Function FloatForCompare

src/OpenColorIO/MathUtils.cpp:397–400

------------------------------------------------------------------------------ Map a floating-point number (already represented as an integer) to an ordered integer representation, allowing for a tolerance-based comparison. Floating-point numbers have their magnitude stored in the bits 0-30 as a pair of exponent and mantissa, while the sign is stored on bit 31. That makes the positive floating-p

//      +------------+-----------------+------------------+------------------+-----------------+-----------+
//
inline int FloatForCompare(const unsigned floatBits)
{
    return floatBits < 0x80000000 ? (0x80000000 + floatBits) : (0x80000000 - (floatBits & 0x7FFFFFFF));
}

//------------------------------------------------------------------------------
//