RE: [boost] Re: Variant implementation change (was re: unions)finalsummary

From: Brock Peabody (brock.peabody_at_[hidden])
Date: 2003-10-09 15:52:13

> -----Original Message-----
> From: boost-bounces_at_[hidden] [mailto:boost-bounces_at_[hidden]]
> On Behalf Of E. Gladyshev

[...]

> Here we go again, Brock. It has alredy been discussed that it
> is not a good generic solution to assign a special meaning to the first
> type.

Yeah, and I agree with the outcome of that discussion - which was about
finding a type with nothrow default constructors, not about singular
variants.  This is not the ideal solution - that would take something like a
policy.  The point is that those who want variants that can be singular can
get them - albeit through a less than ideal syntax.

We have two choices:

(1) People who need variants that can be singular can get them but they have
to use the unusual and possibly confusing syntax of putting boost::empty as
the first type.

(2) All variants can be singular.

Do you agree with that?

Brock