Michael Goldshteyn:
> I think that having a ?configurable? deque class template
> is one of the best things that could be offered by a boost
> Containers library. I have always wondered why noone
> wrote a version of the deque class template that allowed
> the configuration of such things as the page / block size,
> rather than have it be determined by a compile time
> formula based on the size of the data type being stored.
> It would seem to me that the page / block size is one of
> the more critical pieces affecting the size / speed
> tradeoffs in a deque?s performance when there are many
> insertions taking place.
The thread I mentioned has a link to a segmented_array
container which may be of interest. I agree that what you
suggest would be an improvement.