Re: [boost] Re: Variant implementation change (was re: unions) final summary

From: Douglas Paul Gregor (gregod_at_[hidden])
Date: 2003-10-10 08:41:13

On Fri, 10 Oct 2003, E. Gladyshev wrote:
> --- Douglas Gregor <gregod_at_[hidden]> wrote:
> > > 1.7.1
> > > "An object has a storage duration which influences its lifetime"
>
> It says an object *has* storage duration.
>
> My interpretation was that by "reusing" the storage
> you end object's storage duration by ending its storage
> duration you end the object's lifetime.

Okay, but is there any evidence of that? There is evidence to the
contrary:

  3.7 defines the 3 types of storage duration: static, automatic, and
dynamic. Static storage duration is equivalent to the duration of the
program (3.7.1/1), automatic storage duration lasts until the block in
which the object was created exits (3.7.2/1), and dynamic storage duration
lasts from new to delete (somewhere in 3.7.3; I'm lazy).

There is nothing in there that says or hints that storage duration is
affected by reusing storage, yet there are specific descriptions of what
the storage duration is. "Reusing storage" only comes into play when we
talk about lifetimes (in fact, the substring "reus" only shows up in
section 3.8).

Look at the example in 3.8/5. At the line:

  new (this) D2;

This reuses the storage at *this, but the storage duration of the memory
at *this is dynamic (because it was allocated with malloc), so it does not
expire until that memory is freed. If reusing the memory at *this ended
the storage duration, it would contradict 3.7.3.

        Doug