|
Boost : |
From: Aleksey Gurtovoy (alexy_at_[hidden])
Date: 2001-01-07 16:27:07
Daryle Walker wrote:
> I've seen that thread. I came up with my idea a day before I read Bill's
> and Howard's first message. I decided to give my idea a shot on this board
> anyway. The main difference is that I provide a totally separate class;
For me Howard's idea appeals exactly because it reuses the existing name that
is already a part of the standard; although technically it's not very
different from adding a separate class, from users' standpoint providing such
auto_ptr specialization is more like fixing a defect (filling a functionality
gap), then a move to a new revision of the standard. Speaking about the later,
I don't think that making yet another (limited) smart pointer class part of
the language definition is a way to go. If the next version of the standard
would provide any other smart pointer types besides already existing auto_ptr,
I think it should do it along the lines of feature modeling + template
metaprogramming. BTW, should we try to do a next step in working out smart
pointers' feature model posted by Beman some times ago, or just wait till
Andrei's book (and library) will come out? :).
--Aleksey
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk