// $Id$
//
// Copyright (C) 2007
// Fraunhofer Institut fuer offene Kommunikationssysteme (FOKUS)
// Kompetenzzentrum fuer Satelitenkommunikation (SatCom)
// Stefan Bund <g0dil@berlios.de>
//
// This program is free software; you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation; either version 2 of the License, or
// (at your option) any later version.
//
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with this program; if not, write to the
// Free Software Foundation, Inc.,
// 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
/** \file
\brief mpl public header */
#ifndef HH_mpl_
#define HH_mpl_ 1
// Custom includes
#include "../config.hh"
//#include "mpl.mpp"
#include "mpl.ih"
///////////////////////////////hh.p////////////////////////////////////////
namespace senf {
namespace mpl {
/** \defgroup senfmpl Template meta programming helpers
*/
/** \brief Marker class for empty default values etc.
This is like Boosts \c boost::mpl::na just an empty class used as template default argument
to mark missing arguments
\note Don't use this as an empty base class. We may add some informative members to this.
*/
struct nil {};
/** \brief Return-value type used to implement overload selection
The senf::mpl::rv type is used together with \ref SENF_MPL_RV() to select template
specializations based on a set of overloads:
\code
template <unsigned _> struct select {};
// Case 0
template <>
struct select<0> {
static bool const has_int_value = true;
void frobble();
};
template <class T>
senf::mpl::rv<0> select_(int, senf::mpl::take_int<T::value> * = 0);
// Case 1
template <>
struct select<1> {
static bool const has_int_value = false;
void dazzle();
};
template <class T>
senf::mpl::rv<1> select_(...);
template <class T>
struct choice : public select<SENF_MPL_RV( select_<T>(0) )> {};
struct A { static const int value = 0; };
struct B {};
choice<A> a; a.frobble();
choice<B> b; b.dazzle();
\endcode
The selection is always based on two components: A selector class specialized for each of
the possible choices and an overloaded function (only signatures, no implementation needed)
to provide the conditions.
When instantiatinv <tt>choice<T></tt>, we forward \a T to the <tt>select_</tt> set of
overloads. Because of <a href="http://en.wikipedia.org/wiki/SFINAE">SFINAE</a>, the overload
set will only contain those instantiations, for which template expansion does not fail.
So, if \a T has an integer \c value member, both \c select_ overloads are ok and the call
<tt>select_<T>(0)</tt> will choose the first (case 0) variant, since the argument \c 0 is
better matched by \c int than by <tt>...</tt>.
However, if \a T does not have an integer \c value member, expansion for the first overload
fails and the overload set only contains the second case.
\ref SENF_MPL_RV() internally uses \c sizeof to find out, \e which overload was selected
and returns the senf::mpl::rv-argument of that overloads return type. For this to work, the
\c select_ functions need not be implemented since no code is generated and \c select_ is
never called.
This number is than forwarded as template argument to \c select which is specialized for
each case. Therefore, <tt>choice\<A\></tt> has a \c frobble() member whereas
<tt>choice\<B\></tt> has a \c dazzle() member.
\see \ref SENF_MPL_RV
\ingroup senfmpl
*/
template <unsigned n>
struct rv {
char _[SENF_MPL_RV_ALIGNMENT][n+1];
};
/** \brief Get return value of overload selector
Used together with senf::mpl::rv to implement overload selection.
\see \ref senf::mpl::rv
\ingroup senfmpl
\hideinitializer
*/
# define SENF_MPL_RV(expr) (sizeof(expr)/SENF_MPL_RV_ALIGNMENT-1)
/** \brief Take an arbitrary unsigned integer template argument
Used together with <a href="http://en.wikipedia.org/wiki/SFINAE">SFINAE</a>: The expression
<tt>take_uint<</tt> \a expr <tt>></tt> is only valid if \a expr is valid and returns a value
convertible to an unsigned integer.
\ingroup senfmpl
*/
template <unsigned long _> struct take_uint {};
/** \brief Take an arbitrary integer template argument
Used together with <a href="http://en.wikipedia.org/wiki/SFINAE">SFINAE</a>: The expression
<tt>take_int<</tt> \a expr <tt>></tt> is only valid if \a expr is valid and returns a value
convertible to an integer.
\ingroup senfmpl
*/
template <long _> struct take_int {};
/** \brief Take an arbitrary type template argument
Used together with <a href="http://en.wikipedia.org/wiki/SFINAE">SFINAE</a>: The expression
<tt>take_class<</tt> \a expr <tt>></tt> is only valid if \a expr is valid and is a type.
\ingroup senfmpl
*/
template <class _> struct take_class {};
/** \brief Define MPL slot
The slot macros \ref SENF_MPL_SLOT_DEF(), \ref SENF_MPL_SLOT_SET() and \ref
SENF_MPL_SLOT_GET() provide a facility to get the last unsigned integer value assigned to
the slot before the current point in the current class.
\code
struct Foo
{
// Define SLOT slot named 'accum' initialized to 0
SENF_MPL_SLOT_DEF(accum, 0);
// Add 2 to 'accum'
SENF_MPL_SLOT_SET(accum, SENF_MPL_SLOT_GET(accum) + 2);
// Multiply 'accum' by 3
SENF_MPL_SLOT_SET(accum, SENF_MPL_SLOT_GET(accum) * 3);
// Define the result as a constant expression. result is now 6
static unsigned result = SENF_MPL_SLOT_GET(accum);
};
\endcode
Of course, it does not make sense to use these macros for simple arithmetic as in the
example. The SENF_MPL_SLOT macros allow to define macros which pass information from one
macro invocation to the next.
\implementation The implementation is based on __LINE__: We check backwards for a value
defined on a previous line. The check is limited to 80 lines backwards.
\ingroup senfmpl
\hideinitializer
*/
# define SENF_MPL_SLOT_DEF(name,value) \
template <class _> \
static senf::mpl::rv<0> _SENF_MPL_SLOT_ ## name (_); \
SENF_MPL_SLOT_SET(name,value)
/** \brief Define MPL slot initialized to 0
This is like \ref SENF_MPL_SLOT_DEF() but initializes the slot to the fixed value 0. The
advantage over \ref SENF_MPL_SLOT_DEF() is, that this macro may be called from an include
file whereas all the other \\c SENF_MPL_SLOT_ macros must always be called from the relevant
file.
*/
# define SENF_MPL_SLOT_DEF_ZERO(name) \
template <class _> \
static senf::mpl::rv<0> _SENF_MPL_SLOT_ ## name (_);
/** \brief Set MPL slot
\see \ref SENF_MPL_SLOT_DEF()
\ingroup senfmpl
\hideinitializer
*/
# define SENF_MPL_SLOT_SET(name,value) \
static senf::mpl::rv<unsigned(value)+1> \
_SENF_MPL_SLOT_ ## name (senf::mpl::take_int<__LINE__>)
/** \brief Get current MPL slot value
\see \ref SENF_MPL_SLOT_DEF()
\ingroup senfmpl
\hideinitializer
*/
# define SENF_MPL_SLOT_GET(name) \
SENF_MPL_SLOT_I_GET(name)
}}
///////////////////////////////hh.e////////////////////////////////////////
//#include "mpl.cci"
//#include "mpl.ct"
//#include "mpl.cti"
#endif
// Local Variables:
// mode: c++
// fill-column: 100
// comment-column: 40
// c-file-style: "senf"
// indent-tabs-mode: nil
// ispell-local-dictionary: "american"
// compile-command: "scons -u test"
// End: