From: Persson Jonas (jonas.persson_at_[hidden])
Date: 2003-11-05 12:25:20
> From: Eric Friedman [mailto:ebf_at_[hidden]]
> Anyhow, as I just posted on the list, I've changed variant so
> that, at
> least in the 1.31 release, variant<> is not supported at all.
> variant on an empty type sequence will lead to a compile-time error.
> While this change perhaps does not help you with your metaprogramming
> 'termination' condition, it does open up variant for future changes
> without negative effects on backwards compatibility.
I think this is a good decision. No empty variants are much easier to
handle than ones with strange behaviour. I hope that it will be possible
in future releases though, as it would be a very handy feature.
Thanks for listening and for a great library.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk