Re: [boost] Re: In-place construction for variant

From: Douglas Gregor (gregod_at_[hidden])
Date: 2003-10-27 23:00:38

On Monday 27 October 2003 09:28 pm, E. Gladyshev wrote:
> --- Douglas Paul Gregor <gregod_at_[hidden]> wrote:
> [...]
>
> > Presumably, if you were able to copy lhs to the heap backup in the first
> > place, copying back to the stack is probably okay. It's the copy of rhs
> > that fails...
>
> Doug, you are kidding, right?

No, I'm just sleepy :)

> "copying back to the stack" is not Ok, there is no guarantee that it won't
> fail.

I had myself perfectly convinced that attempting to copy back to the stack 
would be okay, because if there were an exception we could just trap it, but 
obviously that's not going to happen from within an exception handler. Time 
to go back to documentation *yawn*.

        Doug