Re: [container] Contributing 'stationary_vector' to Boost

From: Mehrdad Niknami (mniknami_at_[hidden])
Date: 2021-01-18 01:44:54

Would you know if he's had a chance to take a look at this?

Thanks,
Mehrdad

On Wed, Jan 6, 2021 at 2:47 AM Hans Dembinski <hans.dembinski_at_[hidden]>
wrote:

>
> > On 5. Jan 2021, at 23:25, Mehrdad Niknami via Boost <
> boost_at_[hidden]> wrote:
> >
> > Edit: Actually it just occurred to me there might be a solution for the
> > first issue: an additional "start offset" parameter that gets adjusted by
> > push_front could potentially avoid the performance hit to iterator
> > subtraction. I'm a bit tempted to try to implement it and see if it
> breaks
> > any assumptions I have, but it might work.
> >
> > That said though, it still wouldn't substitute for deque, given that the
> > space complexity difference implies some deque users would now need to
> > worry about manually calling reserve & shrink_to_fit (and then deal with
> > having elements get moved), which deque doesn't have any notion of.
>
> Ok, it sounds like it is sufficiently dissimilar from deque to be its own
> thing. It sounds like a useful addition to boost::container, would be great
> to hear what Ion Gaztanaga thinks.
>
> Best regards,
> Hans