I think you meant "up is always the direction you're heading," since what you described is already how it works today.

The added memory will make it easier to keep the necessary pixels duplicated in an appropriate pixbuf without hogging all of the system's memory. And the new CPU is probably enough to support arbitrary rotation. It might work suitably on a 770, albeit obviously a lot slower.

The problem as it exists now is that I have a great deal of both memory and CPU optimization in the code at the moment that depends on the current "up is always north" behavior. Changing that behavior would mean abandoning all of that optimization, which is not a trivial undertaking.

If I ever get an n800, I'll probably experiment with it, but since I don't want to abandon 770 users (of whom myself is one), I am hesitant to make changes that will require the n800 or otherwise ruin the Maemo Mapper experience on the 770.

To be honest, if the n800 catches on and draws a larger customer base, then I'll put higher priority on documentation in order to assist new Maemo Mapper users, since the documentation as it exists now is almost non-existent. n800-specific capability enhancements are definitely not high on my list.