Why Can't I Use Floats as Non-Type Template Parameters in C ?

Understanding Template Parameter Restrictions: Why Float Values are Forbidden

In C , when defining a generic class with template parameters, certain restrictions apply. One such restriction involves the inability to use float values as non-type template parameters. This becomes apparent when attempting to compile code like the following:

template <class T, T defaultValue>
class GenericClass
{
    // ...
};

int main()
{
    GenericClass<float, 4.6f> gcFloat;  // Error
}

Error Message:

'float' is not a valid type for a template constant parameter
Invalid type in declaration before ';' token

Reasoning:

According to the C 11 standard, non-type template arguments must conform to specific requirements. For non-type template parameters of integral or enumeration type, only converted constant expressions of the same type are permitted. Float values do not meet this requirement, as they cannot be represented exactly in memory.

Technical Explanation:

Using float values as template arguments could lead to inconsistent behavior. For example, consider the following:

func<1/3.f>();
func<2/6.f>();

Even though we intended to call the same function twice, the floating-point representation of the two calculations may not be identical. This could lead to different results, defeating the purpose of using template arguments.

Alternative Approach:

To represent floating-point values as template arguments, consider using C 11's constant expressions (constexpr). You can create advanced expressions that calculate the numerator and denominator of a float value at compile time and pass them as separate integer arguments. Additionally, define a threshold to ensure that floating-point values close to each other yield the same numerator and denominator.

The above is the detailed content of Why Can't I Use Floats as Non-Type Template Parameters in C ?. For more information, please follow other related articles on the PHP Chinese website!