|
|
/////////////////////////////////////////////////////////////////////////////////// /// OpenGL Mathematics (glm.g-truc.net) /// /// Copyright (c) 2005 - 2013 G-Truc Creation (www.g-truc.net) /// Permission is hereby granted, free of charge, to any person obtaining a copy /// of this software and associated documentation files (the "Software"), to deal /// in the Software without restriction, including without limitation the rights /// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell /// copies of the Software, and to permit persons to whom the Software is /// furnished to do so, subject to the following conditions: /// /// The above copyright notice and this permission notice shall be included in /// all copies or substantial portions of the Software. /// /// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR /// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, /// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE /// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER /// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, /// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN /// THE SOFTWARE. /// /// @ref core /// @file glm/core/func_common.inl /// @date 2008-08-03 / 2011-06-15 /// @author Christophe Riccio ///////////////////////////////////////////////////////////////////////////////////
namespace glm{ namespace detail { template <typename genFIType, bool /*signed*/> struct Abs_ {};
template <typename genFIType> struct Abs_<genFIType, true> { GLM_FUNC_QUALIFIER static genFIType get(genFIType const & x) { GLM_STATIC_ASSERT( detail::type<genFIType>::is_float || detail::type<genFIType>::is_int, "'abs' only accept floating-point and integer inputs"); return x >= genFIType(0) ? x : -x; // TODO, perf comp with: *(((int *) &x) + 1) &= 0x7fffffff; } };
template <typename genFIType> struct Abs_<genFIType, false> { GLM_FUNC_QUALIFIER static genFIType get(genFIType const & x) { GLM_STATIC_ASSERT( detail::type<genFIType>::is_uint, "'abs' only accept floating-point and integer inputs"); return x; } }; }//namespace detail
// abs template <typename genFIType> GLM_FUNC_QUALIFIER genFIType abs ( genFIType const & x ) { return detail::Abs_<genFIType, std::numeric_limits<genFIType>::is_signed>::get(x); }
VECTORIZE_VEC(abs)
// sign //Try something like based on x >> 31 to get the sign bit template <typename genFIType> GLM_FUNC_QUALIFIER genFIType sign ( genFIType const & x ) { GLM_STATIC_ASSERT( detail::type<genFIType>::is_float || detail::type<genFIType>::is_int, "'sign' only accept signed inputs");
genFIType result; if(x > genFIType(0)) result = genFIType(1); else if(x < genFIType(0)) result = genFIType(-1); else result = genFIType(0); return result; } VECTORIZE_VEC(sign)
// floor template <> GLM_FUNC_QUALIFIER detail::half floor<detail::half>(detail::half const & x) { return detail::half(::std::floor(float(x))); }
template <typename genType> GLM_FUNC_QUALIFIER genType floor(genType const & x) { GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'floor' only accept floating-point inputs");
return ::std::floor(x); }
VECTORIZE_VEC(floor)
// trunc template <typename genType> GLM_FUNC_QUALIFIER genType trunc(genType const & x) { GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'trunc' only accept floating-point inputs"); return x < 0 ? -floor(-x) : floor(x); }
VECTORIZE_VEC(trunc)
// round template <typename genType> GLM_FUNC_QUALIFIER genType round(genType const& x) { GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'round' only accept floating-point inputs");
if(x < 0) return genType(int(x - genType(0.5))); return genType(int(x + genType(0.5))); }
VECTORIZE_VEC(round)
/* // roundEven template <typename genType> GLM_FUNC_QUALIFIER genType roundEven(genType const& x) { GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'roundEven' only accept floating-point inputs");
return genType(int(x + genType(int(x) % 2))); } */ // roundEven template <typename genType> GLM_FUNC_QUALIFIER genType roundEven(genType const & x) { GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'roundEven' only accept floating-point inputs"); int Integer = int(x); genType IntegerPart = genType(Integer); genType FractionalPart = fract(x);
if(FractionalPart > genType(0.5) || FractionalPart < genType(0.5)) { return round(x); } else if((Integer % 2) == 0) { return IntegerPart; } else if(x <= genType(0)) // Work around... { return IntegerPart - 1; } else { return IntegerPart + 1; } //else // Bug on MinGW 4.5.2 //{ // return mix(IntegerPart + genType(-1), IntegerPart + genType(1), x <= genType(0)); //} } VECTORIZE_VEC(roundEven)
// ceil template <typename genType> GLM_FUNC_QUALIFIER genType ceil(genType const & x) { GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'ceil' only accept floating-point inputs");
return ::std::ceil(x); }
VECTORIZE_VEC(ceil)
// fract template <typename genType> GLM_FUNC_QUALIFIER genType fract ( genType const & x ) { GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'fract' only accept floating-point inputs");
return x - ::std::floor(x); }
VECTORIZE_VEC(fract)
// mod template <typename genType> GLM_FUNC_QUALIFIER genType mod ( genType const & x, genType const & y ) { GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'mod' only accept floating-point inputs");
return x - y * floor(x / y); }
VECTORIZE_VEC_SCA(mod) VECTORIZE_VEC_VEC(mod)
// modf template <typename genType> GLM_FUNC_QUALIFIER genType modf ( genType const & x, genType & i ) { GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'modf' only accept floating-point inputs");
return std::modf(x, &i); }
template <typename valType> GLM_FUNC_QUALIFIER detail::tvec2<valType> modf ( detail::tvec2<valType> const & x, detail::tvec2<valType> & i ) { return detail::tvec2<valType>( modf(x.x, i.x), modf(x.y, i.y)); }
template <typename valType> GLM_FUNC_QUALIFIER detail::tvec3<valType> modf ( detail::tvec3<valType> const & x, detail::tvec3<valType> & i ) { return detail::tvec3<valType>( modf(x.x, i.x), modf(x.y, i.y), modf(x.z, i.z)); }
template <typename valType> GLM_FUNC_QUALIFIER detail::tvec4<valType> modf ( detail::tvec4<valType> const & x, detail::tvec4<valType> & i ) { return detail::tvec4<valType>( modf(x.x, i.x), modf(x.y, i.y), modf(x.z, i.z), modf(x.w, i.w)); }
//// Only valid if (INT_MIN <= x-y <= INT_MAX) //// min(x,y) //r = y + ((x - y) & ((x - y) >> (sizeof(int) * //CHAR_BIT - 1))); //// max(x,y) //r = x - ((x - y) & ((x - y) >> (sizeof(int) * //CHAR_BIT - 1)));
// min template <typename genType> GLM_FUNC_QUALIFIER genType min ( genType const & x, genType const & y ) { GLM_STATIC_ASSERT( detail::type<genType>::is_float || detail::type<genType>::is_int || detail::type<genType>::is_uint, "'min' only accept numbers");
return x < y ? x : y; }
VECTORIZE_VEC_SCA(min) VECTORIZE_VEC_VEC(min)
// max template <typename genType> GLM_FUNC_QUALIFIER genType max ( genType const & x, genType const & y ) { GLM_STATIC_ASSERT( detail::type<genType>::is_float || detail::type<genType>::is_int || detail::type<genType>::is_uint, "'max' only accept numbers");
return x > y ? x : y; }
VECTORIZE_VEC_SCA(max) VECTORIZE_VEC_VEC(max)
// clamp template <typename valType> GLM_FUNC_QUALIFIER valType clamp ( valType const & x, valType const & minVal, valType const & maxVal ) { GLM_STATIC_ASSERT( detail::type<valType>::is_float || detail::type<valType>::is_int || detail::type<valType>::is_uint, "'clamp' only accept numbers"); return min(maxVal, max(minVal, x)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec2<T> clamp ( detail::tvec2<T> const & x, typename detail::tvec2<T>::value_type const & minVal, typename detail::tvec2<T>::value_type const & maxVal ) { return detail::tvec2<T>( clamp(x.x, minVal, maxVal), clamp(x.y, minVal, maxVal)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec3<T> clamp ( detail::tvec3<T> const & x, typename detail::tvec3<T>::value_type const & minVal, typename detail::tvec3<T>::value_type const & maxVal ) { return detail::tvec3<T>( clamp(x.x, minVal, maxVal), clamp(x.y, minVal, maxVal), clamp(x.z, minVal, maxVal)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec4<T> clamp ( detail::tvec4<T> const & x, typename detail::tvec4<T>::value_type const & minVal, typename detail::tvec4<T>::value_type const & maxVal ) { return detail::tvec4<T>( clamp(x.x, minVal, maxVal), clamp(x.y, minVal, maxVal), clamp(x.z, minVal, maxVal), clamp(x.w, minVal, maxVal)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec2<T> clamp ( detail::tvec2<T> const & x, detail::tvec2<T> const & minVal, detail::tvec2<T> const & maxVal ) { return detail::tvec2<T>( clamp(x.x, minVal.x, maxVal.x), clamp(x.y, minVal.y, maxVal.y)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec3<T> clamp ( detail::tvec3<T> const & x, detail::tvec3<T> const & minVal, detail::tvec3<T> const & maxVal ) { return detail::tvec3<T>( clamp(x.x, minVal.x, maxVal.x), clamp(x.y, minVal.y, maxVal.y), clamp(x.z, minVal.z, maxVal.z)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec4<T> clamp ( detail::tvec4<T> const & x, detail::tvec4<T> const & minVal, detail::tvec4<T> const & maxVal ) { return detail::tvec4<T>( clamp(x.x, minVal.x, maxVal.x), clamp(x.y, minVal.y, maxVal.y), clamp(x.z, minVal.z, maxVal.z), clamp(x.w, minVal.w, maxVal.w)); }
// mix template <typename genType> GLM_FUNC_QUALIFIER genType mix ( genType const & x, genType const & y, genType const & a ) { GLM_STATIC_ASSERT(detail::type<genType>::is_float , "'genType' is not floating-point type");
return x + a * (y - x); }
template <typename valType> GLM_FUNC_QUALIFIER detail::tvec2<valType> mix ( detail::tvec2<valType> const & x, detail::tvec2<valType> const & y, valType const & a ) { GLM_STATIC_ASSERT(detail::type<valType>::is_float , "'genType' is not floating-point type");
return x + a * (y - x); }
template <typename valType> GLM_FUNC_QUALIFIER detail::tvec3<valType> mix ( detail::tvec3<valType> const & x, detail::tvec3<valType> const & y, valType const & a ) { return x + a * (y - x); }
template <typename valType> GLM_FUNC_QUALIFIER detail::tvec4<valType> mix ( detail::tvec4<valType> const & x, detail::tvec4<valType> const & y, valType const & a ) { return x + a * (y - x); }
template <typename valType> GLM_FUNC_QUALIFIER detail::tvec2<valType> mix ( detail::tvec2<valType> const & x, detail::tvec2<valType> const & y, detail::tvec2<valType> const & a ) { return x + a * (y - x); }
template <typename valType> GLM_FUNC_QUALIFIER detail::tvec3<valType> mix ( detail::tvec3<valType> const & x, detail::tvec3<valType> const & y, detail::tvec3<valType> const & a ) { GLM_STATIC_ASSERT(detail::type<valType>::is_float , "'genType' is not floating-point type");
return x + a * (y - x); }
template <typename valType> GLM_FUNC_QUALIFIER detail::tvec4<valType> mix ( detail::tvec4<valType> const & x, detail::tvec4<valType> const & y, detail::tvec4<valType> const & a ) { return x + a * (y - x); }
//template <typename genTypeT> //GLM_FUNC_QUALIFIER genTypeT mix //( // genTypeT const & x, // genTypeT const & y, // float const & a //) //{ // // It could be a vector too // //GLM_STATIC_ASSERT( // // detail::type<genTypeT>::is_float && // // detail::type<genTypeU>::is_float);
// return x + a * (y - x); //}
template <> GLM_FUNC_QUALIFIER float mix ( float const & x, float const & y, bool const & a ) { return a ? y : x; }
template <> GLM_FUNC_QUALIFIER double mix ( double const & x, double const & y, bool const & a ) { return a ? y : x; }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec2<T> mix ( detail::tvec2<T> const & x, detail::tvec2<T> const & y, bool a ) { GLM_STATIC_ASSERT(detail::type<T>::is_float, "'mix' only accept floating-point inputs");
return a ? y : x; }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec3<T> mix ( detail::tvec3<T> const & x, detail::tvec3<T> const & y, bool a ) { GLM_STATIC_ASSERT(detail::type<T>::is_float, "'mix' only accept floating-point inputs");
return a ? y : x; }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec4<T> mix ( detail::tvec4<T> const & x, detail::tvec4<T> const & y, bool a ) { GLM_STATIC_ASSERT(detail::type<T>::is_float, "'mix' only accept floating-point inputs");
return a ? y : x; }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec2<T> mix ( detail::tvec2<T> const & x, detail::tvec2<T> const & y, typename detail::tvec2<T>::bool_type a ) { GLM_STATIC_ASSERT(detail::type<T>::is_float, "'mix' only accept floating-point inputs");
detail::tvec2<T> result; for ( typename detail::tvec2<T>::size_type i = 0; i < x.length(); ++i ) { result[i] = a[i] ? y[i] : x[i]; } return result; }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec3<T> mix ( detail::tvec3<T> const & x, detail::tvec3<T> const & y, typename detail::tvec3<T>::bool_type a ) { GLM_STATIC_ASSERT(detail::type<T>::is_float, "'mix' only accept floating-point inputs");
detail::tvec3<T> result; for ( typename detail::tvec3<T>::size_type i = 0; i < x.length(); ++i ) { result[i] = a[i] ? y[i] : x[i]; } return result; }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec4<T> mix ( detail::tvec4<T> const & x, detail::tvec4<T> const & y, typename detail::tvec4<T>::bool_type a ) { GLM_STATIC_ASSERT(detail::type<T>::is_float, "'mix' only accept floating-point inputs");
detail::tvec4<T> result; for ( typename detail::tvec4<T>::size_type i = 0; i < x.length(); ++i ) { result[i] = a[i] ? y[i] : x[i]; } return result; }
// step template <typename genType> GLM_FUNC_QUALIFIER genType step ( genType const & edge, genType const & x ) { GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'step' only accept floating-point inputs");
return x < edge ? genType(0) : genType(1); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec2<T> step ( typename detail::tvec2<T>::value_type const & edge, detail::tvec2<T> const & x ) { return detail::tvec2<T>( x.x < edge ? T(0) : T(1), x.y < edge ? T(0) : T(1)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec3<T> step ( typename detail::tvec3<T>::value_type const & edge, detail::tvec3<T> const & x ) { return detail::tvec3<T>( x.x < edge ? T(0) : T(1), x.y < edge ? T(0) : T(1), x.z < edge ? T(0) : T(1)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec4<T> step ( typename detail::tvec4<T>::value_type const & edge, detail::tvec4<T> const & x ) { return detail::tvec4<T>( x.x < edge ? T(0) : T(1), x.y < edge ? T(0) : T(1), x.z < edge ? T(0) : T(1), x.w < edge ? T(0) : T(1)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec2<T> step ( detail::tvec2<T> const & edge, detail::tvec2<T> const & x ) { return detail::tvec2<T>( x.x < edge.x ? T(0) : T(1), x.y < edge.y ? T(0) : T(1)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec3<T> step ( detail::tvec3<T> const & edge, detail::tvec3<T> const & x ) { return detail::tvec3<T>( x.x < edge.x ? T(0) : T(1), x.y < edge.y ? T(0) : T(1), x.z < edge.z ? T(0) : T(1)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec4<T> step ( detail::tvec4<T> const & edge, detail::tvec4<T> const & x ) { return detail::tvec4<T>( x.x < edge.x ? T(0) : T(1), x.y < edge.y ? T(0) : T(1), x.z < edge.z ? T(0) : T(1), x.w < edge.w ? T(0) : T(1)); }
// smoothstep template <typename genType> GLM_FUNC_QUALIFIER genType smoothstep ( genType const & edge0, genType const & edge1, genType const & x ) { GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'smoothstep' only accept floating-point inputs");
genType tmp = clamp((x - edge0) / (edge1 - edge0), genType(0), genType(1)); return tmp * tmp * (genType(3) - genType(2) * tmp); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec2<T> smoothstep ( typename detail::tvec2<T>::value_type const & edge0, typename detail::tvec2<T>::value_type const & edge1, detail::tvec2<T> const & x ) { return detail::tvec2<T>( smoothstep(edge0, edge1, x.x), smoothstep(edge0, edge1, x.y)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec3<T> smoothstep ( typename detail::tvec3<T>::value_type const & edge0, typename detail::tvec3<T>::value_type const & edge1, detail::tvec3<T> const & x ) { return detail::tvec3<T>( smoothstep(edge0, edge1, x.x), smoothstep(edge0, edge1, x.y), smoothstep(edge0, edge1, x.z)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec4<T> smoothstep ( typename detail::tvec4<T>::value_type const & edge0, typename detail::tvec4<T>::value_type const & edge1, detail::tvec4<T> const & x ) { return detail::tvec4<T>( smoothstep(edge0, edge1, x.x), smoothstep(edge0, edge1, x.y), smoothstep(edge0, edge1, x.z), smoothstep(edge0, edge1, x.w)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec2<T> smoothstep ( detail::tvec2<T> const & edge0, detail::tvec2<T> const & edge1, detail::tvec2<T> const & x ) { return detail::tvec2<T>( smoothstep(edge0.x, edge1.x, x.x), smoothstep(edge0.y, edge1.y, x.y)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec3<T> smoothstep ( detail::tvec3<T> const & edge0, detail::tvec3<T> const & edge1, detail::tvec3<T> const & x ) { return detail::tvec3<T>( smoothstep(edge0.x, edge1.x, x.x), smoothstep(edge0.y, edge1.y, x.y), smoothstep(edge0.z, edge1.z, x.z)); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec4<T> smoothstep ( detail::tvec4<T> const & edge0, detail::tvec4<T> const & edge1, detail::tvec4<T> const & x ) { return detail::tvec4<T>( smoothstep(edge0.x, edge1.x, x.x), smoothstep(edge0.y, edge1.y, x.y), smoothstep(edge0.z, edge1.z, x.z), smoothstep(edge0.w, edge1.w, x.w)); }
// TODO: Not working on MinGW... template <typename genType> GLM_FUNC_QUALIFIER bool isnan(genType const & x) { GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'isnan' only accept floating-point inputs");
# if(GLM_COMPILER & (GLM_COMPILER_VC | GLM_COMPILER_INTEL)) return _isnan(x) != 0; # elif(GLM_COMPILER & (GLM_COMPILER_GCC | GLM_COMPILER_CLANG)) # if(GLM_PLATFORM & GLM_PLATFORM_ANDROID) return _isnan(x) != 0; # else return std::isnan(x); # endif # elif(GLM_COMPILER & GLM_COMPILER_CUDA) return isnan(x) != 0; # else return std::isnan(x); # endif }
template <typename T> GLM_FUNC_QUALIFIER typename detail::tvec2<T>::bool_type isnan ( detail::tvec2<T> const & x ) { return typename detail::tvec2<T>::bool_type( isnan(x.x), isnan(x.y)); }
template <typename T> GLM_FUNC_QUALIFIER typename detail::tvec3<T>::bool_type isnan ( detail::tvec3<T> const & x ) { return typename detail::tvec3<T>::bool_type( isnan(x.x), isnan(x.y), isnan(x.z)); }
template <typename T> GLM_FUNC_QUALIFIER typename detail::tvec4<T>::bool_type isnan ( detail::tvec4<T> const & x ) { return typename detail::tvec4<T>::bool_type( isnan(x.x), isnan(x.y), isnan(x.z), isnan(x.w)); }
template <typename genType> GLM_FUNC_QUALIFIER bool isinf( genType const & x) { GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'isinf' only accept floating-point inputs");
# if(GLM_COMPILER & (GLM_COMPILER_INTEL | GLM_COMPILER_VC)) return _fpclass(x) == _FPCLASS_NINF || _fpclass(x) == _FPCLASS_PINF; # elif(GLM_COMPILER & (GLM_COMPILER_GCC | GLM_COMPILER_CLANG)) # if(GLM_PLATFORM & GLM_PLATFORM_ANDROID) return _isinf(x) != 0; # else return std::isinf(x); # endif # elif(GLM_COMPILER & GLM_COMPILER_CUDA) // http://developer.download.nvidia.com/compute/cuda/4_2/rel/toolkit/docs/online/group__CUDA__MATH__DOUBLE_g13431dd2b40b51f9139cbb7f50c18fab.html#g13431dd2b40b51f9139cbb7f50c18fab return isinf(double(x)) != 0; # else return std::isinf(x); # endif }
template <typename T> GLM_FUNC_QUALIFIER typename detail::tvec2<T>::bool_type isinf ( detail::tvec2<T> const & x ) { return typename detail::tvec2<T>::bool_type( isinf(x.x), isinf(x.y)); }
template <typename T> GLM_FUNC_QUALIFIER typename detail::tvec3<T>::bool_type isinf ( detail::tvec3<T> const & x ) { return typename detail::tvec3<T>::bool_type( isinf(x.x), isinf(x.y), isinf(x.z)); }
template <typename T> GLM_FUNC_QUALIFIER typename detail::tvec4<T>::bool_type isinf ( detail::tvec4<T> const & x ) { return typename detail::tvec4<T>::bool_type( isinf(x.x), isinf(x.y), isinf(x.z), isinf(x.w)); }
GLM_FUNC_QUALIFIER int floatBitsToInt(float const & value) { union { float f; int i; } fi;
fi.f = value; return fi.i; }
GLM_FUNC_QUALIFIER detail::tvec2<int> floatBitsToInt ( detail::tvec2<float> const & value ) { return detail::tvec2<int>( floatBitsToInt(value.x), floatBitsToInt(value.y)); }
GLM_FUNC_QUALIFIER detail::tvec3<int> floatBitsToInt ( detail::tvec3<float> const & value ) { return detail::tvec3<int>( floatBitsToInt(value.x), floatBitsToInt(value.y), floatBitsToInt(value.z)); }
GLM_FUNC_QUALIFIER detail::tvec4<int> floatBitsToInt ( detail::tvec4<float> const & value ) { return detail::tvec4<int>( floatBitsToInt(value.x), floatBitsToInt(value.y), floatBitsToInt(value.z), floatBitsToInt(value.w)); }
GLM_FUNC_QUALIFIER uint floatBitsToUint(float const & value) { union { float f; uint u; } fu;
fu.f = value; return fu.u; }
GLM_FUNC_QUALIFIER detail::tvec2<uint> floatBitsToUint ( detail::tvec2<float> const & value ) { return detail::tvec2<uint>( floatBitsToUint(value.x), floatBitsToUint(value.y)); }
GLM_FUNC_QUALIFIER detail::tvec3<uint> floatBitsToUint ( detail::tvec3<float> const & value ) { return detail::tvec3<uint>( floatBitsToUint(value.x), floatBitsToUint(value.y), floatBitsToUint(value.z)); }
GLM_FUNC_QUALIFIER detail::tvec4<uint> floatBitsToUint ( detail::tvec4<float> const & value ) { return detail::tvec4<uint>( floatBitsToUint(value.x), floatBitsToUint(value.y), floatBitsToUint(value.z), floatBitsToUint(value.w)); }
GLM_FUNC_QUALIFIER float intBitsToFloat(int const & value) { union { float f; int i; } fi;
fi.i = value; return fi.f; }
GLM_FUNC_QUALIFIER detail::tvec2<float> intBitsToFloat
( detail::tvec2<int> const & value ) { return detail::tvec2<float>( intBitsToFloat(value.x), intBitsToFloat(value.y)); }
GLM_FUNC_QUALIFIER detail::tvec3<float> intBitsToFloat ( detail::tvec3<int> const & value ) { return detail::tvec3<float>( intBitsToFloat(value.x), intBitsToFloat(value.y), intBitsToFloat(value.z)); }
GLM_FUNC_QUALIFIER detail::tvec4<float> intBitsToFloat ( detail::tvec4<int> const & value ) { return detail::tvec4<float>( intBitsToFloat(value.x), intBitsToFloat(value.y), intBitsToFloat(value.z), intBitsToFloat(value.w)); }
GLM_FUNC_QUALIFIER float uintBitsToFloat(uint const & value) { union { float f; uint u; } fu;
fu.u = value; return fu.f; }
GLM_FUNC_QUALIFIER detail::tvec2<float> uintBitsToFloat ( detail::tvec2<uint> const & value ) { return detail::tvec2<float>( uintBitsToFloat(value.x), uintBitsToFloat(value.y)); }
GLM_FUNC_QUALIFIER detail::tvec3<float> uintBitsToFloat ( detail::tvec3<uint> const & value ) { return detail::tvec3<float>( uintBitsToFloat(value.x), uintBitsToFloat(value.y), uintBitsToFloat(value.z)); }
GLM_FUNC_QUALIFIER detail::tvec4<float> uintBitsToFloat ( detail::tvec4<uint> const & value ) { return detail::tvec4<float>( uintBitsToFloat(value.x), uintBitsToFloat(value.y), uintBitsToFloat(value.z), uintBitsToFloat(value.w)); }
template <typename genType> GLM_FUNC_QUALIFIER genType fma ( genType const & a, genType const & b, genType const & c ) { return a * b + c; }
template <typename genType> GLM_FUNC_QUALIFIER genType frexp ( genType const & x, int & exp ) { return std::frexp(x, exp); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec2<T> frexp ( detail::tvec2<T> const & x, detail::tvec2<int> & exp ) { return std::frexp(x, exp); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec3<T> frexp ( detail::tvec3<T> const & x, detail::tvec3<int> & exp ) { return std::frexp(x, exp); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec4<T> frexp ( detail::tvec4<T> const & x, detail::tvec4<int> & exp ) { return std::frexp(x, exp); }
template <typename genType> GLM_FUNC_QUALIFIER genType ldexp ( genType const & x, int const & exp ) { return std::frexp(x, exp); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec2<T> ldexp ( detail::tvec2<T> const & x, detail::tvec2<int> const & exp ) { return std::frexp(x, exp); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec3<T> ldexp ( detail::tvec3<T> const & x, detail::tvec3<int> const & exp ) { return std::frexp(x, exp); }
template <typename T> GLM_FUNC_QUALIFIER detail::tvec4<T> ldexp ( detail::tvec4<T> const & x, detail::tvec4<int> const & exp ) { return std::frexp(x, exp); }
}//namespace glm
|