为什么这段代码报告 -31 大于 6？

Question

I have a function double max(int count, ...) in my program. This function should return the highest number, but it reports that -31 > 6 . Where is my mistake? I'm trying to learn va_ . How can I fix this?

double max(int count, ...)
{
    double max = INT_MIN, test;

    int i;
    va_list values;
    va_start(values, count);
    for (i = 0; i < count; ++i)
    {
        test = va_arg(values, double);
        if (test > max)
        {
            max = test;
        }
    }
    va_end(values);
    return max;
}

int main()
{
    printf("%ld", max(5, 1, 6, -31, 23, 24));
    return 0;
}

Answer 1

You are invoking undefined behavior . You are not passing in double values to max() , you are passing in int values instead. int and double are different sizes in memory, and va_arg() can't read an int parameter as if it were a double , and vice versa. You need to match the types correctly.

In this example, change all of the double s to int s, eg:

int max(int count, ...)
{
    int max = INT_MIN, test;

    va_list values;
    va_start(values, count);
    for(int i = 0; i < count; ++i)
    {
        test = va_arg(values, int);
        if(test > max)
        {
            max = test;
        }
    }
    va_end(values);
    return max;
}

int main()
{
    printf("%d", max(5,1,6,-31,23,24));
    return 0;
}

Online Demo

Otherwise, change the int s to double s instead, eg:

double max(int count, ...)
{
    double max = -DBL_MAX, test;

    va_list values;
    va_start(values, count);
    for(int i = 0; i < count; ++i)
    {
        test = va_arg(values, double);
        if(test > max)
        {
            max = test;
        }
    }
    va_end(values);
    return max;
}

int main()
{
    printf("%lf", max(5,1.0,6.0,-31.0,23.0,24.0));
    return 0;
}

Online Demo

---

Either way, you might consider a slight change in logic in how you are handling the max value. You should initialize it to the 1st value that is passed in, eg:

<type> max(int count, ...)
{
    if (count <= 0)
        return <default value>;

    va_list values;
    va_start(values, count);

    <type> max = va_arg(values, <type>), test;

    for(int i = 1; i < count; ++i)
    {
        test = va_arg(values, <type>);
        if (test > max)
        {
            max = test;
        }
    }

    va_end(values);
    return max;
}

---

Also, note that all of the above is the C way of handling things. But you also tagged your question as C++, and the C++ way to handle this would be to use a variadic template instead of ellipses (ie, no va_arg() needed), or else take a std::initializer_list or at least a pair of iterator s so that you can then make use of the standard std::max_element() algorithm.