I have come across this several times:

Consider some templated class with many template args:

template <typename A, typename B, typename C>
struct Base
{
 Base(A&, B, C*);
};

template <typename A, typename B, typename C>
struct Derived : Base<Derived<A, B, C>, B, std::pair<A, C> >
{
 typedef Base<Derived<A, B, C>, B, std::pair<A, C> > Inherited;
 Derived(B b, std::pair<A, C>* c) : Inherited(*this, b, c) { }
};


This a contrived example, but you get the idea. The problem is that in the
design phase you will shuffle around those template params quite a bit.
Updating that typedef is rather tedious. This could be avoided if typedefs
were allowed in the baseclass list:

template <typename A, typename B, typename C>
struct Derived : typedef Base<Derived<A, B, C>, B, std::pair<A, C> >
Inherited
{
 Derived(B b, std::pair<A, C>* c) : Inherited(*this, b, c) { }
};

The semantics could be defined by translation to standard C++ this way
(appropriately extended to the MI case):

class <classname> : <access> typedef <base> <name>
{
 ...
};

---->

class <classname> : <access> <base>
{
 <access>: typedef <base> <name>;
 <default-access>:
 ...
};

Any thoughts? I would love to name the base classes this way.

####################
A related question: In my Derived definition I pass Derived<A, B, C> to the
base template as a parameter. I recognize that I can use Derived instead of
Derived<A, B, C> as an alias inside of the class definition. But can I use
this alias in the baseclass list too?
To wit:

Is

template <typename A, typename B, typename C>
struct Derived : Base<Derived, B, std::pair<A, C> >
{
 typedef Base<Derived, B, std::pair<A, C> > Inherited;
 Derived(B b, std::pair<A, C>* c) : Inherited(*this, b, c) { }
};

legal std c++?

####################
Another related question:
Typedef allows the aliasing of typenames. This is very useful for giving
names to type applications in a certain scope. Now with the appearance of
member templates and template template parameters I see the need for
aliasing template names too. One could extend typedef to cover this
problem. Any better ideas?

 Gabor
---
[ comp.std.c++ is moderated.  To submit articles, try just posting with ]
[ your news-reader.  If that fails, use mailto:std-c++@ncar.ucar.edu    ]
[              --- Please see the FAQ before posting. ---               ]
[ FAQ: http://reality.sgi.com/austern_mti/std-c++/faq.html              ]