From: Peter Dimov (pdimov_at_[hidden])
Date: 2002-05-01 05:02:47
From: "Andrei Alexandrescu" <andrewalex_at_[hidden]>
> "Greg Colvin" <greg_at_[hidden]> wrote in message
> news:5.1.0.14.0.20020430132024.02e59c20_at_GMMAIL...
> > An interesting theory. Of course persons comfortable framework
> > can be another persons Procrustean bed. ;->
>
> Alrighty, watch your height folks :o).
>
> Seriously now, I think there's enough evidence to back up the assertion
that
> useful generalizations are possible at least for smart pointers. So I
don't
> understand why we're having this discussion on "doesn't follow" and
> "interesting theory" etc.
I believe that you missed my point.
I am not trying to claim that useful generalizations are impossible. What I
am trying to say is that "users' needs are so varied" doesn't imply "we need
a standard smart pointer." In fact the opposite would be true.
In order to justify the need for a _standard_ (emphasis on standard) smart
pointer, there must be agreement that users' needs, albeit varied, have so
much in common that a standard class, inrastructure, or framework can help.
That agreed upon, the next step is to claim that a single class template
(shared_ptr) doesn't solve users' problems, and a more general, but still
standard, framework is needed. I've no problem with that, but please do try
to start from the "users' problems" part and build from there.