), as in the example.
    The
    <type>date</type> type is actually quite flexible in what it
    accepts, but for this tutorial we will stick to the unambiguous
    format shown here.
   </para>

   <para>
    The <type>point</type> type requires a coordinate pair as input,
    as shown here:
<programlisting>
INSERT INTO cities VALUES ('San Francisco', '(-194.0, 53.0)');
</programlisting>
   </para>

   <para>
    The syntax used so far requires you to remember the order of the
    columns.  An alternative syntax allows you to list the columns
    explicitly:
<programlisting>
INSERT INTO weather (city, temp_lo, temp_hi, prcp, date)
    VALUES ('San Francisco', 43, 57, 0.0, '1994-11-29');
</programlisting>
    You can list the columns in a different order if you wish or
    even omit some columns, e.g., if the precipitation is unknown:
<programlisting>
INSERT INTO weather (date, city, temp_hi, temp_lo)
    VALUES ('1994-11-29', 'Hayward', 54, 37);
</programlisting>
    Many developers consider explicitly listing the columns better
    style than relying on the order implicitly.
   </para>

   <para>
    Please enter all the commands shown above so you have some data to
    work with in the following sections.
   </para>

   <para>
    <indexterm>
     <primary>COPY</primary>
    </indexterm>

    You could also have used <command>COPY</command> to load large
    amounts of data from flat-text files.  This is usually faster
    because the <command>COPY</command> command is optimized for this
    application while allowing less flexibility than
    <command>INSERT</command>.  An example would be:

<programlisting>
COPY weather FROM '/home/user/weather.txt';
</programlisting>

    where the file name for the source file must be available on the
    machine running the backend process, not the client, since the backend process
    reads the file directly.  You can read more about the
    <command>COPY</command> command in <xref linkend="sql-copy"/>.
   </para>
  </sect1>


  <sect1 id="tutorial-select">
   <title>Querying a Table</title>

   <para>
    <indexterm><primary>query</primary></indexterm>
    <indexterm><primary>SELECT</primary></indexterm>

    To retrieve data from a table, the table is
    <firstterm>queried</firstterm>.  An <acronym>SQL</acronym>
    <command>SELECT</command> statement is used to do this.  The
    statement is divided into a select list (the part that lists the
    columns to be returned), a table list (the part that lists the
    tables from which to retrieve the data), and an optional
    qualification (the part that specifies any restrictions).  For
    example, to retrieve all the rows of table
    <structname>weather</structname>, type:
<programlisting>
SELECT * FROM weather;
</programlisting>
    Here <literal>*</literal> is a shorthand for <quote>all columns</quote>.
     <footnote>
      <para>
       While <literal>SELECT *</literal> is useful for off-the-cuff
       queries, it is widely considered bad style in production code,
       since adding a column to the table would change the results.
      </para>
     </footnote>
    So the same result would be had with:
<programlisting>
SELECT city, temp_lo, temp_hi, prcp, date FROM weather;
</programlisting>

    The output should be:

<screen>
     city      | temp_lo | temp_hi | prcp |    date
---------------+---------+---------+------+------------
 San Francisco |      46 |      50 | 0.25 | 1994-11-27
 San Francisco |      43 |      57 |    0 | 1994-11-29
 Hayward       |      37 |      54 |      | 1994-11-29
(3 rows)
</screen>
   </para>

   <para>
    You can write expressions, not just simple column references, in the
    select list.  For example, you can do:
<programlisting>
SELECT city, (temp_hi+temp_lo)/2 AS temp_avg, date FROM weather;
</programlisting>
    This should give:
<screen>
     city      | temp_avg |    date
---------------+----------+------------
 San Francisco |       48 | 1994-11-27
 San Francisco |       50 | 1994-11-29
 Hayward       |       45 | 1994-11-29
(3 rows)
</screen>
    Notice how the <literal>AS</literal> clause is used to relabel the
    output column.  (The <literal>AS</literal> clause is optional.)
   </para>

   <para>
    A query can be <quote>qualified</quote> by adding a <literal>WHERE</literal>
    clause that specifies which rows are wanted.  The <literal>WHERE</literal>
    clause contains a Boolean (truth value) expression, and only rows for
    which the Boolean expression is true are returned.  The usual
    Boolean operators (<literal>AND</literal>,
    <literal>OR</literal>, and <literal>NOT</literal>) are allowed in
    the qualification.  For example, the following
    retrieves the weather of San Francisco on rainy days:

<programlisting>
SELECT * FROM weather
    WHERE city = 'San Francisco' AND prcp &gt; 0.0;
</programlisting>
    Result:
<screen>
     city      | temp_lo | temp_hi | prcp |    date
---------------+---------+---------+------+------------
 San Francisco |      46 |      50 | 0.25 | 1994-11-27
(1 row)
</screen>
   </para>

   <para>
    <indexterm><primary>ORDER BY</primary></indexterm>

    You can request that the results of a query
    be returned in sorted order:

<programlisting>
SELECT * FROM weather
    ORDER BY city;
</programlisting>

<screen>
     city      | temp_lo | temp_hi | prcp |    date
---------------+---------+---------+------+------------
 Hayward       |      37 |      54 |      | 1994-11-29
 San Francisco |      43 |      57 |    0 | 1994-11-29
 San Francisco |      46 |      50 | 0.25 | 1994-11-27
</screen>

    In this example, the sort order isn't fully specified, and so you
    might get the San Francisco rows in either order.  But you'd always
    get the results shown above if you do:

<programlisting>
SELECT * FROM weather
    ORDER BY city, temp_lo;
</programlisting>
   </para>

   <para>
    <indexterm><primary>DISTINCT</primary></indexterm>
    <indexterm><primary>duplicate</primary></indexterm>

    You can request that duplicate rows be removed from the result of
    a query:

<programlisting>
SELECT DISTINCT city
    FROM weather;
</programlisting>

<screen>
     city
---------------
 Hayward
 San Francisco
(2 rows)
</screen>

    Here again, the result row ordering might vary.
    You can ensure consistent results by using <literal>DISTINCT</literal> and
    <literal>ORDER BY</literal> together:
     <footnote>
      <para>
       In some database systems, including older versions of
       <productname>PostgreSQL</productname>, the implementation of
       <literal>DISTINCT</literal> automatically orders the rows and
       so <literal>ORDER BY</literal> is unnecessary.  But this is not
       required by the SQL standard, and current
       <productname>PostgreSQL</productname> does not guarantee that
       <literal>DISTINCT</literal> causes the rows to be ordered.
      </para>
     </footnote>

<programlisting>
SELECT DISTINCT city
    FROM weather
    ORDER BY city;
</programlisting>
   </para>
  </sect1>


  <sect1 id="tutorial-join">
   <title>Joins Between Tables</title>

   <indexterm zone="tutorial-join">
    <primary>join</primary>
   </indexterm>

   <para>
    Thus far, our queries have only accessed one table at a time.
    Queries can access multiple tables at once, or access the same
    table in such a way that multiple rows of the table are being
    processed at the same time.  Queries that access multiple tables
    (or multiple instances of the same table) at one time are called
    <firstterm>join</firstterm> queries.  They combine rows from one table
    with rows from a second table, with an expression specifying which rows
    are to be paired.  For example, to return all the weather records together
    with the location of the associated city, the database needs to compare
    the <structfield>city</structfield>
    column of each row of the <structname>weather</structname> table with the
    <structfield>name</structfield> column of all rows in the <structname>cities</structname>
    table, and select the pairs of rows where these values match.<footnote>
     <para>
      This  is only a conceptual model.  The join is usually performed
      in a more efficient manner than actually comparing each possible
      pair of rows, but this is invisible to the user.
     </para>
    </footnote>
    This would be accomplished by the following query:

<programlisting>
SELECT * FROM weather JOIN cities ON city = name;
</programlisting>

<screen>
     city      | temp_lo | temp_hi | prcp |    date    |     name      | location
---------------+---------+---------+------+------------+---------------+-----------
 San Francisco |      46 |      50 | 0.25 | 1994-11-27 | San Francisco | (-194,53)
 San Francisco |      43 |      57 |    0 | 1994-11-29 | San Francisco | (-194,53)
(2 rows)
</screen>

   </para>

   <para>
    Observe two things about the result set:
    <itemizedlist>
     <listitem>
      <para>
       There is no result row for the city of Hayward.  This is
       because there is no matching entry in the
       <structname>cities</structname> table for Hayward, so the join
       ignores the unmatched rows in the <structname>weather</structname> table.  We will see
       shortly how this can be fixed.
      </para>
     </listitem>

     <listitem>
      <para>
       There are two columns containing the city name.  This is
       correct because the lists of columns from the
       <structname>weather</structname> and
       <structname>cities</structname> tables are concatenated.  In
       practice this is undesirable, though, so you will probably want
       to list the output columns explicitly rather than using
       <literal>*</literal>:
<programlisting>
SELECT city, temp_lo, temp_hi, prcp, date, location
    FROM weather JOIN cities ON city = name;
</programlisting>
      </para>
     </listitem>
    </itemizedlist>
   </para>

   <para>
    Since the columns all had different names, the parser
    automatically found which table they belong to.  If there
    were duplicate column names in the two tables you'd need to
    <firstterm>qualify</firstterm> the column names to show which one you
    meant, as in:

<programlisting>
SELECT weather.city, weather.temp_lo, weather.temp_hi,
       weather.prcp, weather.date, cities.location
    FROM weather JOIN cities ON weather.city = cities.name;
</programlisting>

    It is widely considered good style to qualify all column names
    in a join query, so that the query won't fail if a duplicate
    column name is later added to one of the tables.
   </para>

   <para>
    Join queries of the kind seen thus far can also be written in this
    form:

<programlisting>
SELECT *
    FROM weather, cities
    WHERE city = name;
</programlisting>

    This syntax pre-dates the <literal>JOIN</literal>/<literal>ON</literal>
    syntax, which was introduced in SQL-92.  The tables are simply listed in
    the <literal>FROM</literal> clause, and the comparison expression is added
    to the <literal>WHERE</literal> clause.  The results from this older
    implicit syntax and the newer explicit
    <literal>JOIN</literal>/<literal>ON</literal> syntax are identical.  But
    for a reader of the query, the explicit syntax makes its meaning easier to
    understand: The join condition is introduced by its own key word whereas
    previously the condition was mixed into the <literal>WHERE</literal>
    clause together with other conditions.
   </para>

   <indexterm><primary>join</primary><secondary>outer</secondary></indexterm>

   <para>
    Now we will figure out how we can get the Hayward records back in.
    What we want the query to do is to scan the
    <structname>weather</structname> table and for each row to find the
    matching <structname>cities</structname> row(s).  If no matching row is
    found we want some <quote>empty values</quote> to be substituted
    for the <structname>cities</structname> table's columns.  This kind
    of query is called an <firstterm>outer join</firstterm>.  (The
    joins we have seen so far are <firstterm>inner joins</firstterm>.)
    The command looks like this:

<programlisting>
SELECT *
    FROM weather LEFT OUTER JOIN cities ON weather.city = cities.name;
</programlisting>

<screen>
     city      | temp_lo | temp_hi | prcp |    date    |     name      | location
---------------+---------+---------+------+------------+---------------+-----------
 Hayward       |      37 |      54 |      | 1994-11-29 |               |
 San Francisco |      46 |      50 | 0.25 | 1994-11-27 | San Francisco | (-194,53)
 San Francisco |      43 |      57 |    0 | 1994-11-29 | San Francisco | (-194,53)
(3 rows)
</screen>

    This query is called a <firstterm>left outer
    join</firstterm> because the table mentioned on the left of the
    join operator will have each of its rows in the output at least
    once, whereas the table on the right will only have those rows
    output that match some row of the left table.  When outputting a
    left-table row for which there is no right-table match, empty (null)
    values are substituted for the right-table columns.
   </para>

   <formalpara>
    <title>Exercise:</title>

    <para>
     There are also right outer joins and full outer joins.  Try to
     find out what those do.
    </para>
   </formalpara>

   <indexterm><primary>join</primary><secondary>self</secondary></indexterm>
   <indexterm><primary>alias</primary><secondary>for table name in query</secondary></indexterm>
   <para>
    We can also join a table against itself.  This is called a
    <firstterm>self join</firstterm>.  As an example, suppose we wish
    to find all the weather records that are in the temperature range
    of other weather records.  So we need to compare the
    <structfield>temp_lo</structfield> and <structfield>temp_hi</structfield> columns of
    each <structname>weather</structname> row to the
    <structfield>temp_lo</structfield> and
    <structfield>temp_hi</structfield> columns of all other
    <structname>weather</structname> rows.  We can do this with the
    following query:

<programlisting>
SELECT w1.city, w1.temp_lo AS low, w1.temp_hi AS high,
       w2.city, w2.temp_lo AS low, w2.temp_hi AS high
    FROM weather w1 JOIN weather w2
        ON w1.temp_lo &lt; w2.temp_lo AND w1.temp_hi &gt; w2.temp_hi;
</programlisting>

<screen>
     city      | low | high |     city      | low | high
---------------+-----+------+---------------+-----+------
 San Francisco |  43 |   57 | San Francisco |  46 |   50
 Hayward       |  37 |   54 | San Francisco |  46 |   50
(2 rows)
</screen>

    Here we have relabeled the weather table as <literal>w1</literal> and
    <literal>w2</literal> to be able to distinguish the left and right side
    of the join.  You can also use these kinds of aliases in other
    queries to save some typing, e.g.:
<programlisting>
SELECT *
    FROM weather w JOIN cities c ON w.city = c.name;
</programlisting>
    You will encounter this style of abbreviating quite frequently.
   </para>
  </sect1>


  <sect1 id="tutorial-agg">
   <title>Aggregate Functions</title>

   <indexterm zone="tutorial-agg">
    <primary>aggregate function</primary>
   </indexterm>

   <para>
    Like  most  other relational database products,
    <productname>PostgreSQL</productname> supports
    <firstterm>aggregate functions</firstterm>.
    An aggregate function computes a single result from multiple input rows.
    For example, there are aggregates to compute the
    <function>count</function>, <function>sum</function>,
    <function>avg</function> (average), <function>max</function> (maximum) and
    <function>min</function> (minimum) over a set of rows.
   </para>

   <para>
    As an example, we can find the highest low-temperature reading anywhere
    with:

<programlisting>
SELECT max(temp_lo) FROM weather;
</programlisting>

<screen>
 max
-----
  46
(1 row)
</screen>
   </para>

   <para>
    <indexterm><primary>subquery</primary></indexterm>

    If we wanted to know what city (or cities) that reading occurred in,
    we might try:

<programlisting>
SELECT city FROM weather WHERE temp_lo = max(temp_lo);     <lineannotation>WRONG</lineannotation>
</programlisting>

    but this will not work since the aggregate
    <function>max</function> cannot be used in the
    <literal>WHERE</literal> clause.  (This restriction exists because
    the <literal>WHERE</literal> clause determines which rows will be
    included in the aggregate calculation; so obviously it has to be evaluated
    before aggregate functions are computed.)
    However, as is often the case
    the query can be restated to accomplish the desired result, here
    by using a <firstterm>subquery</firstterm>:

<programlisting>
SELECT city FROM weather
    WHERE temp_lo = (SELECT max(temp_lo) FROM weather);
</programlisting>

<screen>
     city
---------------
 San Francisco
(1 row)
</screen>

    This is OK because the subquery is an independent computation
    that computes its own aggregate separately from what is happening
    in the outer query.
   </para>

   <para>
    <indexterm><primary>GROUP BY</primary></indexterm>
    <indexterm><primary>HAVING</primary></indexterm>

    Aggregates are also very useful in combination with <literal>GROUP
    BY</literal> clauses.  For example, we can get the number of readings
    and the maximum low temperature observed in each city with:

<programlisting>
SELECT city, count(*), max(temp_lo)
    FROM weather
    GROUP BY city;
</programlisting>

<screen>
     city      | count | max
---------------+-------+-----
 Hayward       |     1 |  37
 San Francisco |     2 |  46
(2 rows)
</screen>

    which gives us one output row per city.  Each aggregate result is
    computed over the table rows matching that city.
    We can filter these grouped
    rows using <literal>HAVING</literal>:

<programlisting>
SELECT city, count(*), max(temp_lo)
    FROM weather
    GROUP BY city
    HAVING max(temp_lo) &lt; 40;
</programlisting>

<screen>
  city   | count | max
---------+-------+-----
 Hayward |     1 |  37
(1 row)
</screen>

    which gives us the same results for only the cities that have all
    <structfield>temp_lo</structfield> values below 40.  Finally, if we only care about
    cities whose
    names begin with <quote><literal>S</literal></quote>, we might do:

<programlisting>
SELECT city, count(*), max(temp_lo)
    FROM weather
    WHERE city LIKE 'S%'            -- <co id="co.tutorial-agg-like"/>
    GROUP BY city;
</programlisting>

<screen>
     city      | count | max
---------------+-------+-----
 San Francisco |     2 |  46
(1 row)
</screen>
   <calloutlist>
    <callout arearefs="co.tutorial-agg-like">
     <para>
      The <literal>LIKE</literal> operator does pattern matching and
      is explained in <xref linkend="functions-matching"/>.
     </para>
    </callout>
   </calloutlist>
   </para>

   <para>
    It is important to understand the interaction between aggregates and
    <acronym>SQL</acronym>'s <literal>WHERE</literal> and <literal>HAVING</literal> clauses.
    The fundamental difference between <literal>WHERE</literal> and
    <literal>HAVING</literal> is this: <literal>WHERE</literal> selects
    input rows before groups and aggregates are computed (thus, it controls
    which rows go into the aggregate computation), whereas
    <literal>HAVING</literal> selects group rows after groups and
    aggregates are computed.  Thus, the
    <literal>WHERE</literal> clause must not contain aggregate functions;
    it makes no sense to try to use an aggregate to determine which rows
    will be inputs to the aggregates.  On the other hand, the
    <literal>HAVING</literal> clause always contains aggregate functions.
    (Strictly speaking, you are allowed to write a <literal>HAVING</literal>
    clause that doesn't use aggregates, but it's seldom useful. The same
    condition could be used more efficiently at the <literal>WHERE</literal>
    stage.)
   </para>

   <para>
    In the previous example, we can apply the city name restriction in
    <literal>WHERE</literal>, since it needs no aggregate.  This is
    more efficient than adding the restriction to <literal>HAVING</literal>,
    because we avoid doing the grouping and aggregate calculations
    for all rows that fail the <literal>WHERE</literal> check.
   </para>

   <para>
    Another way to select the rows that go into an aggregate
    computation is to use <literal>FILTER</literal>, which is a
    per-aggregate option:

<programlisting>
SELECT city, count(*) FILTER (WHERE temp_lo &lt; 45), max(temp_lo)
    FROM weather
    GROUP BY city;
</programlisting>

<screen>
     city      | count | max
---------------+-------+-----
 Hayward       |     1 |  37
 San Francisco |     1 |  46
(2 rows)
</screen>

    <literal>FILTER</literal> is much like <literal>WHERE</literal>,
    except that it removes rows only from the input of the particular
    aggregate function that it is attached to.
    Here, the <literal>count</literal> aggregate counts only
    rows with <literal>temp_lo</literal> below 45; but the
    <literal>max</literal> aggregate is still applied to all rows,
    so it still finds the reading of 46.
   </para>
  </sect1>


  <sect1 id="tutorial-update">
   <title>Updates</title>

   <indexterm zone="tutorial-update">
    <primary>UPDATE</primary>
   </indexterm>

   <para>
    You can update existing rows using the
    <command>UPDATE</command> command.
    Suppose you discover the temperature readings are
    all off by 2 degrees after November 28.  You can correct the
    data as follows:

<programlisting>
UPDATE weather
    SET temp_hi = temp_hi - 2,  temp_lo = temp_lo - 2
    WHERE date &gt; '1994-11-28';
</programlisting>
   </para>

   <para>
    Look at the new state of the data:
<programlisting>
SELECT * FROM weather;

     city      | temp_lo | temp_hi | prcp |    date
---------------+---------+---------+------+------------
 San Francisco |      46 |      50 | 0.25 | 1994-11-27
 San Francisco |      41 |      55 |    0 | 1994-11-29
 Hayward       |      35 |      52 |      | 1994-11-29
(3 rows)
</programlisting>
   </para>
  </sect1>

  <sect1 id="tutorial-delete">
   <title>Deletions</title>

   <indexterm zone="tutorial-delete">
    <primary>DELETE</primary>
   </indexterm>

   <para>
    Rows can be removed from a table using the <command>DELETE</command>
    command.
    Suppose you are no longer interested in the weather of Hayward.
    Then you can do the following to delete those rows from the table:
<programlisting>
DELETE FROM weather WHERE city = 'Hayward';
</programlisting>

    All weather records belonging to Hayward are removed.

<programlisting>
SELECT * FROM weather;
</programlisting>

<screen>
     city      | temp_lo | temp_hi | prcp |    date
---------------+---------+---------+------+------------
 San Francisco |      46 |      50 | 0.25 | 1994-11-27
 San Francisco |      41 |      55 |    0 | 1994-11-29
(2 rows)
</screen>
   </para>

   <para>
    One should be wary of statements of the form
<synopsis>
DELETE FROM <replaceable>tablename</replaceable>;
</synopsis>

    Without a qualification, <command>DELETE</command> will
    remove  <emphasis>all</emphasis>  rows from the given table, leaving it
    empty.  The system will not request confirmation before
    doing this!
   </para>
  </sect1>

 </chapter>