Float OR Int范围内的随机数

Question

I'm making a templates function object that will generate a ranged random number. I know how to do that for either ints or floats. But my problem is that I want it to be able to generate a ranged random number of the type T.

So, since I want it to be able to return both floats and integers I can't use the % operator. This seem to work for float values. But if T is INT it will mess up the range.

return (_minValue + (T)rand() / ((T)RAND_MAX / (_maxValue - _minValue)));

There must be some mistake in that algorithm that I just can't seem to find at this hour.

Answer 1

If the type T is an integer type, (T)rand() / (T)RAND_MAX will clearly be zero except for the rare case when rand() returns RAND_MAX (and T happens to be at least as int ) in which case it will be one. Note that rand() is generally not a very good random number generator although the quality differs quite a bit between different implementations. You are probably best of to just use something from the <random> library (for my normal work I have very rarely a need for random numbers, ie, I don't know how to use the components in this library immediately).

If you insist in using rand() your best option is probably to determine random values different for integers and floating points, eg:

template <typename T, bool = std::is_floating_point<T>::value>
struct random;
template <typename T>
struct random<T, true> {
    static T generate() { return /* floating point formula goes here */; }
};
template <typename T>
struct random<T, false> {
    static T generate() { return /* integer formula goes here */; }
};

template <typename T>
T random_range() { return random<T>::generate(); }