From: David Abrahams (dave_at_[hidden])
Date: 2003-10-08 19:40:13
"E. Gladyshev" <egladysh_at_[hidden]> writes:
> --- David Abrahams <dave_at_[hidden]> wrote:
> [...]
>> I doubt you can gain any useful performance for the "no guarantees"
>> case, since it's easy just to tweak the invariant or add empty_type.
>
> I thought that you have been arguing that it is not
> easy to add empty_type especially in generic context?
No, I've been saying that if you add empty_type to types (A,B,C), you
lose the valuable invariant that the variant always contains one of
(A,B,C). The invariant becomes that it contains one of
(A,B,C,empty_type). Your code now has to account for that case, and
may have to include new cases which fail at runtime. Consider
visitors.
-- Dave Abrahams Boost Consulting www.boost-consulting.com