repo_zf1/documentation/manual/en/module_specs/Zend_Cache-Theory.xml

<?xml version="1.0" encoding="UTF-8"?>
<!-- Reviewed: no -->
<sect1 id="zend.cache.theory">
    <title>The Theory of Caching</title>
    <para>
        There are three key concepts in <classname>Zend_Cache</classname>. One is the unique
        identifier (a string) that is used to identify cache records. The second one is the
        <emphasis>'lifetime'</emphasis> directive as seen in the examples; it defines for how long
        the cached resource is considered 'fresh'. The third key concept is conditional execution so
        that parts of your code can be skipped entirely, boosting performance. The main frontend
        function (eg. <methodname>Zend_Cache_Core::get()</methodname>) is always designed to return
        <constant>FALSE</constant> for a cache miss if that makes sense for the nature of a
        frontend. That enables
        end-users to wrap parts of the code they would like to cache (and skip) in
        <emphasis><methodname>if()</methodname>{ ... }</emphasis> statements where the condition is
        a <classname>Zend_Cache</classname> method itself. On the end if these blocks you must save
        what you've generated, however (eg. <methodname>Zend_Cache_Core::save()</methodname>).
    </para>
    <note><para>
        The conditional execution design of your generating code is not necessary in some frontends
        (<emphasis>Function</emphasis>, for an example) when the whole logic is implemented inside
        the frontend.
    </para></note>
    <note><para>
        'Cache hit' is a term for a condition when a cache record is found, is valid and is 'fresh'
        (in other words hasn't expired yet). 'Cache miss' is everything else. When a cache miss
        happens, you must generate your data (as you would normally do) and have it cached. When you
        have a cache hit, on the other hand, the backend automatically fetches the record from cache
        transparently.
    </para></note>
    <sect2 id="zend.cache.factory">
        <title>The Zend_Cache Factory Method</title>
        <para>
            A good way to build a usable instance of a <classname>Zend_Cache</classname> Frontend is
            given in the following example :
        </para>
        <programlisting language="php"><![CDATA[
// We choose a backend (for example 'File' or 'Sqlite'...)
$backendName = '[...]';

// We choose a frontend (for example 'Core', 'Output', 'Page'...)
$frontendName = '[...]';

// We set an array of options for the choosen frontend
$frontendOptions = array([...]);

// We set an array of options for the choosen backend
$backendOptions = array([...]);

// We create an instance of Zend_Cache
// (of course, the two last arguments are optional)
$cache = Zend_Cache::factory($frontendName,
                             $backendName,
                             $frontendOptions,
                             $backendOptions);
]]></programlisting>
        <para>
            In the following examples we will assume that the <varname>$cache</varname> variable
            holds a valid, instantiated frontend as shown and that you understand how to pass
            parameters to your chosen backends.
        </para>
        <note><para>
            Always use <methodname>Zend_Cache::factory()</methodname> to get frontend instances.
            Instantiating frontends and backends yourself will not work as expected.
        </para></note>
    </sect2>

    <sect2 id="zend.cache.tags">
        <title>Tagging Records</title>
        <para>
            Tags are a way to categorize cache records. When you save a cache with the
            <methodname>save()</methodname> method, you can set an array of tags to apply for this
            record. Then you will be able to clean all cache records tagged with a given tag (or
            tags):
        </para>
        <programlisting language="php"><![CDATA[
$cache->save($huge_data, 'myUniqueID', array('tagA', 'tagB', 'tagC'));
]]></programlisting>
        <note><para>
            note than the <methodname>save()</methodname> method accepts an optional fourth
            argument: <varname>$specificLifetime</varname> (if != <constant>FALSE</constant>,
            it sets a specific lifetime for this particular cache record)
        </para></note>
    </sect2>
    <sect2 id="zend.cache.clean">
        <title>Cleaning the Cache</title>
        <para>
            To remove or invalidate in particular cache id, you can use the
            <methodname>remove()</methodname> method :
        </para>
        <programlisting language="php"><![CDATA[
$cache->remove('idToRemove');
]]></programlisting>
        <para>
            To remove or invalidate several cache ids in one operation, you can use the
            <methodname>clean()</methodname> method. For example to remove all cache records :
        </para>
        <programlisting language="php"><![CDATA[
// clean all records
$cache->clean(Zend_Cache::CLEANING_MODE_ALL);

// clean only outdated
$cache->clean(Zend_Cache::CLEANING_MODE_OLD);
]]></programlisting>
        <para>
            If you want to remove cache entries matching the tags 'tagA' and 'tagC':
        </para>
        <programlisting language="php"><![CDATA[
$cache->clean(
    Zend_Cache::CLEANING_MODE_MATCHING_TAG,
    array('tagA', 'tagC')
);
]]></programlisting>
        <para>
            If you want to remove cache entries not matching the tags 'tagA' or 'tagC':
        </para>
        <programlisting language="php"><![CDATA[
$cache->clean(
    Zend_Cache::CLEANING_MODE_NOT_MATCHING_TAG,
    array('tagA', 'tagC')
);
]]></programlisting>
        <para>
            If you want to remove cache entries matching the tags 'tagA' or 'tagC':
        </para>
        <programlisting language="php"><![CDATA[
$cache->clean(
    Zend_Cache::CLEANING_MODE_MATCHING_ANY_TAG,
    array('tagA', 'tagC')
);
]]></programlisting>
        <para>
            Available cleaning modes are: <constant>CLEANING_MODE_ALL</constant>,
            <constant>CLEANING_MODE_OLD</constant>, <constant>CLEANING_MODE_MATCHING_TAG</constant>,
            <constant>CLEANING_MODE_NOT_MATCHING_TAG</constant> and
            <constant>CLEANING_MODE_MATCHING_ANY_TAG</constant>. The latter are, as their names
            suggest, combined with an array of tags in cleaning operations.
        </para>
    </sect2>
</sect1>
<!--
vim:se ts=4 sw=4 et:
-->