Re: class with a map
mosfet schrieb:
DJ a ??crit :
mosfet napisa??(a):
Hi,
I would like to write a class storing http headers so Istarted with
this :
class WebHeaderCollection
{
private:
std::map<tstring, tstring> m_KeyValueMap;
public:
tstring& operator[](const tstring& Keyval)
{
tstring sTmp = Keyval + _T("\r\n");
return m_KeyValueMap[ Keyval ];
}
};
This class appends \r\n to each key entry.
My question is how can I give access to all the standard operations
of a map.
The simple solution would have been to derive from std::map but it
seems to be a bad idea to derive from STL containers.
So does it mean I need to provide an iterator, an overload operator
like I did, ...
1. std::map<tstring, tstring> get_http_headers() member to access
private map member
2. Class WebHeaderCollection : public std::map<tstring, tstring>
{ ... and you have iterators there
Thanks but solution 1 is not suitable in my case because I also need to
prevent user to add specific keys (accept, content-type, ...)
And in solution 2, I was told it's bad to derive from container...
That's why I was thinking of writing a class with a map inside.
You can't change any values when you use the first solution. It returns
a copied version of the map. If this is a performance issue, you could
return a const reference to the internal map. The user can read values
at will.
Why is it a bad idea to derive from map. And why is it a bad idea if you
want to have access to all std::map methods? You could write wrapper
methods, which call the std::map methods. But why should you do that?
If you only want to export a few methods, then try the following
attempt. I am quite sure it won't compile (not tested), but it's just
that you get the point:
template<K,V> class WebHeaderCollection {
private:
std::map<K,V> m_values;
public:
typename std::map<K,V>::iterator iterator;
iterator begin() { return m_values.begin(); }
iterator end() { return m_values.end(); }
};
--
CU,
Daniel
--
Regards,
Daniel