numeric_conversion/doc/conversion_traits.qbk
2009-02-23 18:39:32 +00:00

273 lines
8.2 KiB
Plaintext

[/
Boost.Optional
Copyright (c) 2003-2007 Fernando Luis Cacciola Carballal
Distributed under the Boost Software License, Version 1.0.
(See accompanying file LICENSE_1_0.txt or copy at
http://www.boost.org/LICENSE_1_0.txt)
]
[section conversion_traits<> traits class]
[section Types]
[section enumeration int_float_mixture_enum]
namespace boost { namespace numeric {
enum int_float_mixture_enum
{
integral_to_integral
,integral_to_float
,float_to_integral
,float_to_float
} ;
} } // namespace boost::numeric
[endsect]
[section enumeration sign_mixture_enum]
namespace boost { namespace numeric {
enum sign_mixture_enum
{
unsigned_to_unsigned
,signed_to_signed
,signed_to_unsigned
,unsigned_to_signed
} ;
} } // namespace boost::numeric
[endsect]
[section enumeration udt_builtin_mixture_enum]
namespace boost { namespace numeric {
enum udt_builtin_mixture_enum
{
builtin_to_builtin
,builtin_to_udt
,udt_to_builtin
,udt_to_udt
} ;
} } // namespace boost::numeric
[endsect]
[section template class int_float_mixture<>]
namespace boost { namespace numeric {
template <class T, class S>
struct int_float_mixture : mpl::integral_c<int_float_mixture_enum, impl-def-value> {} ;
} } // namespace boost::numeric
Classifying `S` and `T` as either integral or float, this __MPL_INTEGRAL_CONSTANT__
indicates the combination of these attributes.
Its `::value` is of enumeration type
[link boost_numericconversion.conversion_traits___traits_class.types.enumeration_int_float_mixture_enum `boost::numeric::int_float_mixture_enum`]
[endsect]
[section template class sign_mixture<>]
namespace boost { namespace numeric {
template <class T, class S>
struct sign_mixture : mpl::integral_c<sign_mixture_enum, impl-def-value> {} ;
} } // namespace boost::numeric
Classifying `S` and `T` as either signed or unsigned, this __MPL_INTEGRAL_CONSTANT__
indicates the combination of these attributes.
Its `::value` is of enumeration type
[link boost_numericconversion.conversion_traits___traits_class.types.enumeration_sign_mixture_enum `boost::numeric::sign_mixture_enum`]
[endsect]
[section template class udt_builtin_mixture<>]
namespace boost { namespace numeric {
template <class T, class S>
struct udt_builtin_mixture : mpl::integral_c<udt_builtin__mixture_enum, impl-def-value> {} ;
} } // namespace boost::numeric
Classifying `S` and `T` as either user-defined or builtin, this __MPL_INTEGRAL_CONSTANT__
indicates the combination of these attributes.
Its `::value` is of enumeration type
[link boost_numericconversion.conversion_traits___traits_class.types.enumeration_udt_builtin_mixture_enum `boost::numeric::udt_builtin_mixture_enum`]
[endsect]
[section template class is_subranged<>]
namespace boost { namespace numeric {
template <class T, class S>
struct is_subranged : mpl::bool_<impl-def-value> {} ;
} } // namespace boost::numeric
Indicates if the range of the target type `T` is a subset of the range of the source
type `S`. That is: if there are some source values which fall out of the
Target type's range.
It is a boolean __MPL_INTEGRAL_CONSTANT__.
It does not indicate if a particular conversion is effectively out of range;
it indicates that some conversion might be out of range because not all the
source values are representable as Target type.
[endsect]
[section template class conversion_traits<>]
namespace boost { namespace numeric {
template <class T, class S>
struct conversion_traits
{
mpl::integral_c<int_float_mixture_enum , ...> int_float_mixture ;
mpl::integral_c<sign_mixture_enum , ...> sign_mixture;
mpl::integral_c<udt_builtin_mixture_enum, ...> udt_builtin_mixture ;
mpl::bool_<...> subranged ;
mpl::bool_<...> trivial ;
typedef T target_type ;
typedef S source_type ;
typedef ... argument_type ;
typedef ... result_type ;
typedef ... supertype ;
typedef ... subtype ;
} ;
} } // namespace numeric, namespace boost
This traits class indicates some properties of a ['numeric conversion] direction:
from a source type `S` to a target type `T`. It does not indicate the properties
of a ['specific] conversion, but of the conversion direction. See
[link boost_numericconversion.definitions.subranged_conversion_direction__subtype_and_supertype Definitions] for details.
The traits class provides the following __MPL_INTEGRAL_CONSTANT__\s of enumeration
type. They express the combination of certain attributes of the Source and
Target types (thus they are call mixture):
[table
[[ ][ ]]
[[[*int_float_mixture ]][
Same as given by the traits class
[link boost_numericconversion.conversion_traits___traits_class.types.template_class_int_float_mixture__ int_float_mixture]
]]
[[[*sign_mixture ]][
Same as given by the traits class
[link boost_numericconversion.conversion_traits___traits_class.types.template_class_sign_mixture__ sign_mixture]
]]
[[[*udt_builtin_mixture ]]
[Same as given by the traits class
[link boost_numericconversion.conversion_traits___traits_class.types.template_class_udt_builtin_mixture__ udt_builtin_mixture]
]]
]
The traits class provides the following __MPL_INTEGRAL_CONSTANT__\s of boolean type
which indicates indirectly the relation between the Source and Target ranges
(see [link boost_numericconversion.definitions.range_and_precision Definitions] for details).
[table
[[ ][ ]]
[[subranged ][
Same as given by [link boost_numericconversion.conversion_traits___traits_class.types.template_class_is_subranged__ is_subranged]
]]
[[trivial][
Indicates if both Source and Target, [_without cv-qualifications], are the same type.
Its `::value` is of boolean type.
]]
]
The traits class provides the following types. They are the Source and Target types classified
and qualified for different purposes.
[table
[[ ][ ]]
[[[*target_type]][
The template parameter `T` without cv-qualifications
]]
[[[*source_type]][
The template parameter `S` without cv-qualifications
]]
[[[*argument_type]][
This type is either source_type or `source_type const&`.
It represents the optimal argument type for the
[link boost_numericconversion.converter___function_object converter] member functions.
If S is a built-in type, this is `source_type`, otherwise, this is `source_type const&`.
]]
[[[*result_type]][
This type is either target_type or target_type const&
It represents the return type of the
[link boost_numericconversion.converter___function_object converter] member functions.
If `T==S`, it is `target_type const&`, otherwise, it is `target_type`.
]]
[[[*supertype]][
If the conversion is subranged, it is `source_type`, otherwise, it is `target_type`
]]
[[[*subtype]][
If the conversion is subranged, it is `target_type`, otherwise, it is `source_type`
]]
]
[endsect]
[endsect]
[section Examples]
#include <cassert>
#include <typeinfo>
#include <boost/numeric/conversion/conversion_traits.hpp>
int main()
{
// A trivial conversion.
typedef boost::numeric::conversion_traits<short,short> Short2Short_Traits ;
assert ( Short2Short_Traits::trivial::value ) ;
// A subranged conversion.
typedef boost::numeric::conversion_traits<double,unsigned int> UInt2Double_Traits ;
assert ( UInt2Double_Traits::int_float_mixture::value == boost::numeric::integral_to_float ) ;
assert ( UInt2Double_Traits::sign_mixture::value == boost::numeric::unsigned_to_signed ) ;
assert ( !UInt2Double_Traits::subranged::value ) ;
assert ( typeid(UInt2Double_Traits::supertype) == typeid(double) ) ;
assert ( typeid(UInt2Double_Traits::subtype) == typeid(unsigned int) ) ;
// A doubly subranged conversion.
assert ( (boost::numeric::conversion_traits<short, unsigned short>::subranged::value) );
assert ( (boost::numeric::conversion_traits<unsigned short, short>::subranged::value) );
return 0;
}
[endsect]
[endsect]