Re: [boost] New boost::function allocator semantics?