Re: [boost] [variant2] Andrzej's review

Subject: Re: [boost] [variant2] Andrzej's review
From: Nevin Liber (nevin_at_[hidden])
Date: 2019-04-16 00:11:59

On Fri, Apr 12, 2019 at 7:47 PM Vinnie Falco via Boost <
boost_at_[hidden]> wrote:

> On Fri, Apr 12, 2019 at 5:35 PM Andrzej Krzemienski via Boost
> <boost_at_[hidden]> wrote:
> > The main (and only) reason for rejection is the never-empty basic
> > exception-safety guarantee offered and endorsed by the library.
>
> If I want/need a variant that offers the never-empty basic guarantee
> then Boost should not provide it?
>

Maybe?  How many variants should Boost provide (and I'm not asking that
facetiously)?

These tradeoffs are not the ones I would make.

There are two other variants I would rather use in different circumstances:

   - Strong exception safety
   - Strong performance (never fall back on noexcept move or double
   buffering to avoid a possible exception)

I don't know how to vote for this.  The design tradeoffs Peter is making
are certainly reasonable ones, but they don't cover the things I do with
variant.  I would just fall back on using std::variant over the tradeoffs
being made here.
-- 
 Nevin ":-)" Liber  <mailto:nevin_at_[hidden] <nevin_at_[hidden]>>
[phone]