On the other hand, I don't see the long-term benefit clearly that would justify writing this in C++. Most of these open source mini-movements hit maintenance and motivation barriers after some time, not performance ones. Although this could mean Sass overcame the prototype phase quite successfully.
Also, at first sight it seems to me that the author of this library is the same as of Sass' itself. (Please correct me.) In this case, congratulations! Must be a real pleasure to see such a big improvement to what once was a pet project. :)