Re: fundamental template inheritance problem.

[Federico David Sacerdoti]

Of course I meant to say:

Right, that was a typo on my part. It doesn't change my main point, however. There still is no reason for g++ to force us to name the template type of our *base* class, since it must be the same as the *derived* class.

Dave

Artak Avetyan wrote:

Hi
right code is  (pay attention on "^^^^^"):

template <class T>
class A
{
};

template <class T>
class B : public A<T>
             ^^^^^^^^^
{
};

Regards,
Artak.
 

Federico David Sacerdoti wrote:

g++ (2.95.2, others)
Problem: a derived templated class inherits a templated base class.

template <class T>
class A {

};

template <class T>
class B : public class A {

};

In class B's constructor C++ requires that we explicitly name the type of our
base class

template <class T>
B<T>::B () :
    A<T>()
{
}

This only makes sense if class B's type can be different than class A's. But
since B 'is an' A having different template types is not possible. Therefore
there is NO REASON for us to call A<T>, the compiler should know this
implicitly. This is a deficiency of the language, a mis-definition of
'template' which says a derived class may have a different type than its base.
I suggest we fix it, in the compiler.

Everyone I know who has tried this has made the same mistake. Mis-definitions
such as this make templates harder to use, and cause object-oriented
programmers to turn to other languages.

Federico David Sacerdoti

_______________________________________________
Bug-gplusplus mailing list
address@hidden
http://mail.gnu.org/mailman/listinfo/bug-gplusplus