blob: 5443bd4890a5773ee4bcdeb32df473345db32339 [file] [log] [blame]
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<title>Header boost/utility.hpp Documentation</title>
</head>
<body bgcolor="#FFFFFF" text="#000000">
<h1><img src="../../boost.png" alt="boost.png (6897 bytes)" align="center" WIDTH="277" HEIGHT="86">Header
<a href="../../boost/utility.hpp">boost/utility.hpp</a></h1>
<p>The entire contents of the header <code><a href="../../boost/utility.hpp">&lt;boost/utility.hpp&gt;</a></code>
are in <code>namespace boost</code>.</p>
<h2>Contents</h2>
<ul>
<li>
Class templates supporting the <a href="doc/html/base_from_member.html">
base-from-member idiom</a></li>
<li>
Function templates <a href="../core/doc/html/core/checked_delete.html">checked_delete() and
checked_array_delete()</a> (moved to the Boost.Core library)</li>
<li>
Function templates <a href="#functions_next_prior">next() and prior()</a></li>
<li>
Class <a href="../core/doc/html/core/noncopyable.html">noncopyable</a> (moved to the Boost.Core library)</li>
<li>
Function template <a href="../core/doc/html/core/addressof.html">addressof()</a> (moved to the Boost.Core library)</li>
<li>Class template <a href="#result_of">result_of</a></li>
<li>
Macro <a href="#BOOST_BINARY">BOOST_BINARY</a></li>
<li><a href="index.html">Other utilities not part of <code>utility.hpp</code></a></li>
</ul>
<h2>
<a name="functions_next_prior">Function</a> templates next() and prior()</h2>
<p>Certain data types, such as the C++ Standard Library's forward and bidirectional
iterators, do not provide addition and subtraction via operator+() or
operator-().&nbsp; This means that non-modifying computation of the next or
prior value requires a temporary, even though operator++() or operator--() is
provided.&nbsp; It also means that writing code like <code>itr+1</code> inside
a template restricts the iterator category to random access iterators.</p>
<p>The next() and prior() functions provide a simple way around these problems:</p>
<blockquote>
<pre>template &lt;class T&gt;
T next(T x) { return ++x; }
template &lt;class T, class Distance&gt;
T next(T x, Distance n)
{
std::advance(x, n);
return x;
}
template &lt;class T&gt;
T prior(T x) { return --x; }
template &lt;class T, class Distance&gt;
T prior(T x, Distance n)
{
std::advance(x, -n);
return x;
}</pre>
</blockquote>
<p>Usage is simple:</p>
<blockquote>
<pre>const std::list&lt;T&gt;::iterator p = get_some_iterator();
const std::list&lt;T&gt;::iterator prev = boost::prior(p);
const std::list&lt;T&gt;::iterator next = boost::next(prev, 2);</pre>
</blockquote>
<p>The distance from the given iterator should be supplied as an absolute value. For
example, the iterator four iterators prior to the given iterator <code>p</code>
may be obtained by <code>prior(p, 4)</code>.</p>
<p>Contributed by <a href="http://www.boost.org/people/dave_abrahams.htm">Dave Abrahams</a>. Two-argument versions by Daniel Walker.</p>
<h2><a name="result_of">Class template
result_of</a></h2> <p>The class template
<code>result_of</code> helps determine the type of a
call expression. For example, given an lvalue <code>f</code> of
type <code>F</code> and lvalues <code>t1</code>,
<code>t2</code>, ..., <code>t<em>N</em></code> of
types <code>T1</code>, <code>T2</code>, ...,
<code>T<em>N</em></code>, respectively, the type
<code>result_of&lt;F(T1, T2, ...,
T<em>N</em>)&gt;::type</code> defines the result type
of the expression <code>f(t1, t2,
...,t<em>N</em>)</code>. This implementation permits
the type <code>F</code> to be a function pointer,
function reference, member function pointer, or class
type. By default, <em>N</em> may be any value between 0 and
16. To change the upper limit, define the macro
<code>BOOST_RESULT_OF_NUM_ARGS</code> to the maximum
value for <em>N</em>. Class template <code>result_of</code>
resides in the header <code>&lt;<a
href="../../boost/utility/result_of.hpp">boost/utility/result_of.hpp</a>&gt;</code>.</p>
<p>If your compiler's support for <code>decltype</code> is
adequate, <code>result_of</code> automatically uses it to
deduce the type of the call expression, in which case
<code>result_of&lt;F(T1, T2, ...,
T<em>N</em>)&gt;::type</code> names the type
<code>decltype(boost::declval&lt;F&gt;()(boost::declval&lt;T1&gt;(),
boost::declval&lt;T2&gt;(), ...,
boost::declval&lt;T<em>N</em>&gt;()))</code>, as in the
following example.</p>
<blockquote>
<pre>struct functor {
template&lt;class T&gt;
T operator()(T x)
{
return x;
}
};
typedef boost::result_of&lt;
functor(int)
&gt;::type type; // type is int</pre>
</blockquote>
<p>You can test whether <code>result_of</code> is using
<code>decltype</code> by checking if the macro
<code>BOOST_RESULT_OF_USE_DECLTYPE</code> is defined after
including <code>result_of.hpp</code>. You can also force
<code>result_of</code> to use <code>decltype</code> by
defining <code>BOOST_RESULT_OF_USE_DECLTYPE</code> prior
to including <code>result_of.hpp</code>.</p>
<p>If <code>decltype</code> is not used,
then automatic result type deduction of function
objects is not possible. Instead, <code>result_of</code>
uses the following protocol to allow the programmer to
specify a type. When <code>F</code> is a class type with a
member type <code>result_type</code>,
<code>result_of&lt;F(T1, T2, ...,
T<em>N</em>)&gt;::type</code> is
<code>F::result_type</code>. When <code>F</code> does
not contain <code>result_type</code>,
<code>result_of&lt;F(T1, T2, ...,
T<em>N</em>)&gt;::type</code> is <code>F::result&lt;F(T1,
T2, ..., T<em>N</em>)&gt;::type</code> when
<code><em>N</em> &gt; 0</code> or <code>void</code>
when <code><em>N</em> = 0</code>. Note that it is the
responsibility of the programmer to ensure that
function objects accurately advertise their result
type via this protocol, as in the following
example.</p>
<blockquote>
<pre>struct functor {
template&lt;class&gt; struct result;
template&lt;class F, class T&gt;
struct result&lt;F(T)&gt; {
typedef T type;
};
template&lt;class T&gt;
T operator()(T x)
{
return x;
}
};
typedef boost::result_of&lt;
functor(int)
&gt;::type type; // type is int</pre>
</blockquote>
<p>Since <code>decltype</code> is a new language
feature recently standardized in C++11,
if you are writing a function object
to be used with <code>result_of</code>, for
maximum portability, you might consider following
the above protocol even if your compiler has
proper <code>decltype</code> support. If you wish to continue to
use the protocol on compilers that
support <code>decltype</code>, there are two options:
You can use <code>boost::tr1_result_of</code>, which is also
defined in <code>&lt;<a href="../../boost/utility/result_of.hpp">boost/utility/result_of.hpp</a>&gt;</code>.
Alternatively, you can define the macro
<code>BOOST_RESULT_OF_USE_TR1</code>, which causes
<code>result_of</code> to use the protocol described
above instead of <code>decltype</code>. If you choose to
follow the protocol, take care to ensure that the
<code>result_type</code> and
<code>result&lt;&gt;</code> members accurately
represent the return type of
<code>operator()</code> given a call expression.</p>
<p>Additionally, <code>boost::result_of</code>
provides a third mode of operation, which some users
may find convenient. When
<code>BOOST_RESULT_OF_USE_TR1_WITH_DECLTYPE_FALLBACK</code>
is defined, <code>boost::result_of</code> behaves as
follows. If the function object has a member
type <code>result_type</code> or member
template <code>result&lt;&gt;</code>, then
<code>boost::result_of</code> will use the TR1
protocol. Otherwise,
<code>boost::result_of</code> will
use <code>decltype</code>. Using TR1 with
a <code>declytpe</code> fallback may workaround
certain problems at the cost of portability. For
example:
<ul>
<li>Deficient compiler: If your code
requires <code>boost::result_of</code> to work
with incomplete return types but your
compiler's <code>decltype</code> implementation
does not support incomplete return types, then you
can use the TR1 protocol as a workaround. Support
for incomplete return types was added late in the
C++11 standardization process
(see <a href="http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2011/n3276.pdf">N3276</a>)
and is not implemented by some compilers.</li>
<li>Deficient legacy code: If your existing TR1
function object advertises a different type than
the actual result type deduced
by <code>decltype</code>, then using TR1 with a
<code>decltype</code> fallback will allow you to
work with both your existing TR1 function objects
and new C++11 function object. This situation
could occur if your legacy function objects
misused the TR1 protocol. See the documentation on
known <a href="#result_of_tr1_diff">differences</a>
between <code>boost::result_of</code> and TR1.</li>
</ul>
<a name="BOOST_NO_RESULT_OF"></a>
<p>This implementation of <code>result_of</code>
requires class template partial specialization, the
ability to parse function types properly, and support
for SFINAE. If <code>result_of</code> is not supported
by your compiler, including the header
<code>boost/utility/result_of.hpp</code> will
define the macro <code>BOOST_NO_RESULT_OF</code>.</p>
<p>For additional information
about <code>result_of</code>, see the C++ Library
Technical Report,
<a href="http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2005/n1836.pdf">N1836</a>,
or, for motivation and design rationale,
the <code>result_of</code> <a href="http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2003/n1454.html">proposal</a>.</p>
<a name="result_of_guidelines">
<h3>Usage guidelines for boost::result_of</h3>
</a>
<p>The following are general suggestions about when
and how to use <code>boost::result_of</code>.</p>
<ol>
<li> If you are targeting C++11 and are not concerned
about portability to non-compliant compilers or
previous versions of the standard, then use
<code>std::result_of</code>. If <code>std::result_of</code>
meets your needs, then there's no reason to stop using
it.</li>
<li> If you are targeting C++11 but may port your code
to legacy compilers at some time in the future, then
use <code>boost::result_of</code> with
<code>decltype</code>. When <code>decltype</code> is
used <code>boost::result_of</code>
and <code>std::result_of</code> are usually
interchangeable. See the documentation on
known <a href="#result_of_cxx11_diff">differences</a>
between boost::result_of and C++11 result_of.</li>
<li> If compiler portability is required,
use <code>boost::result_of</code> with the TR1 protocol.</li>
</ol>
<p>Regardless of how you
configure <code>boost::result_of</code>, it is
important to bear in mind that the return type of a
function may change depending on its arguments, and
additionally, the return type of a member function may
change depending on the cv-qualification of the
object. <code>boost::result_of</code> must be passed
the appropriately cv-qualified types in order to
deduce the corresponding return type. For example:
<blockquote>
<pre>struct functor {
int& operator()(int);
int const& operator()(int) const;
float& operator()(float&);
float const& operator()(float const&);
};
typedef boost::result_of&lt;
functor(int)
&gt;::type type1; // type1 is int &
typedef boost::result_of&lt;
const functor(int)
&gt;::type type2; // type2 is int const &
typedef boost::result_of&lt;
functor(float&)
&gt;::type type3; // type3 is float &
typedef boost::result_of&lt;
functor(float const&)
&gt;::type type4; // type4 is float const &</pre>
</blockquote>
<a name="result_of_tr1_protocol_guidelines">
<h3>Usage guidelines for the TR1 result_of protocol</h3>
</a>
<p>On compliant C++11
compilers, <code>boost::result_of</code> can
use <code>decltype</code> to deduce the type of any
call expression, including calls to function
objects. However, on pre-C++11 compilers or on
compilers without adequate decltype support,
additional scaffolding is needed from function
objects as described above. The following are
suggestions about how to use the TR1 protocol.</p>
<ul>
<li>When the return type does not depend on the
argument types or the cv-qualification of the
function object, simply
define <code>result_type</code>. There is no need
to use the <code>result</code> template unless the
return type varies.</li>
<li>Use the protocol specified type when defining
function prototypes. This can help ensure the
actual return type does not get out of sync with
the protocol specification. For example:
<blockquote>
<pre>struct functor {
typedef int result_type;
result_type operator()(int);
};</pre>
</blockquote> </li>
<li>Always specify the <code>result</code>
specialization near the corresponding
<code>operator()</code> overload. This can make it
easier to keep the specializations in sync with the
overloads. For example:
<blockquote>
<pre>struct functor {
template&lt;class&gt; struct result;
template&lt;class F&gt;
struct result&lt;F(int)&gt; {
typedef int& type;
};
result&lt;functor(int)&gt;::type operator()(int);
template&lt;class F&gt;
struct result&lt;const F(int)&gt; {
typedef int const& type;
};
result&lt;const functor(int)&gt;::type operator()(int) const;
};</pre>
</blockquote> </li>
<li>Use type transformations to simplify
the <code>result</code> template specialization. For
example, the following uses
<a href="../type_traits/doc/html/index.html">Boost.TypeTraits</a>
to specialize the <code>result</code> template for
a single <code>operator()</code> that can be called on
both a const and non-const function object with
either an lvalue or rvalue argument.
<blockquote>
<pre>struct functor {
template&lt;class&gt; struct result;
template&lt;class F, class T&gt;
struct result&lt;F(T)&gt;
: boost::remove_cv&lt;
typename boost::remove_reference&lt;T&gt;::type
&gt;
{};
template&lt;class T&gt;
T operator()(T const&amp; x) const;
};</pre>
</blockquote></li>
</ul>
<a name="result_of_tr1_diff">
<h3>Known differences between boost::result_of and TR1 result_of</h3>
</a>
When using <code>decltype</code>, <code>boost::result_of</code>
ignores the TR1 protocol and instead deduces the
return type of function objects directly
via <code>decltype</code>. In most situations, users
will not notice a difference, so long as they use the
protocol correctly. The following are situations in
which the type deduced
by <code>boost::result_of</code> is known to differ depending on
whether <code>decltype</code> or the TR1 protocol is
used.
<ul>
<li> TR1 protocol misusage
<p>When using the TR1
protocol, <code>boost::result_of</code> cannot
detect whether the actual type of a call to a
function object is the same as the type specified
by the protocol, which allows for the possibility
of inadvertent mismatches between the specified
type and the actual type. When
using <code>decltype</code>, these subtle bugs
may result in compilation errors. For example:</p>
<blockquote>
<pre>struct functor {
typedef short result_type;
int operator()(short);
};
#ifdef BOOST_RESULT_OF_USE_DECLTYPE
BOOST_STATIC_ASSERT((
boost::is_same&lt;boost::result_of&lt;functor(short)&gt;::type, int&gt;::value
));
#else
BOOST_STATIC_ASSERT((
boost::is_same&lt;boost::result_of&lt;functor(short)&gt;::type, short&gt;::value
));
#endif</pre>
</blockquote>
<p>Note that the user can
force <code>boost::result_of</code> to use the TR1
protocol even on platforms that
support <code>decltype</code> by
defining <code>BOOST_RESULT_OF_USE_TR1</code>.</p></li>
<li> Nullary function objects
<p>When using the TR1 protocol, <code>boost::result_of</code>
cannot always deduce the type of calls to
nullary function objects, in which case the
type defaults to void. When using <code>decltype</code>,
<code>boost::result_of</code> always gives the actual type of the
call expression. For example:</p>
<blockquote>
<pre>struct functor {
template&lt;class&gt; struct result {
typedef int type;
};
int operator()();
};
#ifdef BOOST_RESULT_OF_USE_DECLTYPE
BOOST_STATIC_ASSERT((
boost::is_same&lt;boost::result_of&lt;functor()&gt;::type, int&gt;::value
));
#else
BOOST_STATIC_ASSERT((
boost::is_same&lt;boost::result_of&lt;functor()&gt;::type, void&gt;::value
));
#endif</pre>
</blockquote>
<p>Note that there are some workarounds for the
nullary function problem. So long as the return
type does not vary,
<code>result_type</code> can always be used to
specify the return type regardless of arity. If the
return type does vary, then the user can
specialize <code>boost::result_of</code> itself for
nullary calls.</p></li>
<li> Non-class prvalues and cv-qualification
<p>When using the TR1
protocol, <code>boost::result_of</code> will
report the cv-qualified type specified
by <code>result_type</code> or
the <code>result</code> template regardless of
the actual cv-qualification of the call
expression. When using
<code>decltype</code>, <code>boost::result_of</code>
will report the actual type of the call expression,
which is not cv-qualified when the expression is a
non-class prvalue. For example:</p>
<blockquote>
<pre>struct functor {
template&lt;class&gt; struct result;
template&lt;class F, class T&gt; struct result&lt;F(const T)&gt; {
typedef const T type;
};
const short operator()(const short);
int const & operator()(int const &);
};
// Non-prvalue call expressions work the same with or without decltype.
BOOST_STATIC_ASSERT((
boost::is_same&lt;
boost::result_of&lt;functor(int const &)&gt;::type,
int const &
::value
));
// Non-class prvalue call expressions are not actually cv-qualified,
// but only the decltype-based result_of reports this accurately.
#ifdef BOOST_RESULT_OF_USE_DECLTYPE
BOOST_STATIC_ASSERT((
boost::is_same&lt;
boost::result_of&lt;functor(const short)&gt;::type,
short
::value
));
#else
BOOST_STATIC_ASSERT((
boost::is_same&lt;
boost::result_of&lt;functor(const short)&gt;::type,
const short
::value
));
#endif</pre>
</blockquote></li>
</ul>
<a name="result_of_cxx11_diff">
<h3>Known differences between boost::result_of and C++11 result_of</h3>
</a>
<p>When using <code>decltype</code>, <code>boost::result_of</code>
implements most of the C++11 result_of
specification. One known exception is that
<code>boost::result_of</code> does not implement the
requirements regarding pointers to member data.</p>
<p>Created by Doug Gregor. Contributions from Daniel Walker, Eric Niebler, Michel Morin and others</p>
<h2><a name="BOOST_BINARY">Macro BOOST_BINARY</a></h2>
<p>The macro <code>BOOST_BINARY</code> is used for the
representation of binary literals. It takes as an argument
a binary number arranged as an arbitrary amount of 1s and 0s in
groupings of length 1 to 8, with groups separated
by spaces. The type of the literal yielded is determined by
the same rules as those of hex and octal
literals (<i>2.13.1p1</i>). By implementation, this macro
expands directly to an octal literal during preprocessing, so
there is no overhead at runtime and the result is useable in
any place that an octal literal would be.</p>
<p>In order to directly support binary literals with suffixes,
additional macros of the form BOOST_BINARY_XXX are also
provided, where XXX is a standard integer suffix in all capital
letters. In addition, LL and ULL suffixes may be used for representing
long long and unsigned long long types in compilers which provide
them as an extension.</p>
<p>The BOOST_BINARY family of macros resides in the header
<a
href="../../boost/utility/binary.hpp">&lt;boost/utility/binary.hpp&gt;</a>
which is automatically included by
<a
href="../../boost/utility.hpp">&lt;boost/utility.hpp&gt;</a>.
<p>Contributed by Matt Calabrese.</p><p>
</p><h3>Example</h3>
<blockquote>
<pre>
void foo( int );
void foo( unsigned long );
void bar()
{
int value1 = BOOST_BINARY( 100 111000 01 1 110 );
unsigned long value2 = BOOST_BINARY_UL( 100 001 ); // unsigned long
long long value3 = BOOST_BINARY_LL( 11 000 ); // long long if supported
assert( BOOST_BINARY( 10010 )
& BOOST_BINARY( 11000 )
== BOOST_BINARY( 10000 )
);
foo( BOOST_BINARY( 1010 ) ); // calls the first foo
foo( BOOST_BINARY_LU( 1010 ) ); // calls the second foo
}
</pre></blockquote>
<hr>
<p>Revised&nbsp; <!--webbot bot="Timestamp" S-Type="EDITED" S-Format="%d %B, %Y" startspan
-->04 September, 2008<!--webbot bot="Timestamp" endspan i-checksum="39369"
-->
</p>
<p>&copy; Copyright Beman Dawes 1999-2003.</p>
<p>Distributed under the Boost Software License, Version 1.0. See
<a href="http://www.boost.org/LICENSE_1_0.txt">www.boost.org/LICENSE_1_0.txt</a></p>
</body>
</html>