C++Talk.NET

shaun roe · Guest

When should a function be a private member, and when simply a standalone
function in the .cpp file?

I'm in the middle of writing a class which bridges between two packages,
and so I need some helper functions to convert between the types used,
in particular between 'unsigned long long' and a ValidityKey which is
like an unsigned long long but only goes up to 2^63 -1.
(Please dont flame me for using unsigned long long, I know its not
standard, but I'm stuck with it).

So I could do the following (this example is not complete - and is just
off the top of my head - , just highlighting the salient points)

in the .h file:
==============

//fwd decl
namespace pkg1{
class ValidityKey;
}

class CoolInterface{
private:
pkg1::ValidityKey convertToKey(const longlongKey);

};

and then put the implementation in the .cpp file as usual:

in the .cpp file
================

#include "pkg1/ValidityKey.h"
#include "pkg1/ValidityKeyException.h"

pkg1::ValidityKey CoolInterface::convertToKey(const longlongKey){
if (longlongKey > ValidityKey::max()) //max() is 2^63 -1
{
std::cerr<<"CoolInterfaceError: key is too big"<<std::endl;
throw(pkg1::ValidityKeyException("convertToKey",longlongKey));
} else {
return ValidityKey(longlongKey);//this constructor exists
}
}

Now... equally I could just put the whole helper function in the .cpp
file as a standalone function and not put any mention of it in my class,
so I wouldnt need a fwd declaration in my .h file for the class. I think
this would be cleaner, but its not a philosophy I see followed very
often... what is the disadvantage? I will only use this helper in the
scope of the .cpp file.

cheers

shaun