Are there plans for a stable_sort option in the toolkit or can anyone suggest a fast parallel stable sort algorithm that could be implemented with the toolkit?
At the moment, implementation of a stable sorting algorithm is not in the near-term plans.
If you are up to implement one for your needs, Iwould suggest you to start with the Wikipedia page for sorting algorithms. If memory footprint is not a concern, you might consider merge sort; its divide-and-conquer nature fits TBB parallel model well.
