Boost logo

Boost :

From: Dietmar Kuehl (dietmar_kuehl_at_[hidden])
Date: 2002-05-04 19:23:49


David B. Held wrote:

> Also, I get the impression that Dietmar is also dissatisfied with a
> heavily templated smart pointer, although he hasn't mentioned anything
> lately (that I've seen).

I'm happy even with a heavily templates smart pointer given two
preconditions:
- It is obvious which policies are to be used in interfaces and
   probably easy to specify those (eg. because they are default in some
   form).
- It is possible to have different policies for different types using
   the same notation where the smart pointer is used. How this is
   achieved is entirely unconstrained.

Basically, my only concern is that everybody "chooses" the same set of
policies when using a typical smart pointer in an interface (that is,
there may be reasons to use a different set of policies for specific
situations but the "unreflected" use of smart pointer in an interface
should always end up with the same type). However, since one default is
almost certainly inappropriate in some places, there has to be an
approach how a special default is established (eg. for COM objects or
types requiring an intrusive count).

-- 
<mailto:dietmar_kuehl_at_[hidden]> <http://www.dietmar-kuehl.de/>
Phaidros eaSE - Easy Software Engineering: <http://www.phaidros.com/>

Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk