Interesting, do you store the length of your voronoi fixed_int separate from the array?
Yes, I do. Even more I also store the sign of the fixed_int in the same value (with the size). This means that voronoi fixed_int doesn't use 2's complement representation. If leading bytes are not set we won't even operate with them (they might be uninitialized). As another small advantage we will always have opposite value for the smallest possible negative integer value representable with the given fixed_int.
Interesting! I'm going to have to think carefully about that - the whole Raison d'être for my fixed_int was to emulate a regular 2's complement integer type as closely as possible. What you have is a variable precision type using fixed storage? Lot's to think about, thanks, John.