We were really happy to read the work of Vadim Tropashko but found during prototyping that we couldn't use it because alternate rows reversed the order of sibli

We were really happy to read the work of Vadim Tropashko but found during prototyping that we couldn't use it because alternate rows reversed the order of siblings. This encoding started out as a simple stab at skipping every second row to correct the sibling ordering problem. But the maths fell out so easily it seemed worth writing it up. Vadim incidentally was particularly approachable at the time. Lovely fellow. By essentially skipping every second row of Vadim's, you end up only being able to go about half as deep before you blow BIGINT. And that wasn't so deep to begin with. I haven't visited this stuff in many years now, but it seems likely that if you were to use the pgmp postgreSQL extension, or something similar that provided arbitrary precision integers/rationals, you would solve the depth limitation. I'd be really interested to hear about it.