FEAT: C++ Persistent Containers about libpmemobj-cpp

[igchor]

I agree that concurrent hashmap/map is quite heavy but right now, there is no plan to implement more containers.

As for deque - I think that our pmem::obj::segment_vector could be a good base for it.

[marcinslusarz]

vector is in 1.6, string is WIP, changing target to 1.7.

[seghcder]

I think the basic ones are done and then some more too ... is there still a plan to do map/unordered map? Concurrent hashmap is good but also seems to consume a lot more space & have more complexity eg with the defrag requirement. If map/unordered map help with that then might be good?

https://github.com/pmem/libpmemobj-cpp/projects/3

Others might be

• list - I think there was a v early example before vector?
• deque - particularly good for rolling timeseries? And can be adapted for other types of queues