/*
 * Mesa 3-D graphics library
 * Version:  6.5
 *
 * Copyright (C) 2006  Brian Paul   All Rights Reserved.
 *
 * 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
 * BRIAN PAUL 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.
 */

//
// This file defines nearly all constructors and operators for built-in data
// types, using extended language syntax. In general, compiler treats
// constructors and operators as ordinary functions with some exceptions.
// For example, the language does not allow functions to be called in
// constant expressions - here the exception is made to allow it.
//
// Each implementation provides its own version of this file. Each
// implementation can define the required set of operators and constructors
// in its own fashion.
//
// The extended language syntax is only present when compiling this file.
// It is implicitly included at the very beginning of the compiled shader,
// so no built-in functions can be used.
//
// To communicate with the implementation, a special extended "__asm" keyword
// is used, followed by an instruction name (any valid identifier), a
// destination variable identifier and a list of zero or more source
// variable identifiers.
//
// A variable identifier is a variable name declared earlier in the code
// (as a function parameter, local or global variable).
//
// An instruction name designates an instruction that must be exported
// by the implementation.  Each instruction receives data from source
// variable identifiers and returns data in the destination variable
// identifier.
//
// It is up to the implementation how to define a particular operator
// or constructor. If it is expected to being used rarely, it can be
// defined in terms of other operators and constructors,
// for example:
//
// ivec2 __operator + (const ivec2 x, const ivec2 y) {
//    return ivec2 (x[0] + y[0], x[1] + y[1]);
// }
//
// If a particular operator or constructor is expected to be used very
// often or is an atomic operation (that is, an operation that cannot be
// expressed in terms of other operations or would create a dependency
// cycle) it must be defined using one or more __asm constructs.
//
// Each implementation must define constructors for all scalar types
// (bool, float, int).  There are 9 scalar-to-scalar constructors
// (including identity constructors). However, since the language
// introduces special constructors (like matrix constructor with a single
// scalar value), implementations must also implement these cases.
// The compiler provides the following algorithm when resolving a constructor:
// - try to find a constructor with a prototype matching ours,
// - if no constructor is found and this is a scalar-to-scalar constructor,
//   raise an error,
// - if a constructor is found, execute it and return,
// - count the size of the constructor parameter list - if it is less than
//   the size of our constructor's type, raise an error,
// - for each parameter in the list do a recursive constructor matching for
//   appropriate scalar fields in the constructed variable,
//
// Each implementation must also define a set of operators that deal with
// built-in data types.
// There are four kinds of operators:
// 1) Operators that are implemented only by the compiler: "()" (function
//    call), "," (sequence) and "?:" (selection).
// 2) Operators that are implemented by the compiler by expressing it in
//    terms of other operators:
//    - "." (field selection) - translated to subscript access,
//    - "&&" (logical and) - translated to "<left_expr> ? <right_expr> :
//      false",
//    - "||" (logical or) - translated to "<left_expr> ? true : <right_expr>",
// 3) Operators that can be defined by the implementation and if the required
//    prototype is not found, standard behaviour is used:
//    - "==", "!=", "=" (equality, assignment) - compare or assign
//      matching fields one-by-one;
//      note that at least operators for scalar data types must be defined
//      by the implementation to get it work,
// 4) All other operators not mentioned above. If no required prototype is
//    found, an error is raised. An implementation must follow the language
//    specification to provide all valid operator prototypes.
//

//bp:
vec4 vec4(const float a1, const float b1, const float c1, const float d1)
{
   __retVal.x = a1;
   __retVal.y = b1;
   __retVal.z = c1;
   __retVal.w = d1;
}

////
//// Assorted constructors
////

int __constructor (const float f) {
    int i;
    __asm float_to_int i, f;
    return i;
}

bool __constructor (const int i) {
    return i != 0;
}

bool __constructor (const float f) {
    return f != 0.0;
}

int __constructor (const bool b) {
    return b ? 1 : 0;
}

float __constructor (const bool b) {
    return b ? 1.0 : 0.0;
}

float __constructor (const int i) {
    float f;
    __asm int_to_float f, i;
    return f;
}

bool __constructor (const bool b) {
    return b;
}

int __constructor (const int i) {
    return i;
}

float __constructor (const float f) {
    return f;
}

vec2 __constructor (const float f) {
    return vec2 (f, f);
}

vec2 __constructor (const int i) {
    float x;
    __asm int_to_float x, i;
    return vec2 (x);
}

vec2 __constructor (const bool b) {
    return vec2 (b ? 1.0 : 0.0);
}

vec3 __constructor (const float f) {
    return vec3 (f, f, f);
}

vec3 __constructor (const int i) {
    float x;
    __asm int_to_float x, i;
    return vec3 (x);
}

vec3 __constructor (const bool b) {
    return vec3 (b ? 1.0 : 0.0);
}

//bp: TODO replace with asm == f.xxxx
vec4 __constructor (const float f) {
    return vec4 (f, f, f, f);
}

vec4 __constructor (const int i) {
    float x;
    __asm int_to_float x, i;
    return vec4 (x);
}

vec4 __constructor (const bool b) {
    return vec4 (b ? 1.0 : 0.0);
}

ivec2 __constructor (const int i) {
    return ivec2 (i, i);
}

ivec2 __constructor (const float f) {
    return ivec2 (int (f));
}

ivec2 __constructor (const bool b) {
    return ivec2 (int (b));
}

ivec3 __constructor (const int i) {
    return ivec3 (i, i, i);
}

ivec3 __constructor (const float f) {
    return ivec3 (int (f));
}

ivec3 __constructor (const bool b) {
    return ivec3 (int (b));
}

ivec4 __constructor (const int i) {
    return ivec4 (i, i, i, i);
}

ivec4 __constructor (const float f) {
    return ivec4 (int (f));
}

ivec4 __constructor (const bool b) {
    return ivec4 (int (b));
}

bvec2 __constructor (const bool b) {
    return bvec2 (b, b);
}

bvec2 __constructor (const float f) {
    return bvec2 (bool (f));
}

bvec2 __constructor (const int i) {
    return bvec2 (bool (i));
}

bvec3 __constructor (const bool b) {
    return bvec3 (b, b, b);
}

bvec3 __constructor (const float f) {
    return bvec3 (bool (f));
}

bvec3 __constructor (const int i) {
    return bvec3 (bool (i));
}

bvec4 __constructor (const bool b) {
    return bvec4 (b, b, b, b);
}

bvec4 __constructor (const float f) {
    return bvec4 (bool (f));
}

bvec4 __constructor (const int i) {
    return bvec4 (bool (i));
}



//// mat2 constructors

mat2 __constructor (const float f) {
    return mat2 (f, 0.0, 0.0, f);
}

mat2 __constructor (const int i) {
    float x;
    __asm int_to_float x, i;
    return mat2 (x);
}

mat2 __constructor (const bool b) {
    return mat2 (b ? 1.0 : 0.0);
}


//// mat3 constructors

mat3 __constructor (const float f) {
    return mat3 (f, 0.0, 0.0, 0.0, f, 0.0, 0.0, 0.0, f);
}

mat3 __constructor (const int i) {
    float x;
    __asm int_to_float x, i;
    return mat3 (x);
}

mat3 __constructor (const bool b) {
    return mat3 (b ? 1.0 : 0.0);
}


//// mat4 constructors

mat4 __constructor (const float f) {
    return mat4 (f, 0.0, 0.0, 0.0, 0.0, f, 0.0, 0.0, 0.0, 0.0, f, 0.0, 0.0, 0.0, 0.0, f);
}

mat4 __constructor (const int i) {
    float x;
    __asm int_to_float x, i;
    return mat4 (x);
}

mat4 __constructor (const bool b) {
    return mat4 (b ? 1.0 : 0.0);
}

mat4 __constructor (const vec4 r0, const vec4 r1, const vec4 r2, const vec4 r3)
{
   __retVal[0] = r0;
   __retVal[1] = r1;
   __retVal[2] = r2;
   __retVal[3] = r3;
}



//// Basic int operators

int __operator + (const int a, const int b)
{
    float x, y;
    int c;
    __asm int_to_float x, a;
    __asm int_to_float y, b;
    __asm float_add    x, x, y;
    __asm float_to_int c, x;
    return c;
}

int __operator - (const int a, const int b)
{
    float x, y;
    int c;
    __asm int_to_float x, a;
    __asm int_to_float y, b;
    __asm float_negate y, y;
    __asm float_add    x, x, y;
    __asm float_to_int c, x;
    return c;
}

int __operator * (const int a, const int b)
{
    float x, y;
    int c;
    __asm int_to_float   x, a;
    __asm int_to_float   y, b;
    __asm float_multiply x, x, y;
    __asm float_to_int   c, x;
    return c;
}

int __operator / (const int a, const int b)
{
    float x, y;
    int c;
    __asm int_to_float x, a;
    __asm int_to_float y, b;
    __asm float_divide x, x, y;
    __asm float_to_int c, x;
    return c;
}


//// Basic ivec2 operators

ivec2 __operator + (const ivec2 v, const ivec2 u)
{
    return ivec2 (v.x + u.x, v.y + u.y);
}

ivec2 __operator - (const ivec2 v, const ivec2 u)
{
    return ivec2 (v.x - u.x, v.y - u.y);
}

ivec2 __operator * (const ivec2 v, const ivec2 u)
{
    return ivec2 (v.x * u.x, v.y * u.y);
}

ivec2 __operator / (const ivec2 v, const ivec2 u)
{
    return ivec2 (v.x / u.x, v.y / u.y);
}


//// Basic ivec3 operators

ivec3 __operator + (const ivec3 v, const ivec3 u)
{
    return ivec3 (v.x + u.x, v.y + u.y, v.z + u.z);
}

ivec3 __operator - (const ivec3 v, const ivec3 u)
{
    return ivec3 (v.x - u.x, v.y - u.y, v.z - u.z);
}

ivec3 __operator * (const ivec3 v, const ivec3 u)
{
    return ivec3 (v.x * u.x, v.y * u.y, v.z * u.z);
}

ivec3 __operator / (const ivec3 v, const ivec3 u)
{
    return ivec3 (v.x / u.x, v.y / u.y, v.z / u.z);
}


//// Basic ivec4 operators

ivec4 __operator + (const ivec4 v, const ivec4 u)
{
    return ivec4 (v.x + u.x, v.y + u.y, v.z + u.z, v.w + u.w);
}

ivec4 __operator - (const ivec4 v, const ivec4 u)
{
    return ivec4 (v.x - u.x, v.y - u.y, v.z - u.z, v.w - u.w);
}

ivec4 __operator * (const ivec4 v, const ivec4 u)
{
    return ivec4 (v.x * u.x, v.y * u.y, v.z * u.z, v.w * u.w);
}

ivec4 __operator / (const ivec4 v, const ivec4 u)
{
    return ivec4 (v.x / u.x, v.y / u.y, v.z / u.z, v.w / u.w);
}


//// Basic float operators

float __operator + (const float a, const float b)
{
//   __asm float_add __retVal, a, b;
   __asm vec4_add __retVal.x, a, b;
}

float __operator - (const float a, const float b)
{
   __asm vec4_subtract __retVal.x, a, b;
}

float __operator * (const float a, const float b)
{
    __asm float_multiply __retVal, a, b;
}

float __operator / (const float a, const float b)
{
   __asm float_divide __retVal, a, b;
}


//// Basic vec2 operators

vec2 __operator + (const vec2 v, const vec2 u)
{
   __asm vec4_add __retVal.xy, v, u;
}

vec2 __operator - (const vec2 v, const vec2 u)
{
    __asm vec4_subtract __retVal.xy, v, u;
}

vec2 __operator * (const vec2 v, const vec2 u)
{
    __asm vec4_multiply __retVal.xy, v, u;
}

vec2 __operator / (const vec2 v, const vec2 u)
{
   vec2 w; // = 1 / u
   __asm float_rcp w.x, u.x;
   __asm float_rcp w.y, u.y;
   __asm vec4_multiply __retVal.xy, v, w;
}


//// Basic vec3 operators

vec3 __operator + (const vec3 v, const vec3 u)
{
   __asm vec4_add __retVal.xyz, v, u;
}

vec3 __operator - (const vec3 v, const vec3 u)
{
    __asm vec4_subtract __retVal.xyz, v, u;
}

vec3 __operator * (const vec3 v, const vec3 u)
{
    __asm vec4_multiply __retVal.xyz, v, u;
}

vec3 __operator / (const vec3 v, const vec3 u)
{
   vec3 w; // = 1 / u
   __asm float_rcp w.x, u.x;
   __asm float_rcp w.y, u.y;
   __asm float_rcp w.z, u.z;
   __asm vec4_multiply __retVal.xyz, v, w;
}


//// Basic vec4 operators

vec4 __operator + (const vec4 v, const vec4 u)
{
   __asm vec4_add __retVal, v, u;
}

vec4 __operator - (const vec4 v, const vec4 u)
{
    __asm vec4_subtract __retVal, v, u;
}

vec4 __operator * (const vec4 v, const vec4 u)
{
    __asm vec4_multiply __retVal, v, u;
}

vec4 __operator / (const vec4 v, const vec4 u)
{
   vec4 w; // = 1 / u
   __asm float_rcp w.x, u.x;
   __asm float_rcp w.y, u.y;
   __asm float_rcp w.z, u.z;
   __asm float_rcp w.w, u.w;
   __asm vec4_multiply __retVal, v, w;
}




//// Basic vec2/float operators

vec2 __operator + (const float a, const vec2 u)
{
   __asm vec4_add __retVal.xy, a.xx, u.xy;
}

vec2 __operator + (const vec2 v, const float b)
{
   __asm vec4_add __retVal.xy, v.xy, b.xx;
}

vec2 __operator - (const float a, const vec2 u)
{
   __asm vec4_subtract __retVal.xy, a.xx, u.xy;
}

vec2 __operator - (const vec2 v, const float b)
{
   __asm vec4_subtract __retVal.xy, v.xy, b.xx;
}

vec2 __operator * (const float a, const vec2 u)
{
   __asm vec4_multiply __retVal.xy, a.xx, u.xy;
}

vec2 __operator * (const vec2 v, const float b)
{
   __asm vec4_multiply __retVal.xy, v.xy, b.xx;
}

vec2 __operator / (const float a, const vec2 u)
{
   vec2 invU;
   __asm float_rcp invU.x, u.x;
   __asm float_rcp invU.y, u.y;
   __asm vec4_multiply __retVal.xy, a.xx, invU.xy;
}

vec2 __operator / (const vec2 v, const float b)
{
   float invB;
   __asm float_rcp invB, b;
   __asm vec4_multiply __retVal.xy, v.xy, invB.xx;
}


//// Basic vec3/float operators

vec3 __operator + (const float a, const vec3 u)
{
   __asm vec4_add __retVal.xyz, a.xxx, u.xyz;
}

vec3 __operator + (const vec3 v, const float b)
{
   __asm vec4_add __retVal.xyz, v.xyz, b.xxx;
}

vec3 __operator - (const float a, const vec3 u)
{
   __asm vec4_subtract __retVal.xyz, a.xxx, u.xyz;
}

vec3 __operator - (const vec3 v, const float b)
{
   __asm vec4_subtract __retVal.xyz, v.xyz, b.xxx;
}

vec3 __operator * (const float a, const vec3 u)
{
   __asm vec4_multiply __retVal.xyz, a.xxx, u.xyz;
}

vec3 __operator * (const vec3 v, const float b)
{
   __asm vec4_multiply __retVal.xyz, v.xyz, b.xxx;
}

vec3 __operator / (const float a, const vec3 u)
{
   vec3 invU;
   __asm float_rcp invU.x, u.x;
   __asm float_rcp invU.y, u.y;
   __asm float_rcp invU.z, u.z;
   __asm vec4_multiply __retVal.xyz, a.xxx, invU.xyz;
}

vec3 __operator / (const vec3 v, const float b)
{
   float invB;
   __asm float_rcp invB, b;
   __asm vec4_multiply __retVal.xyz, v.xyz, invB.xxx;
}


//// Basic vec4/float operators

vec4 __operator + (const float a, const vec4 u)
{
   __asm vec4_add __retVal, a.xxxx, u;
}

vec4 __operator + (const vec4 v, const float b)
{
   __asm vec4_add __retVal, v, b.xxxx;
}

vec4 __operator - (const float a, const vec4 u)
{
   __asm vec4_subtract __retVal, a.xxxx, u;
}

vec4 __operator - (const vec4 v, const float b)
{
   __asm vec4_subtract __retVal, v, b.xxxx;
}

vec4 __operator * (const float a, const vec4 u)
{
   __asm vec4_multiply __retVal, a.xxxx, u;
}

vec4 __operator * (const vec4 v, const float b)
{
   __asm vec4_multiply __retVal, v, b.xxxx;
}

vec4 __operator / (const float a, const vec4 u)
{
   vec4 invU;
   __asm float_rcp invU.x, u.x;
   __asm float_rcp invU.y, u.y;
   __asm float_rcp invU.z, u.z;
   __asm float_rcp invU.w, u.w;
   __asm vec4_multiply __retVal, a.xxxx, invU;
}

vec4 __operator / (const vec4 v, const float b)
{
   float invB;
   __asm float_rcp invB, b;
   __asm vec4_multiply __retVal, v, invB.xxxx;
}





//// Unary negation operator

float __operator - (const float a)
{
    float b;
    __asm float_negate b, a;
    return b;
}

vec2 __operator - (const vec2 v)
{
    return vec2 (-v.x, -v.y);
}

vec3 __operator - (const vec3 v)
{
    return vec3 (-v.x, -v.y, -v.z);
}

vec4 __operator - (const vec4 v)
{
    return vec4 (-v.x, -v.y, -v.z, -v.w);
}

int __operator - (const int a)
{
    float x;
    int b;
    __asm int_to_float x, a;
    __asm float_negate x, x;
    __asm float_to_int b, x;
    return b;
}

ivec2 __operator - (const ivec2 v)
{
    return ivec2 (-v.x, -v.y);
}

ivec3 __operator - (const ivec3 v)
{
    return ivec3 (-v.x, -v.y, -v.z);
}

ivec4 __operator - (const ivec4 v)
{
    return ivec4 (-v.x, -v.y, -v.z, -v.w);
}



//// dot product

float dot(const float a, const float b)
{
   return a * b;
}

float dot(const vec2 a, const vec2 b)
{
   return a.x * b.x + a.y * b.y;
}

float dot(const vec3 a, const vec3 b)
{
    __asm vec3_dot __retVal, a, b;
}

float dot(const vec4 a, const vec4 b)
{
    __asm vec4_dot __retVal, a, b;
}




//// int assignment operators

void __operator += (inout int a, const int b)
{
    a = int (float (a) + float (b));
}

void __operator -= (inout int a, const int b)
{
    a += -b;
}

void __operator *= (inout int a, const int b)
{
    a = int (float (a) * float (b));
}

void __operator /= (inout int a, const int b)
{
    a = int (float (a) / float (b));
}


//// ivec2 assignment operators

void __operator += (inout ivec2 v, const ivec2 u)
{
    v.x += u.x;
    v.y += u.y;
}

void __operator -= (inout ivec2 v, const ivec2 u)
{
    v.x -= u.x;
    v.y -= u.y;
}

void __operator *= (inout ivec2 v, const ivec2 u)
{
    v.x *= u.x;
    v.y *= u.y;
}

void __operator /= (inout ivec2 v, const ivec2 u)
{
    v.x /= u.x;
    v.y /= u.y;
}


//// ivec3 assignment operators

void __operator += (inout ivec3 v, const ivec3 u)
{
    v.x += u.x;
    v.y += u.y;
    v.z += u.z;
}

void __operator -= (inout ivec3 v, const ivec3 u)
{
    v.x -= u.x;
    v.y -= u.y;
    v.z -= u.z;
}

void __operator *= (inout ivec3 v, const ivec3 u)
{
    v.x *= u.x;
    v.y *= u.y;
    v.z *= u.z;
}

void __operator /= (inout ivec3 v, const ivec3 u)
{
    v.x /= u.x;
    v.y /= u.y;
    v.z /= u.z;
}


//// ivec4 assignment operators

void __operator += (inout ivec4 v, const ivec4 u) {
    v.x += u.x;
    v.y += u.y;
    v.z += u.z;
    v.w += u.w;
}

void __operator -= (inout ivec4 v, const ivec4 u) {
    v.x -= u.x;
    v.y -= u.y;
    v.z -= u.z;
    v.w -= u.w;
}

void __operator *= (inout ivec4 v, const ivec4 u) {
    v.x *= u.x;
    v.y *= u.y;
    v.z *= u.z;
    v.w *= u.w;
}

void __operator /= (inout ivec4 v, const ivec4 u) {
    v.x /= u.x;
    v.y /= u.y;
    v.z /= u.z;
    v.w /= u.w;
}


//// float assignment operators

void __operator += (inout float a, const float b)
{
    __asm vec4_add a.x, a, b;
}

void __operator -= (inout float a, const float b)
{
    __asm vec4_subtract a.x, a, b;
}

void __operator *= (inout float a, const float b)
{
    __asm vec4_multiply a.x, a, b;
}

void __operator /= (inout float a, const float b)
{
   float w; // = 1 / b
   __asm float_rcp w.x, b;
   __asm vec4_multiply a.x, a, w;
}


//// vec2 assignment operators

void __operator += (inout vec2 v, const vec2 u)
{
    v.x += u.x;
    v.y += u.y;
}

void __operator -= (inout vec2 v, const vec2 u)
{
    v.x -= u.x;
    v.y -= u.y;
}

void __operator *= (inout vec2 v, const vec2 u)
{
    v.x *= u.x;
    v.y *= u.y;
}

void __operator /= (inout vec2 v, const vec2 u)
{
    v.x /= u.x;
    v.y /= u.y;
}


//// vec3 assignment operators

void __operator += (inout vec3 v, const vec3 u)
{
   __asm vec4_add v.xyz, v, u;
}

void __operator -= (inout vec3 v, const vec3 u)
{
   __asm vec4_subtract v.xyz, v, u;
}

void __operator *= (inout vec3 v, const vec3 u)
{
   __asm vec4_multiply v.xyz, v, u;
}

void __operator /= (inout vec3 v, const vec3 u)
{
// XXX rcp
    v.x /= u.x;
    v.y /= u.y;
    v.z /= u.z;
}


//// vec4 assignment operators

void __operator += (inout vec4 v, const vec4 u)
{
   __asm vec4_add v, v, u;
}

void __operator -= (inout vec4 v, const vec4 u)
{
   __asm vec4_subtract v, v, u;
}

void __operator *= (inout vec4 v, const vec4 u)
{
   __asm vec4_multiply v, v, u;
}

void __operator /= (inout vec4 v, const vec4 u)
{
// XXX rcp
    v.x /= u.x;
    v.y /= u.y;
    v.z /= u.z;
    v.w /= u.w;
}



//// ivec2/int assignment operators

void __operator += (inout ivec2 v, const int a)
{
   v.x += a;
   v.y += a;
}

void __operator -= (inout ivec2 v, const int a)
{
   v.x -= a;
   v.y -= a;
}

void __operator *= (inout ivec2 v, const int a)
{
   v.x *= a;
   v.y *= a;
}

void __operator /= (inout ivec2 v, const int a)
{
// XXX rcp
    v.x /= a;
    v.y /= a;
}


//// ivec3/int assignment operators

void __operator += (inout ivec3 v, const int a)
{
   v.x += a;
   v.y += a;
   v.z += a;
}

void __operator -= (inout ivec3 v, const int a)
{
   v.x -= a;
   v.y -= a;
   v.z -= a;
}

void __operator *= (inout ivec3 v, const int a)
{
   v.x *= a;
   v.y *= a;
   v.z *= a;
}

void __operator /= (inout ivec3 v, const int a)
{
    v.x /= a;
    v.y /= a;
    v.z /= a;
}


//// ivec4/int assignment operators

void __operator += (inout ivec4 v, const int a)
{
   __asm vec4_add v, v, a.xxxx;
}

void __operator -= (inout ivec4 v, const int a)
{
   __asm vec4_subtract v, v, a.xxxx;
}

void __operator *= (inout ivec4 v, const int a)
{
   __asm vec4_multiply v, v, a.xxxx;
}

void __operator /= (inout ivec4 v, const int a)
{
    v.x /= a;
    v.y /= a;
    v.z /= a;
    v.w /= a;
}



//// vec2/float assignment operators

void __operator += (inout vec2 v, const float a)
{
   __asm vec4_add v.xy, v, a.xx;
}

void __operator -= (inout vec2 v, const float a)
{
   __asm vec4_subtract v.xy, v, a.xx;
}

void __operator *= (inout vec2 v, const float a)
{
   __asm vec4_multiply v.xy, v, a.xx;
}

void __operator /= (inout vec2 v, const float a)
{
// XXX rcp
    v.x /= a;
    v.y /= a;
}


//// vec3/float assignment operators

void __operator += (inout vec3 v, const float a)
{
   __asm vec4_add v.xyz, v, a.xxx;
}

void __operator -= (inout vec3 v, const float a)
{
   __asm vec4_subtract v.xyz, v, a.xxx;
}

void __operator *= (inout vec3 v, const float a)
{
   __asm vec4_multiply v.xyz, v, a.xxx;
}

void __operator /= (inout vec3 v, const float a)
{
    v.x /= a;
    v.y /= a;
    v.z /= a;
}


//// vec4/float assignment operators

void __operator += (inout vec4 v, const float a)
{
   __asm vec4_add v, v, a.xxxx;
}

void __operator -= (inout vec4 v, const float a)
{
   __asm vec4_subtract v, v, a.xxxx;
}

void __operator *= (inout vec4 v, const float a)
{
   __asm vec4_multiply v, v, a.xxxx;
}

void __operator /= (inout vec4 v, const float a)
{
    v.x /= a;
    v.y /= a;
    v.z /= a;
    v.w /= a;
}







void __operator += (inout mat2 m, const mat2 n) {
    m[0] += n[0];
    m[1] += n[1];
}

void __operator -= (inout mat2 m, const mat2 n) {
    m[0] -= n[0];
    m[1] -= n[1];
}

vec2 __operator * (const mat2 m, const vec2 v) {
    return vec2 (
        v.x * m[0].x + v.y * m[1].x,
        v.x * m[0].y + v.y * m[1].y
    );
}




mat2 __operator * (const mat2 m, const mat2 n) {
    return mat2 (m * n[0], m * n[1]);
}

void __operator *= (inout mat2 m, const mat2 n) {
    m = m * n;
}

void __operator /= (inout mat2 m, const mat2 n) {
    m[0] /= n[0];
    m[1] /= n[1];
}

void __operator += (inout mat3 m, const mat3 n) {
    m[0] += n[0];
    m[1] += n[1];
    m[2] += n[2];
}

void __operator -= (inout mat3 m, const mat3 n) {
    m[0] -= n[0];
    m[1] -= n[1];
    m[2] -= n[2];
}





mat3 __operator * (const mat3 m, const mat3 n)
{
// XXX fix
//    return mat3 (m * n[0], m * n[1], m * n[2]);
}

void __operator *= (inout mat3 m, const mat3 n) {
    m = m * n;
}

void __operator /= (inout mat3 m, const mat3 n) {
    m[0] /= n[0];
    m[1] /= n[1];
    m[2] /= n[2];
}

void __operator += (inout mat4 m, const mat4 n) {
    m[0] += n[0];
    m[1] += n[1];
    m[2] += n[2];
    m[3] += n[3];
}

void __operator -= (inout mat4 m, const mat4 n) {
    m[0] -= n[0];
    m[1] -= n[1];
    m[2] -= n[2];
    m[3] -= n[3];
}



vec4 __operator * (const mat4 mx, const vec4 v)
{
   __retVal.x = dot(v, mx[0]);
   __retVal.y = dot(v, mx[1]);
   __retVal.z = dot(v, mx[2]);
   __retVal.w = dot(v, mx[3]);
}

//matmul
vec3 __operator * (const mat3 m, const vec3 v)
{
   __retVal.x = dot(v, m[0]);
   __retVal.y = dot(v, m[1]);
   __retVal.z = dot(v, m[2]);
}


// xxx move this
mat4 __operator * (const mat4 m, const mat4 n) {
    return mat4 (m * n[0], m * n[1], m * n[2], m * n[3]);
}

void __operator *= (inout mat4 m, const mat4 n) {
    m = m * n;
}

void __operator /= (inout mat4 m, const mat4 n) {
    m[0] /= n[0];
    m[1] /= n[1];
    m[2] /= n[2];
    m[3] /= n[3];
}


void __operator += (inout mat2 m, const float a) {
    m[0] += a;
    m[1] += a;
}

void __operator -= (inout mat2 m, const float a) {
    m[0] -= a;
    m[1] -= a;
}

void __operator *= (inout mat2 m, const float a) {
    m[0] *= a;
    m[1] *= a;
}

void __operator /= (inout mat2 m, const float a) {
    m[0] /= a;
    m[1] /= a;
}

void __operator += (inout mat3 m, const float a) {
    m[0] += a;
    m[1] += a;
    m[2] += a;
}

void __operator -= (inout mat3 m, const float a) {
    m[0] -= a;
    m[1] -= a;
    m[2] -= a;
}

void __operator *= (inout mat3 m, const float a) {
    m[0] *= a;
    m[1] *= a;
    m[2] *= a;
}

void __operator /= (inout mat3 m, const float a) {
    m[0] /= a;
    m[1] /= a;
    m[2] /= a;
}

void __operator += (inout mat4 m, const float a) {
    m[0] += a;
    m[1] += a;
    m[2] += a;
    m[3] += a;
}

void __operator -= (inout mat4 m, const float a) {
    m[0] -= a;
    m[1] -= a;
    m[2] -= a;
    m[3] -= a;
}

void __operator *= (inout mat4 m, const float a) {
    m[0] *= a;
    m[1] *= a;
    m[2] *= a;
    m[3] *= a;
}

void __operator /= (inout mat4 m, const float a) {
    m[0] /= a;
    m[1] /= a;
    m[2] /= a;
    m[3] /= a;
}

vec2 __operator * (const vec2 v, const mat2 m) {
    return vec2 (
        v.x * m[0].x + v.y * m[0].y,
        v.x * m[1].x + v.y * m[1].y
    );
}

void __operator *= (inout vec2 v, const mat2 m) {
    v = v * m;
}

void __operator *= (inout vec3 v, const mat3 m) {
//    v = v * m;
}



vec4 __operator * (const vec4 v, const mat4 m)
{
   vec4 r1, r2, r3, r4;
   r1.x = m[0].x;
   r1.y = m[1].x;
   r1.z = m[2].x;
   r1.w = m[3].x;
   r2.x = m[0].y;
   r2.y = m[1].y;
   r2.z = m[2].y;
   r2.w = m[3].y;
   r3.x = m[0].z;
   r3.y = m[1].z;
   r3.z = m[2].z;
   r3.w = m[3].z;
   r4.x = m[0].w;
   r4.y = m[1].w;
   r4.z = m[2].w;
   r4.w = m[3].w;
   __asm vec4_dot __retVal.x, r1, v;
   __asm vec4_dot __retVal.y, r2, v;
   __asm vec4_dot __retVal.z, r3, v;
   __asm vec4_dot __retVal.w, r4, v;
}

// matmul
vec3 __operator * (const vec3 v, const mat3 m)
{
   vec3 r1, r2, r3;
   r1.x = m[0].x;
   r1.y = m[1].x;
   r1.z = m[2].x;
   r2.x = m[0].y;
   r2.y = m[1].y;
   r2.z = m[2].y;
   r3.x = m[0].z;
   r3.y = m[1].z;
   r3.z = m[2].z;
   __asm vec3_dot __retVal.x, r1, v;
   __asm vec3_dot __retVal.y, r2, v;
   __asm vec3_dot __retVal.z, r3, v;
}










void __operator *= (inout vec4 v, const mat4 m)
{
// xxx improve codegen for this case
    v = v * m;
}






mat2 __operator + (const mat2 m, const mat2 n) {
    return mat2 (m[0] + n[0], m[1] + n[1]);
}

mat2 __operator - (const mat2 m, const mat2 n) {
    return mat2 (m[0] - n[0], m[1] - n[1]);
}

mat2 __operator / (const mat2 m, const mat2 n) {
    return mat2 (m[0] / n[0], m[1] / n[1]);
}


mat3 __operator + (const mat3 m, const mat3 n)
{
   __retVal[0] = m[0] + n[0];
   __retVal[1] = m[1] + n[1];
   __retVal[2] = m[2] + n[2];
}

mat3 __operator - (const mat3 m, const mat3 n)
{
   __retVal[0] = m[0] - n[0];
   __retVal[1] = m[1] - n[1];
   __retVal[2] = m[2] - n[2];
}

mat3 __operator / (const mat3 m, const mat3 n)
{
    __retVal[0] = m[0] / n[0];
    __retVal[0] = m[1] / n[1];
    __retVal[0] = m[2] / n[2];
}


mat4 __operator + (const mat4 m, const mat4 n)
{
   __retVal[0] = m[0] + n[0];
   __retVal[1] = m[1] + n[1];
   __retVal[2] = m[2] + n[2];
   __retVal[3] = m[3] + n[3];
}

mat4 __operator - (const mat4 m, const mat4 n)
{
   __retVal[0] = m[0] - n[0];
   __retVal[1] = m[1] - n[1];
   __retVal[2] = m[2] - n[2];
   __retVal[3] = m[3] - n[3];
}



mat4 __operator / (const mat4 m, const mat4 n)
{
    __retVal[0] = m[0] / n[0];
    __retVal[0] = m[1] / n[1];
    __retVal[0] = m[2] / n[2];
    __retVal[0] = m[3] / n[3];
}



//next


mat2 __operator + (const float a, const mat2 n) {
    return mat2 (a + n[0], a + n[1]);
}

mat2 __operator + (const mat2 m, const float b) {
    return mat2 (m[0] + b, m[1] + b);
}

mat2 __operator - (const float a, const mat2 n) {
    return mat2 (a - n[0], a - n[1]);
}

mat2 __operator - (const mat2 m, const float b) {
    return mat2 (m[0] - b, m[1] - b);
}

mat2 __operator * (const float a, const mat2 n) {
    return mat2 (a * n[0], a * n[1]);
}

mat2 __operator * (const mat2 m, const float b) {
    return mat2 (m[0] * b, m[1] * b);
}

mat2 __operator / (const float a, const mat2 n) {
    return mat2 (a / n[0], a / n[1]);
}

mat2 __operator / (const mat2 m, const float b) {
    return mat2 (m[0] / b, m[1] / b);
}

mat3 __operator + (const float a, const mat3 n) {
    return mat3 (a + n[0], a + n[1], a + n[2]);
}

mat3 __operator + (const mat3 m, const float b) {
    return mat3 (m[0] + b, m[1] + b, m[2] + b);
}

mat3 __operator - (const float a, const mat3 n) {
    return mat3 (a - n[0], a - n[1], a - n[2]);
}

mat3 __operator - (const mat3 m, const float b) {
    return mat3 (m[0] - b, m[1] - b, m[2] - b);
}

mat3 __operator * (const float a, const mat3 n) {
    return mat3 (a * n[0], a * n[1], a * n[2]);
}

mat3 __operator * (const mat3 m, const float b) {
    return mat3 (m[0] * b, m[1] * b, m[2] * b);
}

mat3 __operator / (const float a, const mat3 n) {
    return mat3 (a / n[0], a / n[1], a / n[2]);
}

mat3 __operator / (const mat3 m, const float b) {
    return mat3 (m[0] / b, m[1] / b, m[2] / b);
}

mat4 __operator + (const float a, const mat4 n) {
    return mat4 (a + n[0], a + n[1], a + n[2], a + n[3]);
}

mat4 __operator + (const mat4 m, const float b) {
    return mat4 (m[0] + b, m[1] + b, m[2] + b, m[3] + b);
}

mat4 __operator - (const float a, const mat4 n) {
    return mat4 (a - n[0], a - n[1], a - n[2], a - n[3]);
}

mat4 __operator - (const mat4 m, const float b) {
    return mat4 (m[0] - b, m[1] - b, m[2] - b, m[3] - b);
}

mat4 __operator * (const float a, const mat4 n) {
    return mat4 (a * n[0], a * n[1], a * n[2], a * n[3]);
}

mat4 __operator * (const mat4 m, const float b) {
    return mat4 (m[0] * b, m[1] * b, m[2] * b, m[3] * b);
}

mat4 __operator / (const float a, const mat4 n) {
    return mat4 (a / n[0], a / n[1], a / n[2], a / n[3]);
}

mat4 __operator / (const mat4 m, const float b) {
    return mat4 (m[0] / b, m[1] / b, m[2] / b, m[3] / b);
}

ivec2 __operator + (const int a, const ivec2 u) {
    return ivec2 (a) + u;
}

ivec2 __operator + (const ivec2 v, const int b) {
    return v + ivec2 (b);
}

ivec2 __operator - (const int a, const ivec2 u) {
    return ivec2 (a) - u;
}

ivec2 __operator - (const ivec2 v, const int b) {
    return v - ivec2 (b);
}

ivec2 __operator * (const int a, const ivec2 u) {
    return ivec2 (a) * u;
}

ivec2 __operator * (const ivec2 v, const int b) {
    return v * ivec2 (b);
}

ivec2 __operator / (const int a, const ivec2 u) {
    return ivec2 (a) / u;
}

ivec2 __operator / (const ivec2 v, const int b) {
    return v / ivec2 (b);
}

ivec3 __operator + (const int a, const ivec3 u) {
    return ivec3 (a) + u;
}

ivec3 __operator + (const ivec3 v, const int b) {
    return v + ivec3 (b);
}

ivec3 __operator - (const int a, const ivec3 u) {
    return ivec3 (a) - u;
}

ivec3 __operator - (const ivec3 v, const int b) {
    return v - ivec3 (b);
}

ivec3 __operator * (const int a, const ivec3 u) {
    return ivec3 (a) * u;
}

ivec3 __operator * (const ivec3 v, const int b) {
    return v * ivec3 (b);
}

ivec3 __operator / (const int a, const ivec3 u) {
    return ivec3 (a) / u;
}

ivec3 __operator / (const ivec3 v, const int b) {
    return v / ivec3 (b);
}

ivec4 __operator + (const int a, const ivec4 u) {
    return ivec4 (a) + u;
}

ivec4 __operator + (const ivec4 v, const int b) {
    return v + ivec4 (b);
}

ivec4 __operator - (const int a, const ivec4 u) {
    return ivec4 (a) - u;
}

ivec4 __operator - (const ivec4 v, const int b) {
    return v - ivec4 (b);
}

ivec4 __operator * (const int a, const ivec4 u) {
    return ivec4 (a) * u;
}

ivec4 __operator * (const ivec4 v, const int b) {
    return v * ivec4 (b);
}

ivec4 __operator / (const int a, const ivec4 u) {
    return ivec4 (a) / u;
}

ivec4 __operator / (const ivec4 v, const int b) {
    return v / ivec4 (b);
}

///foo




mat2 __operator - (const mat2 m) {
    return mat2 (-m[0], -m[1]);
}

mat3 __operator - (const mat3 m) {
    return mat3 (-m[0], -m[1], -m[2]);
}

mat4 __operator - (const mat4 m) {
    return mat4 (-m[0], -m[1], -m[2], -m[3]);
}

void __operator -- (inout float a) {
    a -= 1.0;
}

void __operator -- (inout int a) {
    a -= 1;
}

void __operator -- (inout vec2 v) {
    --v.x;
    --v.y;
}

void __operator -- (inout vec3 v) {
    --v.x;
    --v.y;
    --v.z;
}

void __operator -- (inout vec4 v) {
    --v.x;
    --v.y;
    --v.z;
    --v.w;
}

void __operator -- (inout ivec2 v) {
    --v.x;
    --v.y;
}

void __operator -- (inout ivec3 v) {
    --v.x;
    --v.y;
    --v.z;
}

void __operator -- (inout ivec4 v) {
    --v.x;
    --v.y;
    --v.z;
    --v.w;
}

void __operator -- (inout mat2 m) {
    --m[0];
    --m[1];
}

void __operator -- (inout mat3 m) {
    --m[0];
    --m[1];
    --m[2];
}

void __operator -- (inout mat4 m) {
    --m[0];
    --m[1];
    --m[2];
    --m[3];
}

void __operator ++ (inout float a) {
    a += 1.0;
}

void __operator ++ (inout int a) {
    a += 1;
}

void __operator ++ (inout vec2 v) {
    ++v.x;
    ++v.y;
}

void __operator ++ (inout vec3 v) {
    ++v.x;
    ++v.y;
    ++v.z;
}

void __operator ++ (inout vec4 v) {
    ++v.x;
    ++v.y;
    ++v.z;
    ++v.w;
}

void __operator ++ (inout ivec2 v) {
    ++v.x;
    ++v.y;
}

void __operator ++ (inout ivec3 v) {
    ++v.x;
    ++v.y;
    ++v.z;
}

void __operator ++ (inout ivec4 v) {
    ++v.x;
    ++v.y;
    ++v.z;
    ++v.w;
}

void __operator ++ (inout mat2 m) {
    ++m[0];
    ++m[1];
}

void __operator ++ (inout mat3 m) {
    ++m[0];
    ++m[1];
    ++m[2];
}

void __operator ++ (inout mat4 m) {
    ++m[0];
    ++m[1];
    ++m[2];
    ++m[3];
}

//
// NOTE: postfix increment and decrement operators take additional dummy int parameter to
//       distinguish their prototypes from prefix ones.
//

float __operator -- (inout float a, const int) {
    float b = a;
    --a;
    return b;
}

int __operator -- (inout int a, const int) {
    int b = a;
    --a;
    return b;
}

vec2 __operator -- (inout vec2 v, const int) {
    return vec2 (v.x--, v.y--);
}

vec3 __operator -- (inout vec3 v, const int) {
    return vec3 (v.x--, v.y--, v.z--);
}

vec4 __operator -- (inout vec4 v, const int) {
    return vec4 (v.x--, v.y--, v.z--, v.w--);
}

ivec2 __operator -- (inout ivec2 v, const int) {
    return ivec2 (v.x--, v.y--);
}

ivec3 __operator -- (inout ivec3 v, const int) {
    return ivec3 (v.x--, v.y--, v.z--);
}

ivec4 __operator -- (inout ivec4 v, const int) {
    return ivec4 (v.x--, v.y--, v.z--, v.w--);
}

mat2 __operator -- (inout mat2 m, const int) {
    return mat2 (m[0]--, m[1]--);
}

mat3 __operator -- (inout mat3 m, const int) {
    return mat3 (m[0]--, m[1]--, m[2]--);
}

mat4 __operator -- (inout mat4 m, const int) {
    return mat4 (m[0]--, m[1]--, m[2]--, m[3]--);
}

float __operator ++ (inout float a, const int) {
    float b = a;
    ++a;
    return b;
}

int __operator ++ (inout int a, const int) {
    int b = a;
    ++a;
    return b;
}

vec2 __operator ++ (inout vec2 v, const int) {
    return vec2 (v.x++, v.y++);
}

vec3 __operator ++ (inout vec3 v, const int) {
    return vec3 (v.x++, v.y++, v.z++);
}

vec4 __operator ++ (inout vec4 v, const int) {
    return vec4 (v.x++, v.y++, v.z++, v.w++);
}

ivec2 __operator ++ (inout ivec2 v, const int) {
    return ivec2 (v.x++, v.y++);
}

ivec3 __operator ++ (inout ivec3 v, const int) {
    return ivec3 (v.x++, v.y++, v.z++);
}

ivec4 __operator ++ (inout ivec4 v, const int) {
    return ivec4 (v.x++, v.y++, v.z++, v.w++);
}

mat2 __operator ++ (inout mat2 m, const int) {
    return mat2 (m[0]++, m[1]++);
}

mat3 __operator ++ (inout mat3 m, const int) {
    return mat3 (m[0]++, m[1]++, m[2]++);
}

mat4 __operator ++ (inout mat4 m, const int) {
    return mat4 (m[0]++, m[1]++, m[2]++, m[3]++);
}


// XXX are the inequality operators for floats/ints really needed????
bool __operator < (const float a, const float b)
{
   __asm vec4_sgt __retVal.x, b, a;
}


bool __operator < (const int a, const int b) {
    return float (a) < float (b);
}

bool __operator > (const float a, const float b) {
    bool c;
    __asm float_less c, b, a;
    return c;
}

bool __operator > (const int a, const int b) {
    return float (a) > float (b);
}

bool __operator >= (const float a, const float b) {
    bool g, e;
    __asm float_less  g, b, a;
    __asm float_equal e, a, b;
    return g || e;
}

bool __operator >= (const int a, const int b) {
    return float (a) >= float (b);
}

bool __operator <= (const float a, const float b) {
    bool g, e;
    __asm float_less  g, a, b;
    __asm float_equal e, a, b;
    return g || e;
}

bool __operator <= (const int a, const int b) {
    return float (a) <= float (b);
}

bool __operator ^^ (const bool a, const bool b) {
    return a != b;
}

//
// These operators are handled internally by the compiler:
//
// bool __operator && (bool a, bool b) {
//     return a ? b : false;
// }
// bool __operator || (bool a, bool b) {
//     return a ? true : b;
// }
//

bool __operator ! (const bool a) {
    return a == false;
}

//
// MESA-specific extension functions.
//

void printMESA (const float f) {
    __asm float_print f;
}

void printMESA (const int i) {
    __asm int_print i;
}

void printMESA (const bool b) {
    __asm bool_print b;
}

void printMESA (const vec2 v) {
    printMESA (v.x);
    printMESA (v.y);
}

void printMESA (const vec3 v) {
    printMESA (v.x);
    printMESA (v.y);
    printMESA (v.z);
}

void printMESA (const vec4 v) {
    printMESA (v.x);
    printMESA (v.y);
    printMESA (v.z);
    printMESA (v.w);
}

void printMESA (const ivec2 v) {
    printMESA (v.x);
    printMESA (v.y);
}

void printMESA (const ivec3 v) {
    printMESA (v.x);
    printMESA (v.y);
    printMESA (v.z);
}

void printMESA (const ivec4 v) {
    printMESA (v.x);
    printMESA (v.y);
    printMESA (v.z);
    printMESA (v.w);
}

void printMESA (const bvec2 v) {
    printMESA (v.x);
    printMESA (v.y);
}

void printMESA (const bvec3 v) {
    printMESA (v.x);
    printMESA (v.y);
    printMESA (v.z);
}

void printMESA (const bvec4 v) {
    printMESA (v.x);
    printMESA (v.y);
    printMESA (v.z);
    printMESA (v.w);
}

void printMESA (const mat2 m) {
    printMESA (m[0]);
    printMESA (m[1]);
}

void printMESA (const mat3 m) {
    printMESA (m[0]);
    printMESA (m[1]);
    printMESA (m[2]);
}

void printMESA (const mat4 m) {
    printMESA (m[0]);
    printMESA (m[1]);
    printMESA (m[2]);
    printMESA (m[3]);
}

void printMESA (const sampler1D e) {
    __asm int_print e;
}

void printMESA (const sampler2D e) {
    __asm int_print e;
}

void printMESA (const sampler3D e) {
    __asm int_print e;
}

void printMESA (const samplerCube e) {
    __asm int_print e;
}

void printMESA (const sampler1DShadow e) {
    __asm int_print e;
}

void printMESA (const sampler2DShadow e) {
    __asm int_print e;
}