C ++未知变量类型。 我应该使用模板吗？

Question

I'm trying to create a class that will parse a file in C++. However the file is actually in 2 files, the header and the raw data.

The header contains the data type of the raw data (it can be int, double, float, etc...). I want to create a variable in my class that will store the raw data. The processing will be the same, no matter the data type.

Should I use a template to do this ? Isn't it overkill to use a templated class for just 1 variable ?

Thank you.

Answer 1

What about using function overloading ? You said the processing is the same, so... I think you could. I use them sometimes to do this kind of stuff.

Example:

Header.h

class MyClass {
public:
    void function(int x);
    void function(double x);
    void function(string x);
    void function(char x);
};

The compiler will know what function to use, depending of the value that you are sending (passing) to it.

Answer 2

Templates could certainly help you here. They offer a way to reduce repetition of that processing which you say is the same for all types.

Whether you "should" use them is of course a more difficult question, as is whether it's overkill - both are matters of opinions, but common use would suggest that the use of just one variable would not be the factor that would decide that.

If the contents of the files containing different types need to be processed into the same container or if a file can contain more than one type, then maybe boost variant would be more suitable.

Since you read the types from file, those types are known at runtime, but since the list of options is known at compile time you can check the type in the file and pass in the type.

if(typeFromFile == headerValueForint)
{
    processFile<int>();
}
else if (typeFromFile == headerValueFordouble)
{
    processFile<double>();
}

However as all the types you list are numeric (and I wonder how the processing could really be the same if there's any other non-numeric types) you could decide to read all the values in as doubles - with suitable range checking. It might take up more memory but the impact of that depends on what the processing actually involves.