|
Boost : |
From: David Abrahams (dave_at_[hidden])
Date: 2003-10-08 16:35:14
"E. Gladyshev" <egladysh_at_[hidden]> writes:
> --- David Abrahams <dave_at_[hidden]> wrote:
>> > I guess the problem is I can't really think of a situation where it makes
>> > any difference if the variant contains the prepended empty_type or
>> > a type from the original list. I'm guessing you can though, so please
>> > help me understand.
>>
>> Just go back to the original design rationale for variant. If it
>> really doesn't matter, variant should just prepend empty_type
>> internally and always. Then we can stop having this discussion ;-)
>
> You have my vote on that.
> I am going to be using variant a lot in my library.
> I'd really like to get it right.
Just to be clear, I wasn't seriously suggesting that change. If
that's what you want always, just prepend empty_type yourself; cheap
and easy. It seems to me that Eric's chosen invariant is a useful
one, and I'm not at all convinced that unconditionally preventing
people from having that invariant is more right than the current
design.
-- Dave Abrahams Boost Consulting www.boost-consulting.com
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk