From: Michael Marcin (mmarcin_at_[hidden])
Date: 2007-10-31 01:39:49
> An overwhelming :-) interest in the proposed Pimpl has certainly been
> overwhelming. LOL.
> The Pimpl idiom has certainly been beaten up and its complexity/importance
> is far from, say, a threading library. Havings said that I feel that the
> simple Pimpl deployment technique and the suggested Pimpl interface is
> simple, elegant and pleasant to use with no drawbacks that I can see so far.
> I've updated and uploaded (as Pimpl-02.zip) the pimpl interface that
> implements both value- and pointer-semantics. Any comments are most welcome.
I've downloaded it, read the documentation, and I think it looks promising.
To test it out I'm converting a pimpled class in my code base to use
your library. This class is noncopyable and as such it isn't immediately
obvious to me whether the class should be modeled using the pimpl's
value semantics, pointer semantics, or some other semantic that doesn't
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk