Re: [boost] [Containers] Should flat_* expose implementation vector?