Re: [boost] Publishing results

From: John Max Skaller (skaller_at_[hidden])
Date: 2001-05-30 10:59:01

Ross Smith wrote:

> In that case I must have completely missed your point. I thought you
> were complaining about not being able to instantiate a smart pointer on
> an incomplete type. 

        I was.

> If you were, I don't think that's a reasonable thing
> to expect to be able to do. 

Recursive data structures are natural. Being able to manage them
with smart pointers is a requirement.

The most trivial case proves that:

        template <class T> class SLNode {
                T data;
                smart_ptr<SLNode<T> > next;
        };

This is a singly linked list using a smart pointer.
Obviously, because it is recursive, it cannot possibly
be complete at the point 'smart_ptr<SLNode<T> > next;'
is seen, so it is absolutely essential, no ifs, buts,
or excuses, to be able to 'instantiate' a template
with an incomplete type, provided of course that its
definition will work with an incomplete type.

        Smart pointer class definitions usually work
with incomplete types. They have pointers to T in them,
and there is no requirement T be complete for the
class to be complete. So it cannot be required, per se,
for template 'instantiation' either:

        template<class T>
        class smart_ptr {
                T *ptr;
        public:
                //methods
        };

-- 
John (Max) Skaller, mailto:skaller_at_[hidden]
10/1 Toxteth Rd Glebe NSW 2037 Australia voice: 61-2-9660-0850
checkout Vyper http://Vyper.sourceforge.net
download Interscript http://Interscript.sourceforge.net