gcc/libstdc++-v3/docs/html/ext/pb_assoc/concepts.html

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<html>
<head>
<title>Concepts</title>
<meta name="GENERATOR" content="Microsoft Visual Studio .NET 7.1">
<meta name="vs_targetSchema" content="http://schemas.microsoft.com/intellisense/ie5">
</head>
<body bgcolor = "white">
<h1>Concepts</h1>

<p>
Following are some concepts used throughout the documentation.
</p>

<ol>
    <li><a href = "#concepts_null_policies">Null Policy Classes</a></li>
    <li><a href = "#concepts_find_and_range_iterators">Find and Range Iterators</a></li>
    <li><a href = "#concepts_mapping_levels">Mapping Levels</a></li>
</ol>

<h2><a name = "concepts_null_policies">Null Policy Classes</a></h2>

<p>
    Associative containers are typically parameterized by various policies.
For example, a hash-based associative
container is parameterized by a hash-functor, transforming each key into an non-negative numerical type. Each such value is then further mapped into a position within the table.
The mapping of a key into a position within the table is therefore a two-step process.
</p>

<p>
In some
cases, instantiations are <i>redundant</i>. For example, when the keys are integers, it is possible to use a <i>redundant</i>
hash policy, which transforms each key into its value.
</p>

<p>
    In some other cases, these policies are <i>irrelevent</i>. For example,
a hash-based associative container might transform keys into positions within
a table by a different method than the two-step method described above. In such a case, the hash functor is simply irrelevent.
</p>

<p>
	<tt>pb_assoc</tt> uses special pre-defined &quot;null policies&quot; classes
for these cases. Some null policies in <tt>pb_assoc</tt>
are:
</p>
<ol>
	<li <a href = "null_data_type.html"><tt>null_data_type</tt></a></li>
    <li><a href = "null_node_updator.html"><tt>null_node_updator</tt></a></li>
    <li><a href = "null_hash_fn.html"><tt>null_hash_fn</tt></a></li>
    <li><a href = "null_probe_fn.html"><tt>null_probe_fn</tt></a></li>
</ol>

<p>
A &quot;set&quot; in <tt>pb_assoc</tt> is an associative container with its <tt>Data_Parameter</tt> instantiated by
<a href = "null_data_type.html"><tt>null_data_type</tt></a>.
<a href = "tree_based_containers.html#node_invariants.html">Tree-Based Containers::Node Invariants</a>
explains another case where a null policy is needed.
</p>


<h2><a name = "concepts_find_and_range_iterators">Find and Range Methods and Iterators</a></h2>

<p>
    Associative containers allow access to their elements via iterators. <i>E.g.</i>,
<tt>find</tt> returns an iterator to an element with a given key and
<tt>begin</tt> returns an iterator to the first element in the container.
</p>

<p>
In general, there are two types of methods: <i>find types</i>, and <i>range types</i>.
    Find-type
methods return iterators corresponding to elements which have been found in some sense, as
the container searched for them in order to access them (<i>i.e.</i>, via the
<tt>find</tt> method) or searched for their location in order to insert them
(<i>i.e.</i>, via the <tt>insert</tt> method). Range-type methods return iterators
which can be used to traverse the range of all stored elements, (<i>i.e.</i>, via the
<tt>begin</tt> and <tt>end</tt> methods).
</p>

<p>Correspondingly, in <tt>pb_assoc</tt> there are two types of iterators: <i>find type</i>
iterators are returned by find methods, and range iterators are returned by range methods. For example,
if <tt>T</tt> is any associative container with integer keys, and <tt>t</tt>
is a container of type <tt>T</tt>,
then the following snippet is valid:
</p>

<pre>
<b>typename</b> T::find_iterator it0 = t.find(3);
<b>typename</b> T::const_find_iterator it0 = t.find(3);

<b>typename</b> T::iterator it0 = t.begin();
<b>typename</b> T::const_iterator it0 = t.begin();
</pre>


<p>
    This is motivated and explained further in
<a href = "ds_gen.html#find_range">Data-Structure Genericity::Find-Type and Range-Type Methods and Iterators</a>, which also explains the relationship between find-type and range-type iterators.
</p>

<h2><a href = "#concepts_mapping_levels">Mapping Levels</a></h2>

<p>
	In <tt>pb_assoc</tt> &quot;multimaps&quot; are
&quot;maps&quot; of &quot;sets&quot;. While this design allows efficient
operations, it makes for cumbersome use at points. For example a
&quot;multimap&quot; of integers to characters does not
directly support <tt>inser(std::make_pair(2, 'b')</tt>, since 2 is mapped
to a &quot;set&quot; of characters, and not to a character.
</p>

<p>
	Consequently, <tt>pb_assoc</tt> contains a rebind-like mechanism so that
containers can support such operations. To dispel ambiguity, container types are
assigned mapping levels. &quot;Maps&quot; and &quot;sets&quot; have
a mapping level 1, since they use a single association level. The &quot;multimap&quot;
above has a mapping level 2, since it uses two association levels: one for integers, and one for characters. The rebind mechanism can be used to alter the association level. This is described in
<a href = "ms_gen.html">Mapping Semantics</a>.
</p>

</body>
</html>